Upload
anne-richards
View
215
Download
0
Embed Size (px)
Citation preview
ORIGINAL ARTICLE
Predicting and manipulating the incidence of inattentionalblindness
Anne Richards • Emily M. Hannon •
Nazanin Derakshan
Received: 26 February 2009 / Accepted: 17 December 2009 / Published online: 9 January 2010
� Springer-Verlag 2010
Abstract Inattentional blindness (IB) occurs when an
observer, who is engaged in a resource-consuming task,
fails to notice an unexpected although salient stimulus
appearing in their visual field. The incidence of IB is
affected by changes in stimulus-driven properties, but little
research has examined individual differences in IB pro-
pensity. We examine working memory capacity (WMC),
processing styles (flicker task), inhibition (Stroop task), and
training in predicting IB. WMC is associated with IB
(Experiments 1 and 2) but neither processing style
(Experiment 1) nor inhibition (Experiment 2) was associ-
ated. In Experiment 2, prior training on a task reduced the
incidence of IB compared to no prior training, and this
effect was significantly larger when trained on the same
tracking task as that used in the IB task rather than a
different task. We conclude that IB is related to WMC and
that training can influence the incidence of IB.
Introduction
Inattentional blindness (IB) occurs when an observer who
is engaged in a resource-consuming task fails to notice an
unexpected stimulus appearing in front of their eyes (Mack
& Rock, 1998; Most, Simons, Scholl, & Chabris, 2000;
Most et al., 2001; Neisser & Becklen, 1975; Simons &
Chabris, 1999). IB occurs in everyday life and can be of
minimal importance (e.g. failing to notice your friend at the
cinema as you search for a vacant seat) or can have cata-
strophic consequences (failing to notice a child crossing the
road). In a series of experiments, Mack and Rock (1998)
presented a primary task in which participants were
required to identify which of two lines of a cross, vertical
or horizontal, was longer. The cross was presented for
200 ms and was followed by a pattern mask. On the third
or fourth trial, an unexpected stimulus was simultaneously
presented in one of the four quadrants of the cross for the
same duration (200 ms). Mack and Rock found that many
people were unaware of the unexpected stimulus on the
critical trial, and depending on the type of visual display,
the incidence of IB ranged from 25 to 85%.
Some early research by Neisser and Becklen (1975)
used a dynamic IB task in which one video was superim-
posed upon another. Participants were required to monitor
one video (a hand slapping game) or the other (a basketball
game). An unexpected event occurred (i.e. the ball was
thrown out of play in the basketball game or the two
players shook hands in the hand slapping game), and most
people remained inattentionally blind to these events.
Later research by Neisser (1979) superimposed videos
of two basketball teams where one team wore white shirts
and the other black shirts. Participants were required to
monitor the ball passes of just one team, either the black
or white shirts. An unexpected, semi-transparent video
with a woman carrying an umbrella appeared after 30 s and
walked across the screen. Again, most participants
remained unaware of this unexpected stimulus (22 out of
28 were IB). Becklen and Cervone (1983) argued that this
failure to notice the unexpected stimulus was unlikely to be
due to memory failure, as when they eliminated the time
delay between the unexpected stimulus disappearing from
the screen and the participant being questioned about their
awareness of any unexpected stimulus, the incidence of IB
remained the same. Simons and Chabris (1999) presented
a more standard video display of two teams playing
A. Richards (&) � E. M. Hannon � N. Derakshan
Psychological Sciences, School of Science, Birkbeck College,
University of London, Malet Street, London WC1E 7HX, UK
e-mail: [email protected]
123
Psychological Research (2010) 74:513–523
DOI 10.1007/s00426-009-0273-8
basketball rather than superimposing two videos together.
Participants were required to count the number of ball
bounces during the video. The unexpected stimulus was a
black gorilla that walked through the scene, stopping for a
few seconds in the centre of the display to thump its chest
before carrying on walking to the other side of the scene.
The basketball game continues for a few seconds after the
gorilla has disappeared. Simons and Chabris found that
only 44% of participants reported seeing the gorilla when
questioned at the end.
Most et al. (2000) later developed a video to measure IB
in a more experimentally controlled manner in which
participants were required to track four white letters and
ignore four black letters. An unexpected gray cross
appeared and traversed the screen during the video, and
participants who were unable to report having seen the red
cross were deemed to be inattentionally blind. Most et al.
(2000) found that levels of IB were highest when the
unexpected stimulus moved either the top or bottom third
of the screen compared to when moved across the centre of
the screen.
Other work has examined the effects of attentional set
on implicit attentional capture (Folk, Remington, &
Johnston, 1992) and attentional set on IB (Most et al.,
2001). Recently, perceptual load on the incidence of IB has
been investigated (e.g. Cartwright-Finch & Lavie, 2007;
Lavie, 2006). However, a question as yet unanswered is
when people are presented with exactly the same physical
display, why some individuals are inattentionally blind
whereas others are not? Very little research has examined
stable differences in individuals that make them susceptible
to IB, and the current paper attempts to examine some of
the possible candidates in this area.
One such candidate is that there may be individual
differences in processing styles that might impact on IB.
Visual attention can be directed at different levels of a
visual scene, with focus on the more holistic, global level
or on the more analytic, local level (Navon, 1977). Navon
(1977) observed a ‘global precedence’ when analysing a
visual scene, with global changes in a display being
detected more accurately and quickly than local changes.
However, other researchers propose that a participant’s
attentional set is a more reliable predictor of performance
on a global/local detection task, and have investigated the
interaction between stimulus-driven (exogenous) properties
with top-down (endogenous) mechanisms. Austen and
Enns (2000) used Navon’s (1977) stimulus items in which
small letters (local component) were configured in the
shape of a large letter (global component). Participants
were presented with alternating displays that were either
identical, or there was a change (global or local) in one of
the letters. Austen and Enns (2000) found that they were
able to bias participants’ attention to either the local or
global component by manipulating the probability of target
level change, which resulted in a corresponding enhance-
ment of performance when there was a change that was
congruent with the induced bias. One possibility is that IBs
(i.e. those who do not notice the unexpected stimulus) are
biased towards a more highly focused but narrowly defined
analysis of visual stimuli than not inattentionally blind
(NIB), rendering the IBs as being less likely to notice the
unexpected stimulus. IBs would have a more focused
processing style than NIBs, with a focus on local compo-
nents making them more likely to be IB.
The relationship between processing resources and the
incidence of IB will also be investigated. Working memory
tasks are assumed to reflect executive functioning of
working memory (Baddeley & Hitch, 1974; Cowan, 2001).
One possibility is that if there are insufficient processing
resources available then the unexpected may not be pro-
cessed to a high level (i.e. will be filtered as a result of
early selection) and IB will result. If IB is associated with
reduced processing resources then it is predicted that those
individuals deemed to be IB will have lower working
memory scores than those deemed to be NIBs. Experiment
1 tests these hypotheses using a global–local flicker task to
measure differences in processing style, an operation span
task (OSPAN) to measure working memory, and an IB
task. If IB individuals are characterised by a bias for a local
level of analysis, they should be more sensitive to local
changes in a visual display compared to NIBs, with NIBs
being more sensitive to global changes. If the limited
resources hypothesis is supported, we would expect lower
working memory capacity (WMC) for IBs compared to
NIBs.
Experiment 1
Method
Participants
A total of 77 participants were tested but 9 were excluded
for misunderstanding the OSPAN task or for reporting that
they were aware of IB research when questioned at the end
of the session. Data from 68 participants were analysed (41
females): mean age 27.43, SD 9.45, range 18–56 years.
Design and procedure
OSPAN of Turner and Engle (1989) In this task, partic-
ipants are required to solve simple mathematical equations
while memorising unrelated words, with word lists varying
between 2 and 5 words per set, e.g. ‘does (2*1) - 1 = 1 ?
pipe.
514 Psychological Research (2010) 74:513–523
123
Three sets of each list length are presented in apparent
random order but fixed across participants. List length is
unknown to participants until the cue ‘???’ appears, when
they must then write down the words they can remember
from that set, in the exact order they appeared. Scoring on
this task consists of summing the recalled words for only
those sets recalled completely and in the correct order, with
scores having a possible range of between 0 and 42.
Local/global flicker task This task was based on that of
Austen and Enns (2000). Two global letters (a large E and a
large S) were created using a series of small Ss or small Es.
This produced four stimuli: a large E comprising small Es
and a large S comprising small Ss (global and local com-
ponents are congruent), and a large E comprising small Ss
and a large S comprising small Es (local and global com-
ponents are incongruent). In this task, participants were
presented with two alternating displays in which the two
displays were either identical (no change) or one of the
items on the display had either a global change (e.g. small
Es would be presented in the shape of a large S on one
display and a large E on the alternating display) or a local
change (e.g. a large E would be configured by small Es on
one display and by small Ss on the alternating display). 50%
of the local letters were consistent with the global letter and
50% inconsistent. Each stimulus (48 9 78 pixels) was
shown in displays of alternating frames of one, three or five
items. Each frame (display and blank) appeared for 225 ms
(see Fig. 1). This flickering display continued until the
participant made a response (i.e. change/no-change deci-
sion). A change occurred on 50% of trials, and half of these
were local and the other half global. Items appeared ran-
domly in one of nine squares of an imaginary 3 9 3 matrix.
When there was only a single item in the flickering display,
spatial attention will be focused on that one item. However,
with displays of multiple items (3 or 5 flickering items),
spatial attention has to be distributed widely across the
screen in order to perceive a change in one of the items
(Austen & Enns, 2000). There were 48 practice trials and 3
experimental blocks of 80 trials.
IB task A computerised, sustained IB trial identical to
that reported in earlier work (Most et al., 2001; video clip
courtesy of Simons, 2003) was used. The task comprised
four black and four white letters (Ls and Ts) moving
haphazardly around the screen, frequently ‘hitting’ the
borders of the display. Participants tracked the four white
letters (2 Ls and 2 Ts) but ignored the black letters, and
reported the number of ‘hits’ at the end of the 17-s video.
After 5 s, a red cross took 7 s to move across the centre of
the screen from right to left.
At the end of the task, participants were asked if they
had seen anything else in addition to the Ls and Ts. Those
who said ‘yes, a red cross’ were deemed to be NIBS and
those who said ‘no’ were deemed to be IB. If a participant
reported seeing ‘something’ they were asked to specify
what they had seen, and all such participants described the
red cross. Although many participants said they saw black
Ls and Ts, none reported seeing anything else that they
could not identify.
Participants completed the OSPAN and local/global
flicker task in randomised order with the IB task presented
Fig. 1 Schematic display of a
three-item global change trial
(bottom right item changes from
global E to global S. Local
component (E) remains
constant)
Psychological Research (2010) 74:513–523 515
123
last to prevent participants adopting a divided attentional
set for later tasks.
Results
Of the 68 participants, 36 (53%) failed to notice the
moving red cross and were therefore classified as IB. The
remaining participants reported seeing the red cross, and
were classified as NIB. There were no differences between
IBs and NIBs in their ability to perform the primary
component of the IB task (means of 11.61, SD = 1.104 and
11.61, SD = 1.25 respectively).
Local/global flicker task
Following Austen and Enns (2000), data were analysed
separately for the focal attention task (set size 1) and for
distributed attention task (set sizes 3 and 5). An examina-
tion of response latencies on the focal attentional task
revealed no differences between IB and NIB individuals
for global or local trials. A comparison of change trials
(collapsed over local and global) and no-change trials
showed faster responses for change than no-change
(1,430 ms, SE = 35 and 1,715 ms, SE = 71, respectively;
F(1, 66) = 32, p \ 0.001, g2p ¼ 0:33). Change detection
during distributed attention was examined. An ANOVA
with display type (local, global) and set size (3, 5) as
within-subjects factors and blindness status (IB, NIB) as a
between-subjects factor revealed faster responses for global
than local changes (2,932 ms, SE = 104, and 3,165 ms,
SE = 116, respectively; F(1, 66) = 5.95, p = 0.02,
g2p ¼ 0:08), and for set size 3 than for 5 (mean of 2,620 ms,
SE = 76 and 3,478 ms, SE = 136, respectively;
F(1, 66) = 84.13, p = 0.02, g2p ¼ 0:56.
A comparison of change trials (collapsed over local and
global) with no-change trials revealed that responses were
faster on change than no-change trials (mean of 3,049 and
4,655 ms, respectively; F(1, 66) = 252, p \ 0.001,
g2p ¼ 0:79), responses to set size 3 were faster than set
size 5 (mean of 3,238 and 4,465 ms, respectively;
F(1, 66) = 227, p \ 0.001, g2p ¼ 0:78). In addition, there
was an interaction between set size and change (F(1, 66) =
65.57, p \ 0.001, g2p ¼ 0:50), showing that the difference
in RTs for change and no-change trials was greater for set
size 5 (mean of 3,478 and 5,453 ms for change and no-
change, respectively) than for set size 3 (mean of 2,620 and
3,857 ms, respectively).
A non-parametric Signal Detection Analysis (Snodgrass
& Corwin, 1988) was applied to the accuracy data for set
sizes 3 and 5 combined (see Table 1). Sensitivity was greater
for global than local displays (mean of 4.12, SE = 0.19,
and 3.75, SE = 0.19, respectively; F(1, 66) = 13.60,
p \ 0.001, g2p ¼ 0:17), but there were no differences
involving IB status (F’s \ 1). There were no significant
effects from the analysis of the response bias scores.
These analyses show that performance is better for all
participants when there is a change in the visual display
compared to when there is no change. When there was a
change in the visual display when attention was distributed,
global changes were detected more rapidly than local
changes. Performance was also better on displays with
fewer items. However, there were no main or interaction
effects involving IB in any of the analyses.
IB and WMC
The IBs had lower OSPAN scores than the NIB individuals
(mean of 15.11, SD = 6.62 and 19.56, SD = 7.69,
respectively; t(66) = 2.57, p = 0.013), indicating lower
WMC is associated with IB. One interpretation of these
data is that individuals with lower WMC did not have
sufficient resources to process the unexpected stimulus.
IB, WMC and sensitivity to local and global visual changes
In order to examine the influence of WMC and sensitivity
to global and local changes in predicting the probability of
IB, a simultaneous entry logistic regression was performed.
IB was the outcome variable and age, sex and latency
differences (global minus local latencies) were predictors
(see Table 2). This analysis revealed that only OSPAN was
a significant predictor.
Discussion
Inattentional blindness was observed in 53% of the sample,
and this is consistent with our own research and that of
Table 1 Mean response latencies (ms), sensitivity, response bias for
the global–local flicker task
NIBs IBs
Set size 3 Set size 5 Set size 3 Set size 5
Latency
Global 2,427 (587) 3,368 (919) 2,563 (711) 3,371 (1,453)
Local 2,655 (691) 3,440 (896) 2,834 (889) 3,732 (1,565)
Sensitivity
Global 4.07 (1.48) 4.18 (1.65)
Local 3.57 (1.55) 3.94 (1.62)
Response bias
Global 0.99 (0.01) 0.98 (0.85)
Local 0.99 (0.03) 0.97 (0.15)
Not inattentionally blind (NIB) and inattentionally blind (IB) indi-
viduals (SDs in parentheses)
516 Psychological Research (2010) 74:513–523
123
others (e.g. Hannon & Richards, 2009; Most et al., 2000).
There were no differences between local and global
detection when attention was focused but when attention
was divided all participants showed increased sensitivity
for global than local visual changes on the flicker task (in
line with Austen & Enns, 2000; Navon, 1977). There were
no differences between the IBs and NIBs, showing no
support for the notion that IB individuals are characterised
by an increase in analytical processing compared to NIB
individuals. However, the IBs were shown to have lower
WMC than the NIBs. The logistic regression gives weight
to this conclusion, showing that only OSPAN significantly
predicted the probability of IB, with the latency difference
on local and global displays playing no role. Although the
IBs had significantly lower WMC than the NIBs, we cannot
draw any conclusions regarding causality. Being inatten-
tionally blind might cause a reduction in WMC, low WMC
may cause IB, or there may be a third variable (e.g.
motivational component) that may be causing both lower
WMC as measured by the OSPAN task and IB. There is no
evidence for a difference between IBs and NIBs in their
motivation from the current study, as both IBs and NIBs
were equally good at performing the primary task in the IB
phase.
However, this finding does not rule out a motivational
account, as the task might be sufficiently easy so that all
participants were able to perform the task adequately. This
finding supports that of Simons and Jensen (2009). Simons
and Jensen used Most et al.’s (2001) IB task but equated
the difficulty level of the primary task for each individual
participant. IB tasks with differing demands were then
presented. They found that manipulating the demands of
the IB task influenced the incidence of IB (Experiment 1),
so that more difficult IB tasks were associated with an
increase in the incidence of IB. However, in Experiment 2,
they found that individual differences in the ability to
perform the task were not related to their ability to notice
the unexpected stimulus. A motivational account would
predict that IBs should have poorer overall performance
than the NIBs on the flicker task. There were, however, no
differences related to IB on the global/local flicker task.
Response latencies were numerically larger for the IBs
compared to the NIBs but none of these differences
approached significance. Although these differences were
non-significant, a logistic regression was performed with
IB as the outcome variable and speed (mean overall RT on
flicker task) and accuracy (overall accuracy on flicker task)
and OSPAN as the independent variables was performed.
This revealed that only OSPAN was a significant predictor
of IB, with both speed and accuracy making no contribu-
tion to the equation (p [ 0.5).
WMC is very important in IB, but there may be addi-
tional influential variables. One such proposal is that IB
individuals fail to notice the unexpected stimulus because
they have successfully inhibited it, and this proposal will
be examined in Experiment 2.
We also ask, ‘To what extent does experience with a
primary task influence rates of IB?’. This question has
potential implications for employment and training in
many areas of life, such as that given to drivers, pilots, air
traffic controllers, medical professionals, etc. Specifically,
we predict that training on a primary task will make the
task more automatic and thereby free up processing
resources. If processing resources become available, then
the incidence of IB should be reduced.
Experiment 2
The ability to inhibit an irrelevant stimulus is an extremely
useful process in most circumstances, as it ensures that the
individual maintains attentional focus on the goal of the
task thereby avoiding the disruptive influence of irrelevant
information. However, such a process brings with it
potential costs, as in the case of the appearance of an
ostensibly irrelevant stimulus that is highly significant. High
WMC has been shown to be related to the ability to inhibit
or block distracting information (Conway & Engle, 1994;
Table 2 Simultaneous entry logistic regression for Experiment 1 (IB status = outcome variable, OSPAN, age, sex and global–local
index = predictors)
95% CI for Exp b
B(SE) Lower Exp b Upper
Constant 1.43 (1.13)
OSPAN* -0.09 (0.04) 0.85 0.92 0.99
Age 0.01 (0.03) 0.96 1.01 1.06
Sex -0.23 (0.53) 0.28 0.80 2.24
Global–local response latency difference 0.00 (0.00) 1.00 1.00 1.00
R2 = 0.10 (Cox and Snell), R2 = 0.13 (Nagelkereke). Model = v2(4) 7.19, p = 0.13
* p \ 0.05
Psychological Research (2010) 74:513–523 517
123
Conway, Tuholski, Shisler, & Engle, 1999; Daneman &
Carpenter, 1980). WMC is viewed by many researchers
as involving controlled attention (e.g. Turner & Engle,
1989). Bleckley, Durso, Crutchfield, Engle, and Khanna
(2003) propose that there are differences in attentional
control between high and low WMC individuals, with the
former having a more flexible attentional allocation whereas
the latter have a spotlight of attention, which is a continuous
but less flexible mode of attentional allocation. They argue
that high WMC individuals are more able to inhibit and
control attention. Kane, Bleckley, Conway, and Engle
(2001) found no difference between high and low
WMC individuals in a prosaccade task, but compared to
low WMC individuals, high WMC individuals had
superior performance on the antisaccade task in which the
saccade towards the cue had to be suppressed in favour of a
saccade in the opposite direction. In addition, Conway,
Kincade, and Shulman (2001) found that high WMC par-
ticipants heard their own name less frequently than low
WMCs in a dichotic listening task, suggesting that they
inhibited this information. If IB is as a result of inhibitory
processes then it might be predicted that individuals with
high levels of inhibition will be more likely to be IB. It might
also be expected from this perspective that IB would be
associated with higher levels of WMC, although it is not
what we found in Experiment 1.
In Experiment 2, we used a task that measures inhibitory
ability on a separate task, i.e. a Stroop task. The Stroop task
comprised four colour–word conditions. In the control
condition, strings of Xs were printed in coloured inks. In the
congruent condition, the word and colour of the ink mat-
ched. In the ignored repeated condition, ink colour and
word colour name conflicted, but in addition, the colour of
the ink on trial n corresponded to the word on trial n - 1. In
the incongruent condition, each word again was the name of
a colour, which conflicted with the coloured ink, but there
was no such relationship between successive trials. As
reading is largely an automatic process, it takes consider-
able attentional effort to ignore the words and identify the
colour of the ink. Facilitation is measured by comparing the
congruent with the control trials. Interference is measured
by comparing responses on the incongruent trials with the
control condition. However, inhibition is measured by
comparing responses on the incongruent trials with those on
the ignored repetition trials. The latter trials are the same as
the incongruent trials (in that the colour and the word are
incongruent) but there is an added inhibitory component, as
participants on trial N have to respond with the colour that
they have just inhibited on the previous trial (N - 1). The
greater the inhibition of the colour word on trial N - 1, the
longer the RT on trial N when this colour now has to be
activated and responded to. The Stroop task has been used
to examine inhibitory processes in schizophrenia (Beech,
Powell, McWilliams, & Claridge, 1989; Boucart et al.
1999), people with Parkinson’s disease (Brown & Mardsen,
1988), the elderly (Hasher & Zacks, 1988), and people with
multiple sclerosis (Vitkovitch, Bishop, Dancey, &
Richards, 2002). If individuals are IB because they have a
tendency to inhibit goal-irrelevant stimuli then we might
expect IB individuals to display stronger inhibition on the
Stroop task.
In the current experiment, we directly compare an
inhibition hypothesis with a reduced capacity hypothesis,
and, in addition, we look at the effects of training on IB. As
participants become more practiced there should be a
corresponding increase in available attentional resources.
We therefore predict that training will decrease the inci-
dence of IB.
Whether practice on video games can improve atten-
tional perceptual tasks is a matter for debate. Green and
Bavelier (2007), e.g., found improvements on such tasks,
whereas Boot, Kramer, Simons, Fabiani, and Gratton
(2008) found no effects after 20 h of practice in non-
gamers. Neisser (1979) describes a study where individuals
were presented with a video scene in which a woman with
an umbrella walks through a basketball game. Prior to this,
participants had completed an easier task, a more difficult
task, or no task. Neisser concludes that people fail to see an
unexpected object in situations where they believe the task
to be difficult. Although essential details are missing from
the account of this study, it does suggest that practice may
have a beneficial effect on reducing incidences of IB. The
current study examines the incidence of IB after (a) no
training, (b) after training on the same task as the primary
task in the final IB task (i.e. counting white Ls and Ts) and
(c) after training on a different task (i.e. counting diamonds
and triangles). We predict that training will reduce the
incidence of IB compared to the no training control con-
dition, as processing resources should be made available by
virtue of the primary task becoming more automatised. By
having a same and different training condition, we will be
able to examine whether general training on the task will
transfer to a different but similar task. We will also
examine whether training, inhibition or WMC predicts the
incidence of IB.
Method
Participants
A total of 89 participants were tested, but 5 were omitted
due to failing to complete all the tasks or for reporting that
they were aware of IB research when questioned at the
end of the session. A final sample of 82 participants
(61 females) with a mean age of 32.48 (SD 7.89, range
21–56 years) is reported.
518 Psychological Research (2010) 74:513–523
123
Design and procedure
Participants performed the automatic operation span
(AOSPAN) task of Unsworth, Heitz, Schrock, and Engle
(2005) and a Stroop task (Stroop, 1935). These two tasks
were counterbalanced across participants. Participants were
then randomly allocated to one of three conditions (control,
same training or different training). Participants in the two
training condition completed a tracking task that was either
the same as the primary task on the IB task or different (but
similar) to the IB tracking task. These two groups then
completed the IB task. The control group did not receive
any training and just completed the same IB task as the
trained groups.
AOSPAN task (Unsworth, Heitz, Schrock, & Engle,
2005) This task is designed to measure an individual’s
WMC, and is less dependent on language abilities as letters
rather than words were used. Participants are presented
with a series of maths problems that they need to solve as
quickly as possible, which is then followed by a letter that
needs to be recalled at a later stage. The practice was
divided into three phases. In the first phase, participants
were presented with a number of letters, each presented for
800 ms (with this being the same for all experimental
blocks). At the end of each trial, participants were pre-
sented with a 4 9 3 matrix of letters (F, H, J, K, L, N, P, Q,
R, S, T, Y). Participants were required to click a box next
to the appropriate letter in the exact order that the letters
had appeared. Accuracy feedback was given.
For the second practice phase, participants were
presented with a series of 15 maths problems, e.g.
(1*2) ? 1 = ?, to be solved as quickly as possible. On
the next screen, a possible answer, e.g. 3 and two boxes
with ‘True’ or ‘False’, was presented, and participants
were required to check the appropriate answer. Accuracy
feedback was also given for this section. During this
phase, the participant’s average solution time was also
calculated. This average time (?2.5 SD) was then used as
a time limit for the maths portion of the task. If, in the
experimental trials, a participant took 2.5 SD longer than
their average solution time to solve a maths problem, the
program skipped the ‘True’/‘False’ part of the maths
problem and moved directly onto the letter part but then
coded this trial as having a speed error. In the third and
final practice sessions, participants performed both the
letter recall and the maths problems together. The
experimental trials consisted of three sets each of set sizes
that ranged from three to seven per set, i.e. sets of 3, 4, 5,
6, 7 letters and maths. Therefore, in total, there were 75
letters and 75 maths problems, with possible scores
ranging between 0 and 75. Order of set sizes was
randomised for each participant.
Stroop task There were four types of trials (congruent,
control, incongruent and ignored repetition). There were
eight experimental blocks, each comprising trials of the
same type (e.g. all control trials). There were, therefore,
two blocks for each of the four types of trials. Within each
block, there were 32 trials, and the 4 colours (red, green,
yellow and blue) appeared an equal number of times. After
each of the eight blocks, participants were allowed to take a
short break. Each trial began with a fixation cross for
500 ms, which was replaced by the target (a colour word or
a row of Xs), presented in one of four colours (red, green,
yellow, blue). Participants were required to identify the
colour of the ink as quickly and accurately as possible by
making a manual response using the colour-coded keys on
the keyboard. After a response had been made, the next
trial began. There was a practice block of 20 trials at the
beginning of the experiment.
IB and training tasks To allow for training effects to be
observed, a similar IB task to that developed by Most et al.
(2001) was created using MATLAB, with the difference
being that the letter and moving cross were slightly smaller
than those in the original task, there was a larger number of
‘hits’ (i.e. bounces off the border) and the scanning area
was slightly larger. The starting positions of the Ls and Ts
on the IB task were changed to create a different starting
position for the targets and distractors for the training
sessions. We were also able to substitute the Ls and Ts with
triangles and diamonds, so that participants in the same and
different training conditions received exactly the same
configuration of targets and distractors with the same
starting positions, movement direction and velocities but
with different identities.
In the control condition, participants completed the IB
task with no training (see Fig. 2). Participants in the
different training condition completed two 17-s training
sessions, similar to the primary task in the IB task, but with
moving diamonds and triangles (and no unexpected stim-
ulus; see Fig. 3). Participants in the same training condition
completed two training sessions on the same primary task
as that used in the IB task (i.e. moving Ls and Ts) but
without the unexpected stimulus. The movement of items
in these two sessions was the same as those of the two same
training sessions. Both of the training groups completed the
IB task immediately after the training sessions.
Results
Training effects on incidence of IB
A Chi-square, performed on the number of IBs and NIBs
in the different training conditions (see Fig. 4), revealed
a significant association between the incidence of IB
Psychological Research (2010) 74:513–523 519
123
and training condition (v2 = 17.01, N = 82, df = 2, p \0.001, U = 0.46). The number of IBs relative to NIBs was
higher in the control condition compared to the same
(v2 = 17.04, N = 57, df = 1, p \ 0.001, U = 0.55) and
compared to the different training condition (v2 = 3.86,
N = 60, df = 1, p = 0.05, U = 0.25). There were also
more IBs than NIBs in the different compared to the same
training conditions (v2 = 4.07, N = 47, df = 1, p = 0.03,
U = 0.32). This showed that the more similar the training
is to the primary task on the IB task, the greater the
reduction in the incidence of IB. These findings support the
proposal that training allows the primary task to become
more automatised resulting in an increase in the available
processing resources to enable the unexpected stimulus to
be fully processed.
Consistent with Experiment 1, the IBs and the NIBs did
not differ in their performance on the primary component
of the IB task with mean number of hits reported being
14.19 (SD 2.47) and 13.89 (SD 3.09) for IBs and NIBs,
respectively.
WMC and IB
An analysis of AOSPAN scores in the three training
conditions revealed that overall the IBs had lower AOSPAN
scores than the NIBs (mean of 44.68, CI95 = 39.57, 49.79
and 55.41, CI95 = 48.10, 63.62, respectively; F(1, 76) =
6.04, p = 0.016, g2p ¼ 0:074). There were no differences
between the training conditions or any interactions involv-
ing training or IB. Thus, it appears that IBs tend to have
lower WMC than NIBs but WMC is not influential in
identifying which individuals would benefit from training in
order to reduce their chances of displaying IB.
Stroop task: facilitation; inhibition and IB
An ANOVA of the Stroop RT data was performed, with
condition (congruent, control, incongruent, ignored repeti-
tion) as a within-subjects factor and training (control, same,
different) and IB status (IB, NIB) as the between-subjects
factors. This revealed a main effect of condition
(F(3, 228) = 11.79, p \ 0.001, g2p ¼ 0:013). Further anal-
yses revealed that RTs were not significantly faster for
congruent compared to control trials (mean of 830 ms,
SD = 155 and 836 ms, SD = 174, respectively), showing
that there was no facilitation observed in the current
study. However, significant inhibition was observed over-
all, with RTs to incongruent being faster than those to
ignored repetition trials (mean of 880 ms, SD = 185 and
910 ms, SD = 199, respectively; t(81) = 2.54, p = 0.013,
g2p ¼ 0:074). There were no other significant effects,
showing that IB individuals do not have a greater tendency
to inhibit irrelevant stimuli compared to the NIBs.
Working memory; inhibition; training, and IB
To test the relative contributions of training, WMC and
inhibition in predicting the probability of IB, simultaneous
entry logistic regression was performed where IB was the
outcome variable and training, inhibition, and AOSPAN
scores were the predictors (Table 3). Both AOSPAN and
training predicted the probability of IB, but inhibition did
not. There were significant effects of different training
compared to control, and for different compared to same
training.
To examine for motivational difference between the
IBs and NIBs, an analysis was performed in which overall
Fig. 2 A still frame from the IB task, showing the unexpected cross
Fig. 3 A still frame from the different training task
Fig. 4 Incidence of inattentionally blind (IBs) and not inattentionally
blind (NIB) by training
520 Psychological Research (2010) 74:513–523
123
RT and overall accuracy were entered as additional
independent variables into the analysis. It was found that
only AOSPAN and training predicted IB with overall RTs
and accuracy not accounting for a significant amount of
the variance (p’s [ 0.15). This analysis again argues
against the proposal that motivational differences predict
IB status.
Discussion
Again, we have demonstrated the relationship between IB
and WMC, with low WMC individuals being more likely
to be IB than high WMC individuals. Although inhibition
was observed in the group as a whole, the IB group did not
display greater inhibition on the Stroop task. This finding
argues against the proposal that individuals who fail to
report the unexpected cross do so because they have a
tendency to inhibit goal irrelevant stimuli. Training had
a significant effect on the incidence of IB, with training on
a task that is similar to the IB task reducing IB compared to
both control (no training) and different training. Training
on a task that is the same as the primary task from the IB
task produces the greatest decrease in the incidence of IB,
but training on a different primary task with the basic
configuration as the primary IB task also produces
decreases in the incidence of IB. These findings suggest
that IB can be manipulated by training, and may have
implications for training of drivers, pilots, etc. Training on
similar tasks is hugely beneficial for later detection of
unexpected stimuli, and there is benefit albeit to a lesser
extent, from training on a different task.
There was no support for a motivational account of IB,
as there were no differences in performance on the primary
component of the IB task, and there was no difference in
overall performance on the Stroop task or any contribution
in terms of speed or accuracy of performance on the Stroop
task in predicting IB in the logistic regression.
General discussion
In two experiments, we have demonstrated robust effects of
WMC on predicting the likelihood of IB. We have shown
this using both the standard OSPAN and the AOSPAN
tasks. There were no differences in processing styles on the
flicker task, and no differences in terms of inhibition as
measured by the Stroop task. Of particular interest were the
effects of training on the incidence of IB, and here we
showed clearly that training on a task similar to the IB task
produced the greatest benefits in terms of reducing IB but
there are also some benefits to be gained from training on a
different IB task.
The findings from the two experiments that low WMC is
associated with IB whereas higher WMC is associated with
NIB suggest that IB individuals do not have sufficient
processing resources to fully process goal-irrelevant stim-
uli. We argue that there is a tendency for low WMC
individuals to not process the unexpected stimulus or to
process the unexpected stimulus and then filter it out at an
early stage of processing. This interpretation is supported
by the findings from the training effects in Experiment 2,
where training might make the processing of the tracking
task more automatic thereby freeing up resources to enable
the unexpected stimulus to be processed. The greater the
similarity of the training task to the primary task in the IB
task, the greater the automatisation of performance result-
ing in a release of processing resources leading to a greater
reduction in the incidence of IB.
If performance on the IB task is seen as resulting from
the interplay between goal-driven and stimulus-driven
attentional control (Egeth & Yantis, 1997), then an indi-
vidual’s propensity towards IB may be the result of the
suppression of stimulus-driven attentional control (Todd,
Fougnie, & Marios, 2005). The intraparietal sulcus (IPS) is
located in the dorsal parietal lobe and has been shown to be
related to goal-driven attentional control and is recruited by
task-driven attention and with the task demands of working
memory (e.g. Cohen et al. 1997; Corbetta, Kincade, &
Shulman, 2002; Todd & Marios, 2004). The temporo-
parietal junction (TPJ) is located at the junction between
the temporal and parietal lobes at the posterior end of the
sylvian fissure, and is associated with stimulus-driven
attentional control (e.g. Ansari, Lyons, van Eitneren, & Xu,
2007; Marois, Leung, & Gore, 2000). Todd et al. (2005)
demonstrated that engagement in a resource-consuming
task was associated with the recruitment of the IPS and
simultaneous suppression of the TPJ. The differential
recruitment of the IPS and the corresponding suppression
of TPJ may offer an explanation for the pattern of results
that we have observed. If the primary task is sufficiently
demanding for a low WMC individual, then there will be
increased suppression of the TPJ resulting in IB. High
Table 3 Simultaneous entry logistic regression for Experiment 2
95% CI for Exp b
B(SE) Lower Exp b Upper
Constant 2.67 (0.96)
AOSPAN* -0.051 (0.02) 0.91 0.95 0.99
Inhibition 0.000 1.00 1.00 1.01
Training*
Control vs. different** 4.47 (1.80) 2.58 87.49 2,968
Same vs. different** -2.79 (1.09) 0.01 0.06 1.00
R2 = 0.28 (Cox and Snell), R2 = 0.39 (Nagelkereke). Model = v2(6)
26.53, p \ 0.001
*p \ 0.05
**p \ 0.001
Psychological Research (2010) 74:513–523 521
123
WMC individuals are likely to find the primary task less
demanding, which would be associated with a reduction in
the suppression of the TPJ and reduced likelihood of IB.
An alternative but related explanation is in terms of
individual differences in selection efficiency. Vogel,
McCollough, and Machizawa (2005) measured contralateral
delay activity (CDA) in an event-related potential (ERP)
study. The CDA is a sustained negative voltage over the
hemisphere that is contralateral to the visual field where
visual items are displayed in order to be memorised, and the
amplitude of this wave increases as the number of items held
in memory increases. Using this methodology, Vogel et al.
(2005) varied the number of components (targets and dis-
tractors) in a visual display, with instructions to participants
to pay attention to the targets. By comparing the amplitude
of the CDA in four-item displays where all items were tar-
gets with four-item displays where two of the items were
targets and the other two distractors, it was possible to infer
how many items were stored in working memory. High
WMC individuals were more efficient at representing only
relevant items in memory whereas low WMC inefficiently
represented both relevant and irrelevant items in memory.
These findings suggest that low WMC individuals not only
have fewer resources available but they also tend to use these
resources less efficiently than high WMC individuals.
Lavie (2006) proposes that when the number of items to
be monitored in a display is low, then perception of irrel-
evant items is more likely to happen than when the number
of items to be monitored is high. She argues that a high
perceptual load results in early selection of relevant items
with no resources left for the processing of irrelevant (or
unexpected) items. Our current studies do not enable us to
examine these predictions, as it is not possible to determine
whether our display (4 targets and 4 distractors) qualifies as
a high or low perceptual load. One possibility is that it
would be a high perceptual load for a low WMC individual
but a low perceptual load for a high WMC individual.
However, Lavie (2005) makes a distinction between per-
ceptual load and cognitive load, proposing that whether
distractors will be processed or not depends on both the
level (high or low) and type (perceptual or cognitive) load.
She proposed that whereas a high perceptual load will
reduce the probability that distractors will be processed due
to limited perceptual capacity, a high-cognitive load would
increase the probability that distractors will be processed.
Lavie argues that a load on executive cognitive control
functions such as working memory results in a failure to
keep activated the processing priorities of the task, and ‘it
increases interference by irrelevant low-priority distractors
rather than decreases it.’ On the basis of this, it might be
predicted that fewer available working memory resources,
as would be the case for low WMC individuals, should
result in a decrease in the incidence of IB. Our studies do
not offer any support for this proposal. In fact, we have
consistently found effects in the opposite direction; with IB
rather NIB being associated with reduced working memory
resources.
It is clear, however, that there are limitations associated
with correlational studies. First, there may be a third factor
that is causally related to both the tendency to be IB and
low WMC. One possible ‘third variable’ is that of moti-
vation. It might be the case that a lack of motivation to
perform the task well could produce non-optimal perfor-
mance on both the OSPAN and the IB tasks. If this was the
case, then we might expect performance on other tasks,
such as the local/global flicker task and the Stroop task, to
be performed more slowly and/or less accurately by the IBs
compared to the NIBs. We might also expect differences in
terms of performance on the primary component of the IB
task. We did not find support for any of these predictions.
However, it cannot be asserted that WM is causally related
to the occurrence of IB. Cognitive load needs to be
manipulated directly in order to test for a possible causal
link between WM and IB. Although Experiment 2 indi-
rectly manipulated the resources available by training
participants in order to free up some resources, a more
direct manipulation is needed.
Further research is necessary to examine different types
of training over different time scales. For example, do the
effects endure over long intervals? Systematic variations of
these variables need to be performed to examine their
effects of IB. Despite finding robust support for the rela-
tionship between low WMC and the incidence of IB, there
are a small number of NIBs who have very high WMC.
Clearly, this group of participants have ample resources to
process all the stimuli in the display, so why are these
individuals IB? We put forward a dual route model, which
proposes that individuals may demonstrate IB but for dif-
ferent reasons. On the basis of this model, we predict that
low WMC individuals fail to notice the irrelevant stimulus
because they do not have sufficient resources to process
information outside the goals of the primary task, whereas
high WMC individuals, who do show IB, do so because they
are actively inhibiting the irrelevant stimulus. Research is
currently underway to examine these proposals.
Acknowledgments This work is supported by the Leverhulme
Foundation awarded to the first author under grant number F/07 112/R.
References
Ansari, D., Lyons, I. M., van Eitneren, L., & Xu, F. (2007). Linking
visual attention and number processing in the brain: The role of
the temporo-parietal junction in small and large symbolic and
non-symbolic number comparison. Journal of Cognitive Neuro-science, 19, 1845–1853.
522 Psychological Research (2010) 74:513–523
123
Austen, E., & Enns, J. T. (2000). Change detection. Paying attention
to detail. Psyche, 6(11). http://psyche.cs.monash.edu.au/v6/
psyche-6-11-austen.html.
Baddeley, A. D., & Hitch, G. (1974). Working memory. In G. H.
Bower (Ed.), The psychology of learning and motivation:Advances in research and theory (Vol. 8, pp. 47–89). New
York: Academic Press.
Becklen, R., & Cervone, D. (1983). Selective looking and the noticing
of unexpected events. Memory & Cognition, 11(6), 601–608.
Beech, A. R., Powell, T. J., McWilliams, J., & Claridge, G. S. (1989).
Evidence of reduced ‘‘cognitive inhibition’’ in schizophrenia.
British Journal of Psychology, 78, 349–356.
Bleckley, M. K., Durso, F. T., Crutchfield, J. M., Engle, R. W., &
Khanna, M. M. (2003). Individual differences in working
memory capacity predict visual attentional allocation. Psycho-nomic Bulletin & Review, 10, 884–889.
Boot, W. R., Kramer, A. F., Simons, D. J., Fabiani, M., & Gratton, G.
(2008). The effects of video game playing on attention, memory,
and executive control. Acta Psychologia, 129, 387–398.
Boucart, M., Mobarek, N., Cuervo, C., & Danion, J. M. (1999). What
is the nature of increased Stroop interference in schizophrenia.
Acta Psychologica, 101, 3–25.
Brown, R. G., & Mardsen, C. D. (1988). Internal versus external cues
and the control of attention in Parkinson’s disease. Brain, 111,
323–345.
Cartwright-Finch, U., & Lavie, N. (2007). The role of perceptual load
in inattentional blindness. Cognition, 102(3), 321–340.
Cohen, J. D., Perlstein, W. M., Braver, T. S., Nystrom, L. E., Noll, D.
C., Jonides, J., et al. (1997). Temporal dynamics of brain
activation during a working memory task. Nature, 386, 604–608.
Conway, A. R. A., Cowan, N., & Bunting, M. F. (2001). The cocktail
party phenomenon revisited: The importance of working mem-
ory capacity. Psychonomic Bulletin & Review, 8, 331–335.
Conway, A. R. A., & Engle, R. W. (1994). Working memory and
retrieval: A resource-dependent inhibition model. Journal ofExperimental Psychology: General, 123, 354–373.
Conway, A. R. A., Tuholski, S. W., Shisler, R. J., & Engle, R. W.
(1999). The effect of memory load on negative priming: An
individual differences investigation. Memory & Cognition, 27,
1042–1050.
Corbetta, M., Kincade, J. M., & Shulman, G. L. (2002). Neural systems
for visual orienting and their relationships to spatial working
memory. Journal of Cognitive Neuroscience, 14, 508–523.
Cowan, N. (2001). The magical number four in short-term memory: A
reconsideration of mental storage capacity. Behavioural & BrainSciences, 24, 87–114.
Daneman, M., & Carpenter, P. A. (1980). Individual differences in
working memory and reading. Journal of Verbal Learning andVerbal Behavior, 19, 450–466.
Egeth, H. E., & Yantis, S. (1997). Visual attention: Control,
representation, and time course. Annual Review of Psychology,48, 269–297.
Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary
covert orienting is contingent on attentional control settings.
Journal of Experimental Psychology: Human Perception andPerformance, 18, 1030–1044.
Green, C. S., & Bavelier, D. (2007). Action video game experience
alters the spatial resolution of attention. Psychological Science,18, 88–94.
Hannon, E. M., & Richards, A. (2009). Is inattentional blindness
related to individual differences in visual working memory
capacity or executive control functioning. Perception (in press).
Hasher, L., & Zacks, R. L. (1988). Working memory, comprehension
and aging: A review and a new view. In G. Bower (Ed.), Thepsychology of learning and motivation (Vol. 22, pp. 193–225).
San Diego, CA: Academic Press.
Kane, M. J., Bleckley, K. M., Conway, A. R. A., & Engle, R. W.
(2001). A controlled-attention view of working-memory capac-
ity. Journal of Experimental Psychology: General, 130, 169–183.
Lavie, N. (2005). Distracted and confused? Selective attention under
load. Trends in Cognitive Science, 9, 75–82.
Lavie, N. (2006). The role of perceptual load in visual awareness.
Brain Research, 1080(1), 91–100.
Mack, A., & Rock, I. (1998). Inattentional blindness. Cambridge,
MA: MIT Press.
Marois, R., Leung, H. C., & Gore, J. C. (2000). A stimulus-driven
approach to object identity and location processing in the human
brain. Neuron, 25, 717–728.
Most, S. B., Simons, D. J., Scholl, B. J., & Chabris, C. F. (2000).
Sustained inattentional blindness: The role of location in the
detection of unexpected dynamic events. Psyche, 6(14).
http://psyche.cs.monash.edu.au/v6/psyche-6-14-most.html.
Most, S. B., Simons, D. J., Scholl, B. J., Jimenez, R., Clifford, E., &
Chabris, C. F. (2001). How not to be seen: The contribution of
similarity and selective ignoring to sustained inattentional
blindness. Psychological Science, 12(1), 9–17.
Navon, D. (1977). Forest before trees: The precedence of global
features in visual perception. Cognitive Psychology, 9, 353–383.
Neisser, U. (1979). The control of information pickup in selective
looking. In A. D. Pick (Ed.), Perception and its development (pp.
201–219). Hillsdale, NJ: Lawrence Erlbaum Associates.
Neisser, U., & Becklen, R. (1975). Selective looking: Attending to
visually specified events. Cognitive Psychology, 7(4), 480–494.
Simons, D. J. (2003). Surprising studies of visual awareness (DVD).
Champaign, IL: Viscog Productions. http://www.viscog.com.
Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst:
sustained inattentional blindness for dynamic events. Perception,28, 1059–1074.
Simons, D. J., & Jensen, M. S. (2009). The effects of individual
differences and task difficulty on inatttentional blindness.
Psychonomic Bulletin & Review, 16, 398–403.
Snodgrass, J. G., & Corwin, J. (1988). Pragmatics of measuring
recognition memory: Applications to dementia and amnesia.
Journal of Experimental Psychology: General, 177, 34–50.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions.
Journal of Experimental Psychology, 18, 643–662.
Todd, J. J., Fougnie, D., & Marois, R. (2005). Visual short-term
memory load suppresses temporo-parietal junction activity and
induces inattentional blindness. Psychological Science, 16(12),
965–972.
Todd, J. J., & Marois, R. (2004). Capacity limit of visual short-term
memory in human posterior parietal cortex. Nature, 428, 751–754.
Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task
dependent? Journal of Memory and Language, 28(2), 127–154.
Unsworth, N., Heitz, R. P., Schrock, J. C., & Engle, R. W. (2005). An
automated version of the operation span task. BehaviourResearch Methods, 37(3), 498–505.
Vitkovitch, M., Bishop, S., Dancey, C., & Richards, A. (2002). Stroop
interference and negative priming in patients with multiple
sclerosis. Neuropsychologia, 40, 1570–1576.
Vogel, E. K., McCollough, A. W., & Machizawa, M. G. (2005).
Neural measures reveal individual differences in controlling
access to working memory. Nature, 438, 500–503.
Psychological Research (2010) 74:513–523 523
123