University of Groningen

Through the Eyes of an InfantHunnius, S.

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2005

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Hunnius, S. (2005). Through the Eyes of an Infant: The Early Development of Visual Scanning andDisengagement of Attention [S.l.]: s.n.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 09-05-2018

Through the Eyes of an Infant

The Early Development of Visual Scanning and Disengagement

of Attention

© 2004, Sabine HunniusAll rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without prior permission by the author.

Printed by: Drukkerij Alba, Groningen

ISBN 90-9018907-6

RIJKSUNIVERSITEIT GRONINGEN

Through the Eyes of an Infant

The Early Development of Visual Scanning and Disengagement of Attention

Proefschrift

ter verkrijging van het doctoraat in de Psychologische, Pedagogische en Sociologische Wetenschappen

aan de Rijksuniversiteit Groningenop gezag van de

Rector Magnificus, dr. F. Zwarts,in het openbaar te verdedigen op

maandag 10 januari 2005om 14.45 uur

door

Sabine Hunniusgeboren op 19 juni 1974

te Bonn, Duitsland

Promotores: Prof. dr. P. L. C. van Geert Prof. dr. J. M. Bouma

Copromotor: Dr. R. H. Geuze

Beoordelingscommissie: Prof. dr. R. N. Aslin Prof. dr. A. Johnson Prof. dr. G. J. P. Savelsbergh

Wir sehen in der Natur nicht Wörter, sondern immer nur Anfangsbuchstaben von Wörtern, und wenn wir alsdann lesen wollen, so finden wir, daß die neuen

sogenannten Wörter wiederum bloß Anfangsbuchstaben von andren sind.

Georg Christoph Lichtenberg

CONTENTS

Chapter 1 Introduction 9

Chapter 2 Developmental Changes in Visual Scanning 31 of Dynamic Faces and Abstract Stimuli in Infants

Chapter 3 Gaze Shifting in Infancy: 57 A Longitudinal Study Using Dynamic Faces and Abstract Stimuli

Chapter 4 Associations between the Developmental Trajectories 79 of Visual Scanning and Disengagement of Attention in Infants Chapter 5 A Longitudinal Study of Shifts of Attention and Gaze 99 in Preterm and Full-term Infants

Chapter 6 Summary and General Discussion 127

Samenvatting 147Zusammenfassung 155

Acknowledgements 165Curriculum Vitae 169

Chapter 1

Introduction

10

Chapter 1

11

Introduction

Anybody who has ever observed a baby of a few weeks of age looking around and examining his environment has undoubtedly noticed how different his visual behavior is from that of an adult. A young infant tends to move his eyes slowly, and often it seems as if he gazes vacantly in front of him. He likes to look at patterns with high contrast, so the locations that interest him can be the dark frame of a painting on a light colored wall or the knob of a drawer. From time to time, it seems as if the baby’s gaze gets “stuck” at one location for 10 to 20 seconds, and the infant may keep on staring there, although there may be other equally salient things to look at in his environment, for example a colorful toy. If a moving object catches his eye, he may follow it, as long as it is not going too fast.

Were we to observe the same infant 3 or 4 months later, we would see that his visual behavior has changed dramatically. The infant now examines objects by scanning them quickly and systematically. Complex, colorful, and moving stimuli attract his attention particularly easily. The baby tends to alternate his intense inspections with brief looks away. His eye movements are fast, and he tracks moving objects or persons easily.

Vision plays a crucial role in the daily life of an infant. As young infants are unable to move around or to grasp objects easily, they explore their environment and learn about the world by looking. Looking is also one of the most important ways in which infants communicate with their caretakers. Face-to-face interaction forms the beginning of social communication and plays an important role in the bonding of infant and caretaker.

The ability to carry out fast, accurate shifts of gaze serves as the basis for visual information processing and communication in early infancy. Accordingly, the devel-opmental changes in the visual system sketched above are of great importance. The question which factors enable or hinder its functional development therefore deserves precise investigation.

This dissertation deals with the developmental changes in attention and gaze shift-ing behavior which occur during the first few months of life. It reports the results of an intensive longitudinal study on visual exploration behavior and the disengagement of attention and gaze from a fixated stimulus. The purpose of the study was to examine these issues using dynamic stimuli and to investigate the influence of different stimuli on young children’s visual behavior. Abstract and socially relevant stimuli were con-trasted. In addition to examining a group of healthy full-term infants, it was the goal of this study to describe the nature and extent of differences in the development of these attentional processes between full-term infants and infants born prematurely. A sequence of six measurements during a period of five months, in which rapid develop-ment of the visual system was expected, was supposed to provide exact information on the timing and tempo of development. Precise techniques to measure visual fixations and eye movements were employed.

This introductory chapter begins with an overview on vision, attention, and eye movements in early infancy. Different methods of measuring eye movements are discussed. Then, the biological mechanisms which are thought to underlie shifts of

10

Chapter 1

11

Introduction

gaze and attention and their development are expounded. Next, different aspects of the role of visual input are examined: The impact of exposure to visual stimulation is considered, and the question is raised whether (extra) visual experience influences the early visual and attentional development. Then, the relation between the sort of visual input and its processing is investigated. In this context, the importance of using ecologically valid stimuli in experiments is discussed. The chapter ends with a description of the goals of this study and an outline of the dissertation.

Attention Shifts and Eye Movements in InfancyShifts of gaze and shifts of attention are tightly associated, although they do not

necessarily occur conjointly (Stelmach, Campsall, & Herdman, 1997). Whereas overt shifts of attention involve saccadic movements of the eyes in order to align the fovea of the retina, the focus of attention can also be shifted covertly in absence of eye or head-movements (Wright & Ward, 1998). It has been argued that covert attention shifts occur as premotor activity in the preparation of eye movements but without an actual gaze shift (Rizzolatti, Riggio, Dascola, & Umiltà, 1987). In daily life, however, attention and gaze shifting are usually very closely joined.

Gaze and attention shifting has been studied extensively in diverse research fields. For example, topics as diverse as stimulus discrimination, memory, or concept formation have been investigated using the habituation paradigm. This procedure is based on the idea that the way infants look at and away from a stimulus signifies their processing of information about the stimulus (Bornstein, 1985). In addition, variables measured during infant habituation studies – as duration of the first fixation, the total fixation time, or the duration of the longest fixation – have been shown to predict later cognitive abilities (for reviews, see Colombo, 1993; Slater, 1995).

Also, gaze shifting behavior has been analyzed in studies on social development. Eye contact has been studied as one of the earliest forms of social interaction and as an important factor for the attachment of mother and infant and for the quality of their relationship (Keller & Gauda, 1987; Schölmerich, Leyendecker, & Keller, 1995). Research on face-to-face interaction between mother and infant has examined infants’ looking to or away from their mothers’ face and has revealed that infants tend to regulate the intensity of an interaction by shifting their gaze (Field, 1979; Stifter & Moyer, 1991).

Furthermore, it has been shown that children who are able to disengage attention and move gaze more easily, are less susceptible to distress and are more soothable (Johnson, Posner, & Rothbart, 1991). These various examples emphasize the importance of shifts of attention and gaze in infancy. They also illustrate that attention and gaze shifting behavior has been studied to gain insight into very different areas of infant psychol-ogy, such as visual exploration, early communication, or arousal modulation.

Also diagnostic instruments which assess infants’ developmental status during the neonatal or the infant period are among other tasks based on items which re-quire visual orientation to animate and inanimate stimuli or shifting gaze between

12

Chapter 1

13

Introduction

two objects (Neonatal Behavioral Assessment Scale, NBAS, Brazelton & Nugent, 1995; Bayley Scales of Infant Development, BSID-II, Bayley, 1993) .

Development of VisionFor a long time, it was commonly accepted that newborn and young infants per-

ceive their environment in an extremely impoverished way (see e.g., Dewey, 1935; Pratt, 1954; see also Stone, Smith, & Murphy, 1973, p. 3-4 for more examples). Also Wilhelm Preyer, whose book “Die Seele des Kindes” (1882) is often considered as the beginning of infant psychology, reported the relative immaturity of the visual system of a human newborn and, as one of the first, sought to examine a young infant’s reac-tions to visual stimuli. William James described the early experiences of an infant as “one great blooming, buzzing confusion” (James, 1890, Vol. 1, p. 488), and also Jean Piaget (1936, 1937) emphasized the strong sensory limitations during infancy. It was not before the 1960s that researchers started to specify what exactly the perceptual limitations of early infancy are, and concentrated on describing what young infants are able to see.

It is clear that young infants’ vision falls far short of adult standards. The optical state of the eye, though, seems to be quite good in the newborn. Infants can accom-modate on targets which are relatively close already during the first days of life (Brad-dick, Atkinson, French, & Howland, 1979; Haynes, White, & Held, 1965). Visual acuity in newborns, however, has been shown to be about 1/30 of the acuity of adult levels and to have improved by a factor of 3 to 4 by 5 months of age (Fantz, Ordy, & Udelf, 1962). However, it has been argued that unsharp vision does not handicap the infants, as they can still perceive relevant features at a proper distance, but might instead even help them to prevent excessive visual stimulation (Maurer & Maurer, 1988).

Even very young infants seem to have some form of color vision, and it is probably very similar to that of adults. However, infants of 1 month of age have been shown to be unable to discriminate certain colors (e.g., green and yellow), but by 2 months they succeed in making most color discriminations (Clavadetscher, Brown, Ankrum, & Teller, 1988). In a complex colorful stimulus, young infants thus probably do not distinguish as many different colors as adults would, and colors presumably appear less intense to them than they do to adults. When infants are 3 to 4 months old, their color vision is at a relatively adult level (Teller & Bornstein, 1987).

To sum up, although infants’ vision is not yet fully mature during the first few months after birth, they see well enough to respond appropriately to relevant aspects of the environment and function effectively in their roles as infants (Hainline, 1998; Hainline & Abramov, 1992).

Developmental Changes in Visual Attention and Eye MovementsEye movements emerge already during prenatal development (Prechtl, 1984). They

have been observed in 16 to 18 weeks gestation, but they are naturally not associated

12

Chapter 1

13

Introduction

with visual stimuli until exposure to light after birth. When infants are born, they have been demonstrated to prefer stimuli and objects with simple patterns and high contrast (Fantz & Yeh, 1979). They are able to localize a visual target, although in a rather inaccurate and unreliable way (Atkinson, 1992). Once they look at a stimulus, they scan it actively, but tend to fixate only limited parts of it (Haith, 1980; Bronson, 1990) and to ignore other stimuli in their visual field (Bronson, 1996; Haith, 1980; Salapatek, 1975). During the first weeks of life, infants have also been shown to look especially at edges or outer contours of a stimulus pattern (“contour salience effect”, Bronson, 1991) and not to attend to stationary inner parts of a stimulus (“externality effect”, Salapatek, 1975; Milewski, 1976). However, if the internal elements of a pattern are moving, they are more likely to be fixated even by very young infants (Bushnell, 1979; Girton, 1979). Newborns are able to track moving stimuli (Tauber & Koffler, 1966), although they tend to do it not in a smooth, but in a saccadic, step-wise way and tend to lag behind the movement of the stimulus (Aslin, 1981). Around 2 months of age then, infants have been observed to follow a moving visual target smoothly (von Hofsten & Rosander, 1996), but it is not before about 3 months that infants predict the movement of the stimulus in an anticipatory way and do not lag behind anymore (Aslin, 1981).

Between approximately 1 and 3 months of age, infants have trouble looking away from a stimulus, once their attention has been engaged, and they may exhibit long periods of staring. This phenomenon of disengagement difficulty has been called “sticky fixation” (Hood, 1995) or “obligatory attention” (Stechler & Latz, 1966). It can be observed in a laboratory context (Hood, Murray, King, Hooper, Atkinson, & Brad-dick, 1996; Aslin & Salapatek, 1975), in free looking situations (Stechler & Latz, 1966), during social interaction (Hopkins & van Wulfften Palthe, 1985), and is also reported frequently by mothers and other caretakers as an every-day experience. As infants grow older, disengaging attention and shifting gaze away from a stimulus becomes increasingly efficient. By 4 months of age, infants are able to shift gaze easily and rapidly, and staring behavior becomes rare (Hood & Atkinson, 1993).

When infants become more capable of achieving a balance between engaging and shifting attention, also their scanning behavior changes. Infants of about 3 months have been described to explore the stimulus under examination more consistently and more extensively (Bronson, 1996). They exhibit more brief fixations and scan more rapidly over the entire array of stimulus figures. Salient parts of a stimulus still attract the infants’ gaze, but they have gained volitional, strategic control over their scanning behavior (Bronson, 1994).

From 3 months on, infants not only start anticipating stimulus movement during visual tracking, but also begin to form expectations about the locations of upcoming stimuli and may even initiate an eye movement in advance (Haith, Hazan, & Good-man, 1988; Canfield & Haith, 1991). The increasing intentional control over their eye movements enables infants of 4 to 5 months to examine their environment in an ef-ficient and flexible way. They can shift their gaze fast and reliably between and within

14

Chapter 1

15

Introduction

visual stimuli and are able to direct their gaze to relevant locations. Eye movements are now generated in accordance with the strategic demands of ongoing information processing. When scanning familiar stimuli, recursive scanning patterns can be ob-served (Bronson, 1982). During face-to-face interaction, infants of 3 months and older tend to shift their gaze away more often, either in order to regulate arousal (Stifter & Moyer, 1991) or to explore other locations which are becoming increasingly more interesting to them (Kaye & Fogel, 1980).

Measuring Eye Movements in Infants Vision and human viewing patterns have fascinated researchers for a long time

(see e.g., Müller, 1826; Preyer, 1882). Eye movements and fixations in infants have been observed in order to address very different topics, such as visual scanning (Haith, 1980; Bronson, 1982) and visual processing (Bronson, 1991), the acquisition of object knowledge (Johnson & Johnson, 2000; Johnson, Slemmer, & Amso, 2004), or the forma-tion of categories (McMurray & Aslin, 2004).

Probably the most common method of studying eye movements is simple observa-tion of gaze. There are two more precise methods of measuring eye movements and fixations: electro-oculography (EOG) and corneal-reflection photography. EOG is based on measuring the change in electrical potential which accompanies the rotation of the eye. However, this method has several limitations, especially when used with young infants (Aslin & McMurray, 2004). It is particularly sensitive to artifacts and requires the application of electrodes on the subject’s face. Furthermore, EOG provides only data on the relative displacement of the eye and no information about where on the stimulus the subject is looking.

For corneal reflection eye-tracking, an (infrared) light source is used to create a reflection off the front surface of the eyeball, while the eye is recorded on video. The reflection is displaced when the subject moves fixation, and the information about the relative position of the corneal reflection with respect to the center of the pupil and its change is used to determine whether an eye movement took place. However, to gather information about the location of fixation, the corneal reflection eye-tracking system has to be individually calibrated before the measurement in order to map the output data onto the field the subject is looking at (Bronson, 1983; Harris, Hainline, & Abramov, 1981).

The technique of infrared corneal photography has been applied to human infants first in the 1960s (Salapatek & Kessen, 1966; Haith, 1969) and has been improved in many respects since then. The sampling rate has been increased from 2 - 6 Hz to 30 - 60 Hz (Aslin, 1981; Hainline, 1981; Bronson, 1982) and even to 120 and 250 Hz. The increased temporal resolution has greatly amended the accuracy in determining fixation dura-tions. Another recent improvement is to have the camera mounted on a motor-driven base which moves in order to maintain the image of the eye in the camera’s field-of-view and compensate for head-movements of the infant. However, large or rapid head-

14

Chapter 1

15

Introduction

movements still formed a problem for a corneal photography eye-tracking system. This resulted in experimental setups in which the infant’s head had to be restrained. As this is a quite unnatural situation and can be distressing for the infants, the implementation of a head-tracker, a position-sensing system that monitors head movements and commu-nicates this information to the eye-tracking system, means an important innovation.

Today, it is thus possible to examine how infants look at different stimuli in a more precise and at the same time more natural way than ever before. However, there are still some problems, which form a challenge for the researcher: The accuracy of the measurements of the location of fixations depends largely on the quality of the calibration carried out. For working with infants as young as 6 weeks of age, custom-built calibration procedures have to be developed in order to make the infants fixate a sequence of (preferably many) calibration points. Furthermore, young infants tend to have poor postural control, which requires several accommodations of the experi-mental setup, including the position of the infrared camera and the stimulus display. To sum up, measuring eye movements in infants remains a challenging task for the researcher as it is tried to apply a complex and highly sensitive technique to delicate subjects, who are indifferent to instructions.

Neurophysiological Models of Eye Movement Generation in AdultsAfter this overview on vision and eye movements in infancy and on early visual

and attentional development, the question remains which neurophysiological pro-cesses underlie the described functions and developmental changes. The following two paragraphs are dedicated to this question.

Eye movements are controlled by different cortical and subcortical structures. Many neurobiological models proceed on the assumption of two visual systems, a phylogenetically older retinotectal system and a newer geniculostriate system. Early anatomical studies had already identified two distinct streams from the retina through the brain, but the functional distinction arose from studies in the 1950s en 1960s (see e.g., Sprague & Meikle, 1965). The first models featuring the distinction between two routes were proposed by Trevarthen (1968) and Schneider (1969). Trevarthen (1968) described an “ambient” system for movement control and a “focal” system for object vision. Schneider (1969) suggested that the tectal system defines “where” an object is located and initiates an orienting response and the newer cortical mechanisms are about “what” there is to see precisely in the selected location. In 1982, Ungerleider and Mishkin revised the dichotomy into a ventral versus a dorsal cortical stream. According to this model, the ventral stream, concerned with the “what”-aspects of an object as color, form, or face recognition, is assigned to the inferotemporal cortex. The identification of spatial location, on the other hand, is thought to be subserved by the dorsal stream (“where”), which is anchored by the posterior parietal cortex. The dorsal versus ventral distinction has been associated with a division earlier in the visual pathway, namely between the parallel magnocellular and parvocellular sys-

16

Chapter 1

17

Introduction

tems (Livingstone & Hubel, 1988; Shapley & Perry, 1986; Van Essen & Maunsell, 1983). These two systems are anatomically segregated at the retina and the lateral geniculate nucleus and project to different parts of the primary visual cortex. The parvocellular based system subserves form and color vision, whereas the magno cells are specialized in movement perception and some aspects of stereoscopic vision. However, the simple distinction into these two parallel systems has been questioned recently, as there are many interactions in visual processing between the magnocellular and the parvocel-lular stream (see e.g., Cowey, 1994; Merigan & Maunsell, 1993 for reviews).

One of the currently predominant models of eye movement generation is the one developed by Peter Schiller (Schiller, 1985, 1998). His recent model (Schiller, 1998) is based on adult primate electrophysiological and lesion data and also distinguishes be-tween two different, but partly overlapping, neural systems of eye movement control: the anterior and the posterior eye movement control system. The anterior system is responsible for saccades that are voluntary or planned (as e.g., scanning behavior), whereas the posterior system generates fast, reflex-like eye movements and orient-ing responses, as they occur, for example after the sudden appearance of a salient stimulus in the periphery.

The streams of the anterior system originate in retinal ganglia, which are special-ized for the analysis of fine detail and color (Richards & Hunter, 1998). They project through the lateral geniculate nucleus to the striate cortex, and from there they run through the temporal or the parietal lobe to the frontal eye fields. Then they project via the basal ganglia and the superior colliculus to the eye movement centers of the brain stem. However, these brain stem structures also receive direct input from the frontal eye fields within the anterior eye movement control system.

The posterior eye movement control system receives the majority of its input from the retinal ganglia, which are located in the peripheral retina and are specialized for the detection of sudden changes (Richards & Hunter, 1998). Its pathways run via the lateral geniculate nucleus to the striate cortex. Then, they project – partially via the parietal lobe – through the basal ganglia to the superior colliculus.

The activity of the superior colliculus thus is controlled via both systems, the anterior as well as the posterior eye movement control system. Their excitatory or inhibitory input plays an important role in the generation of eye movements to in-teresting location and at the same time in the inhibition of reflexive eye movements in order to ensure a well-organized input of visual information. However, within the anterior eye movement control system, there is also a pathway which bypasses the superior colliculus and hereby enables the generation or inhibition of eye movements independently from collicular control.

Neuropsychological Models of Attentional Development in InfantsThe visual and attentional behavior of an infant is largely determined by the

developmental state of the brain structures which form the visual system. Changes

16

Chapter 1

17

Introduction

observed in behavior and neural correlate can be due to maturation, but can as well be a response to experience (Greenough, Black, & Wallace, 1987). Also, there are sev-eral examples of neural and behavioral changes that are the result of an interaction between intrinsic factors and environmental aspects (Johnson & Morton, 1991; Gre-enough, Black, & Wallace, 1987).

Anatomical (Conel, 1939-1967) and PET scan studies (Chugani, 1994) have demon-strated that, generally, subcortical brain structures are more mature at birth than cortical visual mechanisms. The superior colliculus is one of the most mature struc-tures involved and is thought to play a crucial role in the generation of eye movements during early infancy.

Gordon Bronson was one of the first to propose a model which applied findings from research on adult neurological systems (e.g., Schneider, 1969; Trevarthen, 1968) to infant visual behavior (Bronson, 1974). According to his model, the early development of visual attention can be viewed as a shift from subcortical to cortical processing. Visual behavior in the newborn thus is mainly controlled by means of a phylogeneti-cally older visual system. It is only by 2 to 3 months of age that the locus of control switches to the primary visual system and its mainly cortical pathways.

However, the original model proposed by Bronson (1974) based on the “two visual systems” (Schneider, 1969) and a subcortical-cortical dichotomy has been criticized as being too simplistic and incomplete (Atkinson, 1984; Johnson, 1990). Also, the early pres-ence of certain perceptual abilities has given rise to the notion that there is at least some degree of cortical functioning at birth (e.g., Slater, Morison, & Somers, 1988). It is now known that several comparatively independent cortical streams of visual processing exist (see e.g., Van Essen, 1985) and that they undergo various forms of developmental changes, such as myelination, synaptic generation, neural innervation, synaptic prun-ing, and neurotransmitter development (see e.g., de Haan & Johnson, 2003).

Bronson’s most recent model (Bronson, 1994, 1996) is based on two pathways – the “striate” and the “poststriate networks” (Bronson, 1996) – which are similar to the posterior and anterior eye movement control system proposed by Schiller (1998, 1985) and the assumption that the changes observed in early visual behavior can be explained by reference to the maturational state of these pathways. During the first few weeks of life, eye movements are mainly controlled by the striate networks. These areas are highly responsive to stimulus salience, accordingly, young infants’ visual behavior tends to be mainly salience-guided. Once the fovea is aligned with an area of high salience, fixations are often concentrated around this area because – due to the anatomical structure of the retina – close salient areas produce higher striate activity than comparable areas further away. As highly salient stimuli produce long lasting activity, fixations tend to be long in young infants. From about 6 weeks of age on, the poststriate networks with their pathways through the parietal and frontal cortex become increasingly effective. This system comprises areas that are able to encode the location and the form of visual stimuli. These pathways project to the superior

18

Chapter 1

19

Introduction

colliculus and to the brain stem centers which directly generate eye movements. Older infants thus can draw on these poststriate capacities to override salience effects and move their eyes intentionally to locations of interest.

Also the model of visual and attentional development by Mark Johnson (1990, 1995; Johnson, Gilmore, & Csibra, 1998) refers to the eye movement control system by Schiller (1985), particularly to four distinct pathways. Johnson assumes that the characteristics of visually guided behavior mirror the degree of functionality of the four pathways – three cortical and one subcortical – and that the developmental state of the primary visual cortex determines which of theses pathways is functional. In correspondence with the inside-out pattern of postnatal development in the cerebral cortex (see e.g., Nowakowski, 1987; Rakic, 1988), he hypothesizes that the lower layers tend to be more capable than more superficial ones. In newborn infants, only the deeper layers of the primary visual cortex are functional, and the visually guided behavior is controlled predominantly by the subcortical pathway. According to Johnson, the saccadic pur-suit tracking observed in young infants (Aslin, 1981) and the phenomenon that young infants do not attend to a stationary pattern within a larger frame or pattern (“ex-ternality effect”) are characteristic of visual behavior that is controlled subcortically. Other behaviors, such as early pattern recognition (Slater, Morison, & Rose, 1982) or orientation discrimination (Atkinson, Hood, Wattam-Bell, Anker, & Tricklebank, 1988) suggest at least some cortical functioning also in the newborn infant. During the first month, the nigral pathway, which is an inhibitory input to the superior colliculus from several deeper layers of the primary visual cortex, becomes increasingly functional. According to Johnson, this unregulated tonic inhibition has as a transient consequence the infants’ disengagement difficulties, known as “sticky fixation” or “obligatory at-tention”. Around 2 months of age, infants begin to show smooth visual tracking. According to Johnson, the onset of this behavior coincidences with the functioning of the middle temporal area pathway. During the third and fourth month then, the pathways involving the frontal eye fields become functional, as the upper layers of the primary visual cortex mature. This leads to a more differentiated regulation of collicular activity and ends the tonic inhibition, which caused the staring behavior. Infants are now able to move their gaze intentionally from fixation to other locations of interest and to generate anticipatory eye-movements.

The models by Johnson and Bronson succeed in explaining the most important developmental changes of visual behavior in infancy, but they differ in the underly-ing processes which they assume to be involved. However, both models have strongly influenced the models proposed by other researchers (see e.g., Atkinson & Braddick, 2003; Richards & Hunter, 1998).

Premature Birth and the Role of Early Visual InputDevelopment emerges from the interaction of many different factors. As described

earlier, maturation of the eye and of the brain play a crucial role in the development

18

Chapter 1

19

Introduction

of visual and attentional skills. However, other factors might be equally important. The next two sections address different aspects of the role of environmental factors. First, early exposure to visual input caused by premature birth is discussed. The next paragraph focuses on the impact of the nature of visual stimulus material on young infants’ visual and attentional performance.

When infants are born prematurely, they are confronted with a very different environment than they experience in utero much earlier than their full-term age-mates. The preterm birth also puts them at risk for severe medical complications such as breathing problems, infections, and brain damage. In neonatal intensive care units (NICUs), many efforts are being made to create the optimal conditions according to the infants’ physiological needs (e.g., temperature and nutrition) and psychological requirements (e.g., stimulation and contact). A large amount of research on the optimal treatment of preterm infants has been carried out to date (see e.g., Holditch-Davis & Black, 2003; Wolke, 1987).

One important difference between full- and preterm infants is that infants born prematurely are confronted with visual input earlier in life than full-term infants. There are contrasting theories about the impact of this extra experience on the vi-sual development of the infant. One theory implies that healthy preterms benefit from their early exposure to the visual world (Fielder, Foreman, Moseley, & Robinson, 1993). This account is supported by Hunt and Rhodes (1977), who found that during the early months preterm infants have higher scores on the mental scale (MDI) of the Bayley Scales of Infant Development (Bayley, 1969), a scale which – when very young infants are tested – relies mainly on the infants’ visual responses. Also, superiority of visual acuity (Sokol & Jones, 1979; Mohn & van Hof-van Duin, 1986) and more mature focusing and tracking of moving stimuli (Dubowitz, Dubowitz, Morante, & Verghote, 1980; Bloch, 1983) have been described in preterm infants compared to full-terms of the same (corrected) age.

On the other hand, it has been suggested that the immature visual system might suffer from early exposure to visual stimulation (Friedman, Jacobs, & Werthmann, 1981; Turkewitz & Kenny, 1985). In accord with this, there are several studies report-ing that preterm infants have longer look durations in habituation studies (Rose, Feldman, McCarton, & Wolfson, 1988; Spungen, Kurtzberg, & Vaughan, 1985) as well as difficulties localizing new stimuli (Landry, Leslie, Fletcher, & Francis, 1985). In free play, they also pay less attention to toys (Landry & Chapieski, 1988).

It has been shown that preterm infants’ problems concerning visual processing and recognition memory persist throughout the first year of life (Rose, Feldman, & Jankowski, 2001; Rose, 1983). In later childhood these infants tend to have lower scores on attention tests (Taylor, Hack, & Klein, 1998) and cognitive scales (Wolke & Meyer, 1999). Even in early adolescence, they have been shown to be at risk concerning their intellectual functioning (Botting, Powls, Cooke, & Marlow, 1998).

To summarize, there are two major reasons to examine preterm infants’ develop-

20

Chapter 1

21

Introduction

ment of vision and attention: As described above, preterm infants tend to perform more poorly on tasks requiring attentional and processing skills, even later in child-hood. Sorting out precisely how the development of vision and attention is different in preterm infants is crucial in order to understand their eventual deficits and to develop suitable and effective interventions. On the other hand, examining healthy preterm and full-term infants is of great scientific interest, as it offers the possibility to learn more about the roles of maturation and experience in visual and attentional development.

The Role of the Nature of the StimuliIn 1977, Urie Bronfenbrenner provokingly postulated that “much of contemporary

developmental psychology is the science of the strange behavior of children in strange situations … for the briefest possible periods of time” (Bronfenbrenner, 1977, p. 513). Still, more than 25 years after Bronfenbrenner’s well-known criticism and about 60 years after Egon Brunswik introduced the concept of ecological validity (Brunswik, 1943), developmental psychologists struggle with the demand of producing research that allows generalizing to real world phenomena.

A prominent topic in the debate around ecological validity of experimental studies is the choice of stimuli (Schmuckler, 2001). The stimulus material used for instance in studies on cognitive processes has been criticized as “abstract, discontinuous and marginally real”, and results of these kind of studies have been suspected to “be ir-relevant to the phenomenon that one would like to explain” (Neisser, 1976, p. 33, 34). Lewkowicz (2001) correctly remarks that setting up an experiment with ecologically valid stimulus material does not necessarily mean creating conditions which are as naturalistic as possible. Instead, it is essential to identify the relevant aspects of the natural environment that control the infant’s responsiveness and to capture them in the stimuli used.

However, when examining the current research on infants’ development of percep-tion and attention, one can still question the ecological validity of a large number of studies. For example, much we know regarding attention, perceptual responsiveness, and information processing during infancy is based on experiments using unimodal stimuli, although it has been shown that multimodal stimulation can elicit enhanced responsiveness (Bahrick, 1992, 1994).

Another example are studies on the perception and scanning of faces. Most of these studies have been carried out with schematic drawings (e.g., Maurer & Maurer, 1988; Caron, Caron, Caldwell, & Weiss, 1973) or photographs of faces (e.g., Hainline, 1978) or manikins (e.g., Carpenter, 1974). When real faces have been used, they often were still faces (e.g., Maurer & Salapatek, 1976; Bronson, 1982). The generalizability of the findings from these studies is unknown. At the same time, there are several reasons to assume that infants’ reactions to a naturally moving, smiling face might be considerably different. First, it has been demonstrated that moving stimuli – both

20

Chapter 1

21

Introduction

faces (Wilcox & Clayton, 1968; Haith, Bergman, & Moore, 1977) and non-face stimuli (Tronick, 1972) – attract more attention in infants. Further, there are indications that moving stimuli are regarded differently than static displays (Bronson, 1990; Girton, 1979; Johnson & Johnson, 2000). Examples like these emphasize the importance of complying with the demands of ecological validity, especially when investigating young infants skills and competences.

Issues of Further Investigation and Goals of the StudyThe developmental changes which occur in the visual attentional behavior of

young infants have been studied in detail. However, nearly all of those studies have used only unnatural, often abstract stimulus material. Information on the early de-velopment of attention and eye movements in the context of natural stimuli is largely missing (but see e.g., Bornstein & Ludemann, 1989). Consequently, the first aim of the current study was to fill in this gap by examining the development of two important attentional skills – disengagement of attention and visual scanning – using two care-fully selected stimulus types. As an ecologically valid stimulus, a video recording of the face of each infant’s mother was used. Further, it was chosen to present the mother’s face in a natural way. Thus, for the video recording she was filmed mov-ing and smiling as she would do during a face-to-face interaction with her baby. In order to test the influence of different sorts of stimuli, it was also chosen to use a second, abstract stimulus which matched the actual mother video.

The majority of the studies on the development of gaze and attention shifting and visual scanning presents only cross-sectional data with broad age intervals. Studies which provide longitudinal data with several dense measurement occa-sions are scarce (but see e.g., Butcher, Kalverboer, & Geuze, 2000). However, only the latter type of research can provide detailed information on developmental trajectories, the timing and tempo of developmental change and inter-infant dif-ferences concerning this change. In this study, an intensive, longitudinal design was used. The intervals between measurements were kept short, because rapid de-velopment was expected during the measurement period. Using this design avoids the random variance that arises when different groups of infants are compared at different ages and allows studying and comparing the developmental trajectories of the two attentional mechanisms chosen as well as the interindividual variance of this development.

As described earlier, infants born prematurely are exposed to visual stimula-tion earlier in their development and, at the same time, have been shown to have an increased risk for later attentional problems. In early infancy both inferior as well as enhanced visual functioning has been described and theoretically under-pinned. Further research on the early development of fundamental visual atten-tional mechanisms is needed. A third goal of this study was thus to compare the development of attention and gaze shifting observed in a group of healthy infants

22

Chapter 1

23

Introduction

to the developmental trajectory found in a group of preterm infants.Only in the recent years, a technique to measure eye movements in very young

infants reliably and non-intrusively has become available. In order to describe scan-ning patterns properly, it is necessary to rest upon precise and solid data on shifts of gaze as well as locations of fixation points. Therefore, the current study makes use of the latest eye-tracking techniques.

Outline of the DissertationThis thesis is divided into six chapters. Following this 1st introductory chapter, the

development of visual scanning is addressed in Chapter 2. The eye movements and fixations while scanning a naturally moving face and a matched abstract stimulus are registered throughout the first few months of infancy. The development of attention and gaze shifting between the two types of stimuli is examined in Chapter 3. In Chapter 4, the association between the development of these skills – visual scanning and gaze shifting – is explored. While the Chapters 2, 3, and 4 are devoted to the development of full-term infants with no history of medical complications, Chapter 5 deals with the comparison of a group of preterm infants and a group of full-terms. Again, the development of gaze shifting between faces and abstract stimuli is analyzed. The last chapter, Chapter 6, presents summaries of the preceding chapters and ends with a discussion of the general conclusions to be drawn from this study, its limitations and its implications for further research.

REFERENCESAslin, R. N. (1981). Development of smooth pursuit in human infants. In D. F. Fisher, R. A.

Monty, & J. W. Senders (Eds.), Eye movements: Cognition and visual perception (pp. 31-41). Hillsdale, NJ: Erlbaum Associates, Inc.

Aslin, R. N., & McMurray, B. (2004). Automated corneal-reflection eye-tracking in infan-cy: Methodological developments and applications to cognition. Infancy, 6, 155-163.

Aslin, R. N., & Salapatek, P. (1975). Saccadic localization of visual targets by the very young human infant. Perception and Psychophysics, 17, 293-302.

Atkinson, J. (1984). Human visual development over the first six months of life. Human Neurobiology, 3, 61-74.

Atkinson, J. (1992). Early visual development: Differential functioning of parvocellular and magnocellular pathways. Eye, 6, 129-135.

Atkinson, J., & Braddick, O. (2003). Neurobiological models of normal and abnormal visual development. In M. de Haan & M. H. Johnson (Eds.), The cognitive neuroscience of development (pp. 43-71). Hove, UK: Psychology Press.

Atkinson, J., Hood, B. M., Wattam-Bell, J., Anker, S., & Tricklebank, J. (1988). Development of orientation discrimination in infancy. Perception, 17, 587-595.

22

Chapter 1

23

Introduction

Bahrick, L. E. (1992). Infants’ perceptual differentiation of amodal and modality specific audio-visual relations. Journal of Experimental Child Psychology, 53, 180-199.

Bahrick, L. E. (1994). The development of infants’ sensitivity to arbitrary intermodal relations. Ecological Psychology, 6, 111-123.

Bayley, N. (1969). Manual for the Bayley Scales of Infant Development. New York: Psychologi-cal Corporation.

Bayley, N. (1993). Bayley Scales of Infant Development (Second Edition). San Antonio, TX: The Psychological Corporation.

Bloch, H. (1983). La poursuite visuelle chez le nouveau-né à terme et chez le prématuré. Enfance, 1, 19-29.

Bornstein, M. H. (1985). Habituation of attention as a measure of visual processing in human infants: Summary, systematization, and synthesis. In G. Gottlieb & N. A. Krasnegor (Eds.), Measurement of audition and vision in the first year of postnatal life: A methodological overview (pp. 253-300). Norwood, NJ: Ablex.

Bornstein, M. H., & Ludemann, P. M. (1989). Habituation at home. Infant Behavior and Development, 12, 525-529.

Botting, N., Powls, A., Cooke, R. W., & Marlow, N. (1998). Cognitive and educational out-come of very-low-birthweight children in early adolescence. Developmental Medicine and Child Neurology, 40, 652-660.

Braddick, O., Atkinson, J., French, J., & Howland, H. C. (1979). A photorefractive study of infant accommodation. Vision Research, 19, 1319-1330.

Brazelton, T. B., & Nugent, J. K. (1995). The Neonatal Behavioral Assessment Scale. London: Mac Keith Press.

Bronfenbrenner, U. (1977). Toward an experimental ecology of human development. American Psychologist, 32, 513-531.

Bronson, G. (1974). The postnatal growth of visual capacity. Child Development, 45, 873-890.

Bronson, G. W. (1982). The scanning patterns of human infants: Implications for visual learning. Monographs on Infancy No. 2. Norwood, NJ: Ablex.

Bronson, G. W. (1983). Potential sources of error when applying a corneal reflex eye-monitoring technique to infant subjects. Behavior Research Methods and Instrumenta-tion, 15, 22-28.

Bronson, G. W. (1990). Changes in infants’ visual scanning across the 2- to 14-week age period. Journal of Experimental Child Psychology, 49, 101-125.

Bronson, G. W. (1991). Infant differences in rate of visual encoding. Child Development, 62, 44-54.

Bronson, G. W. (1994). Infants’ transitions towards adult-like scanning. Child Develop-ment, 65, 1243-1261.

Bronson, G. W. (1996). The growth of visual capacity: Evidence from infant scanning patterns. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 11, pp. 109-141). Norwood, NJ: Ablex.

24

Chapter 1

25

Introduction

Brunswik, E. (1943). Organismic achievement and environmental probability. The Psy-chological Review, 50, 255-272.

Bushnell, I. W. R. (1979). Modification of the externality effect in young infants. Journal of Experimental Child Psychology, 28, 211-229.

Butcher, P. R., Kalverboer, A. F., & Geuze, R. H. (2000). Infants’ shifts of gaze from a cen-tral to a peripheral stimulus: A longitudinal study of development between 6 and 26 weeks. Infant Behavior and Development, 23, 3-21.

Canfield, R. L., & Haith, M. M. (1991). Young infants’ visual expectations for symmetric and asymmetric stimulus sequences. Developmental Psychology, 27, 198-208.

Caron, A. J., Caron, R. F., Caldwell, R. C., & Weiss, S. J. (1973). Infant perception of the structural properties of the face. Developmental Psychology, 9, 385-399.

Carpenter, G. C. (1974). Visual regard of moving and stationary faces in early infancy. Merrill-Palmer Quarterly, 20, 181-194.

Chugani, H. T. (1994). Development of regional brain glucose metabolism in relation to behavior and plasticity. In G. Dawson & K. Fischer (Eds.), Human behavior and the developing brain (pp. 153-175). New York: Guilford.

Clavadetscher, J. E., Brown, A. M., Ankrum, C., & Teller, D. Y. (1988). Spectral sensitivity and chromatic discriminations in 3- and 7-week-old infants. Journal of the Optical Society of America A, 5, 2093-2105.

Colombo, J. (1993). Infant cognition: Predicting later intellectual functioning. Newbury Park, CA: Sage.

Conel, J. L. (1939-1967). The postnatal development of the human cerebral cortex (Vol. 1-8). Cambridge, MA: Harvard University Press.

Cowey, A. (1994). Cortical visual areas and the neurobiology of higher visual processes. In M. J. Farah & G. Radcliff (Eds.), The neuropsychology of high-level vision (pp. 3-31). Hillsdale, NJ: Erlbaum Associates, Inc.

de Haan, M., & Johnson, M. H. (2003). Mechanisms and theories of brain development. In M. de Haan & M. H. Johnson (Eds.), The cognitive neuroscience of development (pp. 1-18). Hove, UK: Psychology Press.

Dewey, E. (1935). Behavior development in infants. New York: Columbia University Press.Dubowitz, L. M., Dubowitz, V., Morante, A., & Verghote, M. (1980). Visual function in

the preterm and fullterm newborn infant. Developmental Medicine and Child Neurology, 22, 465-475.

Fantz, R. L., Ordy, J. M., & Udelf, M. S. (1962). Maturation and pattern vision in infants during the first six months. Journal of Comparative and Physiological Psychology, 55, 907-917.

Fantz, R. L., & Yeh, J. (1979). Configurational selectivities: Critical for development of visual perception and attention. Canadian Journal of Psychology, 33, 277-287.

Field, T. M. (1979). Visual and cardiac responses to animate and inanimate faces by young term and preterm infants. Child Development, 50, 188-194.

24

Chapter 1

25

Introduction

Fielder, A. R., Foreman, N., Moseley, M. J., & Robinson, J. (1993). Prematurity and visual development. In K. Simons (Ed.), Early visual development, normal and abnormal (pp. 485-504). New York: Oxford University Press.

Friedman, S. L., Jacobs, B. S., & Werthmann, M. W. (1981). Sensory processing in pre-and full-term infants in the neonatal period. In S. L. Friedman & M. Sigman (Eds.), Preterm birth and psychological development (pp. 159-178). New York: Academic Press.

Girton, M. R. (1979). Infants’ attention to intrastimulus motion. Journal of Experimental Child Psychology, 28, 416-423.

Greenough, W. T., Black, J. E., & Wallace, C. S. (1987). Experience and brain development. Child Development, 58, 539-559.

Hainline, L. (1978). Developmental changes in visual scanning of face and nonface pat-terns by infants. Journal of Experimental Child Psychology, 25, 90-115.

Hainline, L. (1981). An automated eye movement recording system for use with human infants. Behavior Research Methods and Instrumentation, 13, 20-24.

Hainline, L. (1998). The development of basic visual abilities. In A. Slater (Ed.), Percep-tual development. Visual, auditory, and speech perception in infancy (pp. 5-50). Hove, UK: Psychology Press.

Hainline, L., & Abramov, I. (1992). Assessing visual development: Is infant vision good enough? In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 7, pp. 30-102). Norwood, NJ: Ablex.

Haith, M. M. (1969). Infrared television recording and measurement of ocular behavior in the human infant. American Psychologist, 24, 279-282.

Haith, M. M. (1980). Rules that babies look by. Hillsdale, NJ: Erlbaum Associates, Inc.Haith, M. M., Bergman, T., & Moore, M. J. (1977). Eye contact and face scanning in early

infancy. Science, 198, 853-855. Haith, M. M., Hazan, C., & Goodman, G. S. (1988). Expectation and anticipation of dynamic

visual events by 3.5-month-old babies. Child Development, 59, 467-479.Harris, C. M., Hainline, L., & Abramov, I. (1981). A method for calibrating an eye-monitor-

ing system for use with human infants. Behavior Research Methods and Instrumentation, 13, 11-20.

Haynes, H., White, B. L., & Held, R. (1965). Visual accommodation in human infants. Science, 148, 528-530.

Holditch-Davis, D., & Black, B. (2003). Care of preterm infants: Programs of research and their relationship to developmental science. In J. J. Fitzpatrick (Series Ed.), M. Miles & D. Holditch-Davis (Vol. Eds.), Annual review of nursing research: Vol. 2. Research on child health and pediatric issues (pp. 23-60). New York: Springer.

Hood, B. M. (1995). Shifts of visual attention in the human infant: A neuroscientific approach. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 9, pp. 163-216). Norwood, NJ: Ablex.

Hood, B. M., & Atkinson, J. (1993). Disengaging visual attention in the infant and adult. Infant Behavior and Development, 16, 405-422.

26

Chapter 1

27

Introduction

Hood, B. M., Murray, L., King, F., Hooper, R., Atkinson, J., & Braddick, O. (1996). Habitua-tion changes in early infancy: Longitudinal measures from birth to 6 months. Journal of Reproductive and Infant Psychology, 14, 177-185.

Holditch-Davis, D., & Black, B. P. (2003). Care of preterm infants: programs of research and their relationship to developmental science. Annual Review of Nursing Research, 21, 23-60.

Hopkins, B., & van Wulfften Palthe, T. (1985). Staring in infancy. Early Human Develop-ment, 12, 261-267.

Hunt, J. V., & Rhodes, L. (1977). Mental development of preterm infants during the first year. Child Development, 48, 204-210.

James, W. (1890). Principles of psychology. New York: Holt.Johnson, M. H. (1990). Cortical maturation and the development of visual attention in

early infancy. Journal of Cognitive Neuroscience, 2, 81-95.Johnson, M. H. (1995). The development of visual attention: A cognitive neuroscience

perspective. In M. S. Gazzaniga (Ed.), The cognitive neurosciences (pp. 735-747). Cam-bridge, MA: MIT Press.

Johnson, M. H., Gilmore, R. O., & Csibra, G. (1998). Toward a computational model of the development of saccade planning. In J. E. Richards (Ed.), Cognitive neuroscience of atten-tion: A developmental perspective (pp. 103-130). Hillsdale, NJ: Erlbaum Associates, Inc.

Johnson, M. H., & Morton, J. (1991). Biology and cognitive development: The case of face rec-ognition. Oxford, UK: Blackwell.

Johnson, M. H., Posner, M. I., & Rothbart, M. K. (1991). Components of visual orienting in early infancy: Contingency learning, anticipatory looking, and disengaging. Journal of Cognitive Neuroscience, 3, 335-344.

Johnson, S. P., & Johnson, K. L. (2000). Early perception-action coupling: Eye movements and the development of object perception. Infant Behavior and Development, 23, 461-483.

Johnson, S. P., Slemmer, J. A., & Amso, D. (2004). Where infants look determines how they see: Eye movements and development of object perception. Infancy, 6, 185-201.

Kaye, K., & Fogel, A. (1980). The temporal structure of face-to-face communication be-tween mothers and infants. Developmental Psychology, 16, 454-464.

Keller, H., & Gauda, G. (1987). Eye contact in the first months of life and its developmental consequences. In H. Rauh & H.-Ch. Steinhausen (Eds.), Advances in Psychology, No. 46. Psychobiology and Early Development (pp. 129-143). Amsterdam: Elsevier.

Landry, S. H., Leslie, N. A., Fletcher, J. M., & Francis, D. J. (1985). Visual attention skills of premature infants with and without intraventricular hemorrhage. Infant Behavior and Development, 8, 309-321.

Landry, S. H., & Chapieski, M. L. (1988). Visual attention during toy exploration in pre-term infants: Effects of medical risk and maternal interaction. Infant Behavior and Development, 11, 187-204.

26

Chapter 1

27

Introduction

Lewkowicz, D. J. (2001). The concept of ecological validity: What are its limitations and is it bad to be invalid? Infancy, 2, 437-450.

Livingstone, M., & Hubel, D. H. (1988). Segregation of form, color, movement and depth: anatomy, physiology and perception. Science, 240, 740-749.

Maurer, D., & Maurer, C. (1988). The world of the newborn. New York: Basic Books.Maurer, D., & Salapatek, P. (1976). Developmental changes in the scanning of faces by

young infants. Child Development, 47, 523-527.McMurray, B., & Aslin, R. N. (2004). Anticipatory eye movements as a window on infants’

auditory and visual categories. Infancy, 6, 203-229.Merigan, W. H., & J. H. R. Maunsell (1993). How parallel are the primate visual pathways?

Annual Review of Neuroscience, 16, 369-402.Milewski, A. (1976). Infant’s discrimination of internal and external pattern elements.

Journal of Experimental Child Psychology, 22, 229-246.Mohn, G., & van Hof-van Duin, J. (1986). Development of the binocular and monocular

visual fields of human infants during the first year of life. Clinical Vision Science, 1, 51-64.

Müller, J. (1826). Zur vergleichenden Physiologie des Gesichtssinnes des Menschen und der Thiere nebst einem Versuch über die Bewegungen der Augen und über den menschlichen Blick. Leipzig, Germany: C. Cnobloch.

Neisser, U. (1976). Cognition and reality: Principles and implications of cognitive psychology. San Francisco: Freeman.

Nowakowski, R. S. (1987). Basic concepts of CNS development. Child Development, 58, 568-595.

Piaget, J. (1936). La naissance de l’intelligence chez l’enfant. Neuchâtel, Switzerland: Delachaux et Niestlé.

Piaget, J. (1937). La construction du réel chez l’enfant. Neuchâtel, Switzerland: Delachaux et Niestlé.

Pratt, K. C. (1954). The neonate. In L. Carmichael (Ed.), Manual of child psychology (2nd ed., pp. 215-291). New York: Wiley.

Prechtl, H. F. R. (1984). Continuity and change in early neural development. In H. F. R. Prechtl (Ed.), Continuity of neural functions from prenatal to postnatal life. Clinics in Developmental Medicine No. 94 (pp 1-15). Oxford, UK: Blackwell.

Preyer, W. (1882). Die Seele des Kindes. Beobachtungen über die geistige Entwicklung des Menschen in den ersten Lebensjahren. Leipzig, Germany: Th. Grieben’s Verlag.

Rakic, P. (1988). Specification of cerebral cortical area. Science, 241, 170-176.Richards, J. E., & Hunter, S. K. (1998). Attention and eye movement in young infants:

Neural control and development. In J. E. Richards (Ed.), Cognitive neuroscience of atten-tion: A developmental perspective (pp. 131-162). Mahwah, NJ: Erlbaum Associates, Inc.

Rizzolatti, G., Riggio, L., Dascola, I., & Umiltà, C. (1987). Reorienting attention across the horizontal and vertical meridians: Evidence in favor of a premotor theory of atten-tion. Neuropsychologia, 25, 31-40.

28

Chapter 1

29

Introduction

Rose, S. A. (1983). Differential rates of visual information processing in full-term and preterm infants. Child Development, 54, 1189-1198.

Rose, S. A., Feldman, J. F., & Jankowski, J. J. (2001). Attention and recognition memory in the 1st year of life: A longitudinal study of preterm and full-term infants. Devel-opmental Psychology, 37, 135-151.

Rose, S. A., Feldman, J. F., McCarton, C. M., Wolfson, J. (1988). Information processing in seven-month-old infants as a function of risk status. Child Development, 59, 589-603.

Salapatek, P. (1975). Pattern perception in early infancy. In L. B. Cohen & P. Salapatek (Eds.), Infant perception: From sensation to cognition (Vol. 1, pp. 133-248). New York: Academic Press.

Salapatek, P., & Kessen, W. (1966). Visual scanning of triangles by the human newborn. Journal of Experimental Child Psychology, 3, 155-167.

Schiller, P. (1985). A model for the generation of visually guided saccadic eye movements. In D. Rose & V. G. Dobson (Eds.), Models of the visual cortex (pp. 62-70). Chichester, UK: Wiley.

Schiller, P. (1998). The neural control of visually guided eye movements. In J. E. Richards (Ed.), Cognitive neuroscience of attention (pp. 5-50). Mahwah, NJ: Erlbaum Associates, Inc.

Schmuckler, M. A. (2001). What is ecological validity? A dimensional analysis. Infancy, 2, 419-436.

Schneider, G. E. (1969). Two visual systems. Science, 163, 895-902.Schölmerich, A., Leyendecker, B., & Keller, H. (1995). The study of early interaction in a

contextual perspective: Culture, communication, and eye contact. In J. Valsiner (Ed.), Child development within culturally structured environments. Vol. III: Comparative-cultural and constructivist perspectives (pp. 29-50). Norwood, NJ: Ablex.

Shapley, R., & Perry, V. H. (1986). Cat and monkey retinal ganglion cells and their visual functional roles. Trends in Neurosciences, 9, 229-235.

Slater, A. (1995). Individual differences in infancy and later IQ. Journal of Child Psychology and Psychiatry and Allied Disciplines, 36, 69-112.

Slater, A., Morison, V., & Rose, D. (1983). Locus of habituation in the human newborn. Perception, 12, 593-598.

Slater, A., Morison, V, & Somers, M. (1988). Orientation discrimination and cortical functioning in the human newborn. Perception, 17, 597-602.

Sokol, S., & Jones, K. (1979). Implicit time of pattern evoked potentials in infants: An index of maturation of spatial vision. Vision Research, 19, 747-755.

Sprague, J. M., & Meikle, T. H. (1965). The role of the superior colliculus in visually guided behavior. Experimental Neurology, 11, 115-146.

Spungen, L. B., Kurtzberg, D., & Vaughan, H. G. (1985). Patterns of looking behavior in full-term and low birth weight infants at 40 weeks post-conceptional age. Develop-mental and Behavioral Pediatrics, 6, 287-294.

28

Chapter 1

29

Introduction

Stechler, G., & Latz, E. (1966). Some observations on attention and arousal in the human infant. Journal of the American Academy of Child Psychology, 5, 517-525.

Stelmach, L. B., Campsall, J. M., & Herdman, C. M. (1997). Attentional and ocular move-ments. Journal of Experimental Psychology: Human Perception and Performance, 23, 823-844.

Stifter, C. A., & Moyer, D. (1991). The regulation of positive affect: Gaze aversion activity during mother-infant interaction. Infant Behavior and Development, 14, 111-123.

Stone, L. J., Smith, H. T., & Murphy, L. B. (Eds.) (1973). The competent infant: Research and commentary. New York: Basic Books.

Tauber, E. S., & Koffler, S. (1966). Optomotor responses in human infants to apparent motion: Evidence of innateness. Science, 152, 382-383.

Taylor, H. G., Hack, M., & Klein, N. K. (1998). Attention deficits in children with <750 gm birth weight. Child Neuropsychology, 4, 21-34.

Teller, D. Y., & Bornstein, M. (1987). Infant color vision and color perception. In P. Sala-patek & L. B. Cohen (Eds.), Handbook of infant perception (Vol. 1, pp. 185-232). Orlando, FL: Academic Press.

Trevarthen, C. B. (1968). Two mechanisms of vision in primates. Psychologische Forschung, 31, 299-337.

Tronick, E. (1972). Stimulus control and the growth of the infant’s effective visual field. Perception and Psychophysics, 11, 373-376.

Turkewitz, G., & Kenny, P. A. (1985). The role of developmental limitations of sensory/perceptual organization. Journal of Developmental and Behavioral Pediatrics, 6, 302-306.

Ungerleider, L. G., & Mishkin, M. (1982). Two cortical visual systems. In D. J. Ingle, M. A. Goodale, & R. J. W. Mansfield (Eds.), Analysis of visual behavior (pp. 549-586). Cam-bridge, MA: MIT Press.

Van Essen, D. C. (1985). Functional organization of primate visual cortex. In A. Peters & E. G. Jones (Eds.), Cerebral cortex (Vol. 3, pp. 259-329). New York: Plenum.

Van Essen, D. C., & Maunsell, J. H. R. (1983). Hierarchical organization and functional streams in the visual cortex. Trends in Neurosciences, 6, 370-375.

von Hofsten, C., & Rosander, K. (1996). Development of smooth pursuit tracking in young infants. Vision Research, 37, 1799-1810.

Wilcox, B. M., & Clayton, F. L. (1968). Infant visual fixation on motion pictures of the human face. Journal of Experimental Child Psychology, 6, 22-32.

Wolke, D. (1987). Environmental and developmental neonatology. Journal of Reproductive and Infant Psychology, 5, 17-42.

Wolke, D., & Meyer, R. (1999). Cognitive status, language attainment, and prereading skills of 6-year-old very preterm children and their peers: The Bavarian Longitudinal Study. Developmental Medicine and Child Neurology, 41, 94-109.

Wright, R. D., & Ward, L. M. (1998). The control of visual attention. In R. D. Wright (Ed.), Visual attention (pp. 132-186). New York: Oxford University Press.

Chapter 2

Developmental Changes in Visual Scanning of Dynamic Faces and Abstract Stimuli in Infants

AbstractThe characteristics of scanning patterns between the ages of 6 and 26 weeks were in-vestigated through repeated assessments of 10 infants. Eye movements were recorded using a corneal reflection system while the infants looked at two dynamic stimuli: the naturally moving face of their mother and an abstract stimulus. Results indicated that the way infants scanned these stimuli stabilized only after 18 weeks, which is slightly later than the ages reported in the literature on infants’ scanning of static stimuli. This effect was especially prominent for the abstract stimulus. From the 14-week session on, infants adapted their scanning behavior to the stimulus characteristics. When scanning the video of their mother’s face, infants directed their gaze at the mouth and eye region most often. Even at the youngest age, there was no indication of an edge effect.

This chapter is published as: Hunnius, S., & Geuze, R. H. (2004). Developmental changes in visual scan-ning of dynamic faces and abstract stimuli in infants: A longitudinal study. Infancy, 6, 231-255.

32

Chapter 2

33

Developmental Changes in Visual Scanning

INTRODUCTIONWhen infants are born, their motor skills are very limited, and the fact that they

have very little control over their limbs restricts the way in which they are able to explore the world around them. The oculomotor system – unlike other motor sys-tems – approximates its mature state several months after birth. The infant exercises this system every day from birth on. This makes vision one of the most important channels through which babies learn about the world surrounding them. However, during the first months of life, eye movements and visual acuity are also subject to certain constraints. For example, during the first month, eye movements are often inaccurate and unreliable (Atkinson, 1992), and infants of 1 to 3 months of age who are fixating a salient stimulus often have difficulty shifting their gaze away toward another interesting stimulus (Stechler & Latz, 1966; Butcher, Kalverboer, & Geuze, 2000; see Chapter 3).

Development of ScanningScanning is the pattern of eye movements to fixate different parts of an object of

examination. As the visual system matures, the characteristics of infants’ scanning patterns change. At birth, infants already show complex active scanning behaviors, but they often fail to direct their gaze toward a stimulus present in their field of vi-sion (Haith, 1980; Bronson, 1990a). Around the age of 1 month, babies tend to look especially at edges or outer contours of a stimulus pattern, but they mostly ignore the inner parts of a stimulus (Salapatek, 1975; Milewski, 1976). They also have the tendency to fixate single locations of stimuli for long periods, a behavior that has been referred to as “staring” (Bronson, 1996). Infants of 2 months of age scan more locations and different features of stimuli (Bronson, 1982, 1990a, 1996; Salapatek & Kessen, 1966; Leahy, 1976).

The changes in infants’ scanning behavior can be described as a gradual transi-tion to adult-like scanning (Bronson, 1994). As infants grow older, they engage more frequently in an advanced scanning mode, showing more brief fixations and more extensive scanning, whereas the infant-like way of scanning – characterized by pe-riods with extremely long fixations directed to single salient features – becomes less prominent. Around 3 months, infants can make accurate, efficient eye movements. Salient features in the environment still attract their gaze, but they have gained vo-litional, strategic control over their scanning behavior (Bronson, 1994).

It is known from studies with adults that characteristics of the stimuli influence the way they are scanned. Changes in the way of scanning have been shown to be re-lated to physical components of the stimuli such as luminance or texture (Mackworth & Morandi, 1967; Antes, 1974; Buswell, 1935) but also to semantic aspects (Loftus & Mackworth, 1978) or familiarity (Althoff & Cohen, 1999) and experience (Chapman & Underwood, 1998). There are indications that infants adapt their scanning patterns to stimulus characteristics increasingly more during the first few months of life (Johnson

32

Chapter 2

33

Developmental Changes in Visual Scanning

& Johnson, 2000). Further, it seems that stimulus characteristics can elicit more or less mature scanning behavior in young infants. For example, if the internal elements of a stimulus are moving, they are more likely to attract the gaze of even 1-month-old infants (Bushnell, 1979), whereas in infants of 13 weeks of age a flickering stimulus evokes a scanning pattern similar to that observed with a static stimulus at a much younger age (Bronson, 1990a).

Face ScanningHuman faces are among the most important stimuli in the visual environment

of a baby. Infants are confronted with faces very frequently from birth on and show attraction for faces or facelike stimuli already at an early age (Fantz, 1961; Morton & Johnson, 1991). Moreover, faces are of large psychosocial significance for a baby because mostly they appear in sight and draw the infant’s attention in a situation of interaction and communication.

Infants’ reactions to faces have been studied frequently, and there has also been considerable interest in how exactly infants examine faces. Infants younger than 2 months of age show limited scanning mostly of the perimeter of faces, whereas in-fants older than 2 months become more likely to fixate the internal elements of faces. When looking at the internal features of a face, they pay most attention to the eyes (Haith, Bergman, & Moore, 1977; Maurer & Salapatek, 1976). The borders of the face have been shown to be attractive regions of the face even for infants of 2 or 3 months of age (Haith et al., 1977; Hainline, 1978), whereas the mouth is looked at quite seldom (Haith et al., 1977). Previous studies thus suggest a developmental shift from scan-ning directed at the edges of a stimulus to scanning of the internal elements similar to that described earlier for geometric stimuli. At least for schematic faces, there are indications that the effect of edge attraction might be less prominent if the internal features of the face are moving (Girton, 1979).

Young infants are usually exposed to faces in social situations. Adults’ (especially parents’) interactions with babies have been shown to be highly similar across indi-viduals even from different cultures and to be characterized by elements like attracting attention and eye contact, displaying greeting responses, and making exaggerated facial movements and signals of pleasure (Papoušek & Papoušek, 1987). It is therefore remarkable that only very few studies have made use of moving or talking faces as stimuli (e.g., Haith et al., 1977). Most studies focusing on young infants’ scanning pat-terns of faces have used photographs or drawings of faces (e.g., Hainline, 1978), and when real faces have been used, they often were still faces (e.g., Maurer & Salapatek, 1976; Bronson, 1982).

The use of stimuli that are of limited ecological validity has been addressed and criticized frequently (Neisser, 1976; Schmuckler, 2001), as the generalizability of the obtained results is questionable. This study examines how infants scan naturally moving faces like those they are confronted with in daily life.

34

Chapter 2

35

Developmental Changes in Visual Scanning

Measuring Eye Movements in InfantsThe technique of infrared corneal photography was first applied to human infants

in the 1960s (Salapatek & Kessen, 1966; Haith, 1969) and always had to cope with a number of problems inherent to using a complex and highly sensitive technique with delicate and unpredictable subjects. The technical progress and the experience of the last 40 years have enabled several improvements of the original method, which enhance the quality and accuracy of the measurements. Examples are the increase of the sampling rate from 2 - 6 Hz to 30 - 60 Hz (Aslin, 1981; Hainline, 1981; Bronson, 1982) to improve the accuracy in determining fixation durations and the implemen-tation of a calibration (Bronson, 1983; Harris, Hainline, & Abramov, 1981) to enhance the spatial accuracy of the measurement. Incorporating a head-tracker into the eye-tracking system now has reduced the need to restrain the infant’s head, which was quite unnatural and often distressing for the infant.

This study combines the recent improvements in infant eye-tracking techniques with an intense longitudinal design. It also demonstrates the limitations of measuring eye movements in young infants and reports ways of handling the emerging problems of optimizing the data acquisition and dealing with missing data in a longitudinal design.

Aims of the StudySeveral studies have compared infants’ attending to moving and static stimuli, and

the attention-enhancing effect of stimulus movement has been described frequently (e.g., Wilcox & Clayton, 1968; Tronick, 1972; Carpenter, 1974). There are also indications that infants might scan moving stimuli in a different way than static displays (Bronson, 1990a; Johnson & Johnson, 2000). However, whereas the development of scanning of static stimuli has been studied widely, it is still unknown how infants scan non-static stimuli and how scanning of dynamic stimuli develops throughout infancy. Thus, our first goal was to describe infants’ scanning of dynamic stimuli and its development during the first few months of infancy. We chose to investigate infants’ scanning of two different dynamic stimuli: the infant’s mother’s face as it moved in a natural way and an abstract stimulus. The scanning of these two stimuli was examined between 6 and 26 weeks of age because this age period covers an interval in which the infant’s oculomotor and visual system is changing rapidly (see e.g., Atkinson, 1984). Infants were examined with intervals of 4 weeks between test sessions to provide enough measurements to precisely describe the development on the one hand but prevent possible training and habituation effects on the other.

As sketched above, stimulus characteristics might influence the way stimuli are scanned already in infancy. Thus, our second goal was to examine whether the scan-ning patterns elicited by a moving face differed from the scanning patterns elicited by an abstract dynamic stimulus and from which age on infants tailored their scanning behavior to the characteristics of the stimuli. To obtain an abstract stimulus that was

34

Chapter 2

35

Developmental Changes in Visual Scanning

comparable to the face video with regard to movement, colors, and luminance, but which had no facelike characteristics, the video of the mother’s face was scrambled. As stimulus characteristics might not only influence the way infants scan them but also how these scanning patterns evolve, we also investigated whether there were differences in how the scanning patterns of the mother and the abstract stimulus developed during the first few months of life.

The third goal of the study was to determine which regions of the face were fre-quently looked at and how this developed as infants were growing older. The question was whether the patterns of scanning of faces found in earlier studies that used less ecologically relevant representations of faces or even static faces are also found for naturally moving faces. We expected that the movement of highly salient internal features would reduce the clear effect of edge preference in very young infants and the frequency of edge fixations in the older infants. We also expected that it would lead to a more equal distribution of fixations over the regions of the face around the different internal features in contrast to results of earlier studies that have emphasized the infants’ preference to fixate the eye region.

METHOD

Participants Ten infants (5 girls; 5 boys) participated in the longitudinal study. The mothers of

the infants were contacted through childbirth education classes, midwives, or gym classes. Parents were told about the course and goals of the study and gave their written informed consent. The research was approved by the local Medical Ethics Committee.

All infants had been born after a gestation period between 37 and 42 weeks, had a birth weight above 2800 g, and no history of pre- or perinatal complications. All infants scored within their age range on the Bayley Scales of Infant Development (BSID-II; Bayley, 1993) at 12 and 24 weeks of age. Measurement sessions were conducted at 6, 10, 14, 18, 22, and 26 weeks, calculated from the due date. Mean ages were 46.7 days (SD = 3.8), 73.0 days (SD = 2.4), 102.2 days (SD = 3.4), 130.2 days (SD = 2.3), 158.0 days (SD = 3.4), and 186.7 days (SD = 4.4).

If a measurement session was unsuccessful because the infant was not in the re-quired state of alert wakefulness or other problems occurred, a new appointment was made. Despite attempts to retest, data for an additional 10 infants were not included because fewer than 5 of their 6 test sessions could be analyzed due to fussiness, cry-ing, or technical problems. Nine of the 10 infants included in the analysis completed all 6 test sessions; one infant completed 5 sessions.

Procedure Appointments were made at a time of the day when the mothers expected their

baby to be awake and able to stay alert for about half an hour. After arriving at the

36

Chapter 2

37

Developmental Changes in Visual Scanning

laboratory, the infant was given some time to get used to the new environment. When the infant was in state 3 or 4 of Prechtl’s scale of alertness (awake, eyes open, some spontaneous movements, no crying; Prechtl & Beintema, 1964), the experiment was started.

ApparatusTo carry out the assessments at the different ages under the same circumstances,

a setup suitable for infants from 1 to 6 months of age was developed. The infant was seated in an infant chair in front of a 21 inch monitor at a distance of 35 cm. The seat was tilted backwards (about 45 degrees) to provide enough support for the younger infants and to keep the older ones from leaning forward. The infant’s head was slightly stabilized, especially when the infant was young, but head-movements were not severely restricted to avoid distressing the babies and to allow them a natural way of moving while looking at the stimuli.

Only the screen of the monitor was visible. The frame and the other equipment necessary to run the experiment and to record eye movements were concealed behind a gray curtain, which filled approximately 180 degrees of the infant’s vi-sual field. During the task, one experimenter stood behind the baby to support the infant’s head when necessary, while the second experimenter controlled the presentation of the stimuli and the measuring equipment. The baby’s face and the display visible to the baby were shown on a video monitor, which made it possible for the experimenter to watch the infant’s behavior while running the task.

StimuliTwo dynamic stimuli were presented to the baby: a short video of the baby’s

mother’s face, in which the mother was looking, smiling, and nodding to the baby as she would in a normal interaction, and an abstract moving figure. Both stimuli were in color. The video recording of the mother’s face was made during a preliminary visit of mother and baby to the lab, shortly before the sessions in which the data were collected began. The mothers were asked to start with an attention getting movement (like nodding or greeting) and then to continue in a way that felt natural to them. As infants are sensitive to other persons’ gaze direction during interactions (Hains & Muir, 1996), mothers were asked not to avert their gaze during the record-ing. The videos were digitalized for use in a computerized experimental design.

The abstract stimulus was derived from the digital video of the mother by carry-ing out a number of transformations in a graphic computer program (Corel PHOTO-PAINT 9), such that it no longer resembled a face. During the transformation the image of the mother was rotated, scrambled, and distorted. This frame-by-frame procedure ensured that the two stimuli used were comparable in terms of total dynamics, color, and luminance. One frame from each type of video is presented as a stimulus example in Figure 2.1.

36

Chapter 2

37

Developmental Changes in Visual Scanning

Both stimuli had the same size and appeared against a gray screen. At the dis-tance of 35 cm, the stimuli were 30 by 40 degrees in size. The luminance of both figures was approximately 50 lux in the center of the stimulus. Each stimulus was presented to the baby for 30 s. The order in which they were shown was pseudo-randomized.

Measurement of Eye MovementsThe eye movements of the infants were measured using a corneal reflection

eye-tracking system (ASL, model 504). The infrared remote camera was positioned on a pan/tilt base underneath the monitor at a distance of about 50 cm from the eyes of the infant. Eye position data were sampled at 50 Hz.

As the infants could move their head in a natural, relatively free way, their head position was tracked by a sensor, which was attached to a soft fabric hat and coupled to a magnetic head-tracker (Polhemus Fastrak). The head position data were used

Figure 2.1. Example of (A) the mother stimulus and (B) the abstract stimulus.

38

Chapter 2

39

Developmental Changes in Visual Scanning

constantly during the experiment to direct the remote camera, which recorded an image of the eye. The camera followed slow horizontal and vertical head-move-ments automatically and was also able to recover an eye-image quickly after a rapid movement or even a turn away from and back to the monitor. A video recording of the location of the current fixation superimposed on the video display allowed the experimenter to see where the infant was looking at all times.

As eye geometries have been shown to differ considerably between infants (Bronson, 1990b), every infant’s visual field was calibrated prior to each measure-ment session. The calibration stimulus was a flashing black-and-white concentric square that appeared on the monitor first in the top left and then in the bottom right corner of the area in which the abstract and the face stimulus were presented. The calibration stimulus was accompanied by a short quacking noise. The experi-menters judged whether the baby was fixating the stimulus before carrying out the calibration.

Each calibration was tested to evaluate the quality of measurement during the experimental session. A spiral of about 12 degrees in diameter was presented to the infant on the monitor. The spiral moved across the gray screen and stopped at five different locations (the four corners and the center of the subsequent stimulus) where it shrank to a size of about 3 degrees with a gaudy dot in the center. As the baby followed the moving stimulus and fixated the points at which it was still, the experimenter could watch the infant’s focus of gaze on the monitors and judge the quality of measurement. If necessary, the calibration procedure was repeated.

AnalysisEye position data. To determine the number and length of fixations during stimulus

exposure, saccades, which naturally mark the beginning and end of a fixation, had to be identified from the eye position data. The displacement was calculated from the horizontal and vertical eye position data using Pythagoras’s theorem and smoothed with a 5-point window (100 ms). The onset of a saccade was defined as (a) pupil di-ameter above 0 as an indication of a valid signal and (b) change in fixation position larger than a threshold value. The threshold value adopted for saccade onset in this study was a mean displacement of .6 degrees per sample point over 3 sample points (60 ms). This is equivalent to an eye movement velocity of 30 degrees/s.

Experience in using eye-tracking methods with infants has taught us that the data yielded by an eye-tracker are often incomplete or contain artifacts. Examples of frequently occurring problems are missing data due to pupil loss after a posture change or a rapid head-movement. Artifacts can be caused by fussing, crying, or screwing up the eyes.

Behavioral coding. To correct errors and complete the data set, the video record-ings of each infant’s face were coded off-line. They were played back half-frame by half-frame (20 ms intervals) and compared with the available information from

38

Chapter 2

39

Developmental Changes in Visual Scanning

the eye-tracker data files. Looking with narrowed eyes was included if a part of the pupil was visible. The coding was carried out by two observers. The interobserver reliability (across subjects and across infants’ ages) for the classification of an eye movement indicated by the eye-tracker as a real eye movement versus an artifact was 94.7%. For detecting an eye movement in the absence of an eye-tracker signal, it was 96.9%. The reliability between observers for the onset and the length of an eye movement was 92.5% and 94.6%, respectively. Finally, for judging whether the infant was looking at the stimulus display or not, the agreement was 92.0%. With these corrections and completions on the basis of the videotaped eye movements it was possible to obtain complete data on whether the infant was looking to or away from the stimulus display and on the relative displacement of the eyes. The latter were used to determine the number and duration of fixations.

Coding of location of fixations. As the test of calibration carried out before every stimulus presentation revealed, the quality of eye-tracker measurement differed considerably across sessions. When the calibration was successful and allowed a reliable measurement of absolute eye position, data on the actual location of the fixations on the respective stimulus could be collected.

The location of the fixations was scored off-line by a single observer, using the video recordings of the stimulus display with the sequence of the infant’s fixations superimposed on it. Every stimulus was divided into a number of zones, and each fixation was – if possible – assigned to one of these areas. For the face stimulus, the categories were “eyes”, “mouth”, “edge” (including the hairline and the ears as well as the lower edge of the face), “background”, and “body” (neck, hair lying on the shoulders). The zones were determined separately for each stimulus using grid lines (for an example, see Figure 2.2). As the stimulus was moving, the areas could

Figure 2.2. Scheme of zones for one frame of a mother stimulus. For this frame, the regions of the face consist of the following zones: mouth - 4D, 4E; eyes - 3D, 3E; edge - 1D, 1E, 2C, 2D, 2E, 2F, 3C, 3F, 4C, 4F; body - 5B, 5C, 5D, 5E, 5F, 6A, 6B, 6C, 6D, 6E, 6F, 6G; background - 1A, 1B, 1C, 1F, 1G, 1H, 2A, 2B, 2G, 2H, 3A, 3B, 3G, 3H, 4A, 4B, 4G, 4H, 5A, 5G, 5H, 6H.

40

Chapter 2

41

Developmental Changes in Visual Scanning

change with time. Fixations were categorized as “missing” when it was impossible to determine where on the stimulus the baby was looking, and as “off” when gaze was averted from the stimulus.

The data on the location of fixations of the abstract stimulus were not analyzed. To make the movement, color range, and luminance of the two stimuli as similar as possible, each infant had been presented with an abstract stimulus derived from the video of his or her own mother. The location and movement dynamics of the core features of the resulting stimuli were too different from each other to incorporate them in a useful way into a common coding scheme.

Statistical AnalysisThree types of data were analyzed: data on the infants’ attention for the stimuli

(i.e. percentage of stimulus fixation, number of times turning away), data on the nature of the scanning movements (i.e. duration of fixations, number of fixations), and data on the location of the fixations while looking at the mother stimulus. The data on infants’ attention for the stimuli and the nature of the scanning movements were analyzed using 6 (age) x 2 (stimulus) repeated measures analyses of variance (ANOVAs). Data were missing for one session for one infant. Group means were used for this session. In several cases the correlation matrices between the different measurement points were heterogeneous. In these cases, the degrees of freedom of the F test were corrected according to the Greenhouse-Geisser method.

The data on the locations of the fixations were incomplete for the reasons men-tioned previously. The percentage of time for which information on fixation loca-tions was available is shown in Table 2.1. From this data set, percentages of time spent fixating the different regions of the face were calculated for each session. The change in the duration of looking at different zones of the stimulus across ages was examined using a linear regression of the averages. To test the significance of dif-ferences between the duration of fixating the different face regions, a permutation method, a procedure linked to the group of bootstrap techniques, was implemented (for an introduction see e.g., Good, 1999; Efron & Tibshirani, 1993). Permutation methods make it possible to estimate the probability that the observed values of crucial test statistics (e.g., the difference between averages, the slope of a regression line, etc.) can be explained by the null hypothesis model without having to make assumptions about expected distributions across the “population” of observations (e.g., a normal distribution). Further, they are especially suitable for analyzing data from a longitudinal design as they allow for missing values. Resampling methods are becoming more and more common in the social sciences because of their diversity and flexibility (see e.g., van Geert & van Dijk, 2002; Boosman, van der Meulen, van Geert, & Jackson, 2002).

40

Chapter 2

41

Developmental Changes in Visual Scanning

Table 2.1. Percentage of time for which information on the fixation location was available (presentation of the mother stimulus) by infant and age.

Infant 6 weeks 10 weeks 14 weeks 18 weeks 22 weeks 26 weeks

1 0.00 69.0 0.00 80.7 92.8 84.82 0.00 42.8 52.1 0.00 50.0 95.33 0.00 57.6 0.00 83.0 84.2 100.04 39.7 87.9 57.1 100.0 31.7 0.005 81.8 65.6 100.0 2.2 4.8 79.26 0.00 49.8 73.4 100.0 39.8 86.97 0.00 96.1 100.0 100.0 100.0 93.78 12.4 0.00 0.00 49.6 18.2 94.49 95.5 100.0 10.1 98.9 51.3 81.4

10 0.00 95.6 89.3 100.0 79.9 97.2

Note. If there was less than 3 seconds of reliable measurement, the session was excluded from the analysis of location of fixation. Sessions excluded for this reason are marked.

RESULTSThere are two levels at which one can examine how infants look at a visual display

(Bronson, 1994): At a higher level, it can be assessed how long the stimulus is looked at before attention is shifted away. The lower level is the dimension of single fixations. Here, the number and duration of each fixation is examined. In this study, infants’ attention for the two stimuli was examined first. Therefore, both the total time spent looking at the two stimuli and the number of gaze shifts away from the stimuli were analyzed. These measures provide information on infants’ attentiveness and interest for the two stimuli throughout the research period.

Percentage of Time of Stimulus FixationFirst, it was investigated how long the infants attended to and away from the stim-

uli. Therefore, the total duration of fixations on and off the stimulus (and subsequently the relative amount of time spent fixating on and off the stimuli) were calculated.

Effect of age. The percentage of time infants spent looking at the displayed stimuli changed significantly with age (F(1.16, 10.45) = 4.89, p < .05). The course of develop-ment of the average percentage of time spent fixating the two stimuli is shown in Figure 2.3. Pairwise comparisons with Bonferroni-correction revealed significant

42

Chapter 2

43

Developmental Changes in Visual Scanning

differences between the first three and the last three measurement points, espe-cially between the age of 10 (M = 91.6%, SD = 5.1) and 14 weeks (M = 88.9%, SD = 6.6) versus 22 (M = 72.4%, SD = 9.1) and 26 weeks (M = 71.6%, SD = 11.5). Because of the clear partition of the developmental course, the two parts (6 until 14 weeks and 18 until 26 weeks) were also analyzed separately. For the first three measurement occasions, the ANOVA demonstrated a significant effect of age (F(1.16, 10.45) = 4.74, p < .05) with a significant increase in looking time between 6 (M = 72.7%, SD = 24.0) and 10 weeks of age (M = 91.6%, SD = 5.1).

Figure 2.3. The averaged percentage of time spent fixating the mother and the abstract stimulus between 6 and 26 weeks of age (with standard errors).

Effect of stimulus. There was no main effect of the type of stimulus (abstract versus mother stimulus) and no significant interaction. If the two parts of the developmental course were analyzed separately, a significant Age x Stimulus interaction (F(1.31, 11.81) = 5.22, p < .05) was revealed, indicating that the increase in fixating time was greater for the abstract stimulus (6 weeks, M = 64.8%, SD = 30.7; 10 weeks, M = 93.6%, SD = 4.8) than for the mother’s face (6 weeks, M = 80.6%, SD = 20.1; 10 weeks, M = 89.7%, SD = 9.2). The second part of the developmental trajectory was characterized by stability: No significant effects of age or stimulus were found.

Turning Away from the Stimulus DisplayEffect of age. The analysis of how often the infants turned away from the stimuli at

the different ages (see Figure 2.4) revealed a significant effect of age (F(2.93, 26.39) =

42

Chapter 2

43

Developmental Changes in Visual Scanning

15.67, p < .01). Infants averted gaze from the stimulus more often as they grew older. Significant differences were found between the frequency of turning away at 6 (M = 2.0, SD = 1.3) and 10 weeks (M = 1.6, SD = .9) versus 18 (M = 5.3, SD = 2.4), 22 (M = 7.0, SD = 1.8) and 26 weeks of age (M = 6.6, SD = 2.9).

Effect of stimulus. The analysis of how often infants averted their gaze yielded no significant difference between the two stimuli.

Figure 2.4. The averaged number of times of turning away from the mother and the abstract stimulus between 6 and 26 weeks of age (with standard errors).

Number of Fixations To study the infants’ way of scanning the stimuli at the microlevel, the number and

duration of fixations were examined. In addition, the distributions of fixation lengths were analyzed to assess the frequency of brief or prolonged fixations, as Bronson (1994) characterized them as typical for differently mature modes of scanning.

Effect of age. The development of the number of fixations while fixating on or off the stimuli between 6 and 26 weeks of age is shown in Figure 2.5. Although the overall time spent looking at the stimulus declined as the infants grew older, the number of fixations increased with age (F(5, 45) = 11.14, p < .01). Pairwise comparisons with Bonferroni-correction demonstrated a significant increase in number of fixations between the age of 6 weeks (M = 14.3, SD = 4.9) and 14 weeks (M = 25.0, SD = 6.1) or older. The infants also fixated off the stimulus more often as they grew older (F(5, 45) = 10.11, p < .01).

44

Chapter 2

45

Developmental Changes in Visual Scanning

Effect of stimulus. The analysis revealed a significant effect of stimulus on the number of fixations (F(1, 9) = 7.83, p < .05). The mother’s face elicited significantly more fixa-tions (M = 26.4, SD = 4.3) than the abstract stimulus (M = 23.0, SD = 3.9). T tests used as post-hoc tests revealed that the number of fixations on the stimulus was significantly larger if the stimulus was the mother’s face than if it was abstract at the age of 14 weeks (mother, M = 29.2, SD = 7.9; abstract, M = 20.7, SD = 8.1; t(9) = 2.57, p < .05) and 18 weeks (mother, M = 32.0, SD = 7.9; abstract, M = 27.2, SD = 8.0; t(9) = 1.98, p < .10). With respect to the number of fixations off the stimulus, the two types of stimuli did not differ significantly.

Figure 2.5. The mean numbers of fixations while fixating the mother and the abstract stimulus and while fixating away between 6 and 26 weeks of age (with standard errors).

Duration of FixationsEffect of age. For each infant the median length of fixation was calculated per stimu-

lus and session. The change in median fixation duration while fixating on or off the two stimuli over the measurement period is shown in Figure 2.6. The ANOVA revealed a significant decrease in median fixation duration while looking to the stimuli across the measurement period (F(2.5, 22.52) = 4.44, p < .05), from more than 1 s at 6 weeks (6 weeks, M = 1.18 s, SD = .62) to around .6 s at 18 weeks or older (18 weeks, M = .68 s, SD = .23; 22 weeks, M = .64 s, SD = .10; 26 weeks, M = .60 s, SD = .07). Concerning the fixa-tions off the stimulus, there was also an almost significant decrease in duration as the infants grew older (F(1.74, 15.65) = 3.43, p = .06).

44

Chapter 2

45

Developmental Changes in Visual Scanning

Effect of stimulus. From the age of 14 weeks on, there was a significant difference in median fixation length between the abstract stimulus and the mother stimulus (F(1, 9) = 15.27, p < .01), with the abstract stimulus (M = .78 s, SD = .21) eliciting longer fixation durations than the mother (M = .61 s, SD = .12).

A repeated measures ANOVA (6 x 2; Age x On/off) comparing the median duration of fixations on and off the stimulus revealed a significant effect (F(1, 9) = 37.61, p < .01) due to longer fixations when regarding the stimuli versus looking away from the moni-tor. The fixation duration while fixating off the stimulus did not vary significantly as a function of the type of stimulus.

Figure 2.6. The averaged median fixation durations while looking at the mother and the abstract stimulus between 6 and 26 weeks of age (with standard errors).

Distributions of fixation durations. To gain a clearer picture of the change in length of fixations over the measurement period, the distributions of the fixation durations at the different ages were examined. As representative examples, the distributions at the age of 6, 14, and 26 weeks and the different stimuli are shown in Figure 2.7. The fixation lengths are distributed approximately around .5 s, with most of the fixations lasting between .2 and .9 s. However, this might not apply to the age of 26 weeks: Here there seem to be two distinct distributions for the mother and the abstract stimulus with two different modal values. As the distributions found in this study were broadly similar to those reported in Bronson (1990a), his criterion of labeling those fixations as “brief” that last .7 s or less was applied. The percentage of brief fixations increased as infants grew older, especially in the second half of the measurement period. At 6 and 14

46

Chapter 2

47

Developmental Changes in Visual Scanning

Figure 2.7. The distributions of fixation durations while looking at the mother and the abstract stimulus at (A) 6 weeks, (B) 14 weeks, and (C) 26 weeks of age.

46

Chapter 2

47

Developmental Changes in Visual Scanning

weeks of age, on average 42.8% (SD = 13.4) and 46.6% (SD = 15.3) of the infants’ fixations were brief, compared to 64.4% (SD = 6.1) at 26 weeks. A paired-samples t test revealed a significant increase in the proportion of brief fixations between the third and the last measurement session (t(9) = 4.09, p < .05). Whereas at 6 weeks the percentages of brief fixations for the two different stimuli did not differ (mother, M = 38.8%, SD = 13.2; abstract, M = 46.7%, SD = 23.7), at 14 weeks and 26 weeks the abstract stimulus elicited an almost significantly smaller percentage of brief fixations than the mother’s face (14 weeks, t(9) = 2.44, p < .05; 26 weeks, t(9) = 2.08, p = .067). During the abstract display, on average only 40.9% (14 weeks, SD = 16.6) and 59,0% (26 weeks, SD = 11.4) were brief, versus 52.4% (14 weeks, SD = 17.4) and 69.8% (26 weeks, SD = 8.9) while the infants were regarding their mothers.

To explore the incidence of extremely long fixations, the percentage of fixation lengths greater than 2.5 s were calculated for each infant per session and stimulus. Although very long fixations were frequently observed at 6 weeks (M = 19.0%, SD = 10.7), their occurrence decreased between 14 and 26 weeks of age (14 weeks, M = 11.4%, SD = 7.7; 26 weeks, M = 2.4%, SD = 3.5; t(9) = 5.22, p < .01). Again, differences between the two stimuli were found. While the percentages of long fixations were about the same at 6 weeks (mother, M = 19.0%, SD = 9.4; abstract, M = 19.0%, SD = 17.5) and 26 weeks (mother, M = 2.8%, SD = 3.4; abstract, M = 2.0%, SD = 4.9), at 14 weeks the abstract stimulus elicited a higher percentage of very long fixations (M = 16.0%, SD = 13.7) than the video of the mother (M = 6.9%, SD = 4.3). A paired-samples t test demonstrated an almost significant difference (t(9) = 2.16, p = .059).

Location of FixationsTo examine which regions of their mother’s face infants looked at throughout the

measurement period, the durations of fixations assigned to the same face regions (see Figure 2.2) were added up, and proportions of time spent looking at the different regions of the face were calculated.

Table 2.2. Percentages of time spent fixating the different regions of the face stimulus by age.

Region 6 weeks 10 weeks 14 weeks 18 weeks 22 weeks 26 weeks

Mouth 39.98 48.17 44.88 50.38 51.54 57.19Eyes 38.11 30.10 48.10 38.88 26.95 29.74Edge 18.52 17.82 5.87 8.93 18.36 11.87Body 0.00 2.32 0.22 0.00 0.40 0.00Back-ground

3.38 1.60 0.91 1.81 2.74 1.19

48

Chapter 2

49

Developmental Changes in Visual Scanning

1 In this study, a Monte Carlo approach to the permutation method was applied. It uses the actual data to calculate the level of significance of parameters. Assume we want to test, whether the parameters M and E, each based on a set of observations in a group of participants, are significantly different from each other. Our hypothesis then states that the two parameters are different. That is, we can conceive of the set of observed values, which forms the basis of the first parameter M, that it is drawn from a dif-ferent distribution (with different mean) than the observations that led to the calculation of the second parameter E. The null hypothesis, on the other hand, claims that the observed difference is coincidental and that the two sets of observations have been drawn from a single distribution. This means that the assignment of an observed value to either the first or the second set of observations would be purely random. We simulate this condition by randomly assigning a label of the first or second set to the values collected in the common set. If we carry out a large number of such random permutations (e.g., 10000), we obtain an approximation of the exact chance distribution of the difference between M and E. We can now determine the number of times that the simulated null hypothesis difference between M and E is as large as or larger than the observed difference between M and E. If the number of simulations is big enough, the proportion of this number over the number of simulations provides an estimation of the p value of the observed difference, that is, the probability that the observed difference has been the result of the null hypothesis model.

The distribution of the percentages and its change over time is shown in Table 2.2. To test for a possible effect of age, a regression model was used. A linear effect of age was found only for the mouth region (B = .11, t(8) = 6.99, p < .01). The percentage of time spent looking at the mouth region increased significantly from 39.98% at 6 weeks to 57.19% at 26 weeks of age.

To test whether two percentages of fixation duration were significantly different from each other, a permutation technique was implemented.1 From Table 2.2 it can be seen that the percentage of time spent fixating the edges of the face was smaller than the percentage of time spent fixating the mouth and the eye region throughout all measurement sessions. At 6 weeks of age, none of the differences was significant. This might be due to the very small amount of data available at the first measurement point (see Table 2.1). The difference between the time spent fixating the mouth and the edges reached statistical significance from the age of 10 weeks on (10 weeks, p = .02; 26 weeks, p = .01). The difference between the time spent fixating the eye region and the borders of the face was significant at 14, 16, and 26 weeks (p = .02, p = .004, p = .01, respectively). At 26 weeks, there was also an almost significant difference between the time spent fixating the mouth and the time spent fixating the eyes (p = .08) with more time being spent on the mouth. Percentages of time spent fixating other parts of the stimulus display (such as the mother’s body or the background) were negligible.

DISCUSSION

Development of Attention for the Face and the Abstract StimulusInitially, some results on infants’ attention for the two stimuli will be discussed

to provide the background for the subsequent considerations on scanning patterns. The infants in this study showed considerable interest for the stimuli they were

48

Chapter 2

49

Developmental Changes in Visual Scanning

presented with. During the first measurement session, infants spent on average about three quarters of the time looking at the stimuli. The time spent regarding the dynamic stimuli followed a course of development well-known from earlier studies (see Colombo, 2002): The infants’ attention increased between 6 and 10 weeks of age, reaching its maximum when infants were 10 to 14 weeks old, before decreasing again between 18 and 22 weeks and subsequently stabilizing at the 6-week level. However, although infants spent about the same time looking away from the stimulus when they were 6 weeks old as when they were 5 months and older, there was an important difference in looking behavior at the two ages. At the younger ages, fixations were longer and infants looked away from (and back to) the stimulus less often, which is consistent with the poor gaze shifting control often observed in infants of this age (Aslin & Salapatek, 1975; Hood & Atkinson, 1993). The finding that at 6 weeks infants nevertheless spent more time attending to their mother’s face suggests that within these general limitations they might be able to preferentially direct their attention to a face (see Morton & Johnson, 1991).

From the age of 18 weeks on, infants exhibited a different pattern of regarding. They tended to alternate short fixations of the stimuli with short looks away. This pattern has been observed frequently from 3 months of age on, both during social interaction and during inspection of inanimate stimuli. Babies start shifting their gaze between their partner and the surroundings at this age (Kaye & Fogel, 1980; van Wulfften Palthe, 1986). Such gaze shifting might have a regulatory function both in the social context (Stifter & Moyer, 1991) and during inspections of inanimate stimuli when infants regulate the flow of visual input to enable efficient information process-ing (Colombo, 1993).

Development of Scanning of the Face and the Abstract StimulusThere are some remarkable changes taking place in the way infants scan the two

dynamic stimuli in our study. The number of fixations increased significantly dur-ing the measurement period, while the fixation length decreased. This is in accord with previous research (Bronson, 1994, 1996). The number of fixations increased most rapidly between 6 and 18 weeks of age; the median fixation duration declined most strongly between 10 and 18 weeks of age. Only after 18 weeks did neither parameter change significantly. A stable scanning pattern thus emerged a little later than in the studies of Bronson (1990a, 1994), who concluded that infants’ scanning of static stimuli was similar to that of adults around 14 weeks of age. From 14 weeks on, a stable difference in median fixation length when looking at the abstract or the face stimulus was found.

When the results on number and duration of fixations are considered together with the changes over age in the distributions of fixation durations for each stimulus, three important points come forward. First, the median fixation duration did not become stable until 18 weeks of age, and at 14 weeks scanning behavior was still characterized

50

Chapter 2

51

Developmental Changes in Visual Scanning

by more staring behavior and fewer brief fixations than at the end of the measure-ment period. We attribute this effect to the dynamic nature of both stimuli and their increased sensory salience prolonging the occurrence of subcortically controlled scanning patterns, which are usually found in younger infants. This is also supported by the finding that infants exhibited much shorter average fixation durations when fixating off the stimulus display. These results are consistent with earlier findings of Johnson and Johnson (2000), who also reported a protracted developmental course for infants’ scanning of complex, moving stimuli. Taken together, these findings suggest that the development of scanning of more demanding stimuli is extended compared to relatively simple, achromatic stimuli.

Second, the persistence of less advanced scanning behavior at 14 weeks was espe-cially marked when the infants were examining the abstract stimulus. This is suggested by the higher proportion of extremely long fixations for the abstract compared to the mother stimulus found at 14 weeks but not at 6 and 26 weeks of age. To explain this effect, one has to take into account that there might have been a discrepancy in sa-lience also between the abstract and the face stimulus. Although they were comparable in terms of dynamics, color range, and luminance, the abstract stimulus contained more plain areas of the same color and thus more contrast and less clear structure. Its movements and its change of structure were less predictable. As a consequence, the less advanced scanning behavior might have been more prominent when infants were inspecting the abstract stimulus than their mother’s face. This suggests that before mature scanning patterns have become well established, scanning behavior may be particularly susceptible to the characteristics of a stimulus.

Third, a stable difference between the median fixation durations of the two stimuli was found from 14 weeks on, which is also reflected in the difference in the percent-age of brief fixations during the scanning of the two stimuli at both 14 and 26 weeks of age. This effect indicates the existence of two distinct, rather narrowly confined distributions of short fixation durations as they are depicted in Figure 2.7 B and C. Infants of 14 weeks and older have considerable experience in scanning faces. More-over, their mother’s face is very well-known to them by this time (Barrera & Maurer, 1981). The abstract display, on the other hand, might still have been novel. It was highly salient, moved in an unpredictable way, and did not have the clearly defined structure of meaningful elements found in a face. As it is the function of short fixa-tions to provide the visual system regularly with new input, a scanning pattern with slightly longer fixation durations may have been well adapted and highly functional to examine the abstract stimulus. Thus, one important conclusion from our findings is that infants adjusted their way of scanning to the characteristics of the different stimuli from the age of 14 weeks on.

Compared to some earlier research on infant scanning patterns (e.g., Harris & Hainline, 1987), the median fixation durations found in this study seem to be rather long. On the one hand, one can attribute this finding partly to differences between

50

Chapter 2

51

Developmental Changes in Visual Scanning

the characteristics of the stimuli used in the different studies (e.g., complexity, dy-namic aspects, or salience). On the other hand, these differences can also be due to the methods of processing the raw eye position data as Bronson (1990a) already pointed out. One has to keep in mind that it is of limited value to compare parameters from different studies, as their range depends crucially on how they are derived from the actual eye position data and how a fixation and an eye movement are defined. In this study, the criterion used permitted very small drifting movements or extremely slow eye shifts within a fixation interval because of the dynamic properties of the stimuli. Under these circumstances, it is especially important to concentrate on differences of parameters across different stimuli and ages within the same study.

Development of Location of Fixations on the Mother’s FaceWhen infants were scanning their mother’s naturally moving face, they looked

mainly at her mouth and eyes. The borders of the face did attract some attention, especially at the ages of 6 and 10 weeks, but – at least from 10 weeks on – the inner elements of the face were significantly more interesting to the baby. Even at the young-est age, no indication of an edge preference was found. As mentioned before, previous studies demonstrated strong effects of edge attraction in young infants who were regarding static or moving but less natural representations of faces. This study raises doubts about whether this also holds for natural interaction situations. It is possible that the intense, exaggerated, and slowly paced facial movements characteristic of mother-infant interactions (Stern, 1974) were a more powerful attractor of the infant’s attention than the contours outlining the face. However, a final conclusion concern-ing this question is not possible because the number of successful measurements with the 6-week-olds was limited. Edge preference has been shown to influence the infant’s visual behavior most strongly around 1 month of age, making it likely that any possible effect might have been only mildly present in the 6-week-olds tested for this study. Further research thus is indicated here.

From the first test session on, the mother’s mouth drew much attention. The time infants spent fixating the mouth region increased significantly throughout the mea-surement period, which might indicate their growing interest in language. This finding contrasts with the results of earlier studies, which reported the mouth to be a less (or even the least) regarded area (Maurer & Salapatek, 1976), even when the stimulus was a talking face (Haith et al., 1977). One explanation is that the mothers were perform-ing the facial activities of a natural interaction (such as smiling or slowly articulating relatively simple words), which were particularly attractive and suitable for babies. However, the stimulus videos in this study were displayed silently. The effect thus might also reflect the infants’ growing astonishment to see a moving mouth without hearing the matching sound.

Concerning the reaction to faces, our results indicate a clear advance between 10 and 14 weeks of age. The distribution of durations suggested the overcoming of sticky

52

Chapter 2

53

Developmental Changes in Visual Scanning

fixation while scanning the face stimulus, and concurrently, the proportion of time spent fixating the edges of the face reached its lowest point. Thus, the infants started scanning the face routinely and paid more attention to the particularly meaningful regions of a face: the eyes and the mouth. This transition seems to go together with changes in mother-infant interaction. After about 12 weeks of age, the infant takes an increasingly active part in en face interaction (van Wulfften Palthe, 1986).

REFERENCESAlthoff, R. R., & Cohen, N. J. (1999). Eye-movement-based memory effect: A reprocess-

ing effect in face perception. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 997-1010.

Antes, J. R. (1974). The time course of picture viewing. Journal of Experimental Psychol-ogy, 103, 62-70.

Aslin, R. N. (1981). Development of smooth pursuit in human infants. In D. F. Fisher, R. A. Monty, & J. W. Senders (Eds.), Eye movements: Cognition and visual perception (pp. 31-41). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Aslin, R. N., & Salapatek, P. (1975). Saccadic localization of visual targets by the very young human infant. Perception and Psychophysics, 17, 293-302.

Atkinson, J. (1984). Human visual development over the first 6 months of life. A review and a hypothesis. Human Neurobiology, 3, 61-74.

Atkinson, J. (1992). Early visual development: Differential functioning of parvocellular and magnocellular pathways. Eye, 6, 129-135.

Barrera, M. E., & Maurer, D. (1981). Recognition of mother’s photographed face by the three-month-old infant. Child Development, 52, 714-716.

Bayley, N. (1993). Bayley Scales of Infant Development (Second Edition). San Antonio, TX: The Psychological Corporation.

Boosman, K., van der Meulen, M., van Geert, P., & Jackson, S. (2002). Measuring young children’s perceptions of support, control, and maintenance in their own social networks. Social Development, 11, 386-408.

Bronson, G. W. (1982). The scanning patterns of human infants: Implications for visual learning. Monographs on Infancy No. 2. Norwood, NJ: Ablex.

Bronson, G. W. (1983). Potential sources of error when applying a corneal reflex eye-monitoring technique to infant subjects. Behavior Research Methods and Instrumenta-tion, 15, 22-28.

Bronson, G. W. (1990a). Changes in infants’ visual scanning across the 2- to 14-week age period. Journal of Experimental Child Psychology, 49, 101-125.

Bronson, G. W. (1990b). The accurate calibration of infants’ scanning records. Journal of Experimental Child Psychology, 49, 79-100.

Bronson, G. W. (1994). Infants’ transitions towards adult-like scanning. Child Develop-ment, 65, 1243-1261.

52

Chapter 2

53

Developmental Changes in Visual Scanning

Bronson, G. W. (1996). The growth of visual capacity: Evidence from infant scanning patterns. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 11, pp. 109-141). Norwood, NJ: Ablex.

Bushnell, I. W. R. (1979). Modification of the externality effect in young infants. Journal of Experimental Child Psychology, 28, 211-229.

Buswell, G. T. (1935). How people look at pictures. Chicago: University of Chicago Press.Butcher, P. R., Kalverboer, A. F., & Geuze, R. H. (2000). Infants’ shifts of gaze from a

central to a peripheral stimulus: A longitudinal study of development between 6 and 26 weeks. Infant Behavior and Development, 23, 3-21.

Carpenter, G. C. (1974). Visual regard of moving and stationary faces in early infancy. Merrill-Palmer Quarterly, 20, 181-194.

Chapman, P. R., & Underwood, G. (1998). Visual search of dynamic scenes: Event types and the role of experience in viewing driving situations. In G. Underwood (Ed.), Eye guidance in reading and scene perception (pp. 369-393). Oxford, UK: Elsevier.

Colombo, J. (1993). Infant cognition: Predicting later intellectual functioning. Newbury Park, CA: Sage.

Colombo, J. (2002). Infant attention grows up: The emergence of a developmental cognitive neuroscience perspective. Current Directions in Psychological Science, 11, 196-199.

Efron, B., & Tibshirani, R. J. (1993). An introduction to the bootstrap. Monographs on Statistics and Applied Probability No. 57. New York: Chapman & Hall.

Fantz, R. L. (1961). The origin of form perception. Scientific American, 204, 66-72.Girton, M. R. (1979). Infants’ attention to intrastimulus motion. Journal of Experimental

Child Psychology, 28, 416-423.Good, P. I. (1999). Resampling methods: A practical guide to data analysis. Boston, MA:

Birkhäuser.Hainline, L. (1978). Developmental changes in visual scanning of face and nonface

patterns by infants. Journal of Experimental Child Psychology, 25, 90-115.Hainline, L. (1981). An automated eye movement recording system for use with human

infants. Behavior Research Methods and Instrumentation, 13, 20-24.Hains, S. M. J., & Muir, D. W. (1996). Infant sensitivity to adult eye direction. Child De-

velopment, 67, 1940-1951.Haith, M. M. (1969). Infrared television recording and measurement of ocular behavior

in the human infant. American Psychologist, 24, 279-282.Haith, M. M. (1980). Rules that babies look by. Hillsdale, NJ: Lawrence Erlbaum Associ-

ates, Inc.Haith, M. M., Bergman, T., & Moore, M. J. (1977). Eye contact and face scanning in early

infancy. Science, 198, 853-855. Harris, C. M., & Hainline, L. (1987). Characteristics of fixations in human infants: Du-

rations. In J. K. O’Regan & A. Lévy-Schoen (Eds.), Eye movements: From physiology to cognition (pp. 378-379). Amsterdam: North-Holland.

54

Chapter 2

55

Developmental Changes in Visual Scanning

Harris, C. M., Hainline, L., & Abramov, I. (1981). A method for calibrating an eye-monitoring system for use with human infants. Behavior Research Methods and Instrumentation, 13, 11-20.

Hood, B. M., & Atkinson, J. (1993). Disengaging visual attention in the infant and adult. Infant Behavior and Development, 16, 405-422.

Johnson, S. P., & Johnson, K. L. (2000). Early perception-action coupling: Eye move-ments and the development of object perception. Infant Behavior and Development, 23, 461-483.

Kaye, K., & Fogel, A. (1980). The temporal structure of face-to-face communication between mothers and infants. Developmental Psychology, 16, 454-464.

Leahy, R. L. (1976). Development of preferences and processes of visual scanning in the human infant during the first 3 months of life. Developmental Psychology, 12, 250-254.

Loftus, G., & Mackworth, N. H. (1978). Cognitive determinants of fixation location during picture viewing. Journal of Experimental Psychology: Human Perception and Performance, 4, 565-572.

Mackworth, N. H., & Morandi, A. J. (1967). The gaze selects informative details within a picture. Perception and Psychophysics, 2, 547-552.

Maurer, D., & Salapatek, P. (1976). Developmental changes in the scanning of faces by young infants. Child Development, 47, 523-527.

Milewski, A. (1976). Infant’s discrimination of internal and external pattern elements. Journal of Experimental Child Psychology, 22, 229-246.

Morton, J., & Johnson, M. H. (1991). CONSPEC and CONLEARN: A two-process theory of infant face recognition. Psychological Review, 98, 164-181.

Neisser, U. (1976). Cognition and reality: Principles and implications of cognitive psychology. San Francisco: Freeman.

Papoušek, H., & Papoušek, M. (1987). Intuitive parenting: A dialectic counterpart to the infant’s integrative competence. In J. D. Osofsky (Ed.), Handbook of infant development (pp. 669-720). New York: Wiley.

Prechtl, H. F. R., & Beintema, D. (1964). The neurological examination of the full term infant. Little Club Clinics in Developmental Medicine No. 12. London: Heineman.

Salapatek, P. (1975). Pattern perception in early infancy. In L. B. Cohen & P. Salapatek (Eds.), Infant perception: From sensation to cognition (Vol. 1, pp. 133-248). New York: Academic Press.

Salapatek, P., & Kessen, W. (1966). Visual scanning of triangles by the human newborn. Journal of Experimental Child Psychology, 3, 155-167.

Schmuckler, M. A. (2001). What is ecological validity? A dimensional analysis. Infancy, 2, 419-436.

Stechler, G., & Latz, E. (1966). Some observations on attention and arousal in the human infant. Journal of the American Academy of Child Psychiatry, 5, 517-525.

54

Chapter 2

55

Developmental Changes in Visual Scanning

Stern, D. N. (1974). Mother and infant at play: The dyadic interaction involving facial, vocal, and gaze behaviors. In M. Lewis & L. D. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 187-213). New York: Wiley.

Stifter, C. A., & Moyer, D. (1991). The regulation of positive affect: Gaze aversion activity during mother-infant interaction. Infant Behavior and Development, 14, 111-123.

Tronick, E. (1972). Stimulus control and the growth of the infant’s effective visual field. Perception and Psychophysics, 11, 373-376.

van Geert, P., & van Dijk, M. (2002). Focus on variability: New tools to study intra-individual variability in developmental data. Infant Behavior and Development, 25, 340-374.

van Wulfften Palthe, T. (1986). Neural maturation and early social behavior: A longitudinal study of mother-infant interaction. Doctoral dissertation. University of Groningen, Groningen, The Netherlands.

Wilcox, B. M., & Clayton, F. L. (1968). Infant visual fixation on motion pictures of the human face. Journal of Experimental Child Psychology, 6, 22-32.

Chapter 3

Gaze Shifting in Infancy: A Longitudinal Study Using Dynamic Faces and Abstract Stimuli

AbstractDisengaging from and shifting gaze to a salient stimulus is a prerequisite for early exploration and communication. The efficiency of disengagement increases during the first few months after birth. Little is known about the effect of stimulus charac-teristics on disengagement during the different stages of its development. Twenty infants were studied longitudinally between 6 and 26 weeks of age. The frequency and latency of gaze shifts to peripheral targets were measured in a competition and a non-competition situation. The stimuli were a short video of the baby’s mother’s face and an abstract video, both appearing as central stimulus or peripheral target. In the competition condition, infants were more likely to shift their gaze when the central stimulus was a face and the peripheral target was abstract. They moved their gaze least frequently and with greater latency in the opposite condition (abstract-face). Disengagement developed rapidly between 6 and 22 weeks of age. Differences between stimulus combinations were most marked between 10 and 18 weeks.

This chapter is published as: Hunnius, S., & Geuze, R. H. (2004). Gaze shifting in infancy: A longitudinal study using dynamic faces and abstract stimuli. Infant Behavior and Development, 27, 397-416.

58

Chapter 3

59

Infants’ Shifts of Gaze to a Peripheral Target

INTRODUCTIONOptimal development depends largely on learning about objects and people in the

surrounding world. This is especially true for infants. From birth on, they explore their environment visually. Effective visual exploration requires shifting gaze across differ-ent locations. Adults are very skilled in quick visual search and scanning. Infants, on the other hand, may show long periods of staring. This study addresses the question of how shifts of attention and gaze develop through infancy and whether this devel-opmental trajectory is dependent on characteristics of the stimuli involved.

The Development of Attention ShiftingThe attentional skills which support the visual exploration of the environment

develop during the first months of life. Babies between approximately 1 and 3 months of age have been reported to spend long periods staring and to have difficulty disengag-ing from an object or stimulus they are attending to (Hopkins & van Wulfften Palthe, 1985). Stechler and Latz (1966) named this behavior “obligatory attention”, others no-ticed that infants seemed “stimulus-bound” or “overwhelmed” (Tennes, Emde, Kisley, & Metcalf, 1972, p. 218) or “glued to the pattern” they were examining (Cohen, 1976, p. 235). This effect of a stimulus in central vision impeding the shifting of attention to a second, competing stimulus in the periphery has been demonstrated frequently (e.g., Harris & MacFarlane, 1974; Aslin & Salapatek, 1975; Mohn & van Hof-van Duin, 1986). The phenomenon of obligatory attention seems to be strongest in infants of 1 and 2 months. The frequency and speed of shifts of gaze to a target in the periphery with a stimulus persisting in the center increase substantially around 3 to 4 months of age (Hood & Atkinson, 1990; Johnson, Posner, & Rothbart, 1991; Butcher, Kalverboer, & Geuze, 2000). Saccadic reaction times in situations with competing stimuli tend to be longer than in non-competition situations where no disengagement of attention or gaze has to take place prior to the execution of an eye movement (Atkinson, Hood, Wattam-Bell, & Braddick, 1992). This effect has been shown to be largest in infants younger than 3 months (Matsuzawa & Shimojo, 1997), but has also been demonstrated in older infants and in adults (Hood & Atkinson, 1993).

Unraveling Attention Shifts in InfancyThe phenomenon of obligatory attention is not yet completely understood. Richards

(1997) suggested that it is not caused by a decrease in peripheral sensitivity, but rather by an increased bias against responding when attention is engaged on the focal visual stimulus. Different explanations for this disengagement difficulty have been proposed. Hood (1995) coined the term “sticky fixation” for the phenomenon emphasizing infants’ difficulty in breaking gaze from a stimulus they are fixating. According to Johnson (1990), obligatory attention can be attributed to the inability to produce an orienting eye movement to a peripheral target while processing the stimulus in the central visual field. Rothbart, Posner, and Rosicky (1994), however, suggested that shifts of gaze are

58

Chapter 3

59

Infants’ Shifts of Gaze to a Peripheral Target

preceded by covert shifts of attention and that disengagement problems reflect dif-ficulty orienting covertly to the periphery. Within all three accounts, disengagement problems can be described in terms of two processes competing with each other: a tendency to maintain the focus of attention and gaze, and a tendency to shift atten-tion and gaze to a new location. Accordingly, disengagement difficulty is explained by the relative maturational state of the collicular and cortical areas which subserve these mechanisms (Johnson, 1990; Hood, 1995; Rothbart et al., 1994).

Earlier research has shown that the characteristics of the stimuli used in an atten-tion shifting task influence the reaction of the baby. Non-competition experiments, in which a focal stimulus was replaced by a peripheral target, have found that the size (Cohen, 1972) and the form (Maurer & Lewis, 1979) of the peripheral stimulus affect the probability and latency of localization, while speed of stimulus movement does not (Finlay & Ivinskis, 1982, 1984).

Stimulus attributes also seem to play a role when two stimuli are presented at the same time. Finlay and Ivinskis (1984) demonstrated that a comparatively salient stimulus in the central field makes it more difficult for infants to disengage their gaze. There are also indications that the effect of a salient central stimulus might change with age, being strongest between 9 and 16 weeks (Butcher et al., 2000). Tronick (1972) found that the frequency of detection of a target further in the periphery increased between 2 and 10 weeks of age when the central stimulus was static and the target was moving, but not when the central stimulus was moving and the target was static.

Shifts of gaze seem to be a function not only of what is currently being attended to in the central visual field, but also of what is competing for attention in the periph-ery. Finlay and Ivinskis (1984) demonstrated that infants of 4 months detected and processed information about stimuli in the periphery even when their gaze did not shift from a central stimulus they had been fixating first. In another disengagement experiment, Matsuzawa and Shimojo (1997) found much shorter gaze shifting latencies than Hood and Atkinson (1993) had found before in a similar cross-sectional study with 1.5- to 6-month-olds and attributed this to the fact that they had made use of a less attractive central fixation stimulus and a more attractive target. However, Hicks and Richards (1998) did not find an effect of a dynamic versus a static peripheral stimulus on frequency and latency of gaze shifts between 2 and 6 months of age.

To summarize, the current results concerning the role of stimulus characteristics in shifts of attention are both incomplete and inconsistent. It has not yet been precisely investigated how variation in the characteristics of the central and the peripheral stimulus influences shifts of attention and how this changes throughout infancy.

Disengagement and Social-Emotional Development Vision is crucial not only for exploring and learning about the environment but

also during social interaction. During face-to-face interaction with their mother or another caretaker, infants from the age of 3 months on often shift their gaze away from

60

Chapter 3

61

Infants’ Shifts of Gaze to a Peripheral Target

and back to the other person’s face. These periods of looking away from the mother’s face occur both in situations with negative emotionality (Cohn & Tronick, 1983) and during pleasurable interactions (Stifter & Moyer, 1991). As infants grow older, the at-tention they pay to their mother’s face during social interaction declines, and they start fixating other locations more often (Kaye & Fogel, 1980; van Wulfften Palthe, 1986). Johnson and co-workers interpret this shift as a sign of basic changes in the child’s ability to disengage attention and gaze from a location and to anticipate events that are going to occur at other spatial locations (Johnson et al., 1991).

The distress that has been described as accompanying obligatory attention (Stechler & Latz, 1966; Tennes et al., 1972) suggests that attention and emotion are closely linked. Furthermore, it is a daily observation that a mother can relieve her infant’s unhap-piness by letting him or her orient to something new (e.g., a toy). In accordance with this, it could be shown that babies of 4 months of age who have higher abilities in disengaging from a visual stimulus are more soothable and less susceptible to distress (Johnson et al., 1991). The attention system thus forms a basis for self-regulation al-ready in early infancy (Posner & Rothbart, 1981).

Taking into account the functional importance of attentional behavior and gaze shifting discussed earlier, it is surprising that there have been relatively few attempts to study the development of gaze shifting and the underlying attentional processes in functional contexts, for example, by using different and more socially relevant stimuli.

Aims of the StudyThe goal of this study was to explore how the nature of the stimuli used affects gaze

and attention shifting behavior in infancy. Both abstract and meaningful stimuli were used. Human faces are among the most important stimuli in the visual world of infants. They also play an important role in their social-emotional development. We therefore chose the infant’s mother’s face as it moved in a natural way as social stimulus and a stimulus with similar physical characteristics as abstract stimulus. Since there is some – albeit inconsistent – evidence that the impact of the stimulus characteristics changes with age, the study focused on infants between 6 and 26 weeks of age, the period in which fundamental gaze and attention shifting mechanisms develop.

The aim was to study shifts of gaze both in a non-competition situation when the central and the peripheral stimulus were presented successively and in a competition situation when the central stimulus persisted after the peripheral stimulus appeared. In order to examine the influence of stimulus characteristics on both attention hold-ing and attention getting, the two stimuli were used as central fixation stimulus and peripheral target. The effects of the different stimulus combinations were measured in terms of differences in the frequency and latency of peripheral target localization. We expected no distinct effect on gaze shifting in the non-competition condition, as the two stimuli should be equally easy to detect due to their similar physical charac-

60

Chapter 3

61

Infants’ Shifts of Gaze to a Peripheral Target

teristics. In the competition condition, we expected the characteristics of the stimuli to influence attention holding and attention attraction processes, leading to differ-ences in the frequency and latency of gaze shifts. Around 5 or 6 months of age, we expected reliable disengagement to occur in all stimulus conditions. A possible effect of stimulus characteristics should then still be detectable in differences in the latencies of gaze shifts to the peripheral target. As earlier research has shown that the timing of attentional development may differ substantially between infants (Butcher et al., 2000; Finlay & Ivinskis, 1984), it was also investigated whether there were significant inter-individual differences in the development of gaze and attention shifting, pos-sibly as a function of the influence of stimulus characteristics.

METHOD

ParticipantsTwenty infants (12 girls; 8 boys) took part in the longitudinal study. Their moth-

ers were approached through childbirth education classes, midwives, or gym classes. To be admitted to the study, infants had to meet a set of criteria: a gestation period of 37 to 42 weeks, a birth weight above 2800 g, and no history of pre- and perinatal complications. All infants scored within their age range on the Bayley Scales of Infant Development (BSID-II; Bayley, 1993) at 12 and 24 weeks of age. Parents were informed about the purpose of the study and gave their informed written consent. The research was approved by the local Medical Ethics Committee.

The measurements started when the infants were about 6 weeks old and continued every 4 weeks until they were 26 weeks old. Ages were calculated from the due date. If the infants were unable to carry out the experimental task due to fussing or sleepi-ness, a new appointment was made within 7 days. Despite attempts to retest, one of the measurement sessions was missing for 7 infants. Mean ages at each measurement session were 47.7 days (SD = 4.1), 73.6 days (SD = 4.7), 103.7 days (SD = 4.0), 131.1 days (SD = 3.2), 158.9 days (SD = 3.4), and 188.8 days (SD = 4.8).

ProcedureMeasurement sessions were scheduled for a time when parents expected their baby

to be alert for 20 to 30 minutes. After arriving at the lab, infants were given some time to become used to the new environment. When they were in state 3 or 4 of Prechtl’s scale of alertness (awake, eyes open, some spontaneous movements, no crying; Prechtl & Beintema, 1964), the experiment was started.

The infants carried out a so-called disengagement task. In such a task stimuli can appear at three different positions on a screen: in the center, on the left, or on the right. The stimuli to the left or right were presented at 20 degrees eccentricity. The disengagement task included two different sorts of trials: competition and non-com-petition trials. All trials started with the display of a stimulus in the center of the monitor, the fixation stimulus. To attract the attention of the infant, the onset of this

62

Chapter 3

63

Infants’ Shifts of Gaze to a Peripheral Target

stimulus was accompanied by a short melody. After the infant had been fixating the central stimulus for 1 - 2 seconds, the peripheral stimulus appeared. In competition trials, the first stimulus persisted after the peripheral target appeared; in non-com-petition trials, it disappeared when the peripheral stimulus appeared. After 5 seconds, the stimuli disappeared simultaneously. The screen remained blank for 2.5 seconds, before the following trial began.

Figure 3.1. Schematic representation of a disengagement task with (A) a competition and (B) a non-com-petition trial.

62

Chapter 3

63

Infants’ Shifts of Gaze to a Peripheral Target

Figure 3.1 shows a schematic representation of the disengagement task with (A) the competition condition and (B) the non-competition condition. Competition trials thus required disengagement of attention and gaze from the fixated central stimulus before an eye movement to the peripheral target could be carried out. Non-competition trials did not require disengagement.

ApparatusThe babies carried out the task while sitting in an infant-seat on a table in a re-

clined posture (about 45 degrees) with a light support of their head. The infant-seat was located in front of a 21 inch monitor, which was suspended from the ceiling ap-proximately central and perpendicular to the line of gaze. The distance between the monitor screen and the baby’s eyes was about 35 cm. Only the screen of the monitor was visible to the infant. The monitor itself, the equipment necessary to run the tasks and record eye movements, and the experimenter were hidden behind a gray curtain, which filled 180 degrees of the baby’s visual field. During the experiment, the infant’s face and eye movements were videotaped. The eye movements and the display of the monitor were also shown on a video monitor, allowing the experimenter to run the task on the basis of the infant’s behavior. The babies’ eye movements were scored off-line from the video recording.

Stimulus Material and Stimulus PresentationTwo different dynamic stimuli were used within the disengagement task: an ab-

stract stimulus and a meaningful stimulus. Both stimuli were in color. The meaning-ful stimulus consisted of a short video sequence of the face of the infant’s mother. This video recording was made during a first visit of mother and baby to the lab. The mother’s face was recorded while she was smiling and talking with her baby as she nor-mally would do. The abstract stimulus was then derived from the video of the mother by carrying out several transformations in a graphic computer program (Corel PHOTO-PAINT 9). During the transformation the image of the mother was rotated, scrambled, and distorted. This frame-by-frame procedure ensured that the two stimuli resembled each other regarding their dynamic characteristics, color range, and luminance, but were completely different with respect to their meaning to the infant. One frame from each type of video is given as a stimulus example in Figure 2.1 (see Chapter 2).

At the viewing distance of 35 cm, all stimuli subtended a visual angle of 10 by 10 degrees. Thus, the face was much smaller than it would appear to the baby in a normal interaction. As even newborns can recognize a well-known face after a size change (Walton, Armstrong, & Bower, 1997), we expected that the infants would not have any problems with the size of the face. This was supported by observations during the experiment, which showed the infants’ interest for the stimuli and their smiling to the mother stimulus but not to the abstract stimulus.

The experiment contained 32 competition trials and 8 non-competition trials. As

64

Chapter 3

65

Infants’ Shifts of Gaze to a Peripheral Target

reactions to the peripheral stimulus in the non-competition condition were expected to be less variable, fewer non-competition trials were considered to be necessary to provide reliable measurements. Both stimulus types could appear as central stimulus or as peripheral target. The different combinations of the two stimulus sorts (face-face, face-abstract, abstract-face, abstract-abstract) were presented with equal frequency. Half of the peripheral targets appeared on the left, half on the right side. The order in which the trials were presented was randomized. An experimental run normally lasted about 12 minutes. Whenever the infant started fussing or crying or became sleepy, the testing session was interrupted for some time.

AnalysisBehavioral coding. The eye movements on the video recordings of the infant’s face

were coded off-line from the tape, which was played back half-frame by half-frame (20 ms intervals). The direction and latency of the first eye movement after the pe-ripheral stimulus appeared were scored. Trials in which the infant was not fixating the first stimulus when the peripheral stimulus appeared were excluded from the analysis. This could be trials in which, for example, the infants had already averted their gaze from the central stimulus or had their eyes closed for various reasons (e.g., blinking, yawning, fussing). Eye movements starting less than 200 ms after the second stimulus appeared, were considered anticipatory (Haith, Hazan, & Goodman, 1988) and also dropped from the analyses.

All eye movements following the appearance of the peripheral stimulus but not leading to its localization were considered to be errors. Different sorts of errors were distinguished: errors in which the infant looked up or down instead of to the periph-ery, carried out a horizontal eye movement in the opposite direction of the target, or looked in the direction of the target but went beyond it or did not reach it. If the peripheral target was localized by means of multiple successive saccades, this was scored using a separate code, but counted as a successful shift of gaze, as long as the gaze did not miss the target.

The data were coded by different observers, who had been trained by the first author. About 10% of the sessions were double-coded. Interobserver reliability for the onset of an eye movement was found to be on average 93.5% (range 89.5% to 100%). Cohen’s kappa for the category of first eye movement was on average .82 (range .72 to 1.0).

Statistical analysis. The relative frequency of looks to the peripheral target as well as the frequency and type of error were calculated for each test session and condition. To calculate the latencies of looks, the time differences between the appearance of the peripheral stimulus and the onset of an eye movement to this target were calculated. Then, the median reaction times for each infant, session, and stimulus combination were determined. As a plot of all median reaction times revealed, the distribution of the raw data was positively skewed, and therefore a square root transformation was

64

Chapter 3

65

Infants’ Shifts of Gaze to a Peripheral Target

carried out (Rummel, 1970) before carrying out the statistical analyses. For ease of understanding, the averaged response latencies reported in the text and the reaction times depicted in the figures are given in seconds.

The data were analyzed using a multilevel modeling technique (Snijders & Bosker, 1999; Woodhouse, 1996). Multilevel analysis is a regression procedure which takes into account a possible hierarchical structure of the data set. When applied to longitudinal data, the repeated measures are regarded as “nested” within individuals. Unlike a standard multiple regression model, a multilevel model contains more than one error term: one for every level of the hierarchical data. The model also allows intercept and slope coefficients to vary randomly, which means that the association between session scores and explanatory variables may differ between individuals. Another strength of this approach is that it allows for both the number of observations per individual and the spacing of the observations in time to vary.

Multilevel analysis was used to examine whether (a) the frequency and latency of looks and the frequency of errors changed with age, (b) the combination of stimuli influenced the frequency and latency of looks, and (c) the data was described best by a model which allowed for inter-infant differences. The data of the competition and the non-competition trials were analyzed separately and treated as different subsets of data. The models had three levels: infant, test session, and stimulus combination. Based on earlier research (e.g., Butcher et al., 2000), a model of three piecewise linear functions was fit to the data. Previous studies suggested rapid development for the period from 6 to 9 and from 9 to 16 weeks and a stabilization for the period between 16 and 26 weeks of age. As looks to the competing target were very infrequent at 6 weeks, reliable latency data were not available for the first measurement point. A model of two piecewise linear functions was therefore used for the reaction time data, with one function from 9 to 16 weeks and another from 16 to 26 weeks. Infants’ test ses-sions differed concerning the number of trials carried out. This was controlled for by including a correction term into the models. As the infants’ real ages were entered into the model rather than age categories, the variable age could be treated as a con-tinuous variable, which provided information on infants’ performance also between the measurement points.

For the analyses, the data was centered around 12 weeks of age, which was about the middle of the period in which the largest change was expected. The parameters for the different age periods and the different stimulus combinations were added to the equation in order to predict the frequency of looks or the reaction times. T tests were used to determine the statistical significance of the coefficients. The fit of the model and its improvement was examined using χ2 tests of deviance. Whenever multiple t tests were carried out (e.g., as post-hoc tests), Bonferroni corrections were implemented to keep alpha at .05 (Stevens, 1992).

66

Chapter 3

67

Infants’ Shifts of Gaze to a Peripheral Target

RESULTS

Frequency of Shifts of Gaze to the Peripheral StimulusNon-competition trials. The overall percentage of shifts of gaze to the periphery in-

creased from 66.0% (SD = 40.8) at 6 weeks to 90.7% (SD = 18.8) at 10 weeks (see Figure 3.2). After the age of 10 weeks, there were no significant changes. This was indicated by the variable age having a significant coefficient for the 6 to 9 week function (t(106)= 3.98, p < .001), but not for the 9 to 16 week function and the 16 to 26 week function.

Figure 3.2. The mean frequencies and standard errors of looks to the peripheral target in competition and non-competition trials, and the mean frequency and standard error of errors in competition trials between 6 and 26 weeks of age.

The frequency of looks to the peripheral stimulus in the non-competition condi-tion per stimulus combination is shown in Figure 3.3. No significant effect of stimulus combination was found in any age period. There were no interaction effects of time and stimulus combination. As can be seen from the standard error bars in Figure 3.3, inter-infant differences were rather large. There was also a significant effect for the estimate of the slope variance for the 6 to 9 week function (χ2(1) = 4.54, p < .05) which indicates that infants differed in the rate at which gaze shifts increased during this period. However, it has to be kept in mind that in every test session there were only 2 non-competition trials per stimulus combination. The possible percentage of looks thus could be 0%, 50%, or 100%, which led to the large standard errors and the seem-ingly large inter-infant variance.

66

Chapter 3

67

Infants’ Shifts of Gaze to a Peripheral Target

Figure 3.3. The mean frequency and standard error of looks in non-competition trials per stimulus combination between 6 and 26 weeks of age.

Competition trials. The overall frequencies of looks to the peripheral stimulus on competition trials across ages are depicted in Figure 3.2. Between 6 and 14 weeks of age, the frequencies were significantly lower than in the non-competition condition (6 weeks, t(12) = -7.66, p < .01; 10 weeks, t(19) = -12.41, p < .01; 14 weeks, t(19) = -6.42, p < .01). At 6 weeks, the frequency of looks was only 7.4% (SD = 10.1). It increased significantly throughout the measurement period to 29.9% (SD = 25.0) at 10 weeks, 73.8% (SD = 25.7) at 18 weeks, 88.0% (SD = 12.6) at 22 weeks, and 86.8% (SD = 16.8) at 26 weeks. The model accordingly contained almost significant coefficients of age for the 6 to 9 week function (t(107) = 1.73, p < .10), the 9 to 16 week function (t(107) =10.56, p < .001), and the 16 to 26 week function (t(107) = 4.17, p < .001). However, the coefficient of age for the 9 to 16 week function was significantly larger than the one of the 16 to 26 week function (slope coefficients β9-16 = 1.16 versus β16-26 = .27). This was indicated by the significant gain in fit which was obtained by using separate coefficients of age for the periods of 9 to 16 and 16 to 26 weeks compared to only one age coef-ficient for the period of 9 to 26 weeks (χ2(1) = 31.20, p < .001). Consistent with this, comparisons of mean group frequencies of looks in consecutive sessions of the last age period yielded a significant effect only for 18 and 22 weeks (t(14)= 3.58, p < .01), while between 22 and 26 weeks of age there was no significant increase demonstrable anymore (t(16) = -.86, p > .10).

The proportion of looks to the peripheral target consisting of two or more saccades was large when infants were young (e.g., 6 weeks, M = 27.0%, SD = 1.9; 14 weeks, M =

68

Chapter 3

69

Infants’ Shifts of Gaze to a Peripheral Target

29.6%, SD = 4.0) and decreased significantly between 14 and 26 weeks of age (26 weeks, M = 5.4%, SD = 1.6; t(19) = 3.39, p < .01).

As can be seen from Figure 3.4, the developmental trajectories for the different stimulus combinations diverge. The effect of the different stimulus combinations was tested by adding three of the four possible combinations as a 0-1 dummy variable to the model and contrasting them against the fourth category, the combination face-face. The main effects of the stimulus combinations face-abstract and abstract-face were significant. When controlling for age, the combination face-abstract elicited more frequent shifts of gaze than the reference category face-face (βface-abstract = 13.69; t(427)= 4.64, p < .001). With the combination abstract-face, on the other hand, gaze shifts occurred less frequently than in the reference condition (βabstract-face = -8.74; t(427) = -3.74, p < .001). Also with the stimulus combination abstract-abstract as ref-erence category, there were significant effects of the combinations face-abstract (t(427) = 4.61, p < .001) and abstract-face (t(427) = -3.77, p < .001). There was no significant difference between the stimulus combinations face-face and abstract-abstract.

Figure 3.4. The mean frequency and standard error of looks in competition trials per stimulus combina-tion between 6 and 26 weeks of age.

Three significant interaction effects showed that the frequency of shifts of gaze under the different stimulus conditions was associated with age (see Figure 3.4). In the age period between 6 and 9 weeks, the interactions between the stimulus combination abstract-face and age (t(427) = -2.57, p < .05) and face-abstract and age (t(427) = 1.83, p < .10) became almost significant. This indicates that, for this age period, the increase in

68

Chapter 3

69

Infants’ Shifts of Gaze to a Peripheral Target

disengagement frequency was smaller in the abstract-face condition and greater in the face-abstract condition. While the frequency of looks to the peripheral stimulus between 6 and 10 weeks increased from 9.8% (SD = 21.5) to 44.5% (SD = 25.1) in the face-abstract condition, with an abstract stimulus in the center and a face in the pe-riphery it changed from 9.2% (SD = 16.5) to only 18.3% (SD = 16.0) in the same period of time. The third significant interaction effect was an interaction between the stimulus combination face-abstract and age (t(427) = -3.17, p < .01) for the 16 to 26 week period. In the last age period the increase in the frequency of gaze shifting was smaller for the face-abstract stimulus combination than it was for the combination face-face. As Figure 3.4 also shows, unlike under the other three conditions, the frequency of gaze shifting in the face-abstract condition had almost reached its maximum at 18 weeks and did not increase substantially anymore (18 weeks, M = 85.1%, SD = 18.2; 22 weeks, M = 90.2%, SD = 19.7; 26 weeks, M = 88.8%, SD = 17.3).

In the analysis of random effects, the estimate of intercept variance was significant (χ2(1) = 8.12, p < .01). This indicates that the frequency of looks differed significantly across infants at 12 weeks of age, which has been set as the intercept of the model. The slope variance for the age function 16 to 26 weeks turned out to be nearly significant (χ2(1) = 3.83, p < .10).

Latency of Shifts of GazeNon-competition trials. In Figure 3.5, the development of the averaged median reac-

tion times when shifting gaze to the periphery is depicted for the different stimulus combinations. The median reaction time declined from .58 s (SD = .15) at 10 weeks to .41 s (SD = .09) at 14 and .39 s (SD = .07) at 22 weeks. In the multilevel model, the coef-ficient of the variable age was significant for the two age periods, 9 to 16 weeks (t(102) = - 7.16, p < .001) and 16 to 26 weeks (t(102) = 2.06, p < .05). The age coefficients of the period 9 to 16 weeks (β9-16 = -.0064) and 16 to 26 weeks (β16-26 = .0008; χ2(1) = 47.71, p < .001) were significantly different from each other. The fact that the 16 to 26 week age coefficient was positive and significantly smaller than the coefficient of the age period 9 to 16 weeks suggests that the decrease in gaze shifting latency came to a standstill as infants grew older. Comparisons of averaged median reaction times in consecutive sessions revealed significant decreases only between 10 and 14 weeks (t(18) = 5.56, p < .01), but not after 14 weeks of age, which also indicates a stabilization at the end of the measurement period.

There was no significant effect of stimulus combination. The slope variance es-timate for the age period 9 to 16 weeks reached significance (χ2(1) = 17.94, p < .001), indicating significant inter-infant differences in the rate of decrease of gaze shift-ing latency during this age period. Also, the estimate of the intercept variance was significant (χ2(1) = 18.14, p < .001), which suggests that infants differed in their gaze shifting latency at 12 weeks of age.

70

Chapter 3

71

Infants’ Shifts of Gaze to a Peripheral Target

Competition trials. Until 22 weeks of age, the mean latencies to look to the periph-eral target were significantly higher in the competition than in the non-competition condition (10 weeks, t(17) = 6.60, p < .01; 14 weeks, t(19) = 4.96, p < .01; 18 weeks, t(17) = 4.79, p < .01; 22 weeks, t(16) = 3.19, p < .05). The development of the averaged median latencies of looks for the different stimulus combinations of the competition trials are presented in Figure 3.6. The mean latency of looks in the competition trials decreased from 1.42 s (SD = .62) at 10 weeks to .87 s (SD = .53) at 14 and .49 s (SD = .18) at 26 weeks. Both declines were significant (9 to 16 week age period, t(90) = -7.08, p < .001; 16 to 26 week age period, t(90) = -2.83, p < .01). Coefficients of the two age periods differed significantly (β9-16 = -.0085, β16-26 = -.0017; χ2(1) = 22.59, p < .001), which suggests that the decline during the first part of the measurement period was larger than during the second part. T tests comparing the mean latencies of consecutive sessions accordingly yielded a significant decrease only between 10 and 14 weeks of age (t(17) = 3.50, p < .05) and a trend between 18 and 22 weeks (t(14) = 2.64, p < .10).

Figure 3.5. The mean latency of looks and its standard error in non-competition trials per stimulus combination between 6 and 26 weeks of age.

The analysis also revealed an effect of stimulus combination. The main effect of the combination abstract-face indicates that – when controlling for age – the latency to shift gaze was longer than under the reference combination face-face (t(346) = 2.74, p < .01). The model with stimulus combination abstract-abstract as reference category yielded a trend for the stimulus combination abstract-face (t(346) = 1.73, p < .10). Post-hoc t tests carried out at 22 and 26 weeks to determine whether there was a significant

70

Chapter 3

71

Infants’ Shifts of Gaze to a Peripheral Target

effect of the abstract-face condition on latency of gaze shifting especially at the end of the measurement period failed to demonstrate significant differences.

There were marked inter-infant differences indicated by a significant slope vari-ance for the 16 to 26 week age period (χ2(1) = 6.74, p < .01). The intercept variance also reached significance (χ2(1) = 18.79, p < .001).

Figure 3.6. The mean latency of looks and its standard error in competition trials per stimulus combina-tion between 6 and 26 weeks of age.

Error AnalysisThe mean frequency of errors during competition trials between 6 and 26 weeks of

age is depicted in Figure 3.2. The frequency of errors declined from 29.4% (SD = 33.4) at 6 weeks to 20.3% (SD = 20.8) at 14 and 11.9% (SD = 15.5) at 26 weeks. The decrease during the age periods from 6 to 9 weeks (t(108) = -2.13, p < .05) and from 16 to 26 weeks (t(108) = -2.55, p < .05) turned out to be significant. There was no significant association between the different stimulus combinations and the frequencies of errors. The frequency of errors differed significantly across infants at 12 weeks of age, indicated by a significant intercept variance (χ2(1) = 9.03, p < .01). Furthermore, the estimate of slope variance for the 16 to 26 week age period yielded significance (χ2(1) = 6.15, p <.05), indicating that infants differed significantly in the rate at which the frequency of errors decreased.

The relative frequency of errors which consisted of an eye movement up or down increased over the measurement period from 17.3% (SD = 24.9) at 6 weeks to 49.2% (SD = 38.7) at 26 weeks. This change, however, did not reach significance, probably due to the large standard deviations. Eye movements which did not reach the target

72

Chapter 3

73

Infants’ Shifts of Gaze to a Peripheral Target

were frequent in younger infants (6 weeks, M = 34.0%, SD = 35.9; 14 weeks, M = 29.6%, SD = 37.9), but decreased significantly between 14 and 26 weeks (26 weeks, M = 1.6%, SD = 5.0; t(19) = 3.94, p < .05). Looks to the opposite side of the target were relatively frequent, both when infants were young (6 weeks, M = 28.9%, SD = 38.2) and at the end of the measurement period (26 weeks, M = 34.3%, SD = 35.1) and displayed no signifi-cant changes. Eye movements which went beyond the target were rare in very young infants (6 weeks, M = 3.0%, SD = 7.7), but their frequency increased as infants grew older (14 weeks, M = 14.1%, SD = 16.9; t(12) = 4.00, p < .05).

DISCUSSIONIn this study, which made use of dynamic social and abstract stimuli, we found clear

developmental changes concerning the frequency and latency of gaze shifts between 6 and 26 weeks of age. In the non-competition condition, the infants made frequent shifts of gaze to the peripheral stimulus from 10 weeks of age on. The latency of the eye movements decreased between 6 and 16 weeks of age, leveling off thereafter. These developmental trajectories closely replicate the results of the longitudinal study by Butcher et al. (2000), which used only abstract stimuli. In the competition condition, the presence of the central stimulus reduced the probability of looks to the peripheral stimulus in younger infants, as observed in earlier studies (e.g., Aslin & Salapatek, 1975; Harris & MacFarlane, 1974). Around 6 weeks of age, infants made very few shifts of gaze; between 9 and 18 weeks of age, the frequency of shifts of gaze increased rapidly, stabiliz-ing around 80% at 22 weeks. A course of development with a high degree of discontinuity has also been described by Butcher et al. (2000). The findings concerning the latency of gaze shifts are also consistent with findings of earlier studies: The latency of gaze shifts decreased rapidly until 16 weeks of age and then more slowly between 16 and 26 weeks (Hood & Atkinson, 1993). Latencies were significantly longer in the competition than in the non-competition condition when infants were young, but not when they were older (Matsuzawa & Shimojo, 1997). As the studies mentioned earlier (and this one) made use of very different stimulus types (e.g., checkerboards, moving real faces, schematic faces, etc.), it can be concluded that the general pattern of development of disengagement and simple gaze shifts is largely independent of the stimuli used.

Localization of the peripheral stimulus by means of multiple saccades was frequent in younger, but not in older infants, consistent with the findings of Aslin and Salapatek (1975). The overall frequency of errors decreased throughout the measurement period, and, at the same time, the relative frequency of the different types of error changed. Eye movements which failed to reach the target were also frequent in younger in-fants and can be interpreted as unsuccessful attempts to shift gaze to the peripheral stimulus. Looks beyond the target or up or down were frequently noted errors in older infants. They were typical for the reduced interest for the experimental task, which was observed especially during the last measurement sessions when infants were able to carry out the task easily.

72

Chapter 3

73

Infants’ Shifts of Gaze to a Peripheral Target

Analyses of the inter-infant differences in the frequency and latency of gaze shift-ing revealed considerable variance between infants for both frequency and latency measures, as described by Butcher el al. (2000) before. Infants differed both in their overall frequency and latency of gaze shifting and in the rate at which these pa-rameters developed. The largest inter-individual differences were found during the periods of rapid change.

The primary goal of the study was to examine whether the use of different stimuli as central fixation stimulus and peripheral target influences the developmental trajec-tories of the frequency and latency of gaze shifts. Earlier studies have demonstrated that gross differences in the salience of the peripheral stimulus affect the probabil-ity and latency of localization in a non-competition situation (Cohen, 1972; Maurer & Lewis, 1979). In this study, however, we found no effect of type of stimulus on the probability or latency of shifting gaze to the peripheral stimulus in the non-compe-tition condition. This is consistent with our expectations: In the non-competition condition, the central stimulus disappears when the peripheral one appears. As a consequence, only the physical attractiveness and the detectability of the peripheral target should influence the probability of its localization. The absence of such an ef-fect in this study thus suggests that we were successful in creating stimulus pairs in which the two stimulus types were about equally attractive in terms of brightness, movement dynamics, etc.

In the competition condition on the other hand, the different stimulus com-binations significantly influenced the probability of gaze shifts. There are three important findings: First, there was a strong effect of stimulus combination. Infants were more likely to shift their gaze when the central stimulus was a face and the peripheral target was abstract, while they were least likely and also slower to shift gaze in the opposite condition (abstract-face). This effect is, however, modified by age. At 6 weeks of age, the infants rarely looked away from the central stimulus, regardless of what it was or what was presented in the periphery, but once attend-ing kept on “staring” at it. The rate of increase in gaze shifting frequency after 6 weeks was greatest for the stimulus combination face-abstract and smallest for the stimulus combination abstract-face compared to the reference categories face-face or abstract-abstract. The differences between the four stimulus combinations were most marked then between 10 and 18 weeks of age. At 18 weeks of age, the frequency of gaze shifting in the face-abstract condition reached about its final level. From 22 weeks on, infants reliably shifted their gaze from the central stimulus to the peripheral target under all stimulus combination conditions. Thus, the infants showed the greatest sensitivity to stimulus characteristics while disengagement was developing but not yet well established. This is consistent with the idea that developing systems are most susceptible to context variables (Thelen & Smith, 1998). We had expected that, after efficient and reliable disengagement had emerged, the influence of competing stimuli with different characteristics might persist in dif-

74

Chapter 3

75

Infants’ Shifts of Gaze to a Peripheral Target

ferences in gaze shifting latencies. However, after the period of rapid development, we found no effect of stimulus characteristics on either the frequency or the latency of shifts of gaze.

Second, the data clearly indicate that the central and the peripheral stimulus both influenced the probability of looking to the peripheral stimulus. There were significant differences between the stimulus combinations face-abstract versus face-face and abstract-face versus abstract-abstract, which demonstrates the influence of what was presented to the baby peripherally. This corresponds with the results of Finlay and Ivinskis (1984), who also demonstrated infants’ ability to process stimuli presented in their peripheral visual field. This study shows clearly that stimuli in the periphery are not only processed while the infant is looking to another stimulus in the central visual field, but also influence the infant’s actual looking behavior. The results are also in line with the explanations of sticky fixation as they have been proposed by Hood (1995) or by Johnson (1990). These authors emphasize young infants’ failure in disengaging their gaze (Hood, 1995) and their impoverished orienting to stimuli im-pinging on their peripheral visual field (Johnson, 1990) when attending to a central stimulus. However, the present results also suggest that infants are able to shift their attention covertly and that it is unlikely that difficulties in moving covert attention – as proposed by Rothbart et al. (1994) – form the basis of the phenomenon of obliga-tory attention.

Third, the results suggest that – compared to the abstract stimulus – the stimulus featuring the infant’s mother’s face was less able to hold or attract the attention of the infants. This is consistent with findings of studies on face recognition and mother-infant interaction. Very young infants recognize their mother’s face (Bushnell, Sai, & Mullin, 1989; Barrera & Maurer, 1981) and look at it frequently during face-to-face interaction. As they grow older, their interests broaden to include other persons and objects around them. In face-to-face interaction with their mothers, infants of 12 weeks look away from their mother’s face more often (van Wulfften Palthe, 1986), and mothers have to compete and must interact more actively to hold their infant’s attention (Kaye & Fogel, 1980). At the same time, infants have been shown to be very sensitive to disturbances or non-contingent facial activity of their counterparts (e.g., Cohn & Tronick, 1983; Muir & Hains, 1993). We cannot rule out completely the possibil-ity that the use of a video recording of the mother’s face might have had an aversive effect. However, in the scanning experiment in which infants of the same age range were presented with the same stimuli in larger size, no significant differences in fixa-tion time between the abstract and the face stimulus were found. This indicates that the use of a video recording of the mother’s face did not have an aversive effect on the infants (see Chapter 2). At the same time, this finding from the scanning study again suggests that the two stimuli were approximately comparable in terms of their physical characteristics, as the infants did not pay significantly more attention to one of them.

74

Chapter 3

75

Infants’ Shifts of Gaze to a Peripheral Target

The current results clearly support a model which describes the frequency and latency of shifts of gaze as a function of the attractiveness of what is currently at-tended to in the central visual field as well as of the attractiveness of the competing peripheral target (Hood, Murray, King, Hooper, Atkinson, & Braddick, 1996). Devel-opment from obligatory attention to reliable gaze shifting then can be described as reflecting changes in the relative strength of two opposite processes, one maintaining fixation of attention and gaze and the other enabling shifts to a new target. Both types of process are sensitive to characteristics of the competing stimuli.

REFERENCESAtkinson, J., Hood, B. , Wattam-Bell, J., & Braddick, O. (1992). Changes in infants’ ability

to switch visual attention in the first three months of life. Perception, 21, 643-653.Aslin, R. N., & Salapatek, P. (1975). Saccadic localization of visual targets by the very

young human infant. Perception and Psychophysics, 17, 293-302.Barrera, M. E., & Maurer, D. (1981). Recognition of mother’s photographed face by the

three-month-old infant. Child Development, 52, 714-716.Bayley, N. (1993). Bayley Scales of Infant Development (Second Edition). San Antonio, TX:

The Psychological Corporation.Bushnell, I. W. R., Sai, F., & Mullin, J. T. (1989). Neonatal recognition of the mother’s

face. British Journal of Developmental Psychology, 7, 3-15.Butcher, P. R., Kalverboer, A. F., & Geuze, R. H. (2000). Infants’ shifts of gaze from a

central to a peripheral stimulus: A longitudinal study of development between 6 and 26 weeks. Infant Behavior and Development, 23, 3-21.

Cohen, L. B. (1972). Attention-getting and attention-holding processes of infant visual preferences. Child Development, 43, 869-879.

Cohen, L. B. (1976). Habituation of infant visual attention. In T. J. Tighe & R. N. Leaton (Eds.), Habituation: Perspectives from child development, animal behavior, and neurophysi-ology (pp. 207-238). Hillsdale, NJ: Erlbaum Associates, Inc.

Cohn, J. F., & Tronick, E. Z. (1983). Three-month-old infants’ reaction to simulated maternal depression. Child Development, 54, 185-193.

Finlay, D., & Ivinskis, A. (1982). Cardiac and visual responses to stimuli presented both foveally and peripherally as a function of speed of moving stimuli. Developmental Psychology, 18, 692-698.

Finlay, D., & Ivinskis, A. (1984). Cardiac and visual responses to moving stimuli pre-sented either successively or simultaneously to the central and peripheral visual fields in 4-month-old infants. Developmental Psychology, 20, 29-36.

Haith, M. M., Hazan, C., & Goodman, G. S. (1988). Expectation and anticipation of dy-namic visual events by 3.5-month-old babies. Child Development, 59, 467-479.

Harris, P., & MacFarlane, A. (1974). The growth of the effective visual field from birth to seven weeks. Journal of Experimental Child Psychology, 18, 340-348.

76

Chapter 3

77

Infants’ Shifts of Gaze to a Peripheral Target

Hicks, J. M., & Richards, J. E. (1998). The effect of stimulus movement and attention on peripheral stimulus localization by 8- to 26-week-old infants. Infant Behavior and Development, 21, 571-589.

Hood, B. M. (1995). Shifts of visual attention in the human infant: A neuroscientific approach. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 9, pp. 163-216). Norwood, NJ: Ablex.

Hood, B. M., & Atkinson, J. (1990). Sensory visual loss and cognitive deficits in the se-lective attentional system of normal infants and neurologically impaired children. Developmental Medicine and Child Neurology, 32, 1067-1077.

Hood, B. M., & Atkinson, J. (1993). Disengaging visual attention in the infant and adult. Infant Behavior and Development, 16, 405-422.

Hood, B. M., Murray, L., King, F., Hooper, R., Atkinson, J., & Braddick, O. (1996). Habitu-ation changes in early infancy: Longitudinal measures from birth to 6 months. Journal of Reproductive and Infant Psychology, 14, 177-185.

Hopkins, B., & van Wulfften Palthe, T. (1985). Staring in infancy. Early Human Develop-ment, 12, 261-267.

Johnson, M. H. (1990). Cortical maturation and the development of visual attention in early infancy. Journal of Cognitive Neuroscience, 2, 81-95.

Johnson, M. H., Posner, M. I., & Rothbart, M. K. (1991). Components of visual orienting in early infancy: Contingency learning, anticipatory looking, and disengaging. Journal of Cognitive Neuroscience, 3, 335-344.

Kaye, K., & Fogel, A. (1980). The temporal structure of face-to-face communication between mothers and infants. Developmental Psychology, 16, 454-464.

Matsuzawa, M., & Shimojo, S. (1997). Infants’ fast saccades in the gap paradigm and development of visual attention. Infant Behavior and Development, 20, 449-455.

Maurer, D., & Lewis, T. L. (1979). Peripheral discrimination by three-month-old infants. Child Development, 50, 276-279.

Mohn, G., & van Hof-van Duin, J. (1986). Development of the binocular and monocular visual fields of human infants during the first year of life. Clinical Vision Science, 1, 51-64.

Muir, D. W., & Hains, S. M. J. (1993). Infant sensitivity to perturbations in adult facial, vocal, tactile, and contingent stimulation during face-to-face interactions. In B. de Boysson-Bardies, S. de Schonen, P. Jusczyk, P. MacNeilage, & J. Morton (Eds.), Developmental neurocognition: Speech and face processing in the first year of life (pp. 171-185). Dordrecht, The Netherlands: Kluwer.

Posner, M. I., & Rothbart, M. K. (1981). The development of attentional mechanisms. In H. E. Howe & J. H. Flowers (Eds.), Nebraska symposium on motivation (pp. 1-52). Lincoln, NB: University of Nebraska Press.

Prechtl, H. F. R., & Beintema, D. (1964). The neurological examination of the full term infant. Little Club Clinics in Developmental Medicine No. 12. London: Heineman.

76

Chapter 3

77

Infants’ Shifts of Gaze to a Peripheral Target

Richards, J. E. (1997). Peripheral stimulus localization by infants: attention, age and individual differences in heart rate variability. Journal of Experimental Psychology: Human Perception and Performance, 23, 667-680.

Rothbart, M. K., Posner, M. I., & Rosicky, J. (1994). Orienting in normal and pathological development. Development and Psychopathology, 6, 635-652.

Rummel, R. J. (1970). Applied factor analysis. Evanston, IL: Northwestern University Press.

Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modelling. London: Sage.

Stechler, G., & Latz, E. (1966). Some observations on attention and arousal in the human infant. Journal of the American Academy of Child Psychology, 5, 517-525.

Stevens, J. (1992). Applied multivariate statistics for the social sciences. Hillsdale, NJ: Erl-baum Associates, Inc.

Stifter, C. A., & Moyer, D. (1991). The regulation of positive affect: gaze aversion activity during mother-infant interaction. Infant Behavior and Development, 14, 111-123.

Tennes, K., Emde, R., Kisley, A., & Metcalf, D. (1972). The stimulus barrier in early infancy: An exploration of some formulations of John Benjamin. In R. R. Holt & E. Peterfreund (Eds.), Psychoanalysis and contemporary science (pp. 206-234). New York: Macmillan.

Thelen, E., & Smith, L. B. (1998). Dynamic systems theories. In E. Damon (Series Ed.) & R. M. Lerner (Vol. Ed.), Handbook of child psychology. Vol. 1: Theoretical models of human development (pp. 563-634). New York: Wiley.

Tronick, E. (1972). Stimulus control and the growth of the infant’s effective visual field. Perception and Psychophysics, 11, 373-376.

van Wulfften Palthe, T. (1986). Neural maturation and early social behavior: A longitudinal study of mother-infant interaction. Doctoral dissertation. University of Groningen, Groningen, The Netherlands.

Walton, G. E., Armstrong, E. S., & Bower, T. G. R. (1997). Faces as forms in the world of the newborn. Infant Behavior and Development, 20, 537-543.

Woodhouse, G. (1996). Multilevel modelling applications: A guide for users of MLn. London: University of London, Institute of Education.

Chapter 4

Associations between the Developmental Trajectories of Visual Scanning and Disengagement of Attention in Infants

AbstractThe relation between the developmental trajectories of visual scanning and disen-gagement of attention and gaze were examined throughout early infancy. A sample of 10 infants carried out a scanning and a disengagement task with the same visual stimuli six times between 6 and 26 weeks of age. Frequency and latency measures were analyzed using multivariate multilevel models and Monte Carlo analyses. The results suggest that the ability to scan a face or an abstract stimulus evolves slightly earlier than the ability to shift gaze to a newly appeared target in the periphery. This is con-sistent with the account that the parvocellular stream becomes functional slightly before the magnocellular stream. The study revealed no indications of a positive as-sociation between the development of scanning and disengagement on the level of the individual infant. Scanning and disengagement change scores contrasted more with one another than could be expected on the basis of chance. This implies that the dorsal and the ventral stream develop rather independently up to the age of 26 weeks.

This chapter is based on Hunnius, S., Geuze, R. H., & van Geert, P. L. C. (submitted for publication). As-sociations between the developmental trajectories of visual scanning and disengagement of attention in infants.

80

Chapter 4

81

Associations between the Development of Scanning and Disengagement

INTRODUCTIONThe visual system of an infant undergoes substantial developmental changes

during the first few months of life (see e.g., Atkinson, 1992). Infants of 1 to 2 months of age have been reported to show only very limited scanning with long fixations on few locations of the stimulus under examination (Bronson, 1990; Leahy, 1976). They also have problems shifting their gaze away from a stimulus they are currently at-tending to (Hopkins & van Wulfften Palthe, 1985; Aslin & Salapatek, 1975). A gradual transition to an adult-like scanning pattern during the first few months of life has been described, and from the age of 3 to 4 months on, a controlled, strategic way of scanning emerges (Bronson, 1994). At approximately the same age, the frequency and speed of shifts of gaze from a stimulus to a target appearing in the periphery increase (Hood & Atkinson, 1993; Butcher, Kalverboer, & Geuze, 2000).

Several studies have examined the interrelations between different measures of attention in infancy, for instance between looking time during habituation and disengagement latency or quality of visual scanning (Bronson, 1991; Frick, Colombo, & Saxon, 1999). One motivation for these studies examining associations between different measures of attentional functioning has been to explore the underlying processes which give rise to continuity in early individual differences (Colombo, 1995).

However, to date no study has compared the interrelations of the development of functional scanning and the ability to disengage attention and gaze. In this study, the development of scanning and disengagement during early infancy is analyzed, and the results of a comparison of the two developmental trajectories are reported.

Explanations of the Development of Disengagement and ScanningAlthough the skills of visual scanning and disengagement of gaze and attention

emerge during approximately the same period of time, they are considered to be based on different neurological systems. Shifting gaze away from a stimulus currently under attention to a newly appeared target in the peripheral visual field is – according to Schiller’s model (Schiller, 1998) – mediated by the posterior eye movement control system, as this system is responsible for the generation of fast, reflex-like orienting responses. The pathways of this system run from the retina through the occipital and – partly – the parietal cortex and reach the brain stem eye movement centers through the superior colliculus.

To scan a stimulus, however, requires higher level eye movements. This kind of intentional eye movements to selected or remembered locations and organized sequences of saccades are thought to be generated by the anterior eye movement control system, in which the frontal eye fields are thought to play an important role (Bichot, Schall, Thompson, 1996; Guitton, Buchtel, & Douglas, 1985). Input to the an-terior system is thought to originate in the P cells of the retina and the parvocellular pathways, which subserve detailed form and color vision. The posterior system, on

80

Chapter 4

81

Associations between the Development of Scanning and Disengagement

the other hand, is thought to receive mostly magnocellular input from cells which are specialized for the detection of sudden changes (Richards & Hunter, 1998).

There are indications that the magnocellular and the parvocellular system may have different developmental courses (Atkinson & Braddick, 2003; Atkinson, 1992). They arise mostly from comparisons between the developmental time courses of the cortical streams on basis of visual evoked potential (Braddick, 1993; Hood, Atkinson, Braddick, & Wattam-Bell, 1992) and behavioral measures (see e.g., Atkinson, 1992). Several cortical mechanisms (such as orientation selectivity, direction selectivity, and selectivity to binocular relations) are associated with processing mainly within one of the two streams, and these functions have been shown to emerge at different ages. The order in which these mechanisms develop suggests that the two systems may have different developmental courses with the parvocellular pathway becoming functional slightly ahead of the magnocellular pathway (see Hickey & Peduzzi, 1987, for a review).

Aims of the StudyAs mentioned above, both behaviors – visual scanning and disengagement – undergo

rapid development during the first few months of life. However, to date, there is no evidence available on the same infants’ performance on both behaviors during this period of development, nor have the developmental trajectories of these behaviors been compared. It was the goal of this study to compare the development of visual scanning and disengagement in the same infants, using data obtained in an intense longitudinal investigation. Considering the presumed neurological underpinnings of both behaviors, it was expected that functional scanning would emerge slightly earlier than reliable gaze shifting.

Although the magnocellular and the parvocellular system seem to subserve different functions, there are indications that the two streams cannot be considered to be strictly parallel pathways, but rather that they interact (see for reviews, Cowey, 1994; Merigan & Maunsell, 1993). The second aim of this study was therefore to explore possible as-sociations between the developmental changes in scanning and disengagement.

METHOD

ParticipantsTen infants (5 girls; 5 boys) carried out a scanning task (as described in Chapter 2)

as well as a disengagement task (as described in Chapter 3). They were tested every 4 weeks, starting at the age of 6 weeks, until they were 26 weeks old. Mean ages at the testing dates were 46.7 days (SD = 3.8), 73.0 days (SD = 2.4), 102.2 days (SD = 3.4), 130.2 days (SD = 2.3), 158.0 days (SD = 3.4), and 186.7 days (SD = 4.4). If a measurement session could not be completed due to sleepiness, fussing, or crying, a new appointment was made within a few days. Eight of the infants carried out all 6 measurement sessions; two infants completed 5 sessions.

82

Chapter 4

83

Associations between the Development of Scanning and Disengagement

All participants were healthy, full-term infants with no history of complications during gestation or delivery and a birth weight above 2800 g. All infants scored within their age range on the Bayley Scales of Infant Development (BSID-II; Bayley, 1993) at 12 and 24 weeks of age. The mothers of the infants were contacted through childbirth education classes, midwives, or gym classes. They were told about the course and goals of the study and gave their written informed consent. The study was approved by the local Medical Ethics Committee.

ProcedureAppointments were scheduled for a time of the day when mothers expected their

infants to be able to stay awake for about 30 minutes. After mother and infant had ar-rived at the lab, infants were given some time to get used to the new environment.

The babies carried out a scanning task and a disengagement task. They were seated in an infant seat in a reclined posture (about 45 degrees) in front of a 21 inch computer monitor. Only the screen of the monitor was visible. The frame of the monitor and all the other equipment was concealed behind a gray curtain. The distance between the infant’s eyes and the screen was 35 cm. The infant’s face and the display shown to the infant were visible to the experimenter on a video monitor and were recorded for off-line analysis.

In the tasks, two different dynamic stimuli were used (see Figure 2.1, Chapter 2). The first was a short video of the infant’s mother’s face as she was interacting with her baby. This stimulus was matched with an abstract stimulus that was created to be comparable in terms of physical characteristics, such as movement dynamics, color, and luminance.

The infants were first presented with the scanning task. During this task, the in-fants were shown the two stimuli one after another for 30 seconds each. Each stimulus subtended 30 by 40 degrees. The eye movements infants made while scanning the dynamic displays were measured using an infrared eye-tracking system (ASL, model 504) in combination with a head-tracker (Polhemus Fastrak). In addition, the video-recordings of each infant were coded off-line in order to correct errors and complete the data set. The scanning task and the measurement of the eye movements are fully described in Chapter 2.

Infants then carried out a disengagement task as it is reported in detail in Chapter 3. The experiment contained 32 competition trials, in which shifts of attention and gaze between a central fixation stimulus and a peripheral target were measured (see Figure 3.1 A, Chapter 3). In these trials, a stimulus first appeared in the center of the monitor. When the infant was fixating this stimulus, a second stimulus was added in the periphery (at 20 degrees eccentricity). Half of the peripheral targets appeared on the left, half on the right. Both stimulus sorts – the face and the abstract stimulus – were used as central fixation stimulus and as peripheral target. The order in which the trials were presented was randomized. The infants’ eye movements were videotaped

82

Chapter 4

83

Associations between the Development of Scanning and Disengagement

and the frequency and latency of gaze shifts to the peripheral stimulus were coded off-line. Therefore, the direction and latency of the first eye movement after the peripheral target appeared were scored. All direct eye movements to the peripheral stimulus which started at least 200 ms after the second stimulus had appeared were coded as looks.

Coding was carried out by different observers, who were trained by the first author. The interobserver reliability was determined by double-coding 10% of the sessions. For the disengagement task, it was found to be on average 93.5% (range 89.5% to 100%) for the onset of an eye movement. Cohen’s kappa for the category of first eye movement was on average .82 (range .72 to 1.0). The interobserver reliability for the corrections and completions of the eye-tracker measurement during the scanning task was also good. For identifying an eye movement in the absence of an eye-tracker signal, it was 96.9%. The agreement between observers for the onset and the length of an eye move-ment was 92.5%, respectively 94.6%.

AnalysisTwo indices of performance in the scanning and in the disengagement task were

analyzed: frequency and latency measures. Both measures have been used and shown to be suitable to describe the development of visual scanning and disengagement in earlier studies (see e.g., Bronson, 1994; Butcher et al., 2000). The frequency measures consisted of the relative frequency of looks to the peripheral target in the disengage-ment task and the number of fixations during scanning. Both measures represent the frequency of eye movements during the tasks. The second measures were the averaged median latency of gaze shifts in the disengagement task and the median fixation duration during scanning. These measures both reflect the time before an eye movement to a new location of interest (within a scanning pattern or between two different stimuli) is started.

Multilevel analysis. First, the developmental changes of these pairs of indices were examined using multivariate multilevel models (Snijders & Bosker, 1999). Analyzing the data jointly instead of with several separate models has the advantage that the tests of specific effects tend to be more powerful and that the danger of chance capitaliza-tion is avoided. The models had three levels: infant, test session, and stimulus sort. Based on the results of earlier analyses (see e.g., Chapter 3; Butcher et al., 2000), the development was described using three piecewise linear age functions (age periods 6 to 9, 9 to 16 and 16 to 26 weeks). The data were centered around 12 weeks of age, as this was approximately the middle of the period in which the largest changes were expected. The statistical significance of the coefficients was determined using t tests. T tests were also used as post-hoc tests. With multiple testing, Bonferroni corrections were employed to keep alpha at .05 (Stevens, 1992).

Monte Carlo analyses. Further analyses were carried out using Monte Carlo methods (random permutation or random resampling methods). On the one hand, possible as-

84

Chapter 4

85

Associations between the Development of Scanning and Disengagement

sociations (i.e. similarities) between the development of scanning and disengagement in individual infants were investigated. On the other hand, it was examined whether the changes in the development of visual scanning stabilized earlier than those oc-curring in the emergence of reliable disengagement.

For these analyses, both pairs of indices were standardized. The rescaling had to conserve the characteristic features in the form and relative range of the develop-mental trajectory to enable optimal comparability between these variables. In view of the fact that the data represent developmental curves, which, in principle, show a trajectory of increase between an initial and a final point in time (i.e. the first and the last observation), we opted for the following approach. Based on the group data, averages were calculated for the first observation and the last, for both pairs of vari-ables (see Figure 4.1 and Figure 4.2). The distance between the average of the first and last observation was taken as the unit interval, that means the interval with value 1. Individual scores were standardized by means of the following transformation

ynorm = (yobs – minV) / (maxV – minV)for ynorm the normalized score, yobs the observed score and minV and maxV the aver-age score of the first and last observation respectively. By doing so, individual scores were compared on a unit interval defined by the developmental trajectory based on group data.

The data were examined using a permutation or a resampling method, which is a procedure linked to the group of bootstrap techniques (Todman & Dugard, 2001; Good, 1999; Manly, 1997; permutation boils down to random selection without replacement, resampling is random selection with replacement). By carrying out a large number of randomizations (1000 or 5000, depending on the required accuracy), an approxima-tion of the exact chance distribution of the test statistic used can be obtained. So, with permutation and resampling methods, the probability that an observed test statistic is a result of chance can be estimated without having to make assumptions about expected distributions across the “population” of observations. The probabil-ity is calculated by comparing the observed test statistic to the test statistic that is obtained when the data are randomly assembled. Another advantage of permutation and resampling methods is that they are particularly suited for small datasets with missing data.

Similarity was expressed by means of averaged χ values. The χ value is defined as the average of the distance between corresponding scores. It was chosen as it is an intuitively understandable measure of distance and – other than a χ2 value – does not carry the risk of overestimating the effect of larger differences. Its chance distribution can be obtained by means of the Monte Carlo simulation.1

1 All analyses were double-checked by means of χ2 value analyses. These analyses yielded results that were consistent with those based on the χ value analyses.

84

Chapter 4

85

Associations between the Development of Scanning and Disengagement

RESULTS

Multilevel AnalysisFrequency indices. The development of the averaged frequency of disengagement

and the averaged number of fixations during scanning are presented in Figures 4.1 A and B. It can be seen that both indices first undergo a period of rapid change, before a stabilization takes place. However, this stabilization seems to occur earlier in the development of scanning behavior than of disengagement.

The multivariate multilevel model contained (almost) significant coefficients for all three age periods concerning the development of gaze shifting. Thus, the frequency of disengagement increased almost significantly between 6 and 9 (t(342) = 1.93, p < .10)

Figure 4.1. The development of (A) the mean number (and standard error) of fixations during the scanning task and (B) the mean frequency (and standard error) of looks to the periph-eral target during the disengagement task.

86

Chapter 4

87

Associations between the Development of Scanning and Disengagement

and then significantly between 9 and 16 (t(342) = 6.01, p < .001) and 16 and 26 (t(342) = 3.61, p < .001) weeks of age. Looking at the development of scanning in these infants, the pattern was slightly different. The increase in the number of fixations was (al-most) significant during the age period of 6 to 9 weeks (t(342) = 1.70, p < .10) and 9 to 16 weeks (t(342) = 2.85, p < .001). For the age period of 16 to 26 weeks, the age coefficient did not indicate a significant increase anymore (16 to 26 week period, t(342) = .82, p > .10). Accordingly, post-hoc tests yielded a significant increase in disengagement frequency between 18 (M = 65.5%, SD = 17.0) and 22 weeks of age (M = 87.7%, SD = 13.0; t(9) = 5.65, p < .001), but there was no significant change anymore in the number of fixations during scanning between the same two measurement points (18 weeks, M = 59.2, SD = 13.9; 22 weeks, M = 55.7, SD = 16.7; t(9) = .54, p > .10).

Figure 4.2. The development of (A) the median fixation duration (and standard error) during the scanning task and (B) the mean latency (and standard error) of looks to the peripheral target during the disengagement task.

86

Chapter 4

87

Associations between the Development of Scanning and Disengagement

Latency indices. Figures 4.2 A and B depict the development of the averaged median fixation duration during scanning and of the averaged latency of gaze shifts in the disengagement task. The multivariate multilevel model showed that the latency of gaze shifts during the disengagement task decreased (almost) significantly between 6 and 9 (t(303) = 2.19, p < .05), 9 and 16 (t(303) = 1.91, p < .10) and between 16 and 26 (t(303) = 2.69, p < .01) weeks of age. So the reaction time in the disengagement task decreased throughout the entire measurement period. For the duration of fixations during visual scanning, the multilevel model contained only one significant coef-ficient. The fixation duration declined significantly solely between 9 and 16 weeks of age (t(303) = 2.71, p < .01), while it remained rather stable between 6 and 9 and 16 and 26 weeks of age.

The data thus indicate that both parameters – the frequency and the latency measure – leveled off slightly earlier for the development of scanning than for the development of gaze shifting.

Monte Carlo AnalysesAssociations between the individual developmental trajectories. In this paragraph, the

relation between the developmental trajectories of visual scanning and disengage-ment of attention are investigated at the level of the individual infant. Figure 4.3 and Figure 4.4 depict the individual developmental trajectories of the standardized frequency and latency measures of visual scanning and disengagement. Inspection of the individual curves of the frequency measures (see Figure 4.3) indicates that both the scanning and the disengagement curves show an increase roughly between the minimum and maximum values. Infant 10 is an exception: Here, the scanning curve fluctuates around the maximum and drops back to around the minimum value dur-ing the session at the age of 22 weeks. Some infants show a relatively smooth increase in both curves (infants 1 and 8 in particular), others show salient fluctuations, par-ticularly in the scanning curve. While the averaged curves of the latency measures describe an overall decrease (see Figure 4.2), the graphs of the individual infants show a rather large diversity (see Figure 4.4). Infants 3, 5, and 6 show large fluctuations, especially concerning the median fixation duration during scanning. In two infants (infant 4 and 10), no decrease in the latency measures can be observed concerning their performance either in the scanning or the disengagement task. For infants 1, 2, 3, 7, 8, 9, and 10, one measurement point was missing at 6 weeks. This was mainly due to the fact that infants rarely shifted their gaze during the disengagement task when they were young (as can also be seen from the frequency measures) but rather kept on staring to the central stimulus. However, in infants 1, 2, 7, 8, and 9 the latency measures decrease relatively smoothly throughout the measurement period.

The – eventual – similarity between the development of scanning and disen-gagement was also approached quantitatively. Associations between the individual developmental trajectories were examined by means of inter- and intraindividual

88

Chapter 4

89

Associations between the Development of Scanning and Disengagement

comparisons. First, the question was addressed whether the developmental trajecto-ries of scanning and disengagement of the same infant tended to display similarity. Therefore, the similarity of the two trajectories in the same infant was compared to the similarity of one of these trajectories with one of another infant. Second, it was examined whether the developmental changes concerning the performance on the scanning and the disengagement task were related within the individual infant.

Figure 4.3. The development of the mean number of fixations during the scanning task and the mean fre-quency of looks to the peripheral target during the disengagement task for the 10 individual infants.

88

Chapter 4

89

Associations between the Development of Scanning and Disengagement

Figure 4.4. The development of the median fixation duration during the scanning task and the latency of looks to the peripheral target during the disengagement task for the 10 individual infants.

If the two curves display related developmental course, the within-subject similar-ity should be greater than the between-subject similarity. For instance, the similarity between the disengagement and scanning curve of one infant should on average be greater than the similarity between his own scanning curve and the disengagement curve of another infant. The null hypothesis – that within-subject similarity cannot be distinguished from between-subject similarity – was tested by means of a Monte Carlo procedure. The scanning data were randomly permuted across the subjects, with

90

Chapter 4

91

Associations between the Development of Scanning and Disengagement

conservation of the individual trajectories (i.e. the six measurement points of each infant were kept together during the random permutation procedure). In this way, the developmental trajectory of disengagement of a particular infant was randomly assigned to a trajectory of scanning of another infant. The averaged χ value of the permutated data was compared to the value of the original data. By carrying out a large number of random permutations, an approximation of the exact chance distri-bution of the χ value was obtained. The proportion of the number of times that the within-subject similarity was greater than the between-subject similarity provided an estimation of the p value (i.e. the probability that the observed similarity was due to chance).

For both the frequency and the latency data, the random permutation procedure confirmed the null hypothesis: The within-subject and between-subject similarity could not be distinguished from one another. The p values for the frequency mea-sures were .66 for a random permutation procedure and .59 for a random resampling procedure based on 5000 simulations. For the latency measures, the p values were .31 respectively .40.2

Second, it was examined whether disengagement and scanning levels measured during a particular occasion in a particular infant showed more similarity amongst each other than a disengagement score compared with a scanning score measured at an arbitrarily different moment. That is, it was investigated whether the changes in the variables (e.g., increases or decreases) were more similar in a particular infant at a particular measurement point than could be expected on the basis of chance. To investigate this question, the scores first had to be de-trended, which was done by means of differencing. The difference between successive scores was calculated (score at t+1 minus score at t). Then, the average distance between these difference scores was determined, which is the average of the absolute values of the difference between a disengagement score at time t and a scanning score at time t, a disengage-ment score at time t+1 and a scanning score at time t + 1, and so forth. If there exists any similarity between disengagement and scanning, the average distance between the two should be smaller (i.e. the similarity should be greater) than a distance score based on chance alone. This possibility was tested by means of a random permutation procedure, where, within each individual, scanning scores were randomly permuted and then compared with the (un-permuted) disengagement scores. Probabilities were calculated for each infant separately and for an average of the distance across the 10 subjects.

2 It has been hypothesized that the development of visual scanning stabilizes before the development of disengagement does. This implies that the within-subject similarity with lag 1 (or more) could eventu-ally be greater than the similarity based on simultaneous observations (i.e. disengagement at time t+1 should be more similar to scanning at time t than to scanning at time t+1). However, χ values with lag (1 or 2) resulted in even less similarity than simultaneous comparisons.

90

Chapter 4

91

Associations between the Development of Scanning and Disengagement

For the frequency measures, the resulting p value, based on 5000 simulations, was equal to .95, which shows that in 95% of the cases the random association of scan-ning and disengagement frequency scores provided a better similarity measure than the actual association. That is, the frequency scores of disengagement and scanning tended to differ more strongly from one another than can be expected on the basis of chance alone. This difference eventually refers to a potential contrast or antagonistic relationship between the development of the number of fixations during scanning and the frequency of looks in the disengagement task. Inspection of the individual p values indicated that this overall effect was due to the majority of the infants, as their distance score of the real data was greater than that of the randomly permuted series (infants 2, 3, 4, 5, 7, 9, and 10; for 2 infants, 5 and 10, the difference was highly significant; p < .001). However, concerning the latency measures of scanning and disengagement there was no such effect: The resulting p value was .11 based on 5000 simulations.3

Timing of development. Finally, it was investigated to what extent the individual curves complied with the hypothesis that the development of scanning stabilizes slightly earlier than the development of disengagement. Therefore, the following pro-cedure was carried out, which is based on the specification of a “point of stabilization”. If a process stabilizes at time t, its average change (growth or fluctuation) before t must be (considerably) greater than its change or fluctuation after time t (since it cannot be expected that stabilization means a standstill, stabilization is defined as “little change”). In a time series consisting of n observations, a breaking point can be determined as the point in time t, where the difference in average change before t and average change after t is maximal (i.e. the stable phase is defined as the phase with minimal average change, in comparison with the average change preceding that phase). If – in the present case – the scanning variable stabilized earlier than the dis-engagement variable, the breaking point of the development of visual scanning should fall at an earlier time than that of the development of disengagement.

The statistical test of the null hypothesis went as follows: First, the change scores of the disengagement and the scanning variables were standardized to obtain scores coming from a distribution with a similar average and standard deviation. After the standardization, the only meaningful distinction between the disengagement and scanning change scores is the presumed difference in the way they are ordered, which

3 As it was hypothesized that visual scanning develops earlier than disengagement does, it is possible that the obtained effect is due to a lag in the developmental course of disengagement, as the distance score of two data series with a phase shift might be greater than should be expected on the basis of randomly permuted series. However, it could be shown, that this is not necessarily the case: A simula-tion was carried out with two series of five imaginary scores, which displayed a clear phase shift. The simulation showed that in only 21% of the cases the random association of the two imaginary series provided a better similarity measure than the actual association (compared to the 95% which resulted from the analyses of our data).

92

Chapter 4

93

Associations between the Development of Scanning and Disengagement

corresponds with the similarity or dissimilarity in breaking points. Second, for every infant a “compliance” value was calculated, which was 1 if the breaking point of scan-ning was earlier than that of disengagement, .5 if they occurred at the same time, and 0 if disengagement preceded scanning. The scores of the individual infants were added up to receive an overall “compliance” score. Third, a within-subjects/within-mea-surements random sampling of the change scores was carried out.4 For each random sampling per subject, breaking points and a “compliance” score were calculated, as defined above. By carrying out a large number of randomizations, an estimation of the distribution of breaking points for each individual subject and for the average breaking point over all subjects was obtained, as based on the null hypothesis (“no difference”). With this distribution, the probability was calculated that the observed average breaking point was due to chance.

In infants 1, 2, 3, 7, 9, and 10, the frequency scores of visual scanning stabilized earlier than the frequency of disengagement. In 2 of these infants, the difference between the breaking points was larger than one measurement interval. Scanning and disengagement stabilized at the same time in infants 4 and 8, and in infants 5 and 6 disengagement stabilized before scanning. This resulted in an overall “compli-ance” score of 7. The resampling procedure, based on 5000 randomized runs, yielded a p value of .007. That is, it is highly unlikely that the “compliance” score was based on chance, that means it is highly likely that the observed difference in stabilization between scanning and disengagement is a real phenomenon. It can be concluded

4 The random sampling procedure was carried out on the basis of the following reasoning and hypotheses: The existence of a breaking point implies that, on average, larger change scores occur in the first part of the set of change scores and smaller ones tend occur in the second part (“first” and “second” refer to variable intervals, depending on where the breaking point actually falls). Similarity of the breaking points means that if one finds a large change score at a point in time during the development of scan-ning, there is a fair chance of finding a comparable, large score at the same time for the development of disengagement (or a small score in one and a small score in the other, for that matter). If the breaking points are different, and scanning stabilizes earlier, there must be at least one measurement occasion for which a small change score in scanning, on average corresponds with a large score in disengagement. Thus, if the change scores of scanning and disengagement are standardized (similar average, similar distribution) and the breaking point is the same for both variables, the standardized scores of scanning and disengagement must be statistically interchangeable (for each specific measurement occasion, i.e. measurement t of scanning must be interchangeable with measurement t of disengagement, t+1 of scanning with t+1 of disengagement, and so forth). If the breaking points differ systematically (e.g., scanning stabilizes earlier than disengagement) there must be one measurement (or more) for which the standardized scores are not interchangeable (e.g., at time t+n, where the change score of scanning is on average considerably smaller than that of disengagement). Thus, if the null hypothesis is true and the breaking points of scanning and disengagement are, on average, the same, a resampling test that arbitrarily resamples the change scores of scanning and of disengagement within the first, second, etc. measurement, must on average lead to the same breaking point distribution as that found in the data. If the breaking points are not similar, the arbitrary resampling of the scores within each measurement must lead to a distribution of breaking points that is different from the observed one.

92

Chapter 4

93

Associations between the Development of Scanning and Disengagement

that the number of fixations during scanning stabilized earlier than the frequency of looks during disengagement if stabilization is defined as reduced average change or fluctuation. Thus, concerning the frequency measures, the development of visual scanning came to rest earlier than the development of disengagement.

For the latency measures, there were no indications that the changes in one of the two variables leveled off earlier. The analyses yielded an overall “compliance” score of 5.5 and a p value of .41.

DISCUSSIONIn this study, infants carried out both a scanning task and a gaze shifting task

six times during the first six months after birth. Multilevel modeling revealed that frequency and latency measures continued to change throughout the last age period between 16 and 26 weeks of age for the disengagement task, while there was no change for the scanning task after 16 weeks. This finding was confirmed for the frequency measures by the Monte Carlo analyses: The changes in the number of fixations dur-ing scanning stabilized earlier than the frequency of looks in a disengagement situ-ation.

The results thus suggest that the ability to scan a dynamic stimulus (face or ab-stract) evolves slightly earlier than the ability to disengage attention and gaze and look to a newly appeared target in the periphery. While gaze shifting is thought to depend on the dorsal stream, which receives magnocellular input and projects to the posterior parietal cortex, visual scanning is considered to be mediated by the ven-tral stream, which is parvocellular-based and projects to the inferotemporal cortical areas. Our findings thus are in accord with earlier accounts that the parvocellular stream becomes functional slightly before the magnocellular stream (Atkinson, 1992; Atkinson & Braddick, 2003).

It was the second aim of this study to explore possible associations between the development of scanning and disengagement at the level of the individual subjects. The developmental courses of the scanning and the disengagement scores were com-pared but could not be shown to be more similar within the same infant than between different infants. Although earlier studies suggested that there might be interrela-tions between the dorsal and the ventral system in the adult brain (see e.g., Felleman & Van Essen, 1991; Schiller, Logothetis & Charles, 1990), our study thus revealed no indications for a relation between the developmental trajectories of these two visual abilities until the age of 6 months. However, other studies have also emphasized the independence of different ventral and dorsal mechanisms, as for example in children with Williams syndrome. These children show a profile in which ventral processes (such as face recognition) are unimpaired, while the dorsal function for visual con-trol of action develops abnormally (Bellugi, Wang, & Jernigan, 1994; Atkinson, Anker, Macpherson, Nokes, Andrew, & Braddick, 2001).

Furthermore, it turned out that the developmental changes in a particular infant

94

Chapter 4

95

Associations between the Development of Scanning and Disengagement

at a particular measurement point were even less similar than one would randomly expect. The change scores of the frequency measures of scanning and disengagement thus showed some kind of contrast or complementarity. This contrast effect could be due to the fact that the processes governing scanning and disengagement do not only develop rather independently, but might even compete for similar resources of development or performance.

To sum up, it seems that during the first 6 months of life the two skills examined in this study – visual scanning and gaze shifting – develop independently and follow their own developmental time courses. However, these conclusions are based on a relatively small (but representative) sample. Larger samples are needed not only to make the results more reliable, but also to obtain a better understanding of the nature of the inter- and intra-individual differences. Furthermore, one has to keep in mind that a study of behavioral change cannot provide hard evidence on the developmental status of neurological substrates. However, detailed descriptions of behavioral development provide valuable information and can form the basis for specific future investigations of neurological change.

REFERENCESAslin, R. N., & Salapatek, P. (1975). Saccadic localization of visual targets by the very

young human infant. Perception and Psychophysics, 17, 293-302.Atkinson, J. (1992). Early visual development: Differential functioning of parvocellular

and magnocellular pathways. Eye, 6, 129-135.Atkinson, J., Anker, S., Macpherson, F., Nokes, L., Andrew, R., & Braddick, O. (2001).

Visual and visuo-spatial development in young children with Williams Syndrome. Developmental Medicine and Child Neurology, 43, 330-337.

Atkinson, J., & Braddick, O. (2003). Neurobiological models of normal and abnormal visual development. In M. de Haan & M. H. Johnson (Eds.), The cognitive neuroscience of development (pp. 43-71). Hove, UK: Psychology Press.

Bayley, N. (1993). Bayley Scales of Infant Development (Second Edition). San Antonio, TX: The Psychological Corporation.

Bellugi, U., Wang, P. P., & Jernigan, T. L. (1994). Williams syndrome: An unusual neu-ropsychological profile. In S. H. Broman & J. Grafman (Eds.), Atypical cognitive defi-cits in developmental disorders: Implications for brain function (pp. 23-56). Hillsdale NJ: Erlbaum Associates, Inc.

Bichot, N. P., Schall, J. D., & Thompson, K. G. (1996). Visual feature selectivity in frontal eye fields induced by experience in mature macaques. Journal of Neurophysiology, 48, 338-31.

Braddick, O. J. (1993). Orientation and motion-selective mechanisms in infants. In K. Simons (Ed.), Early visual development: Normal and abnormal (pp. 163-177). New York: Oxford University Press.

94

Chapter 4

95

Associations between the Development of Scanning and Disengagement

Bronson, G. W. (1990). Changes in infants’ visual scanning across the 2- to 14-week age period. Journal of Experimental Child Psychology, 49, 101-125.

Bronson, G. W. (1991). Infant differences in rate of visual encoding. Child Development, 62, 44-54.

Bronson, G. W. (1994). Infants’ transitions towards adult-like scanning. Child Develop-ment, 65, 1243-1261.

Butcher, P. R., Kalverboer, A. F., & Geuze, R. H. (2000). Infants’ shifts of gaze from a central to a peripheral stimulus: A longitudinal study of development between 6 and 26 weeks. Infant Behavior and Development, 23, 3-21.

Colombo, J. (1995). On the neural mechanisms underlying developmental and individual differences in visual fixation in infancy: Two hypotheses. Developmental Review, 15, 97-135.

Cowey, A. (1994). Cortical visual areas and the neurobiology of higher visual processes. In M. J. Farah & G. Radcliff (Eds.), The neuropsychology of high-level vision (pp. 3-31). Hillsdale, NJ: Erlbaum Associates, Inc.

Felleman, D. J., & Van Essen, D. C. (1987). Receptive field properties of neurons in area V3 of macaque monkey extrastriate cortex. Journal of Neurophysiology, 57, 889-920.

Frick, J. E., Colombo, J., & Saxon, T. E. (1999). Individual and developmental differences in disengagement of fixation in early infancy. Child Development, 70, 537-548.

Good, P. I. (1999). Resampling methods: A practical guide to data analysis. Boston, MA: Birkhäuser.

Guitton, D., Buchtel, H. A., & Douglas, R. M. (1985). Frontal lobe lesions in man cause difficulties in suppressing reflexive glances and in generating goal directed sac-cades. Experimental Brain Research, 58, 455-472.

Hickey, T. L., & Peduzzi, J. D. (1987). Structure and development of the visual system. In L. Cohen & P. Salapatek (Eds.), Handbook of infant perception (pp. 1-42). New York, NY: Academic Press.

Hood, B. M., & Atkinson, J. (1993). Disengaging visual attention in the infant and adult. Infant Behavior and Development, 16, 405-422.

Hood, B. M., Atkinson, J., Braddick, O. J., & Wattam-Bell, J. (1992). Orientation selectivity in infancy: Behavioral evidence for temporal sensitivity. Perception, 21, 351-354.

Hopkins, B., & van Wulfften Palthe, T. (1985). Staring in infancy. Early Human Develop-ment, 12, 261-267.

Leahy, R. L. (1976). Development of preferences and processes of visual scanning in the human infant during the first 3 months of life. Developmental Psychology, 12, 250-254.

Manly, B. F. J. (1997). Randomization, bootstrap and Monte Carlo methods in biology (2nd edition). Boca Raton, FL: Chapman & Hall.

Merigan, W. H., & J. H. R. Maunsell (1993). How parallel are the primate visual path-ways? Annual Review of Neuroscience, 16, 369-402.

96

Chapter 4

Richards, J. E., & Hunter, S. K. (1998). Attention and eye movement in young infants: Neural control and development. In J. E. Richards (Ed.), Cognitive neuroscience of at-tention: A developmental perspective (pp. 131-162). Mahwah, NJ: Erlbaum Associates, Inc.

Rose, D., Slater, A, & Perry, H. (1986). Prediction of childhood intelligence from habitu-ation in early infancy. Intelligence, 10, 251-263.

Schiller, P. (1998). The neural control of visually guided eye movements. In J. E. Richards (Ed.), Cognitive neuroscience of attention (pp. 5-50). Mahwah, NJ: Erlbaum Associates, Inc.

Schiller, P. H., Logothetis, N. K., & Charles, E. R. (1990). Role of the color-opponent and broad-bent channels of vision. Visual Neuroscience, 5, 321-346.

Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modelling. London: Sage.

Stevens, J. (1992). Applied multivariate statistics for the social sciences. Hillsdale, NJ: Erl-baum Associates, Inc.

Todman, J. B., & Dugard, P. (2001). Single-case and small-n experimental designs. A practical guide to randomization tests. Mahwah, NJ: Erlbaum Associates, Inc.

96

Chapter 4

Chapter 5

A Longitudinal Study of Shifts of Attention and Gaze in Preterm and Full-term Infants

AbstractThe development of shifts of attention and gaze was examined, between 6 and 26 weeks corrected age, in full-term and preterm infants. Infants carried out a gaze shifting task with competition and non-competition trials. A video of the infant’s mother’s face and an abstract video appeared as central or peripheral stimulus, which resulted in four different combinations (face-face, face-abstract, abstract-face, abstract-abstract). Until 18 weeks, the preterm infants were quicker in shifting gaze from fixation than full-terms. There were, however, no differences in gaze shifting frequency between preterms and full-terms. At the same time, preterm infants also showed less mature gaze shifting behavior: At the age of 6 weeks, they continued to stare at the location of the central stimulus after the peripheral target had appeared more often than full-terms. Preterms as well as full-terms were more likely to shift gaze away from a face to an abstract peripheral target, while they moved their gaze least frequently and more slowly in the opposite condition (abstract-face). However, there were indi-cations that looking away from an abstract stimulus formed a particular challenge for the preterm infants.

This chapter is based on: Hunnius, S., Geuze, R. H., Bos, A. F., & Zweens, M. J. (submitted for publication). A longitudinal study of shifts of attention and gaze in preterm and full-term infants.

100

Chapter 5

101

Pre- and Full-term Infants’ Shifts of Attention and Gaze

INTRODUCTION

The Development of AttentionDuring the first few months of life, infants learn a lot about the world around

them as they explore it visually. Effective visual exploration, however, requires being able to shift gaze flexibly across different locations. Also during early face-to-face interaction, being able to shift gaze to and away from another person plays a crucial role and thus is important for the social-emotional development of an infant. The attentional skills that allow the infant to strike balance between engaging and shift-ing attention develop during the first few months of life. While the sensory-motor processes involved in detecting and shifting gaze to visual targets are functional after 40 weeks gestational age (Butcher, Kalverboer, & Geuze, 2000), infants between 1 and 4 months of age have difficulty disengaging their attention and gaze from an object or stimulus they are currently fixating (Harris & MacFarlane, 1974; Aslin & Salapatek, 1975). This phenomenon of “obligatory attention” (Stechler & Latz, 1966) or “sticky fixation” (Hood, 1995) can be observed in situations of the daily life, such as social interaction (Kaye & Fogel, 1980), but also in a laboratory context during ha-bituation (Hood, Murray, King, Hooper, Atkinson, & Braddick, 1996) or gaze shifting tasks (Matsuzawa & Shimojo, 1997; Hood & Atkinson, 1993). However, around 5 to 6 months of age, infants’ visual behavior is qualitatively similar to that of adults (Hood & Atkinson, 1993; Matsuzawa & Shimojo, 1997).

Visual Development in Preterm InfantsIt has been reported that visual and attentional development is different in preterm

infants and that preterm infants process visual information differently from full-term infants. Studies on habituation demonstrate that preterm infants exhibit longer look durations than full-terms (see e.g., Rose, Feldman, McCarton, & Wolfson, 1988; Spun-gen, Kurtzberg, & Vaughan, 1985; Sigman, Cohen, Beckwith, & Parmelee, 1986; but: Bonin, Pomerleau, & Malcuit, 1998). But preterm infants do not only need more time to process and habituate to stimuli, they also show lower novelty preference scores when confronted with a familiar and a novel stimulus (Sigman, 1983). In a study by Rose, Gottfried, and Bridger (1979), full-term infants discriminated between a familiar and a novel stimulus at 6 and 12 months, whereas preterms did so only at 12 months of age. This suggests that preterm infants also exhibit deficits concerning their visual recognition memory (Rose et al., 1988; Rose et al., 2001). These observations were found to persist throughout the first year of life (Rose, Feldman, & Jankowski, 2001). Preterm infants have also been shown to spend more time off-task (Rose et al., 1988). During free play alone or with their mothers, high-risk preterm infants of 6 months of age shift their gaze less frequently and notice fewer toys (Landry & Chapieski, 1988).

Consistent with these descriptions of preterms’ less efficient performance, Fried-man, Jacobs, and Werthmann (1981) put forward the position that the visual system of the preterm infant may suffer by the premature exposure to visual input and to the

100

Chapter 5

101

Pre- and Full-term Infants’ Shifts of Attention and Gaze

extrauterine environment. However, in the literature on preterm infants conflicting evidence of both delayed and advanced visual development can be found. Preterm infants are reported to show precocious maturity when focusing and when tracking moving stimuli (Dubowitz, Dubowitz, Morante, & Verghote, 1980; Bloch, 1983). Fielder, Foreman, Moseley, and Robinson (1993) thus suggest that preterms with few medical problems might benefit from early visual experience.

Infant attention patterns have been reported to be related to later intellectual functioning (see e.g., Colombo, 1993; Yarrow, Klein, Lomonaco, & Morgan, 1975; Fagan & McGrath, 1981), and this connection has been described for preterm infants as well (Cohen & Parmelee, 1983; Sigman et al., 1986). Preterm infants have been shown to be at risk for problems concerning their intellectual and attentional functioning in later childhood (Breslau & Chilcoat, 2000; Anderson et al., 2003). This underlines the importance of examining the development of attention closely and exploring possible early differences between high-risk infants and healthy controls.

Shifts of Gaze in Preterm InfantsThere are several studies which have compared full- and preterm infants’ local-

ization of stimuli. Tasks which examine these simple shifts of gaze first attract the infant’s gaze to a fixation point. Then this central fixation stimulus is extinguished as a new target appears in the periphery. The frequency and latency of the infant’s localization of the peripheral stimulus is assessed and provides information on the infant’s sensory-motor processing.

Several studies have shown that preterm infants are more efficient in shifting their gaze as they moved their gaze to a peripheral target more frequently (Butcher, Kalverboer, Geuze, & Stremmelaar, 2002) and with shorter latencies (Foreman, Fielder, Price, & Bowler, 1991; Atkinson, 2000; Butcher et al., 2002) than their full-terms age mates. However, these studies also demonstrated that the gain from early visual ex-perience was only temporary. When infants were older than 4 to 6 weeks of corrected age, pre- and full-terms’ gaze shifting behavior was about equally efficient (Foreman et al., 1991; Butcher et al., 2002; Atkinson, 2000). Preterm infants with more severe medical complications did not substantially benefit (Butcher et al., 2002) or performed even more poorly than full-terms (Landry, Leslie, Fletcher, & Francis, 1985). On the other hand, there are also studies which describe low-risk preterm infants being less efficient in their gaze shifting behavior (Friedman, et al., 1981; Masi & Scott, 1983; Vervloed, 1995).

Shifts of Gaze from an Attended Stimulus (Disengagement) in Preterm InfantsThe results of studies addressing the development of gaze shifts away from an

attended stimulus in preterm infants are also rather inconsistent. Disengagement is investigated using tasks in which a fixation stimulus is first presented in the central visual field of the infant, before a target in the periphery is added. In order to look

102

Chapter 5

103

Pre- and Full-term Infants’ Shifts of Attention and Gaze

away from the persisting stimulus in the center and shift gaze to the periphery, in-fants have to disengage their attention and gaze. Such tasks thus measure both the attentional processes required for disengagement and the sensory-motor processing which is necessary to carry out and eye movement to the competing stimulus in the periphery.

We know of only two studies that examined disengagement of attention in pre- and full-term infants by means of such a specifically designed task. Butcher et al. (2002) studied the development of simple gaze shifts and disengagement in high- and low-risk preterm infants and full-term controls. They reported that low-risk preterms were faster than their full-term age mates in disengaging their attention and gaze during the first 10 weeks after the term date, but lagged behind from 16 weeks on and were significantly slower by 6 months. However, there were no significant differences in disengagement latency between high-risk preterms and full-terms. Concerning the frequency of gaze shifts, the three groups did not differ, but staring behavior persisted slightly longer in the preterm groups. Atkinson (2000) also found shorter latencies of gaze shifts in 4- to 5-week old in relatively healthy preterms. However, Butcher et al. (2002) and Atkinson (2000) interpret their results in relation with their findings on simple gaze shifting: As they also found shorter latencies of simple shifts of gaze, they suggest that disengagement latencies might be shorter in preterm infants due to faster sensory-motor processing. The authors accordingly conclude that mainly early developing sensory and motor processes benefited from early visual experience, but not the later maturing attentional processes.

The Underlying Mechanisms of Disengagement of Attention in InfancyResearchers have explained the disengagement difficulty during early infancy in

different ways. Rothbart, Posner, and Rosicky (1994) suppose that gaze shifts are pre-ceded by a covert shift of attention and accordingly explain disengagement difficulty as problems shifting attention covertly. Johnson (1990) on the other hand, suggests that obligatory attention reflects the inability to generate an eye movement to a pe-ripheral target while processing a stimulus in the central visual field. According to Hood (1995), however, disengagement difficulty is caused by problems breaking gaze from fixation.

Previous studies have shown that the characteristics of the stimuli used in a dis-engagement task influence the infants’ gaze shifting behavior. Both, the stimulus currently under attention and the stimulus in the peripheral visual field affect the frequency and latency of looks to the periphery, and this influence has been shown to change with age (see Chapter 3; Butcher et al., 2000). Whether and how quickly infants carry out gaze shifts is influenced by the physical salience of the stimuli (Finlay & Ivinskis, 1984; Tronick, 1972), but also by higher level attributes of the stimuli such as social meaningfulness (see Chapter 3).

102

Chapter 5

103

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Aims of the StudyThe results reported in the literature on gaze and attention shifting in pre- and

full-term infants are far from being exhaustive and consistent. The aim of this study thus was to compare the development of disengagement and simple gaze shifts in preterm and full-term infants.

As both simple gaze shifts and disengagement were examined, an experiment with non-competition and competition tasks was created. In a non-competition situation simple shifts of gaze are studied, elicited by a central and a peripheral stimulus, which are presented successively. In the competition condition, on the other hand, the central stimulus persists after the peripheral one appears, and the infant has to disengage his or her attention first, before the gaze can be shifted to the periphery.

As fundamental gaze and attention shifting mechanisms are known to develop during the first few months after the term date, the infant participants of this study were followed between 6 and 26 weeks of (corrected) age. An intense longitudinal design with 6 measurement points was chosen. Young infants – and especially infants with diverse medical histories – tend to differ considerably in their performance on visual tasks. A longitudinal design, however, reduces the inter-individual variance between measurement points and increases the power.

Earlier research has shown that the sort of stimuli used influenced the gaze shift-ing behavior in a disengagement task. Infants shifted their gaze less frequently and more slowly when looking away from an abstract stimulus to a face in the periphery than in the opposite condition (see Chapter 3). Preterm infants, however, have been shown to be visually less responsive to faces than their full-term age mates. They look less at their mother’s (Field, 1977, 1979; Barratt, Roach, & Leavitt, 1992) and a stranger’s face (Masi & Scott, 1983) and are slower in orienting toward those faces (Masi & Scott, 1983). It was thus the second goal of this study to investigate the infants’ gaze shift-ing between different stimuli and explore whether the influence of the stimuli was different in preterm than in full-term infants.

METHOD

ParticipantsTwenty full-term infants (12 girls; 8 boys) and ten preterm infants (4 girls; 6 boys)

participated in the longitudinal study. The parents of the infants were informed about the purpose of the research and gave their written consent. The study was approved by the local Medical Ethics Committee.

Measurements started when infants were about 6 weeks old and continued every four weeks until the age of 26 weeks. Thus, measurement sessions were conducted at 6, 10, 14, 18, 22, and 26 weeks of age. Dates were determined on basis of the corrected age of every infant, that is, ages were calculated from the due date. If the infants were unable to carry out the experimental task due to fussiness or sleepiness, a second ap-pointment was scheduled within 7 days.

104

Chapter 5

105

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Full-term infants. The mothers of the full-term infants were approached through childbirth education classes, midwives, or gym classes. Their babies were born after a gestation period of 37 to 42 weeks (M = 40.1 weeks; SD = 1.3), had a birth weight above 2800 g (M = 3440 g; SD = 281), and no history of pre- and perinatal complications. All infants of the full-term group scored within their age range on the Bayley Scales of Infant Development (BSID-II; Bayley, 1993) at 12 and 24 weeks of age.

Mean ages at the measurement sessions for the full-term group were 47.7 days, 73.6 days, 103.7 days, 131.1 days, 158.9 days and 188.8 days. Despite attempts to retest after a unsuccessful test date, for 7 infants one test session was missing.

Preterm infants. All preterm infants were admitted to the neonatal intensive care unit (NICU) of the University Hospital Groningen. Clinical data are provided in Table 5.1. The infants’ gestational age ranged from 27.3 to 32.4 weeks (M = 29.6 weeks; SD = 1.8). They had a birth weight between 640 and 2035 g (M = 1267 g; SD = 468). Two of the preterm babies were small for gestational age (SGA, according to Kloosterman, 1970). All preterm infants were screened for retinopathy of prematurity (ROP) around the term age. None suffered of ROP more severe than grade 1. Two infants went through an infection during the neonatal period, indicated by positive blood cultures. To as-sess the possible influence of neonatal illnesses a clinical scoring system was applied at discharge, the Nursery Neurobiologic Risk Score (NBRS; Brazy, Eckerman, Oehler, Goldstein, & O’Rand, 1991). The averaged NBRS of the preterm group was M = 3.3 (SD = 2.0). Two infants had a NBRS of 6, which is associated with an increased risk for de-velopmental problems (Brazy et al., 1991).

Of all preterms brain ultrasound scans were made within the first week after birth and at weekly intervals thereafter until abnormalities had disappeared on two subsequent scans. The scans were interpreted by two radiologists experienced in neo-natal radiology. Five infants had germinal matrix hemorrhages (GMH), four of them grade 1 and one GMH grade 2 (according to Papile, Burstein, Burstein, & Koffler, 1978). In addition, the infant with grade 2 hemorrhage had a venous infarction in the left temporal region. Six infants had periventricular echodensities (PVE) with a duration of more than one week (grade 1 periventricular leukomalacia, PVL, according to de Vries, Eken, & Dubowitz, 1992). The duration of prolonged PVE varied from less than 2 to 5 weeks. None of the infants had cystic periventricular leukomalacia.

Follow-up examinations were performed at the outpatient department at regular intervals. They consisted of a pediatric and a neurological examination. All of the infants developed within the normal range. Two had mild developmental anomalies, especially concerning their motor development. For details of the clinical data of each individual infant (i.e. obstetrical and neonatal variables, findings of the ultrasound scans, outcome), see Table 5.1. Taking into account the developmental outcomes of the preterm infants, the group can be qualified as reasonably homogenous, although the obstetrical and neonatal measures show that the infants have rather diverse clinical histories.

104

Chapter 5

105

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Tabl

e 5.1.

Clin

ical

dat

a of

the

grou

p of

pre

term

infa

nts.

Brai

n ul

tras

ound

find

ings

Case

no

.Se

xGA

in

w

eeks

Birt

h w

eigh

t in

gSG

AAp

gar

scor

e 1’

- 5’

Days

on

IP

PVCL

DN

BRS

Dura

tion

PV

EAd

diti

onal

find

ings

Out

com

e

fron

tal

pari

eto-

occi

pita

l

1F

27.2

964

0+

1 - 1

13+

50

wk

3 w

kGM

H g

r 1

L; P

VL

gr 1

fron

tal c

ysts

(pre

nata

l)CA

27

m: s

trab

ism

us,

gros

s mot

or a

nd c

oord

i-na

tion

pro

blem

s,Ba

yley

MDI

108

, PDI

108

2M

29.7

113

25-

1 - 6

11+

20

wk

0 w

k-

CA 1

4 m

: nor

mal

3M

30.7

117

30-

9 - 8

5-

30

wk

< 1

wk

GMH

gr

1 R

CA 2

0 m

: nor

mal

4F

28.7

194

5-

4 - 8

5-

60

wk

3 w

kGM

H g

r 1

L; P

VL

gr 1

CA 1

9 m

: nor

mal

5M

31.5

718

75-

8 - 9

3-

60

wk

2 w

kGM

H g

r 2;

PV

L gr

1ve

nous

infa

rcti

on L

tem

pora

lCA

14

m: n

orm

al

6F

28.4

386

0-

7 - 8

14+

55

wk

2 w

kGM

H g

r 1;

PV

L gr

1CA

19

m: n

orm

al

7M

28.0

010

60-

9 - 7

2-

2<

1 w

k3

wk

-CA

22

m: n

orm

al

8F

32.4

312

30+

1 - 4

5-

10

wk

< 1

wk

-CA

12

m: n

orm

al

9M

28.1

497

0-

7 - 9

1-

12

wk

2 w

kPV

L gr

1CA

12

m: n

orm

al

10M

31.4

320

35-

9 - 9

2-

20

wk

1 w

kPV

L gr

1?

vent

ricu

lar

asym

met

ry(R

>L, o

f pre

nata

l ons

et)

CA 1

2 m

: min

or n

euro

-lo

gica

l dys

func

tion

, m

ild d

iple

gia?

Note

. F fe

mal

e, M

mal

e; G

A ge

stat

iona

l age

; SGA

sm

all f

or g

esta

tion

al a

ge; I

PPV

inte

rmit

tent

pos

itiv

e pr

essu

re v

enti

lati

on; C

LD c

hron

ic lu

ng d

isea

se (i

.e. s

uppl

em

enta

ry o

xyge

n re

quir

ed a

t 36

wee

ks p

ostm

enst

rual

age

); N

BRS

Nur

sery

Neu

robi

olog

ic R

isk

Scor

e; P

VE

peri

vent

ricu

lar

echo

dens

itie

s; w

k w

eek/

wee

ks; G

MH

ge

rmin

al m

atri

x he

mor

rhag

e; P

VL

peri

vent

ricu

lar

leuk

omal

acia

; gr

grad

e (a

ccor

ding

to P

apile

, Bur

stei

n, B

urst

ein,

& K

offler

(197

8, G

MH

) and

de

Vri

es, E

ken,

Du

bow

itz

(199

2, P

VL)

); L

left

; R r

ight

; CA

corr

ecte

d ag

e; m

mon

ths.

106

Chapter 5

107

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Parents who had indicated their interest in participating in this study together with their baby were contacted after the infants were released from the NICU. For practical reasons, only families who lived up to 50 km away were admitted to the study. They were offered free transportation to the institute and back home. From the group of preterm infants, four infants missed one of the six test sessions. Mean ages at each measurement session were 47.8 days (SD = 5.6), 74.9 days (SD = 6.5), 102.1 days (SD = 5.2), 132.0 days (SD = 6.1), 157.5 days (SD = 3.5), and 183.2 days (SD = 2.6).

ProcedureMeasurement sessions were planned for a time of the day when parents expected

their babies to be alert for 20 to 30 minutes. In the beginning of the test session, infants were given some time to become accustomed to the new environment. When they were in state 3 or 4 of Prechtl’s scale of alertness (awake, eyes open, some spontaneous movements, no crying; Prechtl & Beintema, 1964), the experiment was started.

Infants were presented with a so-called disengagement task. In such a task stimuli can appear at three different positions on a screen: in the center, on the left, or on the right. In this study, the stimuli to the left or right were displayed at 20 degrees eccentricity. The task included two different sorts of trials: competing and non-com-peting trials. All trials started with the appearance of a stimulus in the center of the monitor, the fixation stimulus. The onset of this stimulus was accompanied by a short melody to attract the infant’s attention. After the infant had been fixating the central stimulus for 1 - 2 seconds, the peripheral stimulus was displayed. While in non-com-petition trials the central stimulus disappeared when the peripheral target came up, in competition trials it persisted also after the peripheral target had appeared. After 5 seconds, the stimuli disappeared simultaneously. The screen remained blank for a period of 2.5 seconds, before the following trial began. In Figure 3.1 (see Chapter 3), a schematic representation of the disengagement task is given. Competition trials thus demanded disengagement of attention and gaze from the fixated central stimulus before an eye movement to the peripheral target could be generated. Non-competition trials did not require disengagement, but a simple shift of gaze to the newly appeared stimulus in the periphery. They also serve as control trials for the different stimulus combinations: In a localization task stimuli of about equal physical attractiveness and detectability should not influence the probability or latency of shifting gaze.

ApparatusDuring the tasks the infants were sitting in an infant-seat on a table in a reclined

posture (about 45 degrees) with a light support for their head. In front of the infant-seat a 21 inch monitor was suspended from the ceiling approximately central and perpendicular to the line of gaze. The distance between the monitor and the baby’s eyes was about 35 cm. Only the screen of the monitor was visible. The monitor itself, the equipment necessary to run the tasks and record eye movements, and the experi-

106

Chapter 5

107

Pre- and Full-term Infants’ Shifts of Attention and Gaze

menter were concealed behind a gray curtain, which filled 180 degrees of the infant’s visual field. The eye movements and the display of the monitor were shown on a SVHS monitor, which allowed the experimenter to run the task on the basis of the babies’ behavior. The infant’s face and eye movements were recorded on video during the experiment. The eye movements were scored off-line from the video recording.

Stimulus Material and Stimulus PresentationTwo different dynamic stimuli were used within the disengagement task: an ab-

stract stimulus and a socially relevant stimulus. The social stimulus consisted of a short video sequence of the face of each infant’s mother. This video recording was made during a first visit of mother and baby to the lab. The mother’s face was recorded while she was smiling and talking with her baby as she normally did. The abstract stimulus was derived from the video of the mother by carrying out several transformations in a graphic computer program (Corel PHOTO-PAINT 9). This procedure ensured that the two stimuli resembled each other regarding their dynamics, color range, etc., but were completely different with respect to their meaning to the infant. Some frames from each type of video are given in the schematic representation of the task (see Figure 3.1, Chapter 3). At the viewing distance of 35 cm, all stimuli subtended a visual angle of 10 by 10 degrees. Thus, the face was much smaller than it would appear to the baby in a normal interaction. As even newborns can recognize a well-known face after a size change (Walton, Armstrong, & Bower, 1997), we expected that the infants would not to have any problems with the size of the face. This was supported by observations during the experiment that infants were very interested in the stimuli and were oc-casionally smiling to the mother stimulus but not to the abstract stimulus.

The experiment contained 32 competition trials and 8 non-competition trials. As reactions to the peripheral stimulus in the non-competition condition were expected to be less variable, fewer non-competition trials were considered to be necessary to provide reliable measurements. Both stimulus types could appear as central stimulus or as peripheral target, which resulted in four different stimulus combinations (face-face, face-abstract, abstract-face, abstract-abstract). These stimulus combinations were presented with equal frequency. Peripheral targets appeared as often on the left as on the right side, and the order in which the trials were presented was randomized. An experimental run normally lasted about 12 minutes. Whenever the infant started fuss-ing or crying or became sleepy, the testing session was interrupted for some time.

AnalysisBehavioral coding. The video recording of the infant’s face was played back half-

frame by half-frame (20 ms intervals) in order to code the eye movements. The direc-tion and the latency of the first eye movement after the appearance of the peripheral stimulus were scored. Also, the frequency of trials when the infant did not move his or her eyes was calculated. Trials in which the infant was not fixating the first stimulus

108

Chapter 5

109

Pre- and Full-term Infants’ Shifts of Attention and Gaze

when the peripheral stimulus appeared were excluded from the analysis, for example, when the infants had already averted their gaze from the central stimulus or had their eyes closed for various reasons (e.g., blinking, yawning, fussing). Eye movements starting less than 200 ms after the second stimulus appeared, were considered an-ticipatory (Haith, Hazan, & Goodman, 1988) and also dropped from the analyses. All eye movements following the appearance of the peripheral stimulus but not leading to its localization were considered to be errors.

The data were coded by different observers, who had been trained by the first au-thor. Approximately 10% of the sessions were double-coded. Interobserver reliability for the onset of an eye movement was found to be on average 87.2% (range 62.5% to 100%). Cohen’s kappa for the category of first eye movement was on average .87 (range .63 to 1.0).

Statistical analysis. The relative frequency of looks to the peripheral target and the frequency of no eye movement occurring were calculated for each test session and condition. To calculate the latencies of looks, the time differences between the ap-pearance of the peripheral stimulus and the onset of an eye movement to this target were calculated. The median reaction times for each infant, session, and stimulus combination were determined. A plot of all median reaction times revealed that the distribution of the raw data was positively skewed. Therefore a square root transfor-mation of the latency data was executed (Rummel, 1970) before the statistical analyses were carried out. For ease of understanding, the averaged response latencies reported in the text and the reaction times displayed in the figures are given in seconds.

The data were analyzed using a multilevel modeling technique (Snijders & Bosker, 1999; Woodhouse, 1996). Multilevel analysis is a regression procedure for data with a hierarchical structure and complex patterns of variability. When applied to longitudi-nal data, the repeated measures are regarded as “nested” within individuals. Unlike a standard multiple regression model, a multilevel model contains more than one error term: one for every level of the hierarchical data. The model also allows intercept and slope coefficients to vary randomly which means that the association between session scores and explanatory variables may differ between individuals. Another strength of this approach is that it allows for both the number of observations per individual and the spacing of the observations in time to vary.

Multilevel analysis was used to examine whether (a) the frequency and latency of looks and frequency of trials in which the infant did not perform an eye movement changed with age, (b) the combination of stimuli influenced the frequency and latency of looks, and (c) there were differences between the full- and preterm group in the developmental trajectories and in the effects of stimulus combinations. The data of the competition and the non-competition trials were analyzed separately and treated as different subsets of the data. In order to test whether there were differences be-tween the preterm and the full-term group, the variable “prematurity” was added as a 0-1 dummy variable to the models. Also coefficients for the interaction effects were

108

Chapter 5

109

Pre- and Full-term Infants’ Shifts of Attention and Gaze

included. Furthermore, it was tested whether the residual variance was larger in one of the two infant groups.

All models had three levels: infant, test session, and stimulus combination. Based on earlier research (see e.g., Chapter 3; Butcher et al., 2002), a model of three piecewise linear functions was fit to the data. These studies suggested rapid development for the period from 6 to 9 and from 9 to 16 weeks and a stabilization for the period between 16 and 26 weeks of age. As shifts of gaze to the peripheral target at 6 weeks were very infrequent, reliable latency data were not available for the first measurement point. A model of two piecewise linear functions was therefore used for the reaction time data, with one function from 9 to 16 weeks and another from 16 to 26 weeks. Infants’ test sessions differed in terms of the number of trials carried out as some sessions had to be terminated early due to infants’ fussing or getting sleepy. This was controlled for by adding a correction term to the models. As the infants’ real ages were entered into the model rather than age categories, the variable age could be treated as a con-tinuous variable which provided information on infants’ behavior also between the measurement points.

For the analyses, the data were centered around 12 weeks of age, which was about the middle of the period in which the largest change was expected. Variables for the different age periods and the different stimulus combinations were added to the equation in order to predict the frequency of looks or the reaction times. T tests were implemented to determine the statistical significance of the coefficients. The fit of the model and its improvement was examined using χ2 tests of deviance. T tests were also used as post-hoc tests. Whenever multiple t tests were carried out, Bonferroni corrections were implemented to keep alpha at .05 (Stevens, 1992).

RESULTS

Frequency of Shifts of Gaze to the Peripheral StimulusThe mean group frequencies of looks to the peripheral stimulus in competition

and non-competition trials across ages are depicted in Figure 5.1. In the competi-tion condition, the frequency of looks increased gradually from 6.5% (SD = 8.3) at 6 weeks to 61.9% (SD = 22.8) at 14 and 88.0% (SD = 9.1) at 26 weeks of age. However, in the non-competition condition infants shifted their gaze in 57.2% (SD = 31.0) of the trials already at 6 weeks and in 85.9% (SD = 20.8) of the trials at 10 weeks of age. Between 6 and 14 weeks of age, the frequencies of the non-competition condition were signifi-cantly higher than those of the competition condition (6 weeks, t(19) = -7.89, p < .01; 10 weeks, t(29) = -11.27, p < .01; 14 weeks, t(29) = -6.91, p < .01). From 18 weeks on, the gaze shifting frequencies of the competition situation started to approach those of non-competition situations (18 weeks, t(27) = -2.54, p > .10; 22 weeks, t(26) = -3.00, p < .05; 26 weeks, t(28) = .09, p > .10). However, the developmental trajectories of looks to the periphery were highly similar in preterm and full-term infants.

110

Chapter 5

111

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Figure 5.1. The mean frequency and standard error of looks to the peripheral target in competition and non-competition trials for the preterm (PT) and the full-term (FT) group.

Competition trials. As described earlier, the frequency of gaze shifts to the peripheral target in the competition condition increased throughout the three age periods. The model of the competition trial data correspondingly contained significant age coef-ficients for the age period of 6 to 9 weeks (t(163) = 2.57, p < .05), 9 to 16 weeks (t(163) = 11.64, p < .001), and 16 to 26 weeks (t(163) = 5.23, p < .001). However, the coefficient of age for 9 to 16 week function was significantly larger than the one of the 16 to 26 week function (slope coefficients β9-16 = 1.09 versus β16-26 = .28.), which indicates a more gradual increase in shifts of gaze after 16 weeks of age. This was confirmed by the significant gain in fit which was obtained when using separate age coefficients for the periods of 9 to 16 and 16 to 26 weeks compared to only one age coefficient for the period of 9 to 26 weeks (χ2(1) = 38.20, p < .001). Consistent with this, post-hoc comparisons of mean group frequencies of looks in consecutive sessions were still significant between 18 and 22 weeks (t(24)= -3.87, p < .01), whereas between 22 and 26 weeks of age there was no significant increase demonstrable anymore (t(26) = -.88, p > .10).

In Figure 5.2, the development of the frequency of looks to the peripheral target is depicted for the different stimulus combinations. The effect of the different stimulus combinations was tested by adding three of the four possible combinations as a 0-1 dummy variable to the model and contrasting them against the fourth category, the combination face-face. There were two significant main effects of stimulus combina-tion: When controlling for age, the combination face-abstract elicited significantly more shifts of gaze to the peripheral stimulus than the reference combination face-face (t(648) = 6.15, p < .001), whereas under the condition abstract-face less gaze shifts were observed (t(648) = -4.24, p < .001). No significant difference was found between

110

Chapter 5

111

Pre- and Full-term Infants’ Shifts of Attention and Gaze

the stimulus combinations face-face and abstract-abstract.

Figure 5.2. The mean frequency and standard error of looks in competition trials per stimulus combina-tion between 6 and 26 weeks of age.

There were three significant interactions between stimulus combination and age period: During the age period of 6 to 9 weeks, the increase in gaze shifting frequency was significantly smaller for the stimulus combination abstract-face (t(648) = -2.69, p < .01) and greater for the combination face-abstract (t(648) = 2.43, p < .05) than it was for the reference combination face-face. While the frequency of looks in the stimulus combination face-abstract increased from 7.5% (SD = 18.4) to 46.2% (SD = 27.6) between 6 and 10 weeks, in the abstract-face combination it grew from 11.0% (SD = 16.8) to only 20.8% (SD = 18.8) in the same period of time. The third interaction, the interaction of the stimulus combination face-abstract with the age period of 16 to 26 weeks (t(648) = -3.90, p < .001) implies that in the last age period the frequency of looks changed less when the central stimulus was a face and the peripheral stimulus was abstract than in the face-face condition. As can also be seen in Figure 5.2, the frequency of gaze shifts in the face-abstract condition had almost already reached its maximum by 18 weeks (18 weeks, M = 82.2%, SD = 5.0; 22 weeks, M = 86.2%, SD = 3.5; 26 weeks, M = 92.0%, SD = 2.7), which resulted in the smaller increase throughout the 16 to 26 weeks age period.

The analysis of random effects revealed that the residual variance depended on whether the infant belonged to the preterm or full-term group (χ2(1) = 5.10, p < .05). The effect was caused by a larger variance in performance of preterm compared to the full-

112

Chapter 5

113

Pre- and Full-term Infants’ Shifts of Attention and Gaze

term infants. Further, a significant estimate of intercept variance (χ2(1) = 16.88, p < .001) was found. This indicates that the frequency of gaze shifts differed significantly between infants at 12 weeks of age. Further, there were significant differences between infants between 16 and 26 weeks concerning the slope of the function (χ2(1) = 6.18, p < .05).

Figure 5.3. The mean frequency and standard error of looks in competition trials per stimulus combina-tion for (A) the preterm and (B) the full-term group.

112

Chapter 5

113

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Figure 5.3 shows the developmental trajectories for the different stimulus combi-nations for both (A) the preterm and (B) the full-term group. The common model of the preterm and full-term infants yielded no significant differences between the two groups, and there were no significant interaction effects involving the variable pre-maturity, either. This indicates similar developmental trajectories in both groups.

However, although the group of full-term infants and the group of preterm infants did not differ concerning their frequency of looks to a peripheral target, there are some results which suggest that disengagement of attention was less efficient in the preterm group. The frequency of the lack of an eye movement in reaction to the newly appeared peripheral stimulus in pre- and full-term infants is displayed in Figure 5.4. While 6-week-old preterm infants did not shift their gaze away from the central stimulus in 86.7% (SD = 32.4) of the trials, full-term infants of the same age kept on staring to it in only 61.3% (SD = 8.8). The multilevel model predicting the frequency of trials in which the infant did not perform an eye movement correspondingly contained a significant interaction of the first age period and the variable prematurity (t(162) = -2.11, p < .05). A t test for independent samples as post-hoc test also revealed a significant difference in the frequency of staring at 6 weeks of age (t(19.26) = -2.87, p < .05).

Figure 5.4. The mean frequency and standard error of staring in competition and non-competition trials for the preterm (PT) and the full-term (FT) group.

Non-competition trials. As can be seen in Figure 5.1, the overall percentage of shifts of gaze to the peripheral target in non-competition trials increased significantly be-tween the first measurement session at 6 and the second measurement session at 10

114

Chapter 5

115

Pre- and Full-term Infants’ Shifts of Attention and Gaze

weeks of age. No significant changes in gaze shifting frequency occurred after the age of ten weeks. This was indicated by the variable age having a significant coefficient for the 6 to 9 week function (t(162) = 4.45, p < .001), but not for the 9 to 16 week function and the 16 to 26 week function.

Figure 5.5. The mean frequency and standard error of looks in non-competition trials per stimulus combination for (A) the preterm and (B) the full-term group.

114

Chapter 5

115

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Infants differed in the rate at which gaze shifts increased between 6 and 9 weeks of age as the significant slope variance for the 6 to 9 week function (χ2(1) = 7.63, p < .01) indicates.

However, there was no effect of stimulus combination on the frequency of looks. The frequency of gaze shifts in the non-competition trials did not depend on whether an infant was born prematurely or not, either. This can be seen in Figure 5.5, which depicts the development of gaze shifts to the peripheral target for both (A) the preterm and (B) the full-term group.

Like in the competition condition, there were also in this simple gaze shifting task indications that preterm infants were performing less efficiently than full-term infants during the first measurement session. As Figure 5.4 shows, at 6 weeks of age, the pre-term infants kept on staring to the center of the monitor even after the stimulus was gone and the peripheral target had appeared in 15% (SD = 9.57) of the trials, whereas the full-term infants did so only in 9.53% (SD = 7.53) of the trials. This occurred mainly when the young preterm infants were supposed to look from the mother stimulus to the abstract one (t(595) = -4.08, p < .001) or the other way around (t(595) = -2.52, p < .05).

Figure 5.6. The mean latency and standard error of looks to the peripheral target in competition and non-competition trials for the preterm (PT) and the full-term (FT) group.

Latency of Shifts of GazeThe median latency of looks to the peripheral target in the competition and

non-competition situation for both infant groups is displayed in Figure 5.6. The first

116

Chapter 5

117

Pre- and Full-term Infants’ Shifts of Attention and Gaze

measurement point was excluded from the analysis, as it was based on only very few data points.

As can also be seen from Figure 5.6, infants shifted their gaze faster in non-com-petition trials than in competition trials. Although the median reaction times of the competition and non-competition condition were converging throughout the testing period, they were significantly different at young ages (10 weeks, t(25) = 5.46, p < .01; 14 weeks, t(26) = 4.67, p < .01) as well as at the later measurement points (18 weeks, t(26) = 6.87, p < .01; 22 weeks, t(26) = 4.11, p < .01; 26 weeks, t(28) = 2.99, p < .05).

Competition trials. The median reaction time of both preterm and full-term infants dropped throughout the entire testing period. While it took infants of 10 weeks on average more than a second to shift their gaze, this time decreased throughout the measurement period to about 500 ms. The coefficient of the variable age was significant for the age period of 9 to 16 (t(136) = -5.41, p < .001) and 16 to 26 weeks (t(136) = -3.49, p < .001). The two age coefficients (β9-16 = -.0072, β16-26 = -.0022) differed significantly from each other (χ2(1) = 14.26, p < .001), suggesting a less rapid decline during the last age period.

The multilevel analysis revealed a significant effect of group (t(28) = -2.52, p < .05) with the preterm infants exhibiting shorter latencies than the full-term infants when controlling for age. In the age period between 16 and 26 weeks, the interaction of the variables prematurity and age was significant (t(136) = 2.43, p < .05), indicating that the effect of preterm infants exhibiting shorter latencies disappeared as infants grew older. However, when three-way interactions were added to the model, the interaction of prematurity, the stimulus combination abstract-abstract, and the age period from 16 to 26 weeks yielded a significant effect (t(524) = 2.30, p < .05), while the two-way interaction between the variables age and prematurity remained a trend (t(136) = 1.91, p < .10). In Figure 5.7 the change in median reaction times of the full-term and the preterm group is depicted for the four different stimulus combinations. In all stimulus combination conditions the preterm infants showed faster gaze shifts than their full-term age mates during the first few months of the measurement period, but after 16 weeks this effect diminished. This was particularly the case for the stimulus combination abstract-abstract: Here, the gaze shifting latency in preterm infants from the age of 16 weeks on age tended to be even larger than in full-term infants of the same age.

The multilevel analysis also revealed a main effect of the stimulus combination abstract-face (t(525) = 3.41, p < .001). It indicates that – when controlling for age and prematurity – infants were slower in shifting their gaze away from an abstract stimulus to a face in the periphery compared to the reference condition face-face.

The estimate of slope variance was significant for the 9 to 16 (χ2(1) = 7.05, p < .01) and the 16 to 26 weeks period (χ2(1) = 3.22, p < .10). Infants thus differed considerably concerning the rate at which their gaze shifting latency decreased. Again, the residual variance depended on whether the infant belonged to the preterm or full-term group

116

Chapter 5

117

Pre- and Full-term Infants’ Shifts of Attention and Gaze

(χ2(1) = 19.48, p < .001). However, in this case preterm infants showed a smaller variance in performance than the full-terms.

Figure 5.7. The mean latency and standard error of looks to the peripheral target per stimulus combination in competition trials for the preterm (PT) and the full-term (FT) group.

118

Chapter 5

119

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Non-competition trials. As can be seen in Figure 5.6, also in the non-competition situation the median reaction time decreased significantly throughout the measure-ment period. The age coefficients of the 9 to 16 (t(154) = 4.18, p < .001) and the 16 to 26 weeks period (t(154) = 4.73, p < .001) yielded significance. However, post-hoc t tests revealed that the gaze shifting latency decreased between 14 and 18 weeks of age (t(26) = 4.22, p < .01), but not between 18 and 22 (t(23) = -1.16, p > .10) and 22 and 26 weeks (t(25) = -1.52, p > .10) of age anymore. This indicates that the decrease in median reaction time came to an end as infants grew older. There were no main effects of stimulus combination or group effect and no interaction effects, either. In the analysis of the random effects, the estimate of the intercept variance was significant (χ2(1) = 5.59, p < .05).

DISCUSSION

The Development of Simple Shifts of Gaze and DisengagementThe overall developmental changes of attention and gaze shifting are – taken by

and large – in accord with earlier results. While infants shifted their gaze effectively in the non-competition condition already from 10 weeks on (see e.g., Butcher et al., 2000), the persistence of a central stimulus as in the competition condition lowered the probability of looks to the peripheral stimulus in younger infants (see e.g., Aslin & Salapatek, 1975; Harris & MacFarlane, 1974). At 18 and 26 weeks of age, there were no differences in gaze shifting frequency anymore between the competition and the non-competition condition. The latency of gaze shifts decreased throughout the measure-ment period in both conditions. Saccadic reaction times in situations with competing stimuli were longer than in non-competing situations both in infants younger than 3 months as demonstrated also by Matsuzawa and Shimojo (1997) and in older infants as shown before by Hood and Atkinson (1993).

Differences in Gaze Shifting Behavior between Preterm and Full-term InfantsIt was the goal of this study to examine differences in gaze shifting behavior

between the preterm and the full-term group under different conditions: in situa-tions with competing and non-competing stimuli and with different combinations of stimuli.

Simple gaze shifting (without disengagement). Concerning simple shifts of gaze, there were few differences found between the full-term and the preterm group. Both groups showed about the same frequencies of looks to the periphery in the non-competition condition throughout the measurement period. The latencies of gaze shifts and their developmental trajectory were not significantly different for full- and preterms, either. However, at 6 weeks of age preterm infants kept on looking to the center of the display more often, even after the stimulus had disappeared and a peripheral one had come up. Although this effect was relatively small and occurred only in the face-abstract and the abstract-face condition, it suggests that young preterm infants were performing

118

Chapter 5

119

Pre- and Full-term Infants’ Shifts of Attention and Gaze

somewhat less efficiently than their full-term age mates. These results are in accord with studies which report gaze shifting problems in preterms, although mostly in the form of longer latencies (Masi & Scott, 1983; Landry et al., 1985; Friedman et al., 1981; Vervloed, 1995). However, the small disadvantage of the preterms was only temporary, and for the most part their performance and developmental trajectories were similar to full-term infants.

Shifts of gaze which require disengagement of attention. When infants had to disengage their attention before shifting their gaze to the peripheral stimulus, there were again no differences between full- and preterm infants concerning the frequency of gaze shifts and its development. But the young preterm infants had clearly more problems with sticky fixation than their full-term age mates: At 6 weeks, preterm infants tended to keep on looking to the central stimulus more often than their full-term age-mates did. This effect had disappeared by 10 weeks of age. In contrast, the preterm infants also showed an advantage in latency when they shifted their gaze to the periphery. They were faster in disengaging and shifting their gaze until about 16 weeks of age. Thus when young preterm infants overcame sticky fixation, which they did less fre-quently than their full-term age-mates, they tended to shift their gaze more quickly to the peripheral stimulus.

Variance in performance. The variance in task performance was different for the preterm and the full-term group. In the preterm group, the variance in performance concerning the frequency of disengagement was larger, while it was smaller for the latency of disengagement. As a larger variance can also be indicative of a worse per-formance, these findings fit with the rest of results: Preterm infants had difficulty in overcoming staring behavior, but performed more efficiently in generating eye movements.

The analyses also revealed significant differences in the frequency and latency of gaze shifting and especially in the rate at which these parameters developed between infants which is in accord with earlier findings (Butcher et al., 2000). Large inter-individual differences tended to occur during periods of rapid change.

Effects of Stimulus CombinationIn this study, different combinations of stimuli were used to examine the effect of

socially relevant and abstract stimuli on gaze and attention shifting. We have described in another study how healthy full-term infants reacted to a gaze shifting task with different stimulus combinations (see Chapter 3). Preterm infants have been reported to react differently to faces (Field, 1977, 1979; Masi & Scott 1982; Barratt et al., 1992) and unknown stimuli (Sigman, 1983; Rose et al., 1979) compared to full-term infants. However, in this study, a remarkable consistency of preterm infants’ reactions to the different stimulus combinations with the full-term infants was found:

No effects of stimulus combination were found in the non-competition condition concerning frequencies or latencies of gaze shifts. This was in agreement with the

120

Chapter 5

121

Pre- and Full-term Infants’ Shifts of Attention and Gaze

expectations, as in a task which only requires simple gaze shifts stimuli of compa-rable physical salience should not elicit differences in the frequency and latency of stimulus localization. However, in the competition condition infants disengaged their gaze more frequently when the central stimulus was a face and the peripheral stimu-lus was abstract and less frequently and also more slowly in the opposite condition (abstract-face). The interaction effect of the factors age and stimulus indicated that these differences were most pronounced when the ability to disengage gaze and at-tention was developing, but not yet well established. At 6 weeks of age, infants kept on looking to the central stimulus once they fixated it, regardless of what was ap-pearing in the periphery. From 22 weeks on, they shifted their gaze reliably between all four stimulus combinations. The frequency of looks for the stimulus combination face-abstract increased most rapidly and reached its final level already at 18 weeks, while for the stimulus combination abstract-face full capacity of disengagement was reached approximately 4 weeks later. Both stimuli – the central fixation stimulus and the peripheral target – influenced the frequency of looks to the periphery, which is concordant with results described by Finlay and Ivinskis (1984). The – by this time well-known (Barrera & Maurer, 1981) – infant’s mother’s face was less attractive than the relatively new, salient abstract stimulus for the preterm as well as the full-term group. However, while preterm infants in general tended to shift their gaze between two competing stimuli as fast or even faster than full-term infants, from 16 weeks of age on, they seemed to have some difficulties in looking away from an abstract stimu-lus. This can be interpreted as a sign of slower habituation to a new stimulus, which has been reported earlier in preterm infants (e.g., Sigman et al., 1986).

Extra Visual Input and Attentional DevelopmentWe know of two earlier studies which examined the development of simple gaze

shifts and disengagement in pre- and full-term infants (Butcher et al., 2002; Atkinson, 2000). They found shorter latencies of preterm infants’ gaze shifting in competition and non-competition situations and interpreted this effect as a benefit of experience in sensory-motor processing. Following their account, later developing skills, such as disengagement, should not benefit from extra visual input. But the results of this study suggest that the changes in the development of attention due to a preterm birth and early visual experience are not so clear-cut. While preterm and full-term infants did not differ concerning the latency of simple gaze shifts, additional visual input was associated with a short-lasting facilitation of the execution of shifts of gaze from fixation. However, in this study preterm infants were also shown to have more difficulty in overcoming staring behavior. This finding suggests that the triggering of a gaze shift and its execution are not indissolubly associated, but might be two rather independent processes.

120

Chapter 5

121

Pre- and Full-term Infants’ Shifts of Attention and Gaze

ConclusionAs described above, there have been differences in opinion on whether early visual

input is beneficial for an infant or whether it might be detrimental to the developing visual system. The results of this study do not support one of the two accounts, but suggest that extra visual experience changes the visual and attentional development of an infant in general with positive effects on the one hand and a negative impact on the other. However, the effects were shown to be temporary – by the end of the measurement period there were no relevant differences in attention and gaze shifting performance detectable anymore between the two groups of infants.

REFERENCESAnderson, P., Doyle, L. W., Callanan, C., Carse, E., Casalaz, D., Charlton, M. P., Davis, N.,

Duff, J., Ford, G., Fraser, S., Hayes, M., Kaimakamis, M., Kelly, E., Opie, G., Watkins, A., Woods, H., & Yu, V. (2003). Neurobehavioral outcomes of school-age children born extremely low birth weight or very preterm in the 1990s. Journal of the American Medical Association, 289, 3264-3272.

Aslin, R. N., & Salapatek, P. (1975). Saccadic localization of visual targets by the very young human infant. Perception and Psychophysics, 17, 293-302.

Atkinson, J. (2000). The developing visual brain. Oxford, UK: Oxford University Press.Barratt, M. S., Roach, M. A., & Leavitt, L. A. (1992). Early channels of mother-infant

communication: Preterm and term infants. Journal of Child Psychology and Psychiatry and Allied Disciplines, 33, 1193-1204.

Barrera, M. E., & Maurer, D. (1981). Recognition of mother’s photographed face by the three-month-old infant. Child Development, 52, 714-716.

Bayley, N. (1993). Bayley Scales of Infant Development (Second Edition). San Antonio, TX: The Psychological Corporation.

Bloch, H. (1983). La poursuite visuelle chez le nouveau-né à terme et chez le prématuré. Enfance, 1, 19-29.

Bonin, M., Pomerleau, A, & Malcuit, G. (1998). A longitudinal study of visual attention and psychomotor development in preterm and full-term infants during the first six months of life. Infant Behavior and Development, 21, 103-118.

Brazy, J. E., Eckerman, C. O., Oehler, J. M., Goldstein, R. F., O’Rand, A. M. (1991). Nursery Neurobiologic Risk Score: Important factor in predicting outcome in very low birth weight infants. Journal of Pediatrics, 118, 783-792.

Breslau, N., & Chilcoat, H. D. (2000). Psychiatric sequelae of low birth weight at 11 years of age. Biological Psychiatry, 47, 1005-1011.

Butcher, P. R., Kalverboer, A. F., & Geuze, R. H. (2000). Infants’ shifts of gaze from a central to a peripheral stimulus: A longitudinal study of development between 6 and 26 weeks. Infant Behavior and Development, 23, 3-21.

122

Chapter 5

123

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Butcher, P. R., Kalverboer, A. F., Geuze, R. H., & Stremmelaar, E. F. (2002). A longitudi-nal study of the development of shifts of gaze to a peripheral stimulus in preterm infants with transient periventricular echogenicity. Journal of Experimental Child Psychology, 82, 116-140.

Cohen, S. E., & Parmelee, A. H. (1983). Prediction of five-year Stanford-Binet scores in preterm infants. Child Development, 54, 1242-1253.

Colombo, J. (1993). Infant cognition: Predicting later intellectual functioning. Newbury Park, CA: Sage.

de Vries, L. S., Eken, P., & Dubowitz, L. M. S. (1992). The spectrum of leukomalacia using cranial ultrasound. Behavioural Brain Research, 49, 1-6.

Dubowitz, L. M., Dubowitz, V., Morante, A., & Verghote, M. (1980). Visual function in the preterm and fullterm newborn infant. Developmental Medicine and Child Neurol-ogy, 22, 465-475.

Fagan, J. F., & McGrath, S. K. (1981). Infant recognition memory and later intelligence. Intelligence, 5, 121-130.

Field, T. (1977). Effects of early separation, interactive deficits, and experimental manipulations on infant-mother face-to-face interaction. Child Development, 48, 763-771.

Field, T. (1979). Visual and cardiac responses to animate and inanimate faces by young term and preterm infants. Child Development, 2, 179-184.

Fielder, A. R., Foreman, N., Moseley, M. J., & Robinson, J. (1993). Prematurity and visual development. In K. Simons (Ed.), Early visual development, normal and abnormal (pp. 485-504). New York: Oxford University Press.

Finlay, D., & Ivinskis, A. (1984). Cardiac and visual responses to moving stimuli pre-sented either successively or simultaneously to the central and peripheral visual fields in 4-month-old infants. Developmental Psychology, 20, 29-36.

Foreman, N., Fielder, A., Price, D., & Bowler, V. (1991). Tonic and phasic orientation in full-term and preterm infants. Journal of Experimental Child Psychology, 51, 407-422.

Friedman, S. L., Jacobs, B. S., & Werthmann, M. W. (1981). Sensory processing in pre-and full-term infants in the neonatal period. In S. L. Friedman & M. Sigman (Eds.), Pre-term birth and psychological development (pp. 159-178). New York: Academic Press.

Haith, M. M., Hazan, C., & Goodman, G. S. (1988). Expectation and anticipation of dy-namic visual events by 3.5-month-old babies. Child Development, 59, 467-479.

Harris, P., & MacFarlane, A. (1974). The growth of the effective visual field from birth to seven weeks. Journal of Experimental Child Psychology, 18, 340-348.

Hood, B. M. (1995). Shifts of visual attention in the human infant: A neuroscientific approach. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 9, pp. 163-216). Norwood, NJ: Ablex.

Hood, B. M., & Atkinson, J. (1993). Disengaging visual attention in the infant and adult. Infant Behavior and Development, 16, 405-422.

122

Chapter 5

123

Pre- and Full-term Infants’ Shifts of Attention and Gaze

Hood, B. M., Murray, L., King, F., Hooper, R., Atkinson, J., & Braddick, O. (1996). Habitu-ation changes in early infancy: longitudinal measures from birth to 6 months. Journal of Reproductive and Infant Psychology, 14, 177-185.

Johnson, M. H. (1990). Cortical maturation and the development of visual attention in early infancy. Journal of Cognitive Neuroscience, 2, 81-95.

Kaye, K., & Fogel, A. (1980). The temporal structure of face-to-face communication between mothers and infants. Developmental Psychology, 16, 454-464.

Kloosterman, G. J. (1970). On intrauterine growth: The significance of prenatal care. International Journal of Gynaecology and Obstetrics, 8, 895-912.

Landry, S. H., & Chapieski, M. L. (1988). Visual attention during toy exploration in preterm infants: Effects of medical risk and maternal interaction. Infant Behavior and Development, 11, 187-204.

Landry, S. H., Leslie, N. A., Fletcher, J. M., & Francis, D. J. (1985). Visual attention skills of preterm infants with and without intraventricular hemorrhage. Infant Behavior and Development, 8, 309-321.

Masi, W. S., & Scott, K. G. (1983). Preterm and full-term infants’ visual responses to mothers’ and strangers’ faces. In T. M. Field & A. Sostek (Eds.), Infants born at risk: Physiological, perceptual and cognitive processes (pp. 173-180). New York: Grune & Stratton.

Matsuzawa, M., & Shimojo, S. (1997). Infants’ fast saccades in the gap paradigm and development of visual attention. Infant Behavior and Development, 20, 449-455.

Papile, L., Burstein, J., Burstein, R., & Koffler, H. (1978). Incidence and evolution of subependymal and intraventricular hemorrhage: A study of infants with birth weights less than 1,500 gm. Journal of Pediatrics, 92, 529-534.

Prechtl, H. F. R., & Beintema, D. (1964). The neurological examination of the full term infant. Little Club Clinics in Developmental Medicine No. 12. London: Heineman.

Rothbart, M. K., Posner, M. I., & Rosicky, J. (1994). Orienting in normal and pathological development. Development and Psychopathology, 6, 635-652.

Rose, S. A., Feldman, J. F., & Jankowski, J. J. (2001). Attention and recognition memory in the 1st year of life: A longitudinal study of preterm and full-term infants. Devel-opmental Psychology, 37, 135-151.

Rose, S. A., Feldman, J. F., McCarton, C. M., Wolfson, J. (1988). Information processing in seven-month-old infants as a function of risk status. Child Development, 59, 589-603.

Rose, S. A., Gottfried, A. W., & Bridger, W. H. (1979). Effects of haptic cues on visual recognition memory in fullterm and preterm infants. Infant Behavior and Develop-ment, 2, 55-67.

Rummel, R. J. (1970). Applied factor analysis. Evanston, IL: Northwestern University Press.

Sigman, M. D. (1983). Individual differences in infant attention: Relations to birth status and intelligence at five years. In T. M. Field & A. Sostek (Eds.), Infants born at

124

Chapter 5

risk: Physiological, perceptual and cognitive processes (pp. 271-293). New York: Grune & Stratton.

Sigman, M., Cohen, S. E., Beckwith, L., & Parmelee, A. H. (1986). Infant attention in rela-tion to intellectual abilities in childhood. Developmental Psychology, 22, 788-792.

Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modelling. London: Sage.

Spungen, L. B., Kurtzberg, D., & Vaughan, H. G. (1985). Patterns of looking behavior in full-term and low birth weight infants at 40 weeks post-conceptional age. Develop-mental and Behavioral Pediatrics, 6, 287-294.

Stechler, G., & Latz, E. (1966). Some observations on attention and arousal in the human infant. Journal of the American Academy of Child Psychology, 5, 517-525.Stevens, J. (1992). Applied multivariate statistics for the social sciences . Hillsdale, NJ:

Lawrence Erlbaum Associates, Inc.Tronick, E. (1972). Stimulus control and the growth of the infant’s effective visual field.

Perception and Psychophysics, 11, 373-376.Vervloed, M. P. J. (1995). Learning in preterm infants: Habituation, operant conditioning, and

their associations with motor development. Doctoral dissertation. University of Gron-ingen, Groningen, The Netherlands.

Walton, G. E., Armstrong, E. S., & Bower, T. G. R. (1997). Faces as forms in the world of the newborn. Infant Behavior and Development, 20, 537-543.

Woodhouse, G. (1996). Multilevel modelling applications: A guide for users of MLn. London: University of London, Institute of Education.

Yarrow, L. J., Klein, R. P., Lomonaco, S., Morgan, G. A. (1975). Cognitive and motivational development in early childhood. In B. Z. Friedlander, G. M. Sterritt, & G. E. Kirk (Eds.), Exceptional infant (Vol. 3, pp. 491-503). New York: Brunner-Mazel.

124

Chapter 5

Chapter 6

Summary and General Discussion

128

Chapter 6

129

Summary and General Discussion

THEORETICALBACKGROUNDANDGOALSOFTHESTUDYThis thesis investigates the development of two different attentional skills dur-

ing early infancy: visual scanning and shifting attention and gaze. To examine the world around them, infants have to be able to carry out small, systematic shifts of gaze and to pick up information from different parts of an object they are scanning. This information then has to be integrated with the image already stored in memory and should form the basis for the next eye movements. Besides the relatively small eye movements observed during scanning behavior, larger gaze shifts also play a weighty role in infants’ early visual development. Being able to shift gaze away from a stimulus in the central visual field to the periphery is important in order to monitor the environment and identify other potentially interesting targets. While inspecting an object, infants have been observed to shift their gaze away and back regularly. This gaze shifting has been interpreted as a regulation of the flow of visual input to enhance visual information processing (Colombo, 1993). Also during the interaction with another person, infants have been reported to avert their gaze from the other person’s face. Again, these gaze shifts are thought to fulfill a regulatory function: When the interaction becomes too intense, arousal is controlled by looking away from the interaction partner (Field, 1981; Stifter & Moyer, 1991).

Both behaviors – visual scanning and shifting of attention and gaze – undergo substantial changes during the first few months after birth. Younger infants have been observed to exhibit long periods of staring once they fixate a stimulus even when there are new stimuli appearing in their peripheral visual field (see e.g., Hood & Atkinson, 1993; Johnson, Posner, & Rothbart, 1991; Butcher, Kalverboer, & Geuze, 2000). They have also been reported to show only very limited scanning of visual stimuli (see e.g., Bronson, 1990, 1996). Around the age of 3 to 6 months, their visual scanning (Bronson, 1990, 1994) and their disengagement behavior (Johnson et al., 1991; Matsuzawa & Shi-mojo, 1997) have been described as similar to what can be observed in adults.

In experiments on visual processing and attention, the nature of the stimuli used seems to influence the visual behavior of the subjects. This is suggested, for example, by research on visual scanning in adults (see e.g., Mackworth & Morandi, 1967; Antes, 1974) and on gaze shifting in infants (Finlay & Ivinskis, 1984; Butcher et al., 2000). However, the impact of the nature of the stimulus on gaze shifting and visual scanning during infancy has not yet been systematically investigated. This thesis examines the effect of different stimuli on the development of these attentional processes. As stated in Chapter 1, nearly all studies to date have examined visual and attentional development using abstract, mostly unnatural stimuli. So one of the goals was to study attentional development using ecologically relevant as well as abstract stimulus material.

The aim of these studies was to examine the above-mentioned developmental processes during the first half year of infancy. During this period the visual and ocu-lomotor systems mature rapidly, and several visual skills develop (see e.g., Atkinson, 1984). An intensive longitudinal design was chosen. Between the ages of 6 and 26 weeks,

128

Chapter 6

129

Summary and General Discussion

infants were tested six times. So it was possible to describe and compare the timing and tempo of the deployment on the basis of individual developmental trajectories rather than on the basis of isolated snapshots of different infants’ performances. Precise registration techniques were adopted to measure eye movements, which served as the indicator for attentional processes: an infrared eye-tracking system and observation accurate to 20 milliseconds.

In addition to studying the “normal” development of attention, it was the goal of this study to examine a group of infants who were born prematurely. Preterm infants have been described on the one hand as being at risk for problems concerning their intellectual functioning during childhood (Breslau & Chilcoat, 2000; Anderson et al., 2003) and adolescence (Botting, Powls, Cooke, & Marlow, 1998). On the other hand, a possibly positive effect of the early visual stimulation that preterm infants are exposed to has been discussed (Fielder, Foreman, Moseley, & Robinson, 1993). The finding that attention patterns in early infancy are related to later cognitive functioning (see e.g., Colombo, 1993; Fagan & McGrath, 1981) emphasizes the importance of comparing the attentional development of healthy full-term and preterm infants.

In the next section, summaries of the chapters 2 to 5 are presented. The first two summaries describe the results of the investigation of the development of visual scan-ning and gaze shifting respectively in a group of healthy full-term infants. The third summary discusses the associations that were found between the developmental tra-jectories of visual scanning and shifts of attention and gaze. In the last summary, a comparison of the gaze shifting behavior in full-term and preterm infants is described. The subsequent paragraph contains the conclusions and a general discussion in which I will come back to a number of issues which formed the rationale for the studies. I will explore the theoretical implications of the studies, and present implications and suggestions for further research.

SUMMARIESOFTHESTUDIES

Developmental Changes in Visual Scanning of Dynamic Faces and Abstract StimuliThe way infants examine stimuli in their environment changes considerably dur-

ing the first months of life. When infants are born, they already show complex active scanning, but they are often unable to promptly shift their gaze to interesting stimuli in their visual field (Haith, 1980; Bronson, 1990). Younger babies have been observed to exhibit long fixations of relatively few locations and to examine only the perimeter of visual patterns (Salapatek, 1975; Milewski, 1976). Infants from about 3 months on start to overcome this non-flexible visual behavior and routinely scan various distant parts of a stimulus (Bronson, 1990).

Faces are important and frequent stimuli in the visual environment of a baby. When young infants scan faces, they seem to look at clear contrasts within the face like the edges and the hairline, while infants of 2 months and older look at internal features of a face increasingly frequently (Maurer & Salapatek, 1976). However, most previous

130

Chapter 6

131

Summary and General Discussion

studies of face scanning in infants have used still faces or photographs, which lack functional relevance for the infant.

This study had three aims: First, we wanted to examine how the visual scanning of dynamic stimuli – faces and abstract stimuli – develops throughout early infancy. Although there are indications that moving stimuli are examined in a different way from static displays, infants’ scanning of dynamic displays has rarely been investigated. The two stimuli used were the infant’s mother’s face and an abstract stimulus. The second goal of the study was to investigate whether the scanning patterns elicited by a moving face differed from those elicited by a moving abstract stimulus. As there are hardly any studies which examined face scanning with naturally moving faces as stimuli, the third aim of the study was to investigate which regions of a smiling, mov-ing face infants look at most frequently throughout the first few months of infancy.

The characteristics of scanning patterns were investigated through repeated as-sessments of 10 infants (5 girls; 5 boys). The infants were examined every 4 weeks from 6 to 26 weeks of age. Sitting in an infant-seat, they watched two different dynamic stimuli displayed one after another on a computer monitor. For the first stimulus a video recording of the infant’s mother was made as she was smiling and talking to her baby. This stimulus was combined with an abstract one. To ensure that the abstract stimulus was equivalent to the stimulus featuring the mother’s face with regard to its physical characteristics (such as movement dynamics, color range, luminance, etc.) but at the same time contained no meaning or recognizable structures, it was derived from the mother stimulus by carrying out several transformations in a graphic com-puter program. The videos were presented to the infant on the monitor screen for 30 seconds each. The infants’ eye movements during exposure to the stimulus were registered by an infrared eye-tracking system. Number and duration of the infants’ fixations were measured. The locations of the infants’ fixations while watching their mother’s face were also determined. The changes in number, duration, and location of the fixations were examined as a function of age and stimulus.

Infants of all ages attended to the stimuli most of the time they were present. Whereas young babies engaged in single, long fixations, infants from 14 weeks onwards showed more and shorter fixations. They also alternated their scanning of the stimulus with more short looks away from the screen. The way infants scanned the stimuli did not stabilize until 18 weeks, which is slightly later than has been reported in the litera-ture on infants’ scanning of static stimuli (Bronson, 1990, 1994). The dynamic nature of the stimuli increased their salience, which may have made them more demanding for the infants. This effect was especially marked for the abstract stimulus. From the 14-week session on, a stable difference in the median fixation duration during inspec-tion of the two stimuli was found. This suggests that infants from this age onwards tended to adapt their scanning behavior to the stimulus characteristics.

When scanning their mother’s face, infants directed their gaze at the mouth and eye region most often. Even at the youngest age, infants fixated their mother’s eyes

130

Chapter 6

131

Summary and General Discussion

and mouth. There was no indication of an edge preference as has been found in stud-ies using still faces. Between 10 and 14 weeks of age, the quality of the scanning of faces increased significantly, as sticky fixation was overcome, and infants looked at the particularly meaningful regions of the face – the eyes and the mouth – almost the entire time.

The Development of Gaze Shifting between Dynamic Faces and Abstract StimuliBeing able to shift attention and gaze in a strategic and flexible way forms a pre-

requisite for many behaviors which play an important role in development during infancy. Young infants explore and monitor their environment by looking around. They regulate the flow of visual input by alternating intense inspections of a stimulus with short looks away and control their arousal by regularly shifting their gaze away from an interaction partner.

However, when the infant is looking at something – for example, a toy or a person’s face – the actual gaze shift to a new location is preceded by the disengagement of at-tention and gaze. Infants of 1 to 2 months of age have been reported to have difficulty shifting attention and gaze away from fixation (Harris & MacFarlane, 1974; Aslin & Salapatek, 1975). This staring behavior has been frequently observed in young infants under diverse circumstances (Hood, Murray, King, Hooper, Atkinson, & Braddick, 1996; Stechler & Latz, 1966; Hopkins & van Wulfften Palthe, 1985) and has been named “obligatory attention” (Stechler & Latz, 1966) or ”sticky fixation” (Hood, 1995). The emergence of reliable and rapid disengagement around 3 to 4 months of age (Hood & Atkinson, 1993; Butcher et al., 2000) is an important step towards efficient functioning in different areas such as visual attention, cognition, and self-regulation.

Past research has shown that attributes of the stimuli used influence the infant’s visual reaction (see e.g., Cohen, 1972; Butcher et al., 2000; Finlay & Ivinskis, 1984). There are indications that the attributes of both the stimulus which is currently being attended to and the stimulus which appears in the peripheral visual field influence the infant’s visual reaction. However, this question has not been systematically addressed to date.

The aim of this study was to examine how the nature of the stimuli affects gaze and attention shifting behavior and how this changes throughout early infancy. Therefore, both a socially meaningful and an abstract stimulus were used. Again, a video of the infant’s mother’s face and a matched abstract stimulus were used.

Twenty healthy infants (12 girls; 8 boys) carried out a gaze shifting experiment. The measurements started when the infants were 6 weeks old and continued every 4 weeks until the infants reached the age of 26 weeks. The frequency and latency of shifts of gaze to peripheral targets were measured in a competition situation (the fixation stimulus persisted after the target appeared) and in a non-competition situation (the fixation stimulus disappeared when the target appeared). Both stimuli were used as central stimulus or peripheral target, which resulted in four different conditions (face-face, face-abstract, abstract-face, abstract-abstract).

132

Chapter 6

133

Summary and General Discussion

In the non-competition condition, infants shifted their gaze to the peripheral target frequently from the age of 10 weeks on. The latency of the eye movements de-creased between 6 and 16 weeks of age and leveled off thereafter. There was no effect of stimulus combination for the non-competition condition.

In the competition condition, however, younger infants looked away from the central to the peripheral stimulus less frequently and more slowly than in the non-competition condition. The disengagement frequency increased rapidly between 6 and 22 weeks of age and stabilized thereafter. The disengagement latency decreased throughout the entire measurement period.

There was a strong effect of stimulus combination in the competition condition: Infants were more likely to shift their gaze when the central stimulus was a face and the peripheral target was abstract, while they moved their gaze least frequently and most slowly in the opposite condition (abstract-face). The differences between the four stimulus combinations were most marked between 10 and 18 weeks of age. This suggests that the sensitivity to context variables is highest when disengagement is not yet well established. The results indicated that – compared to the abstract stimulus – the mother’s face was less able to hold and attract the infants’ attention. This is consistent with the finding that even very young infants know their mother’s face well (Barrera & Maurer, 1981) and that their interest broadens to include novel faces and objects in their environment as they grow older (Kaye & Fogel, 1980). The results also demonstrate that the attributes of both the central and the peripheral stimulus influenced the infants’ gaze shifting behavior. The transition from sticky fixation to reliable, flexible control over gaze shifts thus can be described in terms of changes in the relative strength of two processes. While one process is maintain-ing fixation and the other one is enabling gaze shifts to a new target, both seem to be sensitive to the characteristics of the competing stimuli.

Associations between the Developmental Trajectories of Visual Scanning and Disengagement

The visual system of the infant goes through substantial changes during the first few months of life. While young infants’ gaze shifting behavior is dominated by the occurrence of “sticky fixation” (Hood, 1995), infants of 3 to 4 months of age start exhibiting fast, flexible disengagement of attention and gaze. In the same period, the ability to scan stimuli in a quick, functional way emerges. Although these mechanisms develop during approximately the same age period, they seem to be based on differ-ent neurological structures. While shifts of gaze to a peripheral target are thought to be mediated by the magnocellular pathway, visual scanning has been assigned to the parvocellular pathway. As there are indications that the two pathways may have different developmental courses (Hickey & Peduzzi, 1987), it was one of the goals of this study to compare the developmental trajectories of disengagement and scanning, and it was hypothesized that functional scanning would emerge slightly earlier than

132

Chapter 6

133

Summary and General Discussion

reliable disengagement. There are, however, indications that the magnocellular and the parvocellular pathway are not strictly parallel but have anatomical and functional interrelations (Merigan & Maunsell, 1993). So the second aim of the study was to ex-plore possible associations between the development of disengagement and scanning at the level of the individual infant.

The data of 10 infants who participated in the scanning study and the disengage-ment experiment were analyzed. Ten healthy infants (5 girls; 5 boys) carried out the scanning and the disengagement task six times between the age of 6 and 26 weeks. In the tasks, two different dynamic stimuli were used: the infant’s mother’s face and a matched abstract stimulus.

Frequency and latency measures of the eye movements during scanning and gaze shifting were examined using multivariate multilevel models and Monte Carlo analyses. The analyses revealed that both frequency and latency measures continued to change between 16 and 26 weeks for the disengagement task, while there was no significant change for the scanning task after the age of 16 weeks. For the frequency measures, a breaking point after which development stabilized occurred earlier for the scanning than for the disengagement. This is consistent with the account that the parvocellular stream becomes functional slightly before the magnocellular stream. There were no indications for positive associations between the development of scan-ning and disengagement. Results rather suggested that scanning and disengagement change scores contrasted more with one another than could be expected on the basis of chance. This implies that the dorsal and the ventral stream develop rather independently and do not interact but might even compete for similar resources of development or performance during the first 26 weeks of age.

The Development of Shifts of Attention and Gaze in Preterm InfantsInfants who are born preterm are exposed to visual stimulation earlier in their

development than full-terms. Some researchers have suggested that this early visual experience could have a positive influence on the visual development of preterm in-fants (see e.g., Fielder et al., 1993), others have argued that this early stimulation could be harmful to the immature visual system (see e.g., Friedman, Jacobs, & Werthmann, 1981).

Concerning visual and attentional mechanisms in preterm infants, evidence for an accelerated as well as for a disturbed development has been reported. Findings on gaze and attention shifting in preterm infants are sparse and inconsistent (Butcher, Kalver-boer, Geuze, & Stremmelaar, 2002; Atkinson, 2000). As early visual and attentional pat-terns are related to intellectual functioning in later life (Fagan & McGrath, 1981; Cohen & Parmelee, 1983), it is of great importance to examine early normal development of attention as well as possibly deviant developmental trajectories in high-risk infants.

The first aim of this study was to compare the development of attention and gaze shifting in pre- and full-term infants. As the use of face and non-face stimuli influ-

134

Chapter 6

135

Summary and General Discussion

ences the gaze shifting behavior of infants in a disengagement task (see Chapter 3) and preterm infants have been shown to be visually less responsive to faces (Masi & Scott, 1983), the second goal of the study was to explore possible differences in the impact of the stimulus sort in preterm and full-term infants.

Twenty healthy full-term infants (12 girls; 8 boys) and 10 preterm infants (4 girls; 6 boys) participated in the study. The preterm infants were born after a gestation period between 27.3 and 32.4 weeks and were – despite their rather diverse clinical backgrounds – considered to be developing normally. The infants carried out a disen-gagement task identical to the one used in the studies described previously. Again, the stimuli consisted of a video of the infant’s mother’s face and an abstract video. Both stimuli appeared as central fixation stimulus and peripheral target, which resulted in four possible stimulus combinations. The infants were tested every 4 weeks from 6 to 26 weeks of (corrected) age.

The infants’ visual reaction to the appearance of the peripheral stimulus was reg-istered (whether infants kept on staring to the center of the monitor, shifted their gaze to the peripheral target, or shifted their gaze away but did not look at the newly appeared target). The latency of gaze shifts was also determined.

The development of the frequency of disengagement was similar in both groups: Infants shifted their gaze effectively in the non-competition condition from 10 weeks on. In the competition condition, the frequency of looks to the peripheral stimulus was significantly lower than in the non-competition condition when infants were young. At 18 and 24 weeks of age, there were no differences in gaze shifting frequency anymore between the competition and the non-competition condition. The gaze shifting latency of both the preterm and the full-term infants decreased throughout the measurement period. Saccadic reaction times were longer in the competition than in the non-com-petition condition at younger ages as well as at the later measurement points.

There were, however, several significant differences between full- and preterm infants concerning their gaze shifting behavior. At 6 weeks of age, preterm infants had more difficulty overcoming staring behavior in both the competition and the non-competition tasks. Prematurity, however, was also associated with a short-lasting advantage in the latency of disengagement. Thus, preterm infants were on the one hand more likely to show staring behavior but on the other hand quicker in shifting their gaze. This suggests that the triggering of a gaze shift and the actual execution of the eye movement are not necessarily associated. Preterm and full-term infants reacted similarly to the different stimulus combinations. In the competition condition, infants shifted their gaze more frequently when the central stimulus was a face and the peripheral stimulus was abstract. They were also significantly slower in looking away from an abstract stimulus to a face.

One of the aims of the study was to examine whether early visual input is beneficial for early attentional development or whether it might be damaging to the immature visual system. The results of the study do not support either one of the two accounts,

134

Chapter 6

135

Summary and General Discussion

but suggest that prematurity is associated with small differences in visual and at-tentional development in general. However, the effects were temporary – by the end of the measurement period, the differences in full- and preterm infants’ gaze shifting behavior had disappeared.

CONCLUSIONSANDGENERALDISCUSSIONThe aim of this thesis was to investigate the development of visual scanning and

shifts of attention and gaze in early infancy. In Chapter 1, several key issues to be examined were set out. In the next section, these issues will be discussed in the light of the results of our investigations. Contextual aspects of development such as the impact of different stimuli or the role of early visual experience in the case of prema-ture birth are treated. The tempo and timing of development of visual scanning and shifts of attention will also be addressed. The use of precise eye movement registration will be evaluated. Finally, theoretical implications and some suggestions for further research are described.

The Role of Different StimuliIn the experiments described in this thesis, two carefully selected stimuli were

used: the infant’s mother’s face as she was interacting with her baby and a moving abstract stimulus. One reason for this stimulus selection was that many studies on infants’ attentional development still neglect the issue of the ecological (ir-)relevance of the stimuli. As a consequence, very little is known about young infants’ gaze shift-ing and scanning behavior when they are confronted with natural stimuli, although there are several reasons to expect that the findings might differ from what has been described on basis of studies which lack ecological validity.

Our results on the development of visual scanning of faces confirmed these ex-pectations. While earlier studies which had used still faces as stimuli reported that infants younger than 2 months show only limited scanning and rarely look at the in-ner features of the face, in our study, no indication of an edge effect was found. This implies that in a natural face-to-face interaction even infants as young as 6 weeks of age might be able to look at the inner features of the other person’s face. The slow, exaggerated way in which mothers tend to talk and move their face when interacting with their babies (Stern, 1974) might help the infants to overcome the edge attraction, which impedes their visual exploration behavior at this young age.

Furthermore, it was our aim to investigate the influence of different sorts of stimuli on the infants’ visual behavior. It turned out that different developmental trajectories of visual scanning and gaze shifting could be observed for the abstract and the face stimulus. Less advanced scanning behavior persisted longer when infants were ex-amining the abstract stimulus than when they were scanning their mother’s face. In the gaze and attention shifting task, infants were more likely to shift their gaze away from their mother’s face and look to an abstract stimulus, and less likely to shift their

136

Chapter 6

137

Summary and General Discussion

gaze from the abstract stimulus to their mother’s face in the periphery. Shifts of gaze were also slower in this condition. This effect was found for a group of normal control subjects as well as for preterm infants. The impact of the different stimuli on infants’ scanning as well as on their gaze shifting behavior was most marked around the age of 14 weeks, which corresponds approximately with the period of the most rapid de-velopment. This finding is consistent with a dynamic systems theoretical approach, which predicts that a skill which is not yet well established is especially sensitive to context variables (van Geert, 1997, 2002).

From the age of about 3 months on, a stable difference in the median fixation lengths while scanning the different stimuli was observed. This finding implies that from the age of about 3 months on, infants started to tailor their scanning behavior to the characteristics of the different stimuli.

These results are particularly interesting as they provide insight into underly-ing attentional processes. The findings on the development of gaze shifting between different sorts of stimuli revealed that both stimuli – the stimulus presented in the central visual field of the infant and the stimulus used as peripheral target – affected the infants’ gaze shifting behavior. While the infants were fixating the stimulus in their central visual field, they were also processing the peripheral target, and the peripheral target also influenced their gaze shifting behavior.

As mentioned above, the two stimuli were designed to be as comparable as possible in terms of their physical characteristics, for example in terms of luminance, color range, and overall movement dynamics. In the gaze and attention shifting task, the abstract stimulus held and attracted the infants’ attention better than the face. This is in accord with findings from studies on face-to-face interaction between moth-ers and infants. While their mother’s face is a powerful attractor for infants when they are young (Morton & Johnson, 1991), from the age of about 3 months on, other objects become increasingly more interesting (Kaye & Fogel, 1980), and infants tend to look away from their mother’s face more often (van Wulfften Palthe, 1986). The abstract stimulus, however, might still have been new and therefore more attractive. In 3-month-olds, it also elicited a less advanced scanning behavior. It is also possible that the abstract stimulus was slightly more salient due to the fact that it contained more plain areas of the same color and thus more contrast. However, in any case, it was especially challenging for the infant, as its structure and way of movement was still unfamiliar compared to the – by this time – well-known and familiar face of the mother.

Timing, Tempo, and Individual Differences of Development One of the aims of these studies was to describe the timing and tempo of develop-

ment of visual scanning and disengagement and the inter-infant differences con-cerning this change. Therefore, a longitudinal design with six measurement points during the first six months of life was implemented. This design of relatively dense

136

Chapter 6

137

Summary and General Discussion

assessments allowed us to determine more precisely when a stable scanning pattern and reliable disengagement emerged with the different stimuli. In order to explore a possible ordering in the emergence of functional scanning versus the development of disengagement, infants were presented with both tasks throughout the measure-ment period. Investigating frequency and latency measures, the analyses showed that visual scanning and disengagement improved during the measurement period. The efficiency of disengagement increased rapidly until 16 weeks of age and more gradually thereafter. However, there were no significant changes in scanning behavior anymore during the last measurement period. Furthermore, it could be shown that – for the frequency measures – a stabilization of the developmental changes occurred earlier for visual scanning than for the disengagement of attention and gaze.

Analyses of the inter-infant differences in the development of the frequency and latency of disengagement revealed that there were considerable differences between the performances of different infants. These differences concerned both frequency and latency measures and occurred mainly during periods of rapid development.

The Impact of PrematurityAnother aim of this thesis was to compare the attentional development of a group

of infants who were born preterm and a group of full-term infants with uneventful pre- and perinatal histories. As described above, there are two opposite theories on the impact of preterm birth and the resulting extra visual experience. One theory argues that the early visual input could have a positive influence on the infants’ visual development (see e.g., Fielder et al., 1993). The other one suggests that this premature visual stimulation might be detrimental to the visual system of the young infant (see e.g., Friedman et al., 1981).

In this thesis, the differences in performance of a group of low-risk preterm infants and a group of healthy full-term infants on a disengagement task were examined. It turned out that young preterm infants had more problems overcoming staring behavior than their full-term age mates, which is indicative of an adverse impact of preterm birth on early attentional processes. At the same time, preterm infants were quicker when shifting their gaze away from one stimulus to a new one in the periphery. This, however, is an indication of a – though short-lasting – advantage in attentional development of the preterm compared to the full-term infant group.

Thus, the findings of this thesis do not provide the basis for a definite decision in favor of either one of these two theories. In my opinion, however, these results rather suggest that neither of these accounts describes the visual and attention develop-ment in preterm infants properly. Visual and attentional development depend on the complex interaction of maturational and experience-dependent processes (see e.g., Greenough, Black, & Wallace, 1987). In preterm infants, the timing and sequence of this interaction is disturbed by the preterm birth. Furthermore, they face a high risk for medical complications (such as hemorrhagic and ischemic brain injury or infections)

138

Chapter 6

139

Summary and General Discussion

which might also interfere with their early development. The early visual develop-ment of preterm infants is thus affected in many ways, with the result that while some developmental processes may be enhanced, others may be disturbed or delayed.

Studying early attentional development in preterm infants is important for several reasons. As mentioned above, vision and attention have been shown to develop differ-ently in infants born preterm and in healthy full-term infants. However, instead of a having clearly beneficial or harmful impact, the effect of prematurity on infants’ early development of vision and attention seems to be complex and heterogeneous. It is of great importance to describe the development of different groups of preterm infants carefully as this forms the basis to understand their abilities and special needs and to develop custom-built interventions. On the other hand, studying abnormal develop-ment provides insight into vision and attentional mechanisms in general.

Eye Movement RegistrationThe first studies using infrared eye-trackers with young infants appeared in the

1960’s (Salapatek & Kessen, 1966; Haith, 1969). Since then, this technique has sub-stantially improved. The eye-tracking techniques now available make it possible to measure eye movements and gaze direction with great accuracy and relative ease. Compared to earlier methods, the precision of the measurements has been improved with regard to both the time resolution and the spatial accuracy.

Recent studies have yielded interesting results, not only on the development of infants’ scanning of dynamic stimuli, but also on infants’ cognitive and perceptual abilities, such as for example the formation of categories (McMurray & Aslin, 2004) or the perception of object unity (Johnson, Slemmer, & Amso , 2004). In the future, the new eye-tracking techniques will probably be implemented by more researchers. Promising topics are processes which to date have mainly been studied using simple habituation measures, such as action perception or object permanence. The use of precise methods to measure eye movements and fixation locations offers the possibil-ity to gain more insight into the actual perceptional processes and early cognitive development.

Aside from the great power and large advantages of eye-tracking methods, research-ers should realize that this technique is highly complex and sensitive (see also Haith, 2004). An accurate and reliable measurement of gaze position and eye movements in infants requires a careful positioning of the infant, of the camera and of the stimulus display and a precise calibration. The application of this recording technique with in-fants as young as 1 to 2 months still forms a great challenge to the researcher because of young infants’ poor postural control and limited capability to complete all the steps required for a successful calibration. Table 2.1 in Chapter 2, which presents the amount of time during the stimulus presentation with a complete and reliable measurement at different ages of the infant subjects, illustrates this fact well. However, the diverse oppor-tunities to apply the method and the promising results make the efforts worthwhile.

138

Chapter 6

139

Summary and General Discussion

Theoretical ImplicationsExisting theories about the underlying mechanisms and the neurological basis of

early attentional development differ in how they explain the development of disengage-ment and the development of visual scanning. Hood (1995) attributes the occurrence of staring behavior to young infants’ inability in breaking gaze from a stimulus they are fixating. Johnson (1990), on the other hand, argues that young infants have dif-ficulty generating an eye movement while processing a stimulus which is currently in their central visual field. Rothbart, Posner, and Rosicky (1994), however, suggest that shifts of gaze are preceded by covert shifts of attention, and disengagement problems reflect difficulty shifting attention covertly.

As mentioned above, the results of these studies allow some conclusions concerning the different theoretical accounts. The studies on disengagement showed that infants’ reactions in a gaze shifting task depended on the type of stimulus in the central visual field as well as in the periphery. This indicates that infants attended to the peripheral targets even when their gaze was still “stuck” on the central stimulus. This finding makes interpretations which emphasize the role of covert attention less likely but provides support for the theoretical accounts of Hood (1995) or Johnson (1990).

In Chapter 4, the development of visual scanning and of disengagement were com-pared. As existing theories suggest that these two skills are mediated by different neurological structures, slightly different developmental trajectories were expected. The results confirmed these expectations, and analyses of associations on the level of individual subjects suggested that the two mechanisms seem to develop rather independently.

It has to be kept in mind, though, that a study focusing on behavioral change cannot provide conclusive evidence about underlying neurological substrates (Butcher, 2000). Harder evidence could be provided by studies using imaging methods, as a seemingly identical behavior, for instance, can be mediated by different mechanisms throughout development. However, experimental behavioral studies help sharpen the focus for future research and so form the connecting link between developmental changes observed in daily life and investigations of neurological development.

Implications for Further ResearchUnderstanding attentional processes. There is a large body of research on the early

development of visual attention (see e.g., Johnson, 1994; Hood, 1995 for reviews). Un-derstanding how different attentional mechanisms develop is important not only to expand our knowledge about early cognitive development and to detect possibly deviant developmental trajectories early in life, but also to increase our insight into attentional processes in general.

The difficulty disengaging attention and gaze, which is characteristic of the early months of infancy, has been shown to diminish with age (Butcher et al., 2000; Johnson et al., 1991). However, there are indications that also older infants (see Chapter 5; Hood

140

Chapter 6

141

Summary and General Discussion

& Atkinson, 1993) and even adults (Hood & Atkinson, 1993; Fischer, 1986) shift their gaze more slowly between two competing stimuli. There seems to be some resemblance between the performances of infants and adults on disengagement tasks. This suggests that the effect of a stimulus in central vision impeding shifts of gaze to a new location reflects a general characteristic of the human attentional system. However, there are only very few studies which examine the performance on attention tasks at different ages throughout development and which also include adult subjects.

In a pilot study, Butcher and Hunnius (2001) examined the disengagement of atten-tion and gaze using different sorts of stimuli in a small sample of 6 adult participants. The experimental setup and procedure were the same as in the infant studies described in chapters 3 and 4 of this thesis. In the gaze shifting task with competing stimuli, two types of stimuli were used: a video of a still female face and a matched abstract stimulus. It could be shown that the type of stimulus presented in the center and in the periphery affected the speed of gaze shifting. Thus, higher level characteristics – such as meaningfulness – of both the stimulus currently under attention and the target in the periphery are processed and have an effect on the ease with which at-tention and gaze are disengaged. This effect holds especially for young infants whose ability to disengage attention and gaze is still developing (see Chapter 2), but it is also present in adult gaze shifting behavior.

The above-mentioned examples of research demonstrate how studies on attentional processes in infants and adults can provide information on the underlying maxims of the visual and attentional system in general. To date, this kind of studies are scarce. More research which investigates attentional mechanisms using comparable tasks with very different age groups is needed.

Attentional processes and their development are frequently studied by means of looking behavior. However, in these studies the underlying mechanisms can only be measured indirectly, as changes in looking behavior function as indicators of these mechanisms. So definite conclusions about, for instance, different theoretical accounts or underlying neurological substrates are not always possible. These considerations emphasize the importance of using different methods in studies on attentional de-velopment in order to obtain a picture which is as complete as possible. Examples of additional measures are heart rate (see e.g., Finlay & Ivinskis, 1984; Richards, 1997) or imaging techniques (see for reviews, Thomas & Casey, 2003; Richards, 2003). Studies which combine behavioral and more direct measures are needed and will provide new insights into early attentional and visual development.

Exploring the importance of attentional mechanisms for development. Attentional pro-cesses play an important role in many daily activities. Reading, recognizing someone you know in a group of people, or walking through a crowded mall are only a few examples of skills which depend on fast, accurate shifts of attention and gaze.

For infants, gaining control over attention and gaze shifts is crucial to be able to explore the environment and to learn about the surrounding world. As mentioned

140

Chapter 6

141

Summary and General Discussion

earlier, the vast majority of studies on infants’ cognitive development has made use of global measures such as habituation or preferential looking. These methods were suitable to describe the time course and the circumstances of the emergence of cog-nitive skills, but reveal rather little about the underlying attentional and perceptual processes. A few studies have tried to link cognitive development and looking behavior. Bronson (1991), for instance, compared scanning patterns of infants who were fast or slow in processing a stimulus. In a recent study, Johnson et al. (2004) examined the relation between infants’ way of scanning and their ability to perceive object unity. However, we are only at the beginning of understanding how basic attentional and perceptual processes contribute to complex cognitive skills. Further research promises to provide valuable information on, for instance, visual processing, face recognition or action perception.

The ability to shift attention and gaze swiftly and reliably plays a role not only in early cognitive development, but is also important for young infants’ developing emo-tion regulation skills. Looking away from a stimulus which is annoying or too intense is a way of regulating sensation, and thus it is not surprising that sticky fixation has often been described to cause infants distress (Stechler & Latz, 1966; Tennes, Emde, Kisley, & Metcalf, 1972). Only a few studies have addressed the associations between the development of attention and the development of emotion regulation. Johnson et al. (1991) found that 4-month-olds who disengaged more easily were – according to their mothers’ reports – more easily soothed. For infants of 13.5 months of age, Rothbart, Ziaie, and O’Boyle (2003) were able to show that the ability to redirect attention away from distressing stimuli was related to lower levels of negative affect.

As mentioned earlier, early face-to-face interaction also calls on the infants’ regulation skills. Infants shift their gaze regularly (Stifter & Moyer, 1991) in order to regulate the visual input, to turn away from an unpleasant (Cohn & Tronick, 1983) or excessively intense interaction. The still-face procedure (Tronick, Als, Adamson, Wise, & Brazelton, 1978) has been used to assess parent-infant interaction, coping, and the regulation of arousal in a situation of face-to-face interaction. Abelkop and Frick (2003) have investigated infants’ looking behavior during a still-face situation as well as their performance on attention tasks and found that attentional measures showed moderate stability within cognitive and social contexts. A few studies thus have provided support for the notion that attentional processes affect infants’ regula-tory skills and social behavior. However, future research should continue to explore the interrelations of attentional and emotional development in infancy.

REFERENCESAbelkop, B. S., & Frick, J. E. (2003). Cross-task stability in infant attention: New per-

spectives using the still-face procedure. Infancy, 4, 567-588.

142

Chapter 6

143

Summary and General Discussion

Anderson, P., Doyle, L. W., Callanan, C., Carse, E., Casalaz, D., Charlton, M. P., Davis, N., Duff, J., Ford, G., Fraser, S., Hayes, M., Kaimakamis, M., Kelly, E., Opie, G., Watkins, A., Woods, H., & Yu, V. (2003). Neurobehavioral outcomes of school-age children born extremely low birth weight or very preterm in the 1990s. Journal of the American Medical Association, 289, 3264-3272.

Antes, J. R. (1974). The time course of picture viewing. Journal of Experimental Psychol-ogy, 103, 62-70.

Aslin, R. N., & Salapatek, P. (1975). Saccadic localization of visual targets by the very young human infant. Perception and Psychophysics, 17, 293-302.

Atkinson, J. (1984). Human visual development over the first six months of life. Human Neurobiology, 3, 61-74.

Atkinson, J. (2000). The developing visual brain. Oxford, UK: Oxford University Press.Barrera, M. E., & Maurer, D. (1981). Recognition of mother’s photographed face by the

three-month-old infant. Child Development, 52, 714-716.Botting, N., Powls, A., Cooke, R. W. I, & Marlow, N. (1998). Cognitive and educational

outcome of very-low-birthweight children in early adolescence. Developmental Medicine and Child Neurology, 40, 652-660.

Breslau, N., & Chilcoat, H. D. (2000). Psychiatric sequelae of low birth weight at 11 years of age. Biological Psychiatry, 47, 1005-1011.

Bronson, G. W. (1990). Changes in infants’ visual scanning across the 2- to 14-week age period. Journal of Experimental Child Psychology, 49, 101-125.

Bronson, G. W. (1991). Infant differences in rate of visual encoding. Child Development, 62, 44-54.

Bronson, G. W. (1994). Infants’ transitions towards adult-like scanning. Child Develop-ment, 65, 1243-1261.

Bronson, G. W. (1996). The growth of visual capacity: Evidence from infant scanning patterns. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 11, pp. 109-141). Norwood, NJ: Ablex.

Butcher, P. R. (2000). Longitudinal studies of visual attention in infants: The early develop-ment of disengagement and inhibition of return. Doctoral dissertation. University of Groningen, Groningen, The Netherlands.

Butcher, P. R., & Hunnius, S. (2001, April). The influence of abstract and meaningful stimuli on shifts of visual attention. Poster presented at the Voorjaarsconferentie Nederlandse Vereniging voor Neuropsychologie, Utrecht, The Netherlands.

Butcher, P. R., Kalverboer, A. F., & Geuze, R. H. (2000). Infants’ shifts of gaze from a central to a peripheral stimulus: A longitudinal study of development between 6 and 26 weeks. Infant Behavior and Development, 23, 3-21.

Butcher, P. R., Kalverboer, A. F., Geuze, R. H., & Stremmelaar, E. F. (2002). A longitudi-nal study of the development of shifts of gaze to a peripheral stimulus in preterm infants with transient periventricular echogenicity. Journal of Experimental Child Psychology, 82, 116-140.

142

Chapter 6

143

Summary and General Discussion

Cohen, L. B. (1972). Attention-getting and attention-holding processes of infant visual preferences. Child Development, 43, 869-879.

Cohen, S. E., & Parmelee, A. H. (1983). Prediction of five-year Stanford-Binet scores in preterm infants. Child Development, 54, 1242-1253.

Cohn, J. F., & Tronick, E. Z. (1983). Three-month-old infants’ reaction to simulated maternal depression. Child Development, 54, 185-193.

Colombo, J. (1993). Infant cognition: Predicting later intellectual functioning. Newbury Park, CA: Sage.

Fagan, J. F., & McGrath, S. K. (1981). Infant recognition memory and later intelligence. Intelligence, 5, 121-130.

Field, T. M. (1981). Infant arousal, attention, and affect during early interactions. In L. P. Lipsitt & C. Rovee-Collier (Eds.), Advances in infancy research (Vol. 1, pp. 57-100). Norwood, NJ: Ablex.

Fielder, A. R., Foreman, N., Moseley, M. J., & Robinson, J. (1993). Prematurity and visual development. In K. Simons (Ed.), Early visual development, normal and abnormal (pp. 485-504). New York: Oxford University Press.

Finlay, D., & Ivinskis, A. (1984). Cardiac and visual responses to moving stimuli pre-sented either successively or simultaneously to the central and peripheral visual fields in 4-month-old infants. Developmental Psychology, 20, 29-36.

Fischer, B. (1986). The role of attention in the preparation of visually guided eye move-ments in monkey and man. Psychological Research, 48, 251-257.

Friedman, S. L., Jacobs, B. S., & Werthmann, M. W. (1981). Sensory processing in pre-and full-term infants in the neonatal period. In S. L. Friedman & M. Sigman (Eds.), Pre-term birth and psychological development (pp. 159-178). New York: Academic Press.

Greenough, W. T., Black, J. E., & Wallace, C. S. (1987). Experience and brain development. Child Development, 58, 539-559.

Haith, M. M. (1969). Infrared television recording and measurement of ocular behavior in the human infant. American Psychologist, 24, 279-282.

Haith, M. M. (1980). Rules that babies look by. Hillsdale, NJ: Erlbaum Associates, Inc.Haith, M. M. (2004). Progress and standardization in eye movement work with human

infants. Infancy, 6, 257-265.Harris, P., & MacFarlane, A. (1974). The growth of the effective visual field from birth

to seven weeks. Journal of Experimental Child Psychology, 18, 340-348.Hickey, T. L., & Peduzzi, J. D. (1987). Structure and development of the visual system.

In L. Cohen & P. Salapatek (Eds.), Handbook of infant perception (pp. 1-42). New York: Academic Press.

Hood, B. M. (1995). Shifts of visual attention in the human infant: A neuroscientific approach. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 9, pp. 163-216). Norwood, NJ: Ablex.

Hood, B. M., & Atkinson, J. (1993). Disengaging visual attention in the infant and adult. Infant Behavior and Development, 16, 405-422.

144

Chapter 6

145

Summary and General Discussion

Hood, B. M., Murray, L., King, F., Hooper, R., Atkinson, J., & Braddick, O. (1996). Habitu-ation changes in early infancy: Longitudinal measures from birth to 6 months. Journal of Reproductive and Infant Psychology, 14, 177-185.

Hopkins, B., & van Wulfften Palthe, T. (1985). Staring in infancy. Early Human Develop-ment, 12, 261-267.

Johnson, M. H. (1990). Cortical maturation and the development of visual attention in early infancy. Journal of Cognitive Neuroscience, 2, 81-95.

Johnson, M. H., Posner, M. I., & Rothbart, M. K. (1991). Components of visual orienting in early infancy: Contingency learning, anticipatory looking, and disengaging. Journal of Cognitive Neuroscience, 3, 335-344.

Johnson, M. H. (1994). Visual attention and the control of eye movements in early in-fancy. In C. Umiltà & M. Moscovitch (Eds.), Attention and performance XV: Conscious and nonconscious information processing. Attention and performance series (pp. 291-310). Cambridge, MA: MIT Press.

Johnson, S. P., Slemmer, J. A., & Amso, D. (2004). Where infants look determines how they see: Eye movements and object perception performance in 3-month-olds. Infancy, 6, 185-201.

Kaye, K., & Fogel, A. (1980). The temporal structure of face-to-face communication between mothers and infants. Developmental Psychology, 16, 454-464.

Mackworth, N. H., & Morandi, A. J. (1967). The gaze selects informative details within a picture. Perception and Psychophysics, 2, 547-552.

Masi, W. S., & Scott, K. G. (1983). Preterm and full-term infants’ visual responses to mothers’ and strangers’ faces. In T. M. Field & A. Sostek (Eds.), Infants born at risk: Physiological, perceptual and cognitive processes (pp. 173-180). New York: Grune & Stratton.

Matsuzawa, M., & Shimojo, S. (1997). Infants’ fast saccades in the gap paradigm and development of visual attention. Infant Behavior and Development, 20, 449-455.

Maurer, D., & Salapatek, P. (1976). Developmental changes in the scanning of faces by young infants. Child Development, 47, 523-527.

McMurray, B., & Aslin, R. N. (2004). Anticipatory eye movements reveal infants’ audi-tory and visual categories. Infancy, 6, 203-229.

Merigan, W. H., & J. H. R. Maunsell (1993). How parallel are the primate visual path-ways? Annual Review of Neuroscience, 16, 369-402.

Milewski, A. (1976). Infant’s discrimination of internal and external pattern elements. Journal of Experimental Child Psychology, 22, 229-246.

Morton, J., & Johnson, M. H. (1991). CONSPEC and CONLEARN: A two-process theory of infant face recognition. Psychological Review, 98, 164-181.

Richards, J. E. (1997). Peripheral stimulus localization by infants: Attention, age, and individual differences in heart rate variability. Journal of Experimental Psychology: Human Perception and Performance, 23, 667-680.

144

Chapter 6

145

Summary and General Discussion

Richards, J. E. (2003). The development of visual attention and the brain. In M. de Haan & M. H. Johnson (Eds.), The cognitive neuroscience of development (pp. 73-98). Hove, UK: Psychology Press.

Rothbart, M. K., Posner, M. I., & Rosicky, J. (1994). Orienting in normal and pathological development. Development and Psychopathology, 6, 635-652.

Rothbart, M. K., Ziaie, H., & O’Boyle, C. G. (1992). Self-regulation and emotion in in-fancy. In N. Eisenberg & R. A. Fabes (Eds.), Emotion and its regulation (pp. 7-23). San Francisco: Jossey-Bass.

Salapatek, P. (1975). Pattern perception in early infancy. In L. B. Cohen & P. Salapatek (Eds.), Infant perception: From sensation to cognition (Vol. 1, pp. 133-248). New York: Academic Press.

Salapatek, P., & Kessen, W. (1966). Visual scanning of triangles by the human newborn. Journal of Experimental Child Psychology, 3, 155-167.

Stechler, G., & Latz, E. (1966). Some observations on attention and arousal in the human infant. Journal of the American Academy of Child Psychology, 5, 517-525.

Stern, D. N. (1973). Mother and infant at play: The dyadic interaction involving facial, vocal, and gaze behaviors. In M. Lewis & L. D. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 187-213). New York: Wiley.

Stifter, C. A., & Moyer, D. (1991). The regulation of positive affect: Gaze aversion activity during mother-infant interaction. Infant Behavior and Development, 14, 111-123.

Tennes, K., Emde, R., Kisley, A., & Metcalf, D. (1972). The stimulus barrier in early infancy: An exploration of some formulations of John Benjamin. In R. R. Holt & E. Peterfreund (Eds.), Psychoanalysis and contemporary science (pp. 206-234). New York: Macmillan.

Thomas, K. M., & Casey, B. J. (2003). Methods for imaging the developing brain. In M. de Haan & M. H. Johnson (Eds.), The cognitive neuroscience of development (pp. 19-41). Hove, UK: Psychology Press.

Tronick, E. Z., Als, H., Adamson, L., Wise, S., & Brazelton, T. B. (1978). The infant’s re-sponse to entrapment between contradictory messages in face-to-face interaction. Journal of the American Academy of Child Psychology, 17, 1-13.

van Geert, P. (1997). Variability and fluctuation: A dynamic view. In E. Amsel & K. A. Renninger (Eds.), Change and development: Issues of theory, method, and application (pp. 193-212). Mahwah, NJ: Erlbaum.

van Geert, P. (2002). Developmental dynamics, intentional action and fuzzy sets. In N. Granott & J. Parziale (Eds.), Microdevelopment: Transition processes in development and learning (pp. 319-343). Cambridge, UK: Cambridge University Press.

van Wulfften Palthe, T. (1986). Neural maturation and early social behavior: A longitudinal study of mother-infant interaction. Doctoral dissertation. University of Groningen, Groningen, The Netherlands.

Samenvatting

148

Samenvatting

149

Samenvatting

INLEIDINGIedereen die ooit heeft gezien hoe een baby van enkele weken oud zijn omgeving

verkent en bekijkt, heeft ongetwijfeld opgemerkt dat het kijkgedrag van een pasgeborene zich sterk onderscheidt van dat van een volwassene. Zeer jonge baby’s zijn visueel al actief en kijken naar stimuli in hun visuele veld. Hierbij komen echter vaak lange periodes voor, waarin hun blik “vastgeplakt” lijkt te zijn aan een stimulus. Dikwijls beperkt hun blik zich dan ook tot één of twee van de stimuli die in hun visuele veld aanwezig zijn. Wanneer een stimulus wordt bekeken, scant de baby deze vaak slechts gedeeltelijk. Tijdens de eerste levensmaanden vindt er echter een snelle ontwikkeling van het kijkgedrag en de aandachtsprocessen plaats, en rond de leeftijd van 5 maanden beschikken baby’s over effectieve controle over hun oogbewegingen. Deze ontwikkeling wordt onder meer bepaald door de rijping van verschillende hersenstructuren en door omgevingsfactoren.

In het dagelijkse leven van een baby vervult het kijken een belangrijke rol. Omdat jonge baby’s zich nog niet kunnen voortbewegen of makkelijk een voorwerp kunnen vastpakken, ontdekken ze hun omgeving vooral door er naar te kijken. Ook vormt het maken van oogcontact de basis voor de vroege interactie en communicatie tussen de baby en zijn verzorgers. Daarbij wordt de intensiteit van de visuele stimulatie tijdens de interactie of tijdens het kijken naar een stimulus gereguleerd door telkens kort de blik af te wenden. Gezien de belangrijke functie van het kijkgedrag in het dagelijkse leven, is inzicht in de ontwikkeling van visuele aandachtsprocessen en kijkgedrag ook van belang voor ons begrip van de bredere cognitieve en sociale ontwikkeling.

Aanleiding en doel van het onderzoekDoel van dit onderzoek was de ontwikkeling van twee fundamentele aspecten van het

kijkgedrag – het scannen en het losmaken van aandacht en blik (disengagement) – tijdens de eerste 6 levensmaanden te beschrijven. Om de timing van deze ontwikkelingen in kaart te brengen, is gekozen voor een longitudinaal design. Op deze manier is het mogelijk om zowel individuele als groepsontwikkelingstrajecten te onderzoeken.

Het doel van dit onderzoek was tevens om de invloed van verschillende factoren op de vroege ontwikkeling van visueel gedrag te bestuderen. Ten eerste is het effect van de aard van de visuele stimuli onderzocht. Omdat de meeste onderzoeken tot nu toe gebruik hebben gemaakt van abstracte afbeeldingen als stimuli, is ervoor gekozen om naast een abstracte stimulus ook een natuurlijke, bewegende stimulus aan te bieden. Aangezien gezichten in het dagelijkse leven van een baby een belangrijke rol spelen, is daartoe een video-opname van het gezicht van de moeder van de baby gebruikt.

Ten tweede is het effect van vroege visuele ervaring op de ontwikkeling van het kijkgedrag van jonge baby’s onderzocht. Baby’s die te vroeg geboren zijn, worden vroeger in hun ontwikkeling aan visuele stimulatie blootgesteld dan op tijd geboren baby’s. Omdat vroege patronen van visuele aandacht sterk gerelateerd zijn aan de intellectuele ontwikkeling op latere leeftijd, is het van groot belang om naast de normale

148

Samenvatting

149

Samenvatting

ontwikkeling van visuele aandachtsprocessen ook de ontwikkeling bij risicokinderen te onderzoeken.

In de laatste jaren zijn de mogelijkheden om oogbewegingen, fixaties en blikrichting nauwkeurig te meten sterk verbeterd. In vergelijking met eerdere methodes bestaat er nu de gelegenheid om metingen met een betere tijdsresolutie en betere ruimtelijke precisie uit te voeren, die tegelijkertijd een veel kleinere ingreep in het natuurlijke gedrag van de baby betekenen. In dit onderzoek werden daarom oogbewegingen met behulp van een eye-tracker gemeten en deze metingen door nauwkeurige observaties gecompleteerd.

Het onderzochte kijkgedragDit proefschrift richt zich op de ontwikkeling van twee vormen van kijkgedrag:

scannen en disengagement. Scannen is het patroon van oogbewegingen waarmee een stimulus verkend wordt. Disengagement heeft betrekking op het loskoppelen van aandacht en blik van een stimulus om vervolgens aandacht en blik op een andere stimulus te richten.

De werkwijzeAan het onderzoek namen 20 op tijd geboren en 10 te vroeg geboren baby’s deel.

De premature baby’s waren na een zwangerschapsduur van 27-32 weken geboren. Bij de op tijd geboren baby’s waren er geen medische complicaties tijdens zwangerschap en geboorte opgetreden. In totaal werden de baby’s 6 keer onderzocht. Om de 4 weken bezochten de baby’s het laboratorium; bij het eerste onderzoek waren zij 6 weken oud en bij het laatste onderzoek 26 weken. Hierbij werd uitgegaan van gecorrigeerde leeftijden, dat wil zeggen dat de leeftijd bepaald werd aan de hand van de uitgerekende geboortedatum.

Bij elk bezoek aan het laboratorium voerden de baby’s twee experimentele taken uit. Om het scangedrag te onderzoeken, keken de baby’s naar verschillende stimuli die op een monitor werden aangeboden. Daarbij werden hun oogbewegingen geregistreerd. De ene stimulus was een korte video-opname van het gezicht van hun moeder. In deze opname knikte, glimlachte en keek de moeder zoals zij dit tijdens de interactie met haar baby zou doen. De tweede stimulus was abstract en met behulp van een grafiekprogramma zo gemaakt, dat hij op vrijwel alle fysieke kenmerken (zoals kleuren, lichtheid en bewegingsdynamiek) vergelijkbaar was met de eerste video.

Dezelfde stimuli werden, in kleiner formaat, gebruikt in de taak waarin disengagement gemeten werd. In dit experiment werd eerst de blik van de baby naar een stimulus in zijn centrale visuele veld getrokken. Daarna verscheen een tweede stimulus in het perifere gezichtsveld van de baby. Om naar deze stimulus te kijken, moest de baby zijn aandacht en zijn blik loskoppelen van de eerste stimulus. Deze conditie werd vergeleken met een controleconditie, waarin de eerste stimulus verdween op het moment dat de tweede verscheen. Deze controleconditie doet dus geen beroep op de vaardigheid om

150

Samenvatting

151

Samenvatting

de aandacht en de blik los te maken om naar de perifere stimulus te kunnen kijken. Twee keer tijdens de onderzoeksperiode, namelijk op het moment dat de baby’s 12

en 24 weken oud waren, werd met behulp van de Bayley Ontwikkelingsschalen het ontwikkelingsniveau van de op tijd geboren en de premature baby’s gemeten, om vast te stellen of er geen sprake was van een afwijkende ontwikkeling.

DEBEVINDINGEN

De ontwikkeling van het scannen bij op tijd geboren baby’sIn Hoofdstuk 2 wordt onderzocht hoe jonge baby’s de twee in dit onderzoek gebruikte

stimuli scannen en hoe hun scangedrag tijdens de eerste 6 levensmaanden verandert. Wanneer jonge baby’s stimuli bekijken, fixeren zij meestal maar enkele locaties en kijken niet naar het binnengedeelte van een stimulus, maar vooral naar de randen. Vanaf ongeveer 3 maanden beginnen baby’s een flexibeler scangedrag te vertonen en bekijken zij in toenemende mate ook verschillende en verder uit elkaar liggende onderdelen van een stimulus. Vergelijkbare resultaten zijn gevonden voor de wijze waarop baby’s gezichten scannen. In deze experimenten heeft men echter haast altijd gebruik gemaakt van voor de baby minder natuurlijke afbeeldingen van gezichten, zoals niet-bewegende gezichten of foto’s.

Gedurende het experiment werden de oogbewegingen van de baby’s gemeten, zodat aantal, duur en locatie van de fixaties bepaald konden worden. Wanneer de baby’s nog jong waren, vertoonden ze vaak weinig en lange fixaties, maar na 14 weken werden er meer en kortere fixaties gemeten. Ook keken zij tijdens de inspectie van een stimulus vaker kort weg. In dit onderzoek stabiliseerde de manier van scannen zich pas rond de leeftijd van 18 weken, later dan wat in ander onderzoek gerapporteerd is met betrekking tot het scannen van niet-bewegende stimuli. Het is dus aannemelijk dat het scannen van een bewegende stimulus voor de baby’s een grotere uitdaging vormt. Dit effect was voornamelijk aanwezig voor de abstracte stimulus. Vanaf 14 weken trad er een verschil op in de gemiddelde fixatieduur tijdens het scannen van de moeder ten opzichte van de abstracte stimulus. Dit duidt erop dat de baby’s vanaf deze leeftijd hun scangedrag aan de kenmerken van de stimuli aanpasten.

Tijdens het bekijken van het gezicht van hun moeder keken de baby’s het vaakst naar de mond en de ogen. In tegenstelling tot resultaten uit eerder onderzoek met niet-bewegende of niet-realistische afbeeldingen van gezichten keken de baby’s in dit onderzoek zelfs rond de leeftijd van 6 weken al naar de ogen en de mond en niet alleen naar de gezichtsomtrekken. Tussen 10 en 14 weken verbeterde de kwaliteit van het scannen sterk. Staren trad niet meer op, en de baby’s keken vrijwel de gehele tijd naar de meest betekenisvolle onderdelen van het gezicht, namelijk de ogen en de mond.

De ontwikkeling van disengagement bij op tijd geboren baby’sIn Hoofdstuk 3 staat de ontwikkeling van de vaardigheid om aandacht en blik tussen

twee stimuli te verschuiven centraal. Baby’s van 1 à 2 maanden hebben er moeite mee

150

Samenvatting

151

Samenvatting

hun blik van een stimulus los te maken en blijven vaak langdurig naar een stimulus staren. Rond 3 à 4 maanden ontwikkelt disengagement zich. Dit vormt een belangrijke stap in de richting van efficiënt gedrag op de terreinen van visuele aandacht, cognitie en zelfregulatie. In bestaand onderzoek komt naar voren, dat de kenmerken van de gebruikte stimuli van invloed zijn op het kijkgedrag van een baby, maar dit verband tussen stimuluskenmerken en kijkgedrag is nooit eerder systematisch onderzocht. Het doel van deze studie was dus om te onderzoeken hoe stimuluskenmerken disengagement bij jonge baby’s beïnvloeden en hoe dit verandert naarmate de baby’s ouder worden.

Zoals reeds beschreven werden twee verschillende stimuli gebruikt, namelijk een korte film van het gezicht van de moeder van de baby en een eveneens bewegende abstracte stimulus. De films werden aangeboden in een computertaak, waarin een blikverschuiving werd uitgelokt. In de disengagementtaak bleef de eerste stimulus zichtbaar nadat de tweede perifere stimulus verscheen, terwijl in de controleconditie de twee stimuli nooit tegelijkertijd zichtbaar waren en de baby zijn blik dus niet hoefde los te maken. Beide stimulussoorten werden als centrale stimulus en als perifere stimulus gebruikt. Dit leidde tot vier verschillende combinaties van stimuli (moeder-moeder, moeder-abstract, abstract-moeder en abstract-abstract). De frequentie en de snelheid van blikverschuivingen van de centrale naar de perifere stimulus werden gemeten.

In de controleconditie verschoven de baby’s al vanaf 10 weken geregeld hun blik naar de perifere stimulus. De snelheid waarmee de oogbewegingen inzetten, verbeterde significant tussen de leeftijd van 6 tot 22 weken en stabiliseerde zich erna. In deze conditie was er geen effect van de verschillende stimuluscombinaties.

In de conditie waarin de centrale stimulus zichtbaar bleef en die het loskoppelen van aandacht en blik vereiste, keken de jonge baby’s minder vaak en langzamer naar de perifere stimulus dan in de controleconditie. De frequentie en de snelheid van de blikverschuiving namen echter tussen 6 en 22 weken toe. In deze conditie werd een duidelijk effect van de stimuluscombinaties gevonden. De baby’s keken vaker naar de perifere stimulus, wanneer de centrale stimulus het gezicht van hun moeder vertoonde en de perifere stimulus abstract was. Minder vaak en trager keken de baby’s naar de perifere stimulus bij de omgekeerde stimuluscombinatie (abstract-moeder). De verschillen tussen de stimuluscombinaties waren het meest markant tussen 10 en 18 weken – de periode waarin disengagement nog niet volledig ontwikkeld is. Hierbij bleek het filmpje van het gezicht van de moeder voor de baby’s minder aantrekkelijk te zijn. Dit heeft er waarschijnlijk mee te maken, dat voor baby’s van deze leeftijd het gezicht van de eigen moeder reeds erg goed bekend is en nieuwe stimuli interessanter worden. Voorts tonen de bevindingen aan dat de kenmerken van zowel de centrale als de perifere stimulus het kijkgedrag van de baby’s beïnvloedden.

Verbanden tussen het ontwikkelingsverloop van het scannen en disengagement Hoofdstuk 4 is gewijd aan de relatie en eventuele verbanden tussen het ontstaan

van efficiënt scangedrag en de ontwikkeling van disengagement. Hoewel deze twee

152

Samenvatting

153

Samenvatting

ontwikkelingstrajecten in ongeveer dezelfde periode van de vroege kindertijd vallen, wijst eerder onderzoek erop dat scannen en disengagement op verschillende hersenstructuren gebaseerd zijn. Disengagement wordt in verbinding gebracht met de magno-cellulaire route en de dorsale stroom, terwijl de georganiseerde patronen van oogbewegingen tijdens het scannen gemedieerd worden door de parvo-cellulaire route en de ventrale stroom.

Er zijn aanwijzingen dat de hersenstructuren waardoor het scannen gemedieerd wordt zich enigszins eerder ontwikkelen dan de structuren die de basis vormen voor disengagement. Aan de andere kant lijkt het erop dat de twee hersensystemen – in ieder geval bij volwassenen – niet strikt gescheiden zijn, maar in bepaalde mate met elkaar in verband staan.

Om de timing van de ontwikkelingstrajecten te vergelijken en eventuele verbanden te onderzoeken, werden de gegevens van de oogbewegingen tijdens het scannen en tijdens de disengagementtaak samen geanalyseerd. Hieruit bleek dat de veranderingen wat betreft het scannen – conform de verwachting – eerder tot rust kwamen dan de ontwikkeling van disengagement. Er waren geen aanwijzingen voor een positieve samenhang tussen de ontwikkeling van scangedrag en disengagement; de veranderingen in de twee gedragingen van één en dezelfde baby vertoonden zelfs een groter verschil dan men op basis van toeval zou verwachten. Dit wijst erop dat de hersenstructuren waarop het scannen en disengagement gebaseerd zijn zich in hoge mate onafhankelijk van elkaar ontwikkelen.

De ontwikkeling van disengagement bij te vroeg geboren baby’sDe ontwikkeling van disengagement bij op tijd en te vroeg geboren baby’s wordt

in Hoofdstuk 5 vergeleken. Te vroeg geboren baby’s zijn vroeger in hun ontwikkeling aan visuele indrukken blootgesteld dan op tijd geboren baby’s. Aan de ene kant zou deze extra ervaring een positieve invloed op de visuele ontwikkeling van de te vroeg geboren baby’s kunnen hebben, aan de andere kant zou een te vroege visuele stimulatie ook schadelijk kunnen zijn voor het nog onrijpe visuele systeem.

De ontwikkeling van de frequentie en de snelheid van blikverschuivingen bij te vroeg geboren baby’s kwam in grote lijnen overeen met de ontwikkeling bij de op tijd geboren baby’s. Ook de effecten van de verschillende stimuluscombinaties waren bij de premature en de op tijd geboren baby’s vergelijkbaar.

Er waren echter ook verschillen tussen de twee groepen: Met 6 weken vertoonden de premature baby’s meer staren dan de op tijd geboren baby’s, zowel in de disengagement- als in de controletaak. Tegelijkertijd waren de te vroeg geboren baby’s op jonge leeftijd echter sneller wanneer zij hun blik naar de perifere stimulus verschoven.

De bevindingen laten dus geen eenduidige conclusies met betrekking tot een potentieel positief of schadelijk effect van vroege visuele ervaring toe, maar tonen aan dat prematuriteit samengaat met een licht afwijkende karakteristiek van de

152

Samenvatting

153

Samenvatting

ontwikkeling van disengagement. De verschillen waren echter tijdelijk; tegen het einde van de onderzoeksperiode waren er geen verschillen meer aantoonbaar tussen te vroeg en op tijd geboren baby’s.

CONCLUSIESDe doelstelling van dit onderzoek was de ontwikkeling van twee vormen van

visueel gedrag – scannen en disengagement – tijdens de eerste 6 levensmaanden in kaart te brengen. Beide gedragingen ontwikkelden zich snel in de periode tussen 6 en 16 weken. Terwijl er voor het scannen hierna geen veranderingen meer gevonden werden, bleef disengagement zich ook na de leeftijd van 16 weken, in afnemende mate, ontwikkelen.

In dit proefschrift werd in het bijzonder ook de invloed van verschillende stimuli en stimuluscombinaties onderzocht. Wat betreft het scannen, kon door het gebruik van een realistische gezichtsstimulus aangetoond worden, dat ook baby’s van 6 weken oud al geregeld naar de belangrijke binnengebieden van een gezicht kijken. De langzame, bijna overdreven manier waarop moeders vaak met hun baby interageren en praten maakt het waarschijnlijk ook voor de nog jonge baby’s gemakkelijker, de sterke aantrekkingskracht van de omtrekken van een gezicht te overwinnen. Tevens waren er aanwijzingen dat baby’s vanaf de leeftijd van 14 weken hun kijkgedrag aan de kenmerken van de stimulus aanpasten.

Ook verschilden de geobserveerde ontwikkelingstrajecten van het scannen en van disengagement voor de verschillende stimuli en stimuluscombinaties. De ontwikkeling van meer gevorderd scangedrag verliep trager voor de abstracte stimulus, en in de disengagementtaak hadden de op tijd geboren baby’s, evenals de te vroeg geboren baby’s, langer moeite om van de abstracte stimulus naar de moederstimulus te kijken dan andersom. Deze bevindingen laten zien dat een vaardigheid die nog niet goed ontwikkeld is bijzonder gevoelig is voor omgevingsinvloeden.

Verder laten deze resultaten ook conclusies over onderliggende aandachts-mechanismen toe. In de disengagementtaak beïnvloedden zowel de centrale als de perifere stimulus het kijkgedrag van de baby. Terwijl de baby’s naar de stimulus in hun centrale visuele veld keken, werd dus ook de stimulus in de periferie verwerkt en kon deze het kijkgedrag mede bepalen.

De vergelijking van de ontwikkeling van disengagement bij op tijd en te vroeg geboren baby’s leverde geen duidelijke aanwijzing op voor een gunstig of een schadelijk effect van vroege visuele ervaring. Het werd echter duidelijk dat premature baby’s kleine, niet eenduidige verschillen vertonen in hun ontwikkeling.

In Hoofdstuk 6 worden naast een samenvatting van de bevindingen van dit onderzoek suggesties voor vervolgonderzoek gegeven. Deze hebben met name betrekking op het beter begrijpen van aandachtsprocessen en kijkgedrag en de relevantie van deze processen voor de vroege ontwikkeling.

Zusammenfassung

156

Zusammenfassung

157

Zusammenfassung

EINLEITUNGJeder, der einmal beobachtet hat, wie ein wenige Wochen altes Baby seine Umgebung

erkundet, hat sicherlich bemerkt, daß sich das Blickverhalten eines Babys stark von dem eines Erwachsenen unterscheidet. Bereits Neugeborene sind visuell aktiv und betrachten Stimuli, die sich in ihrem Blickfeld befinden. Hierbei treten jedoch häufig lange Perioden auf, während derer ihr Blick an einem Stimulus zu „haften“ scheint. Oft betrachten sie jedoch nur einen oder zwei der Stimuli in ihrem Blickfeld, und wenn sie einen Stimulus anschauen, dann erkunden sie ihn oft nur partiell. Während der ersten Lebensmonate findet eine schnelle Entwicklung des Blickverhaltens und der Aufmerksamkeitsprozesse statt, und mit ungefähr 5 Monaten kann ein Baby seine Augenbewegungen auf effiziente Art und Weise kontrollieren. Diese Entwicklung wird unter anderem von der Reifung verschiedener Hirnstrukturen und von Umgebungsfaktoren bestimmt.

Im täglichen Leben eines Babys spielt das Sehen eine wichtige Rolle. Da junge Babys sich noch nicht selbständig fortbewegen oder einfach einen Gegenstand in die Hand nehmen können, entdecken sie ihre Umgebung vor allem mit den Augen. Das Aufnehmen von Blickkontakt bildet die Grundlage für die frühe Kommunikation und Interaktion mit anderen Menschen wie z.B. den Eltern. Dabei wird die Intensität der visuellen Stimulation dadurch reguliert, daß der Blick regelmäßig kurz abgewendet wird. Angesichts der bedeutenden Rolle, die Sehen und Augenbewegungen im täglichen Leben spielen, sind Erkenntnisse über die Entwicklung von visuellen Aufmerksamkeitsprozessen und Blickverhalten auch für unser Verständnis der allgemeinen kognitiven und sozialen Entwicklung wichtig.

Anlass und Ziel der UntersuchungZiel der Dissertation ist es, die Entwicklung von zwei fundamentalen Aspekten des

Blickverhaltens während der ersten sechs Lebensmonate zu beschreiben, nämlich die Entwicklung des Scanverhaltens und des Lösens von Aufmerksamkeit und Blick (Disengagement). Um den Verlauf dieser Entwicklungen beschreiben zu können, wurde ein Längsschnittdesign verwendet. Auf diese Art und Weise ist es möglich, sowohl individuelle Entwicklungsverläufe als auch Gruppenverläufe zu untersuchen.

Ziel war es außerdem, den Einfluß unterschiedlicher Faktoren auf die frühe Entwicklung des Blickverhaltens zu erforschen. Darum wurde zum einen die Wirkung verschiedener Stimuli untersucht. Da bisherige Studien zur Aufmerksamkeitsentwicklung bei Babys zumeist lediglich abstrakte Abbildungen als Stimuli verwendeten, legt die vorliegende Untersuchung neben einem abstrakten auch einen natürlichen, sozial relevanten Stimulus zugrunde. Menschliche Gesichter gehören zu den wichtigsten Stimuli im täglichen Leben eines Babys. Sie betrachten Gesichter bereits als Neugeborene mit Interesse, bald darauf reagieren sie mit einem Lächeln auf menschliche Gesichter. Darum wurden in dieser Untersuchung Videoaufnahmen des Gesichts der Mutter als Stimuli verwendet.

156

Zusammenfassung

157

Zusammenfassung

Zum anderen wurde die Wirkung früher visueller Erfahrung auf die Entwicklung des Blickverhaltens untersucht. Frühgeborene sind eher in ihrer Entwicklung visueller Stimulation ausgesetzt als Reifgeborene. Da das frühe Aufmerksamkeitsverhalten mit der späteren kognitiven Entwicklung im Zusammenhang steht, ist es von besonderer Bedeutung, die „normale“ Entwicklung von Aufmerksamkeitsprozessen mit der bei Risikokindern zu vergleichen.

Während der letzten Jahre haben sich die Möglichkeiten, Augenbewegungen, Fixationen und Blickrichtung auch bereits bei sehr jungen Kindern präzise zu messen, stark verbessert. Im Vergleich zu früheren Vorgehensweisen haben die Messungen nun eine bessere Zeitresolution und eine größere räumliche Genauigkeit und erfordern gleichzeitig einen geringeren Eingriff in das natürliche Verhalten des Babys. In den vorliegenden Studien wurden daher Augenbewegungen mit Hilfe eines eye-trackers gemessen. Diese Messungen wurden durch Videoaufnahmen ergänzt.

Das untersuchte BlickverhaltenDie vorliegende Dissertation hat die Entwicklung von zwei Formen des

Blickverhaltens zum Thema: Scannen und Disengagement. Unter Scannen wird das Muster von Augenbewegungen während der Betrachtung eines Stimulus verstanden. Disengagement bezeichnet das Lösen von Aufmerksamkeit und Blick, um den Blick von einem Stimulus auf einen anderen zu richten.

Die VorgehensweiseAn den Untersuchungen nahmen 20 reifgeborene und 10 frühgeborene Babys teil.

Die Frühgeborenen waren nach einer Gestationszeit von 27 bis 32 Wochen geboren. Bei den Reifgeborenen hingegen durften keinerlei medizinische Komplikationen während Schwangerschaft oder Geburt aufgetreten sein. Alle Babys wurden sechsmal getestet; hierzu kamen die Mütter mit ihnen alle 4 Wochen ins Babylab. Beim ersten Besuch waren die Babys 6 Wochen alt, beim letzten 26 Wochen. Zur Bestimmung des Alters wurde vom korrigierten Alter ausgegangen, d.h. für das Alter des Kindes wurde der errechnete Geburtstermin zugrundegelegt.

Bei jedem der Untersuchungstermine im Babylab wurden zwei Experimente durchgeführt. Um das Scanverhalten zu untersuchen, wurden den Babys zwei verschiedene Stimuli auf einem Computermonitor gezeigt. Während die Babys diese betrachteten, wurden ihre Augenbewegungen registriert. Bei dem einen Stimulus handelte es sich um eine kurze Videoaufnahme des Gesichts der Mutter. In diesem Film lächelte, schaute und nickte die Mutter, wie sie es auch in einer direkten Interaktion mit ihrem Kind tun würde. Der zweite Stimulus war abstrakt und mit Hilfe eines Graphikprogramms so hergestellt, daß er in nahezu allen physikalischen Merkmalen (wie Farbe, Helligkeit und Bewegungsdynamik) mit dem ersten vergleichbar war.

Dieselben Stimuli wurden – in einem kleineren Format – auch im zweiten Experiment, in dem Disengagement untersucht wurde, verwendet. In diesem

158

Zusammenfassung

159

Zusammenfassung

Experiment wurde dem Baby auf dem Computermonitor zunächst ein Stimulus in seinem zentralen Blickfeld gezeigt. Wenn es diesen anschaute, erschien ein zweiter Stimulus im peripheren Gesichtsfeld des Babys. Um diesen Stimulus anschauen zu können, mußte das Baby also seine Aufmerksamkeit und seinen Blick vom ersten Stimulus lösen. Diese Bedingung wurde verglichen mit einer Kontrollbedingung, in der der erste Stimulus in dem Moment, in dem der zweite erschien, verschwand. In dieser Kontrollbedingung brauchte das Baby seine Aufmerksamkeit und seinen Blick nicht zu lösen, um den neuen Stimulus anzuschauen, sondern lediglich zu verschieben.

Zweimal während des Untersuchungszeitraums, nämlich im Alter von 12 und 24 Wochen, wurden die Babys mit dem Bayley-Entwicklungstest untersucht, um sicher zu stellen, daß weder bei einem der reifgeborenen noch der frühgeborenen Kinder eine abweichende Entwicklung vorlag.

DIEERGEBNISSE

Die Entwicklung des Scanverhaltens bei reifgeborenen KindernIn Kapitel 2 wird untersucht, wie Babys die beiden Stimuli scannten und wie ihr

Scanverhalten sich während der ersten sechs Lebensmonate veränderte. Wenn junge Babys einen Stimulus betrachten, fixieren sie zumeist nur einige Punkte und betrachten auch nicht das Innere eines Stimulus, sondern vor allem seine Umrisse. Im Alter von ungefähr 3 Monaten beginnen Babys, ein flexibleres Scanverhalten zu zeigen, und betrachten in zunehmendem Maß auch verschiedene, weiter auseinander liegende Teile eines Stimulus. Vergleichbares wird in der Literatur für das Scannen von Gesichtern beschrieben. Allerdings beruhen diese Aussagen größtenteils auf Experimenten, die nicht natürlicheAbbildungen von Gesichtern verwendet haben wie z.B. Photos oder Zeichnungen.

Während des Experiments wurden die Augenbewegungen der Babys gemessen, so daß Anzahl, Dauer und Ort der Fixationen bestimmt werden konnten. Sehr junge Babys zeigten häufig wenige und lange Fixationen, ab dem Alter von 14 Wochen jedoch wurden mehr und kürzere Fixationen gemessen. Auch wendeten sie ihren Blick immer häufiger kurz ab, während sie den Stimulus betrachteten. In dieser Studie stabilisierte sich das Scanverhalten der Babys erst ab dem Alter von 18 Wochen, um einiges später also, als in der Literatur über das Scannen von sich nicht bewegenden Stimuli beschrieben wird. Es ist also wahrscheinlich, daß das Scannen von sich bewegenden Stimuli eine größere Herausforderung für die Babys darstellte. Dieser Effekt war am stärksten für den abstrakten Stimulus. Auch ergab sich mit dem Alter von 14 Wochen ein Unterschied in der mittleren Dauer der Fixationen während des Scannens des abstrakten und des Mutterstimulus. Dies deutet darauf hin, daß die Babys ab diesem Alter ihr Scanverhalten an die Merkmale der Stimuli anpaßten.

Während die Babys das Gesicht ihrer Mutter betrachteten, schauten sie am häufigsten auf den Mund und die Augen. Im Gegensatz zu Befunden aus früheren Studien mit sich nicht bewegenden oder nicht realistischen Bildern von Gesichtern,

158

Zusammenfassung

159

Zusammenfassung

betrachteten die Babys in der vorliegenden Arbeit bereits im Alter von 6 Wochen häufig die Augen und den Mund ihrer Mutter und nicht lediglich die Umrisse ihres Gesichts. Zwischen 10 und 14 Wochen verbesserte sich die Qualität des Scanverhaltens deutlich. Perioden, in denen die Babys auf eine Stelle des Gesichts starrten, traten nicht mehr auf. Statt dessen betrachteten sie nahezu die ganze Zeit die bedeutsamsten Bereiche des Gesichts, nämlich die Augen und den Mund.

Die Entwicklung von Disengagement bei reifgeborenen BabysIn Kapitel 3 steht die Entwicklung von Disengagement im Mittelpunkt. Babys haben

im Alter von 1 bis 2 Monaten Schwierigkeiten, ihren Blick von einem Stimulus zu lösen, und starren darum häufig lange auf ein und denselben Punkt. Mit ungefähr 3 bis 4 Monaten entwickelt sich Disengagement. Dies bedeutet einen wichtigen Schritt hin zu einem effizienten Verhalten auf den Gebieten der visuellen Aufmerksamkeit, aber auch der Kognition und Selbstregulation. In früheren Studien hat sich gezeigt, daß die Merkmale der verwendeten Stimuli Einfluß auf das Blickverhalten haben; dieser Zusammenhang zwischen Stimuluscharakteristika und Blickverhalten ist jedoch bisher noch nicht systematisch erforscht worden. Das Ziel dieser Untersuchung war es deshalb, zu klären, inwieweit Stimulusmerkmale Disengagement bei Babys beeinflussen und wie sich dieser Einfluß verändert, wenn die Babys älter werden.

Wie bereits beschrieben, wurden in dieser Studie zwei verschiedene Stimuli verwendet, nämlich ein kurzer Film, in dem das Gesicht der Mutterzu sehen war, und ein ebenfalls dynamischer, abstrakter Stimulus. Die Filme wurden im Rahmen eines Computerexperiments gezeigt, das so aufgebaut war, daß die Babys herausgefordert wurden, ihren Blick zwischen zwei Stimuli zu verschieben. In der Disengagementbedingung blieb der erste Stimulus sichtbar, nachdem der zweite erschienen war, während in der Kontrollbedingung die zwei Stimuli nie gleichzeitig zu sehen waren und das Baby seinen Blick also nicht zu lösen brauchte. Beide Stimuli wurden als zentraler und als peripherer Stimulus verwendet, was zu vier verschiedenen Kombinationen führte (Mutter-Mutter, Mutter-Abstrakt, Abstrakt-Mutter und Abstrakt-Abstrakt). Die Häufigkeit und Geschwindigkeit der Blickverschiebungen zum peripheren Stimulus hin wurden gemessen.

In der Kontrollbedingung schauten die Babys bereits im Alter von 10 Wochen regelmäßig zu dem neu erschienenen peripheren Stimulus. Die Geschwindigkeit, mit der diese Augenbewegungen einsetzten, erhöhte sich signifikant zwischen 6 und 22 Wochen und stabilisierte sich danach. In dieser Bedingung hatten die verschiedenen Stimuluskombinationen keinen Effekt.

Wenn der zentrale Stimulus sichtbar blieb und das Lösen von Aufmerksamkeit und Blick erforderlich war, schauten die Babys zunächst weniger oft und weniger schnell zum peripheren Stimulus als in der Kontrollbedingung. Die Häufigkeit und Geschwindigkeit der Blickverschiebungen nahmen jedoch zwischen 6 und 22 Wochen zu. In dieser Bedingung wurde auch ein deutlicher Effekt der Stimuluskombinationen

160

Zusammenfassung

161

Zusammenfassung

sichtbar. Die Babys schauten häufiger zum peripheren Stimulus, wenn der zentrale das Gesicht der Mutter zeigte und der periphere abstrakt war. Weniger häufig und langsamer schauten die Babys zum peripheren Stimulus bei der entgegengesetzten Kombination (Abstrakt-Mutter). Diese Unterschiede zwischen den verschiedenen Stimuluskombinationen waren am markantesten in der Zeit zwischen 10 und 18 Wochen – dem Zeitraum also, in dem Disengagement noch nicht vollständig entwickelt war. Der Film mit dem Gesicht ihrer Mutter war für die Babys weniger attraktiv als der abstrakte Stimulus. Dies hat seinen Grund wahrscheinlich in der Tatsache, daß den Babys diesen Alters das Gesicht ihrer Mutter bereits sehr vertraut ist und neue, unbekannt Stimuli für sie interessanter werden. Weiterhin zeigen die Ergebnisse, daß sowohl der zentrale als auch der periphere Stimulus das Blickverhalten der Babys bestimmten.

Zusammenhänge zwischen der Entwicklung von Scannen und DisengagementKapitel 4 ist dem Zusammenhang zwischen der Entstehung von effizientem

Scanverhalten und der Entwicklung von Disengagement gewidmet. Obwohl die beiden Entwicklungsverläufe in ungefähr dieselbe Altersperiode der frühen Kindheit fallen, weisen Studien darauf hin, daß sich Scannen und Disengagement auf unterschiedliche Hirnstrukturen gründen. Disengagement wird wahrscheinlich vom magnozellulären System und dem dorsalen Strang unterstützt. Die organisierten Muster von Augenbewegungen während des Scannens werden dagegen mit dem parvozellulären System und dem ventralen Strang in Zusammenhang gebracht.

Forschungsergebnisse legen nahe, daß sich die Hirnstrukturen, die mit Disengagement in Zusammenhang stehen, geringfügig früher entwickeln als die Strukturen, die die Basis für Disengagement bilden. Auf der anderen Seite scheinen die beiden Systeme – zumindest bei Erwachsenen – nicht streng unabhängig voneinander zu funktionieren, sondern zu interagieren.

Um die Entwicklungsverläufe vergleichen und eventuelle Zusammenhänge untersuchen zu können, wurden die Ergebnisse des Scannens und des Disengagement zusammen analysiert. Hierbei stellte sich heraus, daß die Entwicklung des Scanverhaltens – entsprechend den Erwartungen – schneller abgeschlossen war als die des Disengagement. Es ergaben sich keine Hinweise auf einen positiven Zusammenhang zwischen den beiden Entwicklungsverläufen. Die Veränderungen bezüglich Scannen und Disengagement wiesen auf dem Niveau der individuellen Versuchspersonen sogar größere Unterschiede auf, als zufällig zu erwarten wären. Dies deutet darauf hin, daß sich die beiden Hirnsysteme in dieser Zeit in hohem Maße unabhängig von einander entwickeln.

Die Entwicklung von Disengagement bei frühgeborenen BabysDie Entwicklung von Disengagement bei frühgeborenen und reifgeborenen Babys

wird in Kapitel 5 verglichen. Frühgeborene Babys sind früher in ihrer Entwicklung

160

Zusammenfassung

161

Zusammenfassung

visuellen Eindrücken ausgesetzt als reifgeborene. Auf der einen Seite könnte diese zusätzliche Erfahrung einen positiven Einfluß auf ihre visuelle Entwicklung haben, auf der anderen Seite könnte die vorzeitige visuelle Stimulation dem noch unreifen visuellen System auch Schaden zufügen.

Die Entwicklung der Häufigkeit und der Geschwindigkeit der Blickverschiebungen bei Frühgeborenen stimmte im großen und ganzen mit der der Reifgeborenen überein. Auch die Effekte der verschiedenen Stimuluskombinationen waren vergleichbar.

Es ergaben sich jedoch auch Unterschiede zwischen den beiden Gruppen: Mit 6 Wochen war bei den Frühgeborenen – sowohl in der Disengagementbedingung als auch in der Kontrollbedingung – mehr Starren zu beobachten als bei den Reifgeborenen. Andererseits waren die Frühgeborenen mit 6 Wochen wiederum schneller darin, ihren Blick zu verschieben, als die Reifgeborenen.

Die Ergebnisse lassen also keine eindeutigen Schlüsse in Bezug auf einen potentiell positiven oder schädigenden Effekt früher visueller Erfahrung zu, zeigen jedoch, daß eine Frühgeburt mit Veränderungen in der Entwicklung von Disengagement einhergeht. Die Unterschiede waren aber an die ersten Monate gebunden; zum Ende des Untersuchungszeitraumes hin waren keine Unterscheide zwischen Frühgeborenen und Reifgeborenen mehr nachzuweisen.

SCHLUSSFOLGERUNGENEs war das Ziel dieser Dissertation, die Entwicklung des Blickverhaltens – zugespitzt

auf Scannen und Disengagement – während der ersten sechs Lebensmonate zu untersuchen. Sowohl Scannen als auch Disengagement entwickelten sich im Alter zwischen 6 und 16 Wochen schnell. Während für das Scanverhalten nach dem Alter von 16 Wochen keine Veränderungen mehr auftraten, waren für Disengagement auch danach noch – wenn auch weniger hervorstechende – Veränderungen feststellbar.

Ferner wurde der Einfluß verschiedener Stimuli und Stimuluskombinationen untersucht. Was das Scannen betrifft, so konnte durch die Verwendung natürlicher Gesichter gezeigt werden, daß auch Babys im Alter von 6 Wochen bereits regelmäßig die wichtigsten inneren Regionen eines Gesichts betrachten. Die langsame, beinahe übertriebene Art und Weise, mit der Mütter häufig mit ihrem Baby interagieren und sprechen, erleichtert es wahrscheinlich den jüngeren Babys, die starke Anziehungskraft, die die Umrisse eines Gesichtes auf sie ausüben, zu überwinden. Es ergaben sich auch Hinweise darauf, daß Babys von 14 Wochen an ihr Blickverhalten an die Merkmale der betrachteten Stimuli anpassen.

Außerdem wurden unterschiedliche Entwicklungsverläufe des Scanverhaltens wie des Disengagements für die verschiedenen Stimuli und Stimuluskombinationen gefunden. Die Entwicklung von reiferem Scanverhalten verlief langsamer für den abstrakten Stimulus, und im Disengagementexperiment hatten sowohl die Reifgeborenen als auch die Frühgeborenen länger Schwierigkeiten, von einem abstrakten Stimulus zum Gesicht ihrer Mutter zu schauen als umgekehrt. Diese

162

Zusammenfassung

Ergebnisse zeigen, daß eine Fähigkeit, die noch nicht vollständig entwickelt ist, besonders empfindlich auf Umgebungseinflüsse reagiert.

Die Ergebnisse lassen weiterhin auch Schlüsse auf die zugrunde liegenden Aufmerksamkeitsprozesse zu. Im Disengagementexperiment beeinflußten sowohl der zentrale als auch der periphere Stimulus das Blickverhalten der Babys. Während das Baby den Stimulus in seinem zentralen Blickfeld betrachtete, wurde also auch der Stimulus in der Peripherie verarbeitet und konnte sich ebenfalls auf das Blickverhalten auswirken.

Der Vergleich der Entwicklung von Disengagement bei reif- und frühgeborenen Babys erbrachte keine eindeutigen Hinweise für einen positiven oder schädlichen Effekt früher visueller Erfahrung. Es wurde jedoch deutlich, daß Frühgeborene kleine, nicht eindeutige Unterschiede in der Entwicklung ihres Blickverhaltens aufwiesen.

In Kapitel 6 wird neben einer Zusammenfassung der Ergebnisse der vorliegenden Dissertation auch ein Ausblick auf mögliche Folgeuntersuchungen gegeben. Hierbei stehen ein besseres Verständnis von Aufmerksamkeitsprozessen und Blickverhalten sowie die Relevanz dieser Prozesse innerhalb der frühkindlichen Entwicklung im Mittelpunkt.

162

Zusammenfassung

165

Acknowledgements

165

Acknowledgements

ACKNOWLEDGEMENTSDuring the work on this PhD project, I received a lot of support, help, and advice

from the people around me. I have been looking forward for a long time to this op-portunity to thank everybody who has contributed to this dissertation.

First of all, I would like to thank my supervisor Reint Geuze. He guided me through all the steps of my project with great patience, and his sharp, deep way of thinking pushed my own insight steadily further.

My Doktorvater Paul van Geert provided creative criticism and helped me get a grip on my from time to time seemingly unmanageable data. His innovative view on developmental processes was a great inspiration to me and broadened my mind. I am deeply grateful to Anke Bouma, my second promotor, for many enlightening and fruitful discussions and her careful, critical comments on my manuscript.

Phillipa Butcher provided indispensable advice all along the way. I would like to thank her for sharing with me her incredible knowledge of early development and her experience with testing infants and for putting a new heart into me so many times.

I also thank Tom Snijders for introducing me to multilevel analysis and taking the time to help me with my models and to answer all my questions patiently.

When I traveled to the United States to learn more about eye-tracking with infants, I received a warm welcome there. I would like to thank Dick Aslin for his valuable advice and for being a member of my PhD committee. I am very grateful that I had the chance to visit his lab at the University of Rochester. During this time I learned a lot about eye-tracking and research with infants, particularly during the week I spent in the lab together with Bob McMurray, figuring out how to combine eye- and head-tracker in an experimental setup suitable for young infants. My thanks also go to Scott P. Johnson for helpful discussions and the opportunity to visit his lab at the time at Cornell University. I would also like to thank the two other members of the PhD committee, Addie Johnson and Geert Savelsbergh.

The financial support of the Graduate School of Behavioral and Cognitive Neurosci-ences (BCN) is also gratefully acknowledged.

Further, I would like to thank the pediatricians and neonatologists at the Univer-sity Hospital Groningen (AZG) and the Martini Hospital (MZH), especially Arie Bos, Mar Zweens, Lily van Doormaal, and Albert Martijn for putting together the sample of preterm infants and providing their clinical data.

Working at the Department of Developmental and Experimental Clinical Psychology of the University of Groningen was a wonderful experience. I enjoyed the stimulating, cooperative atmosphere and felt very much at home. I would like to thank my col-leagues for the interesting discussions and the gezellige Christmas lunches, department days out, and coffee breaks. Especially, I thank my fellow PhD students: Marijn van Dijk, Martine Verheul, Kitty Boosman, Cor van Halen, Els Blijd-Hoogewys, Norbert Boerger, Koen van Braeckel, Georgios Vleioras, Anna Lichtwark-Aschoff, Laura Ballato, Marieke Visser, Raphaela Carriere, Henderien Steenbeek (who shared the office and the daily

166

Acknowledgements

joy and sorrow with me), and Evelien Krikhaar (who taught me much more than my first words of Dutch and is by my side during the defense as my paranimf ).

I am very grateful to the staff of the instrumentation service of the Psychology Department (IDP): Hans Veldman, Jan R. Smit, Edwin Kiers, Pieter Zandbergen, Joop Clots, Eise Hoekstra, Bas Kortman, and Peter Albronda. They provided structural sup-port and were always willing to drop whatever they were doing to help in cases of emergency at the babylab.

I would also like to thank the students Ellin Simon, Thalia Albracht, Jenny Vogel, Lieke Drukker, Mieke van der Horn, and Femke Groen, who assisted during the mea-surement sessions, helped coding data, and set me thinking many times by asking tricky questions.

Also my new colleagues of the Psychology and Health Department at Tilburg Uni-versity were very supportive during the last few months, and I would like to thank in particular Anneloes van Baar for her lively interest, understanding, and profound advice.

I owe a great debt of gratitude to the little infants who form the sample of this study and especially to their mothers and fathers. Without their trust, their enthusiasm, and their sustained cooperation this research would not have been possible. It was not always as easy to get in contact with parents and infants who could participate in my study, so I thank everybody who helped me find babies: midwives, child birth education teachers, acquaintances and friends. My thanks also go to the dependable people of the transportation service of the University of Groningen, who made sure that parents and infants were picked up on time to come to the lab and brought home safely after the measurement sessions.

I am very grateful to Dick Visser for designing the cover and the layout of this dis-sertation, and I also thank him and Marjolijn de Wijk for their warm hospitality.

I thank all my friends, far away or close by, for going through these years with me, for their interest in my research, but much more important, for their efforts to let me forget about it. I also would like to thank my family for all their sympathy and encouragement. Especially, I thank my parents for reading every chapter of this the-sis I sent them with great interest, for correcting my German, and for their constant loving support during all this time.

And finally, there is Jacques Dane, who supported and helped me in so many ways that I don’t even know where to start. So I confine myself to thanking him from the bottom of my heart for loving me over these years.

166

Acknowledgements

169

Curriculum Vitae

169

Curriculum Vitae

CURRICULUMVITAESabine Hunnius was born in Bonn (Germany) in 1974. She studied Psychology at the

Free University Berlin and at UC Berkeley. Between 1995 and 1998 she also worked as a teaching assistant in statistics and methods. She wrote her MA thesis on children’s sociometric status and their expression of emotions during conflict situations and graduated in 1998. In the same year, she started her PhD project on the early develop-ment of functional visual attention at the Department of Developmental and Experi-mental Clinical Psychology at the University of Groningen for which she received a grant from the Ubbo Emmius Foundation.

Sabine Hunnius is currently working at the Department of Pediatric Psychology at Tilburg University.