1

Emotional Stroop interference in trauma-exposed individuals:

A contrast between two accounts

Serge Caparos1 & Isabelle Blanchette2

1 Université de Nîmes, CHROME EA 5372, France

2 Université du Québec à Trois-Rivières, Canada

Contact: serge.caparos@unimes.fr

Publication information

Published in Consciousness and Cognition. Elsevier copyright. This article may not

exactly replicate the final version published in the Elsevier journal. It is not the copy of

record.

Published version of the article may be found at this link:

http://www.sciencedirect.com/science/article/pii/S1053810014000956

Article DOI:

10.1016/j.cognition.2014.10.006

©2015. This manuscript version is made available under the CC-BY-NC-ND 4.0 license

http://creativecommons.org/licenses/by-nc-nd/4.0/

2

Abstract

In the Emotional Stroop task, trauma-exposed victims are slowed when naming the

colour print of trauma-related words, showing the presence of interference. This

interference has been assumed to reflect emotional reactions triggered by experience-

relevant emotional content which interfere with the task. However, it may equally reflect

the activation of task-competing thoughts triggered by experience-relevant semantic

content, thus resulting from cognitive- rather than emotion-driven processes. This study

contrasted these possibilities by measuring the relationship between Emotional Stroop

interference, on the one hand, and severity of sexual-abuse experience, subjective ratings

of emotionality, and working-memory measures, on the other. Whereas there was no

relationship between working-memory measures and interference, providing no support

for the cognitive-based account, experience severity, emotionality ratings and abuse-

related interference were all positively related, providing support for the emotion-based

account. These findings support the idea that the Emotional Stroop task can be used as a

diagnostic tool for emotion-filtering impairment.

Key words

Emotional Stroop, Trauma, Attention, Emotion, Cognition, Interference

3

1. Introduction

Traumatic experiences are common with a lifetime prevalence rate of more than

60% in the general population (e.g., Resnick, Kilpatrick, Dansky, Saunders, & Best,

1993). People subject to traumatic experiences, such as sexual abuse, are exposed to

intense emotions and this has cognitive consequences not only during the traumatic

experience but also long after it, notably on short-term verbal memory, working memory

and sustained attention (Jenkins, Langlais, Delis, & Cohen, 1998; Navalta, Polcari,

Webster, Boghossian, & Teicher, 2006; Stein, Hanna, Vaerum, & Koverola, 1999; Stein,

Kennedy, & Twamley, 2002). Importantly, trauma-exposed victims present a long-term

alteration of selective attention characterised by an impaired attentional filtering for

trauma-related emotional information (Cisler et al., 2011; Williams, Mathews, &

MacLeod, 1996). A good understanding of the effect of trauma on attention is important,

not least because this effect might play a role in the causation and maintenance of

trauma-related psychopathologies such as anxiety and posttraumatic stress disorder

(PTSD; Williams et al., 1996). Thus far, however, this effect has been studied mostly

using the Emotional Stroop task (Ray, 1979), a task that remains controversial with

regard to the processes it indexes. Specifically, the Emotional Stoop task might index

general cognitive effects, such as semantic interference, rather than emotion-specific

effects as classically assumed (Becker, Rinck, Margraf, & Roth, 2001; Williams et al.,

1996). The present study aimed to shed new light on this issue. It contrasted and tested

two accounts, one emotional and one cognitive, for the impact of trauma on attention as

observed with the Emotional Stroop task.

The Emotional Stroop task was designed by analogy to the Cognitive Stroop task

(Stroop, 1935). In the Cognitive Stroop task, participants are asked to name the colour

print of a word; reaction times (RTs) are slowed when colour print and semantic meaning

of a colour word are inconsistent (e.g., the word ‘green’ presented in red). The extent to

which participants are slowed by inconsistent semantic information, compared to neutral

semantic information, provides an index of participant’s attentional inefficiency, that is,

their difficulty in focusing on the primary colour-naming task and ignoring task-unrelated

distracting stimuli. The Emotional Stroop task is similar to the Cognitive Stroop task

except that it uses emotional words instead of colour words. In the Emotional Stroop task,

participants are slower when they name the colour of an emotional word compared to

when they name the colour of a neutral word (Ray, 1979). The amount of interference,

namely, the extent to which participants are slowed by emotional semantic information,

provides an index of participant’s difficulty in ignoring task-unrelated emotional content.

A crucial aspect of this task is that the effect of emotional words largely depends on

experience: the more an emotional word is relevant to the personal experience of an

individual, the stronger its effect on this individual. For instance, while sexual-abuse

words produce strong interference for rape victims (e.g., Cassiday, McNally &, Zeitlin,

4

1992) and war-related words exert strong influence in Vietnam-war veterans (e.g., Litz et

al., 1996), the same words exert little interference in controls.

The effect of emotional words in the Emotional Stroop task is believed to reflect a

failure of selective attention to filter out experience-relevant emotional content. The exact

processes involved however remain disputed despite over thirty years of research using

the task (Becker et al., 2001; Gilboa-Schechtman, Revelle, & Gotlib, 2000; Williams et

al., 1996). On the one hand, experience-relevant emotional words might exert their

impact through their high emotional salience (e.g., their threat value), suggesting that

attention is affected by the emotional reactions triggered by semantic content. On the

other hand, experience-relevant emotional words might exert their impact through their

high conceptual salience (e.g., they match participant’s current thoughts), suggesting that

attention is affected by interfering thoughts triggered by salient semantic content.

If the Emotional Stroop effect results from conceptual salience, the Emotional

Stroop task may not actually index emotional processing. Words relevant to an individual

relate to ‘current concerns’ and might more readily generate task-unrelated thoughts that

compete for cognitive resources and slow the primary colour-naming task (Williams et

al., 1996). This cognitive account is supported by findings of a large overlap in the brain

regions activated by Emotional and Cognitive Stroop tasks (Whalen et al., 1998). It is

also supported by findings showing that words that are not emotional can nevertheless

generate interference when they match personal interests (Gronau, Cohen, & Ben-

Shakhar, 2003). For instance, in one study (Dalgleish, 1995), bird names were found to

produce interference in ornithologists but not in controls. In sum, the cognitive account

suggests that Emotional and Cognitive Stroop tasks tap similar cognitive processes.

Numerous studies have shown that Cognitive Stroop interference is predicted by

participants’ working-memory efficacy (e.g., Long & Prat, 2002; Kane & Engle, 2003;

Kiefer, Ahlegian, & Spitzer, 2005). Differences in the efficiency of two subcomponents

of working memory, the central executive and the phonological loop (Baddeley, 2003),

are likely to account for interindividual differences in Cognitive Stroop interference.

Participants with a less efficient central executive, which is involved in controlling and

regulating information (Baddeley, 2003), are less effective at inhibiting task-irrelevant

semantic information (Long & Prat, 2002). In addition, participants with a less efficient

phonological loop, which is involved in rehearsing processed information and has a

limited capacity (Baddeley, 2003), might be more handicapped by the involuntary storage

of task-irrelevant semantic information (Saeki, 2007). If the Emotional Stroop effect is in

fact a Cognitive Stroop effect, then interindividual differences in Emotional Stroop

interference should be predicted by differences in working-memory efficacy, and more

specifically by differences in the efficiency of the central-executive subcomponent and/or

of the phonological-loop subcomponent. We tested these predictions in the present study.

5

We measured working-memory efficacy using the running-span task (Pollack Johnson, &

Knaff, 1959), a task which allowed us to isolate the respective contribution of central

executive and phonological loop (Vieillard & Bougeant, 2005).

In contrast, or in addition, to the cognitive account, a more widely-held view

suggests that the Emotional Stroop task measures emotion-driven interference (Ben-

Haim, Mama, Icht, & Algom, 2013). Words reminiscent of past emotional experiences

are relevant to people’s goals, core beliefs or identity and trigger emotional reactions

(e.g., increases in arousal levels; Dresler, Mériau, Heekeren, & Van der Meer, 2009).

This idea is compatible with findings that, unlike the Cognitive Stroop task, the

Emotional Stroop task modulates brain activity in the amygdala and occipito-temporal

areas, regions involved in emotional processing (Compton et al., 2003). However it is

impossible to know whether this activation of emotional brain areas is causal or

necessary to the resulting interference. If the effect of emotional words relies on

emotional rather than cognitive processes, it should bear little relation to differences in

working-memory efficacy. Instead, it should be linked to the ‘emotionality’ of the stimuli

used.

Emotionality refers to the emotional impact of a stimulus, regardless of its valence

(i.e., both positive and negative stimuli can be highly emotional; Gilboa-Schechtman et

al., 2000; Schimmack & Derryberry, 2005). Emotional stimuli have an impact both on

individuals’ physiological state (e.g., increase in arousal; Lang, Greenwald, Bradley, &

Hamm, 1993) and on their subjective state (e.g., mood; Ben-Haim et al., 2013), effects

which are intimately related (Greenwald, Cook, & Lang, 1989). Previous findings

suggest that emotionality is central to predicting Emotional Stroop interference (e.g.,

Ben-Haim et al., 2013; Hart, Green, Casp, & Belger, 2010; Salters‐Pedneault, Gentes, &

Roemer, 2007). Accordingly, some data showed that emotional stimuli, which were more

arousing to participants with high than with low state anxiety, also caused more

Emotional Stroop interference in participants with high than with low state anxiety

(Dresler et al., 2009). In addition, non-words that had been conditioned to have a negative

connotation generated longer reaction times than non-words that had been conditioned to

have a neutral connotation, thus highlighting the role of emotionality in Emotional Stroop

interference (Richards & Blanchette, 2004). In sum, an emotional account suggests it is

the emotional reactions triggered by the trauma-related contents, rather than trauma-

related semantic contents per se, that are responsible for Emotional Stroop interference.

The present study contrasted the cognitive and emotional accounts for Emotional

Stroop interference. Though they are not mutually exclusive, we attempt to investigate

their relative influence. Participants were recruited among a female undergraduate

population who had a varying experience of sexual abuse, from no abuse to high-severity

abuse. We used a standard Emotional Stroop task (Ray, 1979) in which target words

6

pertained to one of three semantic contents (neutral, generally emotional or sexual-abuse

related) and were presented in one of four possible colours (red, green, yellow or blue).

Participants’ reaction times at discriminating colour print (using a four-alternative forced

choice) were measured. Emotional Stroop interference was indexed by subtracting RTs

for neutral words from RTs for sexual-abuse and generally emotional words. After

performing the Emotional Stroop task, participants rated the emotionality of the target

words. In addition, two working-memory indices were extracted using a running-span

task (Bunting, Cowan, & Saults, 2006; Pollack et al., 1959; Van der Linden et al., 1999);

the first one measured phonological-loop capacity (rehearsing) and the other central-

executive efficiency (attentional flexibility; Vieillard & Bougeant, 2005). We measured

the effect on Emotional Stroop interference of severity of sexual-abuse experience,

emotionality ratings, phonological-loop capacity and central-executive efficiency.

We expected a positive relationship between severity of abuse experience and

sexual-abuse interference (Cassiday et al., 1992). In addition, two possibilities were

considered in terms of the relationships between sexual-abuse interference, on the one

hand, and emotionality and working-memory measures, on the other. First, an absence of

relationship between interference and working-memory measures would make the

cognitive account unlikely. Second, in the absence of a relationship between interference

and working-memory measures, a relationship between interference and emotionality

ratings would make the emotional account likely.

2. Method

2.1 Participants

Participants were undergraduate Psychology students recruited from the

Université du Québec à Trois-Rivières, Canada, using a participant database. Sexual

abuse was not mentioned at the recruitment stage; participants were informed that they

would answer questions about past experiences of abuse during the information and

consent stage. Only women were recruited since sexual abuse is more relevant to this

group (e.g., in 2008, more than 80% of sexual-abuse victims were women in Canada; see

www.statcan.gc.ca). Twenty-three participants had experienced no abuse (Mean age 21.7

years, SEM =1.1; 14.7 years of education in average, SEM = 0.4; see Table 1) and 35

participants had experienced at least one type of sexual abuse (Mean age 23.5 years, SEM

= 0.9; 14.6 years of education in average, SEM = 0.2). None of the participants had

reported being diagnosed with Posttraumatic Stress Disorder (PTSD) at the time of

testing. All participants had normal or corrected-to-normal vision and spoke French as

their first language. They received $10 as a compensation for their time. Participants

were debriefed at the end of the experiment and were provided with local and national

7

contact details should they require psychological help following the experiment. No

participant reported feeling distressed or requiring counselling.

Table 1. Demographics, PCL-C score and working memory

Group

(n)

Age (yrs)

M

(SD)

Edu (yrs)

M

(SD)

PCL-C

M

(SD)

PL capacity

M

(SD)

CE efficicency

M

(SD)

Abuse

victims

(35)

21.7

(4.6)

14.6

(1.2)

28.9

(10.5)

2.0

(1.0)

0.4

(0.4)

No-abuse

controls

(23)

23.5

(5.2)

14.7

(1.9) -

2.9

(1.1)

0.3

(0.3)

yrs = Years; Edu = Education; PCL-C = Posttraumatic Checklist – Civilian version; PL

= Phonological Loop; CE = Central Executive; n = Number of participants; M = Mean;

SD = Standard Deviation.

2.2 Experimental Setup

Testing took place in dimly-lit and quiet testing room. Questionnaires and stimuli

were presented on an LCD 22-in monitor, operating at a resolution of 1280 x 1024 pixels.

Viewing distance was 60 cm. The stimuli were generated and the experiment was run

using EPrime (Schneider, Eschman, & Zuccolotto, 2002).

2.3 Procedure

2.3.1 Questionnaires

Participants first completed a French-translated 6-item adaptation of the Stressful

Life Events Screening Questionnaire (SLESQ; Goodman, Corcoran, Turner, Yuan, &

Green, 1998) which screened for the occurrence of traumatic sexual event(s). Participants

were asked whether they had ever (1) been touched against their will on intimate parts of

their body, (2) been rubbed by someone else’s private parts against their will, (3) been

forced to touch the intimate parts of someone else’s body, (4) been forced to take part in a

genital sexual intercourse, (5) been forced to take part in an oral sexual intercourse,

and/or (6) been forced to kiss someone in a sexual way. They gave yes/no answers to

each of these six possibilities. ‘No’ answers were coded as zero and ‘yes’ answers as one;

each participant was thus given a score from zero to six. This score was used as a

measure of the ‘objective’ severity of the experience of abuse. Participants scoring above

8

zero filled a second questionnaire, namely, a French-translated version of the

Posttraumatic Stress Disorder CheckList – Civilian Version (PCL-C; Blanchard, Jones-

Alexander, Buckley, & Forneris, 1996). This 17-item questionnaire indexed the incidence

of posttraumatic stress symptoms related to the traumatic sexual event(s). Participants

gave an answer from one (not at all) to five (extremely) to each item. Items addressed

symptoms often associated with PTSD, such as the occurrence of disturbing dreams, the

inability to concentrate or the intrusion of disturbing thoughts related to the event.

Answers were added to produce a score from 17 (no post-traumatic stress symptoms) to

85 (maximum level of post-traumatic stress symptoms). The PCL-C questionnaire was

used to confirm that SLESQ scores indexed experience severity; higher SLESQ scores

were expected to be related to higher PTSD symptoms. The questionnaire was also used

to identify participants with clinical levels of PTSD symptoms since, in these

participants, cognitive differences may pre-exist trauma exposure (Navalta et al., 2006).

2.3.2 Emotional Stroop Task

After filling in the questionnaire(s), participants performed an adaptation of the

Emotional Stroop task. At the beginning of each trial, a fixation cross was displayed for a

varying duration, selected randomly between 100 and 300 ms. Following fixation-cross

offset, a target word (1° in height and 4-12° in width) printed in red, blue, green or

yellow was presented at fixation. Participants indicated the word colour in a speeded way

using four colour-stickered keys: the ‘1’, ‘2’, ‘9’ and ‘0’ keys (from the top side of the

keyboard) indicated red, green, yellow and blue, respectively. Responses were given

using the middle finger and forefinger of the left hand for red (‘1’) and green (‘2’),

respectively, and using the forefinger and the middle finger of the right hand for yellow

(‘9’) and blue (‘0’), respectively. Following key press, the target word disappeared and a

200-ms blank followed, after which the next trial started. At the beginning of the

experiment, participants were informed that the meaning of the target word was irrelevant

to the task.

The words used in the task pertained to three types of semantic content; ten words

were neutral (i.e., the French words for ‘turnip’, ‘board’, ‘fan’, ‘shelf’, ‘bag’,

‘discussion’, ‘keyboard’, ‘typical’, ‘lettuce’ and ‘formulate’), ten words were generally

emotional (i.e., the French words for ‘death’, ‘murder’, ‘mourning’, ‘funeral’, ‘cancer’,

‘tumour’, ‘stress’, ‘abandonment’, ‘panic’ and ‘hatred’), and ten words were related to

sexual abuse (i.e., the French words for ‘erection’, ‘exploited’, ‘pawed’, ‘penetrated’,

‘victim’, ‘incest’, ‘molested’, ‘abused’, ‘rape’). The words of each type of semantic

content were presented in separate blocks of trials (because blocked presentation has been

found to potentiate the Emotional Stroop effect; e.g., Kaspi, McNally, & Amir, 1995).

Within each block of trials, each word was presented in each of the four colours once,

giving a total of 40 trials. Words were selected at random within each block of trials and

9

the three blocks (‘neutral’, ‘generally emotional’ and ‘sexual abuse’) were performed in a

random order across participants.

Before starting the task, participants performed a 12-trial practice using twelve

neutral words drawn from a pool other than the one used in the task. After completing the

task, participants rated the emotionality of the 30 words used in the task (presented in a

random order) using a scale ranging from one (completely neutral) to five (intensely

emotional).

2.3.3 Running span

After performing the Emotional Stroop task, participants performed an adaptation

of the Running Span task (Bunting, Cowan, & Saults, 2006; Van der Linden et al., 1999;

Vieillard & Bougeant, 2005) which comprised two blocks. In the first block, on each

trial, a list of four, six or eight randomly-selected consonants were presented serially at

fixation. There was no repetition within a list. Each consonant was presented for 850 ms

and followed by a 650 ms blank. Participants were told that they would have to recall

(type in) the last four consonants of each list upon hearing a 900ms beep that followed

the presentation of the last consonant. The block consisted of 12 lists and was preceded

by a two-list practice. The second block was identical to the first, except that (1) the lists

contained five, seven or nine consonants (instead of four, six or eight) and (2) participants

were asked to remember the last five consonants (instead of the last four) of each list. A

trial was counted as correct when all four/five consonants were correctly recalled

(sequence knowledge was required).

For each participant and for each list type (i.e., for the four-, five-, six-, seven-,

eight- and nine-item lists), a memory span was calculated by multiplying the number of

item in the list by the average accuracy for that list. Two indices were then extracted.

First, for each participant, the spans for the four- and five-item lists were averaged.

Because the four- and five-item lists did not necessitate any information manipulation (all

the items of the list had to be maintained in memory), this first index was hypothesized to

measure phonological-loop capacity (Vieillard & Bougeant, 2005). The higher the value

of this index (minimum 0; maximum 4.5), the better participants were at holding

information in their phonological loop. Second, for each participant, the six-, seven-,

eight- and nine-item spans were averaged and the resulting average was divided by the

average of the four- and five-item spans. Because the 6/7/8/9-item spans involved

information manipulation (inhibiting old information and replacing it with new

information), this second index was posited to measure central-executive efficiency

(Vieillard & Bougeant, 2005). The higher the value of this second index was, the greater

the central executive resources (minimum 0, maximum 1; note that, while participants

might have an index greater than one if they memorised more items in the 6/7/8/9-item

10

lists than in the 4/5-item lists, there is no theoretical justification for predicting this

possibility).

3. Results

3.1 Questionnaires

Participants’ questionnaire scores were examined. In the Stressful Life Events

Screening Questionnaire (sexual trauma subscale), 23 participants scored zero (i.e., they

reported no traumatic sexual event). Thirty-five participants reported one or more

traumatic sexual event; 17 scored one, eight scored two, three scored three, three scored

four, three scored five and one scored six. Of these 35 participants, 32 filled the

Posttraumatic Stress Disorder Checklist – Civilian Version (three participants did not fill

this second questionnaire due to a technical issue). Their average score was 28.9 (SEM =

1.8; range 17 to 57). Scores on the PCL-C were highly correlated with their scores on the

SLESQ, r(32) = .72, p < .001, confirming that SLESQ scores were an indicator of

experience severity (Dickinson, DeGruy, Dickinson, & Candib, 1999). Only one

participant had a PCL-C score higher than the PTSD cut-off score (50; Blanchard et al.,

1996).

3.2 Effect of severity of sexual-abuse experience

In a first set of analyses, we examined the effect of severity of sexual-abuse

experience on working-memory spans, emotionality ratings and Emotional Stroop

interference.

3.2.1 Running Span

We tested whether there were significant correlations between severity of abuse

experience (SLESQ scores), on the one hand, and phonological-loop capacity and

central-executive efficiency, on the other. There was a significant negative relationship

between severity of abuse and phonological-loop capacity, r(58) = -.26, p = .051,

describing the fact that participants who had experienced a more severe form of abuse

had a smaller phonological-loop capacity. In contrast, there was no relationship between

severity of abuse and central-executive efficiency, r(58) = .11, p = .419, suggesting that

the experience of abuse did not affect the central executive.

3.2.2 Emotionality Ratings

Participants’ emotionality ratings for the thirty words used in the Emotional

Stroop task were examined (see Table 2). First, a bivariate correlation showed that

emotionality ratings for neutral words (M = 1.1, SEM = 0.04) did not vary as a function

of SLESQ scores (i.e., severity of abuse experience), r(58) = -.08, p = .567. Ratings for

11

neutral words were used as a baseline from which relative emotionality ratings could be

extracted for each participant. Relative emotionality ratings were indexed by partialling

out emotionality ratings for neutral words from emotionality ratings for generally

emotional and sexual-abuse words. A mixed-design analysis of covariance (ANCOVA)

tested for the effect on relative emotionality ratings of semantic content (generally

emotional or sexual abuse; within-subject variable) and SLESQ scores (severity of

experience, entered as a between-subject continuous covariable). It showed a significant

interaction between the two factors, F(1,56) = 5.49, p = .023, ηp2 = .089. To understand

the origin of this interaction, the relationship between SLESQ scores and emotionality

ratings was tested separately for each semantic content. Whereas there was a positive

relationship between SLESQ scores and emotionality ratings for sexual-abuse words,

r(58) = .36, p = .006 (see Figure 1a), there was no significant relationship between

SLESQ scores and emotionality ratings for generally emotional words, r(58) = .20, p =

.147 (see Figure 1b).

Table 2. Emotional Stroop task: reaction times, errors and subjective emotionality

ratings

Emotional Stroop task

RT (ms) Errors (%) Subjective ratings

Group

(n)

AR

M

(SD)

GE

M

(SD)

N

M

(SD)

AR

M

(SD)

GE

M

(SD)

N

M

(SD)

AR

M

(SD

)

GE

M

(SD)

N

M

(SD)

Abuse

victims

(35)

788.8

(205.4)

774.3

(199.6)

753.4

(177.6)

2.9

(4.0)

2.4

(4.5)

2.7

(2.2)

2.6

(1.0

)

2.8

(0.9)

1.0

(0.1)

No-

abuse

controls

(23)

685.6

(90.6)

693.0

(116.1)

662.6

(92.4)

2.2

(2.4)

1.7

(1.9)

0.9

(1.5)

3.1

(1.2

)

2.9

(1.1)

1.0

(0.1)

AR = Abuse Related; GE = Generally Emotional; N = Neutral; n = Number of

participants; M = Mean; SD = Standard Deviation.

Other findings from the main analysis were (1) significant main effect of semantic

content, F(1,56) = 4.46, p = .039, ηp2 = .074, describing the fact that relative emotionality

12

ratings were slightly higher for sexual-abuse content (M = 1.9, SEM = 0.16) than for

generally emotional content (M = 1.8, SEM = 0.12), and (2) a significant main effect of

SLESQ scores, F(1,56) = 5.50, p = .023, ηp2 = .090, describing an overall positive

relationship between relative emotionality ratings and SLESQ scores.

3.2.3 Emotional Stroop Task

Participants’ mean reaction times (RTs) were averaged for each condition of the

Emotional Stroop task (see Table 2). Only accurate responses and reaction times above

300 ms and below 2000 ms were included in the calculation (3.6% of RTs were excluded

under these criteria). First, a bivariate correlation showed that RTs in the neutral

condition (M = 715 ms, SEM = 19.8) did not vary as a function of SLESQ scores (i.e.,

severity of abuse experience), r(58) = .08, p = 547, showing that overall performance was

not affected by the experience of abuse. Neutral-word RTs were used as a baseline from

which interference from task-irrelevant emotional information could be calculated for

each participant. Interference was indexed by partialling out neutral RTs from generally

emotional and sexual-abuse RTs. A one-sample t-test showed that overall interference

(i.e., pooled across semantic contents) was significantly higher than zero (M = 27.6 ms,

SEM = 8.4), t(57) = 3.27, p = .002, d = 0.621, confirming that emotional words had a

deleterious effect on performance. Then, a mixed-design ANCOVA tested for the effect

on interference of semantic content (generally emotional or sexual-abuse; within-subject

variable) and SLESQ scores (severity of experience; between-subject continuous

covariable). While neither the main effect of semantic content nor that of SLESQ scores

was significant, respectively, F(1,56) = 0.84, p = .363, ηp2 = .015 and F(1,56) = 0.67, p =

.416, ηp2 = .012, there was a significant interaction between the two factors, F(1,56) =

4.90, p = .031, ηp2 = .080. To understand the origin of this interaction, the relationship

between SLESQ scores and interference was tested separately for each semantic content.

Whereas there was a relationship between SLESQ scores and sexual-abuse interference

which was close to significance, r(58) = .26, p = .055 (see Figure 1c), there was no

significant relationship between SLESQ scores and generally emotional interference,

r(58) = -.04, p = .757 (see Figure 1d).

13

Figure 1. Emotional Stroop task. Relationship between severity of abuse (SLESQ scores;

min. 0, max. 6) and (1a) relative emotionality ratings for abuse-related words (min. 0,

max. 4), (1b) relative emotionality ratings for generally emotional words (min. -4, max.

4), (1c) interference with abuse-related words (ms), and (1d) interference with generally

emotional words (ms). Each dot represents the data of one participant.

In summary, the severity of the abuse experience predicted the strength of the

interference generated by sexual-abuse words. This relationship between severity and

interference could be explained by the fact that subjective emotionality for sexual-abuse

words increased as severity increased, and/or by the fact that phonological-loop capacity

decreased as severity increased. A last set of analyses was performed to explore the

relationship between Emotional Stroop interference, emotionality and working memory.

3.3 Effects of emotionality and working memory

14

A mixed-design ANCOVA tested for the effect on interference of semantic

content (generally emotional or sexual-abuse; within-subject variable) and of

phonological-loop capacity, central-executive efficiency, relative emotionality ratings for

sexual-abuse words and relative emotionality ratings for generally emotional words (all

between-subject continuous covariables).

There was a significant two-way interaction between semantic content and

relative emotionality ratings for sexual-abuse words, F(1,53) = 12.79, p = .001, ηp2 =

.194. To understand the origin of this interaction, the relationship between emotionality

ratings and interference was tested separately for each semantic content. Whereas there

was a significant relationship between sexual-abuse emotionality ratings and sexual-

abuse interference, r(58) = .27, p = .045 (see Figure 2a), there was no such relationship

between sexual-abuse emotionality ratings and generally emotional interference, r(58) = -

.14, p = .277 (see Figure 2b). No other main effect or interaction was significant (no p

value < .150). Notably, neither the main effect of phonological-loop capacity nor that of

central-executive efficiency was significant, respectively, F(1,53) = 0.15, p = .902, ηp2 =

.001, and F(1,53) = 0.76, p = .387, ηp2 = .014. In addition, neither the interaction between

phonological-loop capacity and semantic content nor that between central-executive

efficiency and semantic content was significant, respectively, F(1,53) = 0.15, p = .700,

ηp2 = .003 and F(1,53) = 0.46, p = .479, ηp

2 = .009. This showed that phonological-loop

capacity and central-executive efficiency did not predict Emotional Stroop interference

(see Figures 2c and 2d depicting this absence of relationship).

4. Discussion

In this study, we contrasted a cognitive and an emotional account for the origin of

Emotional Stroop interference. We measured interference generated by sexual-abuse

content in participants who had a varying experience of sexual abuse, from no abuse to

high-severity abuse, and we tested the effect on this interference of (1) severity of sexual-

abuse experience, (2) subjective emotionality, and (3) two indices of working-memory

efficacy, namely, phonological-loop capacity and central-executive efficiency.

Participants who had experienced more severe abuse showed greater sexual-abuse

interference. Given that relevance is posited to increase with severity, this finding was

consistent with the idea that Emotional Stroop interference strongly relies on experience

relevance (e.g., Williams et al., 1996). Emotionality ratings for sexual-abuse content were

also positively related to severity, suggesting that the more relevant an experience was,

the more emotional it was. Finally, greater emotionality ratings for sexual-abuse content

predicted stronger sexual-abuse interference. This finding was consistent with an

emotional account for Emotional Stroop interference (e.g., Richards & Blanchette, 2004);

words reminiscent of past emotional experiences trigger emotional reactions (e.g.,

15

increases in arousal levels; Dresler et al., 2009) that interfere with the primary colour-

naming task.

Figure 2. Emotional Stroop task. Relationship between relative emotionality ratings for

abuse-related words (min. 0, max. 4) and (2a) interference with abuse-related words

16

(ms), and (2b) interference with generally emotional words (ms). Relationship between

interference with abuse-related words (ms) and (2c) phonological-loop capacity (min. 0,

max. 4.5), and (2d) central-executive efficiency (min. 0, max 1). Relationship between

interference with generally emotional words (ms) and (2e) phonological-loop capacity

(min. 0, max. 4.5), and (2f) central-executive efficiency (min. 0, max 1). Each dot

represents the data of one participant.

Unlike emotionality ratings, neither phonological-loop capacity nor central-

executive efficiency predicted sexual-abuse interference. Thus, unlike Cognitive Stroop

interference (Long & Prat, 2002; Kane & Engle, 2003; Kiefer et al., 2005), Emotional

Stroop interference was not related to working memory. This null-effect was observed

despite the fact that participants who had experienced more severe sexual abuse had a

lower phonological-loop capacity. Thus, the positive relationship between severity and

sexual-abuse interference, on the one hand, and the negative relationship between

severity and phonological-loop capacity, on the other, appeared to be orthogonal. In sum,

the present findings did not provide support for the cognitive account for Emotional

Stroop interference.

It is noteworthy that, because the running-span task was always performed after

the Emotional Stroop task, the lower phonological-loop capacity observed in high-

severity participants might have resulted from a stronger impact of the Emotional Stroop

task in these participants (e.g., a stronger increase in arousal). If this is true, lower

phonological-loop capacity in these participants would have been only temporary.

However, this suggestion is inconsistent with previous findings showing lower working-

memory capacity in victims of sexual abuse (compared to no-abuse controls) even in the

absence of a task that promotes trauma-related thoughts (Stein et al., 2002). In any case,

reduced phonological loop capacity in high-severity participants could account neither for

the presence of a relationship between emotionality ratings and interference, nor for the

absence of such relationship between working-memory measures and interference. It is

interesting to note that, in the present study, participants exposed to abuse did not show

lower central-executive efficiency. This finding is somehow at odds with the idea that

trauma impairs executive functioning (Stein et al., 2002).

The last finding that we discuss concerns interference generated by generally

emotional content. Generally emotional interference was not related to emotionality

ratings for generally emotional words. This null effect might have been due to the low

relevance of generally emotional words (Becker et al., 2001; Blanchette & Richards,

2012) and/or to the tighter distribution of ratings (SD = 0.9 for generally emotional

ratings vs. SD = 1.2 for sexual-abuse ratings). Another possibility is linked to task order.

Participants always filled questionnaires before performing the Stroop task. This might

have primed emotions and memories in participants who have previously experienced

17

abuse, thus potentiating abuse-related interference and strengthening the relationship

between abuse-related interference and emotionality ratings for abuse-related words.

Such effect would not have occurred for generally emotional interference.

In conclusion, the present findings show that relevant emotional stimuli, which

are central to participants’ core beliefs and identity, are more difficult to ignore and that

they impact performance on a simple perceptual task. This effect on participants’

attention and voluntary control involves emotional processes, and not only cognitive

processes. These results are good news given all the studies which have assumed, despite

the absence of much evidence, that the Emotional Stroop task measures processes which

are emotion specific and which are to a certain extent distinct from the processes

measured in the Cognitive Stroop task, at least in their origin. Because this study was

performed on a particular population with a specific trauma – students with an experience

of sexual abuse – it will be important to replicate its findings with other populations (e.g.,

clinical) and other types of trauma (e.g., political violence). In any case, however, the

results suggest that the Emotional Stroop task can be used as an index of emotional

filtering efficiency.

More work is needed to understand fully the mechanisms underlying Emotional

Stroop interference. Even though cognitive and emotional processes are not mutually

exclusive and are in fact strongly connected (as illustrated by the overlap in brain regions

activated by Emotional and Cognitive Stroop tasks; Whalen et al., 1998), it should be

possible to isolate their relative contribution.

5. References

Baddeley, A. (2003). Working Memory: Looking back and looking forward.

Nature Neuroscience Reviews, 4, 829-839. doi:10.1038/nrn1201

Becker, E. S., Rinck, M., Margraf, J., & Roth, W. T. (2001). The emotional

Stroop effect in anxiety disorders: General emotionality or disorder specificity? Anxiety

Disorders, 15, 147-159. doi:10.1016/S0887-6185(01)00055-X

Ben-Haim, M. S., Mama, Y., Icht, M., & Algom, D. (2013). Is the emotional

Stroop task a special case of mood induction? Evidence from sustained effects of

attention under emotion. Attention, Perception, & Psychophysics, 76, 81-97. doi:

10.3758/s13414-013-0545-7

Blanchard, E. B., Jones-Alexander, J., Buckley, T. C., & Forneris, C. A. (1996).

Psychometric properties of the PTSD Checklist (PCL). Behaviour Research and Therapy,

34, 669-673.

18

Blanchette, I., & Richards, A. (2012). Is Emotional Stroop interference linked to

affective responses? Evidence from skin conductance and facial electromyography.

Emotion, 13, 129-138. doi:10.1037/a0029520

Bunting, M., Cowan, N., & Saults, J. S. (2006). How does running memory span

work? The Quarterly Journal of Experimental Psychology, 59, 1691-1700. doi:

10.1080/17470210600848402

Cassiday, K. L., McNally, R. J., & Zeitlin, S. B. (1992). Cognitive processing of

trauma cues in rape victims with posttraumatic stress disorder. Cognitive Therapy and

Research, 16, 283–295. doi:10.1007/BF01183282

Cisler, J. M., Wolitzky-Taylor, K. B., Adams, T. G. Jr., Babson, K. A., Badour, C.

L., & Willems, J. L. (2011). The emotional Stroop task and posttraumatic stress disorder:

A meta-analysis. Clinical Psychology Review, 31, 817–828. doi:10.1016/j.cpr.2011.03

.007

Compton, R. J., Banich, M. T., Mohanty, A., Milham, M. P., Herrington, J.,

Miller, G. A., Scalf, P. E., et al. (2003). Paying attention to emotion: An fMRI

investigation of cognitive and emotional Stroop tasks. Cognitive, Affective &Behavioral

Neuroscience, 3, 81–96. doi:10.3758/CABN.3.2.81

Dalgleish, T. (1995). Performance on the emotional Stroop task in groups of

anxious, expert and control participants: A comparison of computer and card presentation

formats. Cognition and Emotion, 9, 341-362. doi:10.1080/02699939508408971

Dickinson, L. M., DeGruy, F. V., Dickinson, W. P., & Candib, L. M. (1999).

Health related quality of life and symptom profiles of female survivors of sexual abuse.

Archives of Family Medicine, 8, 35-43. doi:10.1001/archfami.8.1.35

Dresler, T., Mériau, K., Heekeren, H. R., & Van der Meer, E. (2009). Emotional

Stroop task: effect of word arousal and subject anxiety on emotional interference.

Psychological Research, 73, 364-371. doi:10.1007/s00426-008-0154-6

Gilboa-Schechtman, E., Revelle, W., & Gotlib, I. H. (2000). Stroop interference

following mood induction: Emotionality, mood congruence, and concern relevance.

Cognitive Therapy & Research, 24, 491-502. doi:10.1023/A:1005517326981

Goodman, L. A., Corcoran, C., Turner, K., Yuan, N., & Green, B. L. (1998).

Assessing traumatic event exposure: General issues and preliminary findings for the

Stressful Life Events Screening Questionnaire. Journal of Traumatic Stress, 11, 521-542.

19

Greenwald, M. K., Cook, E. W., & Lang, P. J. (1989). Affective judgment and

psychophysiological response: Dimensional covariation in the evaluation of pictorial

stimuli. Journal of Psychophysiology, 3, 51–64.

Gronau, N., Cohen, A., & Ben-Shakhar, G. (2003). Dissociations of personally

significant and task-relevant distractors inside and outside the focus of attention: a

combined behavioral and psychophysiological study. Journal of Experimental

Psychology: General, 132, 512-529. doi:10.1037/0096-3445.132.4.512

Hart, S. J., Green, S. R., Casp, M., & Belger, A. (2010). Emotional priming effects

during Stroop task performance. Neuroimage, 49, 2662-2670. doi:

10.1016/j.neuroimage.2009.10.076

Kane, M. J., & Engle, R. W. (2003). Working-memory capacity and the control of

attention: The contributions of goal neglect, response competition, and task set to Stroop

interference. Journal of Experimental Psychology: General, 132, 47–70. doi:

10.1037/0096-3445.132.1.47

Kaspi, S. P., McNally, R. J., & Nader, A. (1995). Cognitive processing of

emotional information in posttraumatic stress disorder. Cognitive Therapy and Research,

19, 433-444. doi:10.1007/BF02230410

Kiefer, M., Ahlegian, M., & Spitzer, M. (2005). Working memory capacity,

indirect semantic priming, and Stroop interference: Pattern of interindividual prefrontal

performance differences in healthy volunteers. Neuropsychology, 19, 332-344. doi:

10.1037/0894-4105.19.3.332

Jenkins, M. A., Langlais, P. J., Delis, D., & Cohen, R. (1998). Learning and

memory in rape victims with posttraumatic stress disorder. American Journal of

Psychiatry, 155, 278-279.

Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O. (1993). Looking at

pictures: Affective, facial, visceral, and behavioral reactions. Psychophysiology, 30, 261-

273. doi:10.1111/j.1469-8986.1993.tb03352.x

Litz, B. T., Weathers, F. W., Monaco, V., Herman, D. S., Wulfsohn, M., Marx, B.,

& Keane, T. M. (1996). Attention, arousal, and memory in posttraumatic stress disorder.

Journal of Traumatic Stress, 9, 497–519. doi:10.1002/jts.2490090308

Long, D. L., & Prat, C. S. (2002). Working memory and Stroop interference: An

individual differences investigation. Memory & Cognition, 30, 294-301.

doi:10.3758/BF03195290

20

Navalta, C. P., Polcari, A., Webster, D. M., Boghossian, A., & Teicher, M. H.

(2006). Effects of childhood sexual abuse on neuropsychological and cognitive function

in college women. Journal of Neuropsychiatry and Clinical Neuroscience, 18, 45–53.

doi:10.1176/appi.neuropsych.18.1.45

Pollack, I., Johnson, I. B., & Knaff, P. R. (1959). Running memory span. Journal of

Experimental Psychology, 57, 137–146. doi:10.1037/h0046137

Ray, C. (1979). Examination stress and performance on a color word interference

test. Perceptual and Motor Skills, 49, 400-402. doi:10.2466/pms.1979.49.2.400

Resnick, H. S., Kilpatrick, D. G., Dansky, B. S., Saunders, B. E., & Best, C. L.

(1993). Prevalence of civilian trauma and posttraumatic stress disorder in a representative

national sample of women. Journal of Consulting and Clinical Psychology, 61, 984-991.

doi:10.1037/0022-006X.61.6.984

Richards, A., & Blanchette, I. (2004). Independent manipulation of emotion in an

emotional Stroop task using classical conditioning. Emotion, 4, 275-281.

doi:10.1037/1528-3542.4.3.275

Saeki, E. (2007). Phonological loop and goal maintenance: Effect of articulatory

suppression in number size consistency task. Psychologia, 50, 122-132.

doi:10.2117/psysoc.2007.122

Salters‐Pedneault, K., Gentes, E., & Roemer, L. (2007). The role of fear of emotion

in distress, arousal, and cognitive interference following an emotional stimulus. Cognitive

Behaviour Therapy, 36, 12-22. doi:10.1080/16506070600874281

Schneider, W., Eschman, A., &Zuccolotto, A. (2002). E-Prime user’s guide.

Pittsburgh: PA: Psychology Software Tools.

Stein, M. B., Hanna, C., Vaerum, V., & Koverola, C. (1999). Memory functioning

in adult women traumatized by childhood sexual abuse. Journal of Traumatic Stress, 12,

527-534. doi:10.1023/A:1024775222098

Stein, M. B., Kennedy, C. M., & Twamley, E. W. (2002). Neuropsychological

function in female victims of intimate partner violence with and without posttraumatic

stress disorder. Biological Psychiatry, 52, 1079–1088. doi:10.1016/S0006-

3223(02)01414-2

Schimmack, U., & Derryberry, D. E. (2005). Attentional interference effects of

emotional pictures: threat, negativity, or arousal? Emotion, 5, 55-66. doi:10.1037/1528-

3542.5.1.55

21

Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of

Experimental Psychology, 18, 643-662. doi:10.1037/h0054651

Van der Linden, M., Hupet, M., Feyereisen, P., Schelstraete, M. A., Bestgen, Y.,

Bruyer, R., Lories, G., El Ahmadi, A., and Seron, X. (1999). Cognitive mediators of age-

related differences in language comprehension and verbal memory performance. Aging,

Neuropsychology, and Cognition, 6, 32-55. doi: 10.1076/anec.6.1.32.791

Vieillard, S., & Bougeant, J.-C. (2005). Performance à une tâche de mémoire de

travail sous induction émotionnelle négative : Influence modulatrice de l’état émotionnel

sur les processus exécutifs. L’Année Psychologique, 2005, 63-104.

doi:10.3406/psy.2005.3820

Whalen, P. J., Bush, G., McNally, R. J., Wilhelm, S., McInerney, S. C., Jenike, M.

A., & Rauch, S. L. (1998). The emotional counting Stroop paradigm: a functional

magnetic resonance imaging probe of the anterior cingulate affective division. Biological

Psychiatry, 44, 1219–28. doi:10.1016/S0006-3223(98)00251-0

Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop

task and psychopathology. Psychological Bulletin, 120, 3–24. doi:10.1037/0033-

2909.120.1.3