Stroop Interference following Mood Induction:
Emotionality, Mood Congruence, and Concern Relevance

Eva Gilboa-Schechtman,1 William Revelle,2 and Ian H. Gotlib3

Research using the emotion Stroop task has established that individuals with various
emotional disorders exhibit increased interference for stimuli specifically related to
their disorder due to the concern relevance and negativity of these stimuli.

Our
study expands this research to normal populations. We examined the influence of
emotionality, mood congruence, and concern relevance following experimental mood-
induction procedures (MIPs) on emotion Stroop performance of college students.


Participants completed a computerized emotion Stroop task following positive and
negative MIPs. Results supported the mood congruence and concern relevance but
not the emotionality hypotheses.

The implication of these findings for theories of
emotional breakdown and the importance of studying the idiographic aspects of
affective experience are discussed.


A robust finding based on the emotion Stroop effect is that individuals suffering
from an emotional disorder exhibit selective processing of stimuli that are idiosyn-
cratic to their disorder (for reviews, see Logan & Goetsch, 1993; Mathews & MacLeod
1994; Williams, Mathews, & MacLeod, 1996).

This selective processing could be at-
tributed to three distinct factors: emotionality (i.e., the ‘‘emotional impact’’ of the
stimulus, regardless of its valence), concern relevance (i.e., the association between
the stimulus and the participant’s current concerns), and mood congruence (i.e., the
concordance between a person’s affective state and the valence of the stimulus).

For
certain stimuli, all three factors coincide. For example, social phobics are likely to
perceive such stimulus words as humiliation or rejection as extremely emotional, con-
gruent with their dysphoric affect, and highly related to their current concerns.

In
contrast, other stimuli may differentiate among these factors. For instance, the words
cancer or accident, although high in emotionality and negativity, are not related to
social phobics’ concerns. Similarly, words such as acceptance, although possibly high
in emotionality and in concern relevance, are not negative.

On the basis of an extensive review of the literature on emotional Stroop and
psychopathology, Williams et al. (1996) conclude that individuals with emotional
disturbance show disproportionate color-naming interference for negative stimuli
and for stimuli related to personally relevant themes.

It seems likely that the
interference of negative materials in clinical populations (which is over and above
the interference due to these materials’ concern relevance) is due to mood-congru-
ence effects. Stroop interference in highly anxious participants is influenced by
positive as well as negative mood (Richards, French, Johnson, Naparstek, &
Williams, 1992, Experiment 2).

Combined, these findings suggest that emotion
Stroop interference in clinical and subclinical participants is mediated by mood
congruence and concern relevance, but not by emotionality.

The extent to which each of the three factors—emotionality, mood congruence,
and concern relevance—affects Stroop interference in normal populations is not
yet clear.

Williams et al. (1996) hypothesize that concern relevance but not other
factors may characterize selective processing of nonclinical populations.

Indeed,
Stroop interference with concern relevant stimuli has been found in nonclinical
populations (e.g., Giles & Cairns, 1989; Riemann & McNally, 1995). In contrast,
there are mixed findings regarding the effects of mood congruence on selective
attention to valenced stimuli in nonclinical and subclinical populations.

Whereas
some researchers observed mood-congruent effects on Stroop interference following
an experimental mood induction (e.g., Mogg, Mathews, Bird, & Macgregor-Morris,
1990, Experiment 1; Richards et al., 1992, Experiment 2), others have not (e.g.,
Gotlib & McCann, 1984, Experiment 2; Riemann & McNally, 1995).

Most of these
studies did not attempt to separate the effects of mood congruence and concern
relevance. Finally, the effect of emotionality has received little attention in the
nonclinical literature.

The goal of our study is to investigate the effects of emotionality, mood congru-
ence, and concern relevance in a nonclinical population within a unified framework.
Specifically, consistent with the findings of Richards et al. (1993), we postulated that
individuals exhibit greater Stroop interference with mood-congruent stimuli in both
negative and positive mood states than with either neutral or mood-incongruent
words.

Second, we hypothesized that words highly related to individuals’ concerns
would elicit more interference than would words with low concern relevance.

No
specific predictions were made with respect to the effects of word emotionality. Differ-
entiating among the contributions of these three factors is essential to the understand-
ing of the general mechanisms underlying the emotion Stroop effect. Furthermore,
such an investigation is also crucial to the understanding of the similarities and differ-
ences in selective attention between clinical and nonclinical individuals, and thus may
potentially elucidate cognitive processes that are specific to emotional disorders.

Gilboa-Schechtman, Revelle, and Gotlib

METHOD

Participants

Eighty-four (84) college students participated in the experiment as part of a
course requirement. Participants were run in groups of 1 to 4. The experimental

Stroop Following Mood Induction

session lasted about 50 min. Data from three participants were lost due to equip-
ment failure.

Materials

There were six types of stimuli: experimenter-provided neutral words, experi-
menter-provided positive words, experimenter-provided negative words, partici-
pant-generated words involving neutral experiences, participant-generated words
involving positive experiences, and participant-generated words involving negative
experiences. Experimenter-provided neutral words were global, metric, tangible,
verbal, genetic, and temporal. Experimenter-provided negative words were hurt,
upset, lonely, depressed, helpless, and miserable. Experimenter-provided positive
words were excited, friendly, cheerful, joyful, tender, and carefree. The positive and
negative emotion words were equated for frequency, length, and emotionality based
on John’s (1988) ratings. Participant-generated words included personal ‘‘codes’’
related to a particular neutral, positive, or negative experience. For example, a
participant might provide the words message, physics, professor, and Tuesday as
his or her ‘‘codes’’ for a neutral experience.

Procedure

Participants were told that the experiment was designed to test the effects of
mood on simple cognitive processing. The stimuli and the instructions were pre-
sented on an Apple Macintosh IIci computer with a color monitor. The controlling
software handled the mood-induction procedure (MIP) and stimuli presentation.
It also recorded participants’ mood ratings and their reaction times (RTs) to
Stroop stimuli.

Introductory Phase

Participants were asked to recall a neutral, positive, and negative experiences
and provide four ‘‘key words’’ for these experiences. The order of recall of positive
and negative experiences was counterbalanced across participants. Next, partici-
pants practiced the Stroop task by identifying the color of a single stimulus (the
word practice) by pressing an appropriately labeled key. After reaching a criterion
performance level on this task,4 participants proceeded to the first mood-induction
phase: negative mood induction in the NP condition and positive mood induction
in the PN condition.

4

Criterion performance was defined as a sequence of 10 consecutive error-free trials, with the difference
between the mean time to respond to the preceding 10 trials and the mean of 10 final trials being no
more than 10%. The minimum number of trials (regardless of performance) was 40, and the maximum
number of trials was 100.
5
Other data-reduction strategies (such as eliminating response latencies 2 SD above each participant’s
mean) produced essentially identical results.
6
Analyses using the overall mood ratings are presented for the sake of simplicity. For both experiments,
analyses using individual mood scales produced essentially identical results.

494

Gilboa-Schechtman, Revelle, and Gotlib

Experimental Phase

The experimental phases consisted of three steps: mood induction, mood rating,
and the Stroop task. During the mood-induction procedure, participants were first
instructed to reexperience sad/happy events using ‘‘relive’’ instructions adapted
from Salovey (1992). Next, participants concentrated on their feelings while listening
to music. An excerpt from Beethoven’s string quartet op. 131 was played for the
negative MIP. An excerpt from Vivaldi’s ‘‘Spring’’ concerto of the ‘‘Four Seasons,’’
op. 12, was played for the positive MIP. The concentration step lasted about 1 min.
Finally, before proceeding to the Stroop task, participants rated their current mood
on six unipolar (sad, frustrated, anxious, happy, content, and optimistic) and one
bipolar (overall) visual analog scales. The unipolar scales were anchored with not
at all (coded as intensity rating of 0) and extremely (coded as intensity rating of
100). The overall scale was anchored with very positive (coded as intensity rating
50) and very negative (coded as intensity rating of 50).
During the Stroop task, participants were presented with 150 stimuli. The
intertrial interval was 2 sec.

The stimuli for this task were six words from each of
the experimenter-provided categories and four words from each of the participant-
generated categories. Participants saw five presentations of each word from this
30-word stimulus list, resulting in a total of 150 presentations. The order of presenta-
tion and the color of stimuli were randomly determined for each participant subject
to the constraint that no two consecutive words were displayed in the same color.

Following the completion of the Stroop task, participants rated their moods.
Between the two experimental phases, participants completed a 5-min long
filler task evaluating the computer program they were using.

This filler task was
intended to ensure that the effects of the first MIP would dissipate before the
beginning of the second MIP. After the completion of the filler task, participants
proceeded to the next experimental phase. Apart from the valence of the mood
induced (positive in the NP condition, negative in the PN condition), the structure
of the second experimental phase was identical to that of the first. Specifically, after
completing an MIP, participants rated their mood and performed the Stroop task.

Neutralizing Phase

In order to stabilize participants’ mood before the end of the experiment,
participants were asked to relive the previously recalled neutral experience. Their
mood ratings after the completion of the neutral MIP were taken to represent their
baseline mood. Finally, the participants were debriefed and thanked for their partici-
pation.

RESULTS

Data Analysis

Trials involving incorrect key pressing or extreme scores [i.e., response time
(RT) longer than 2500 ms or shorter than 333 ms] were eliminated from further

Stroop Following Mood Induction

analyses. Excluded responses occurred on 5% of the trials, and their number did
not differ as a function of word type.

Efficacy of MIPs

To determine the efficacy of the positive and negative MIPs, participants’
overall mood rating were analyzed using two separate two-way ANOVAs. We
computed mood change scores as the differences between participants’ overall
mood ratings after an induction and their mood ratings at baseline (i.e., after the
neutral MIP). Mean negative and positive change scores were 12.3 (SD 18.1)
and 22.5 (SD 19.4), respectively. Absolute change scores were analyzed using a
three-way ANOVA, with order of recall (NP vs. PN) and order of MIP (NP vs.
PN) as between-subjects factors and valence of MIP as a within-subject factor. No
effects or interactions involving either order of recall or order of MIP were found.
Test of simple effects indicated that both negative and positive change scores
38.4, p
.001, and F(1,79)
102.7,
differed significantly from zero (F(1,79)
p .001, respectively). The positive MIP produced greater changes in self-ratings
than did the negative MIP, F(1,79) 8.9, p .05. To ensure that the PN and the
NP groups did not differ in the baseline mood scores, we examined overall baseline
mood scores. Mean baseline scores of the PN and NP groups were 5.7 (SD
14.7) and 4.6 (SD 11.3), respectively. A one-way ANOVA on baseline overall
mood ratings was conducted. No differences between the groups’ baseline moods
was revealed, F(1,79) 1.

Mood Congruence, Concern Relevance, and Emotionality Effects

Table I presents the means and standard deviations of RTs to different stimulus
types in the positive and negative experimental conditions.
Preliminary analyses of reaction times were conducted with order of emotional
experience recall (PN vs. NP) as a between-subjects variable and word frequencies
and word length as covariates. No main effects or interactions involving any of
these variables were detected in these analyses, nor did they qualitatively modify
any other main effect or interaction. Therefore, these variables were omitted from
the final analyses.
A potential problem in using participant-generated vs. experimenter-provided

Table I. Means and Standard Deviations of Response Latencies for Stimulus Identification

Condition

Negative

Positive

Note: N

80. Standard deviations (in milliseconds) are shown in parentheses.

Experimenter

Neutral

Negative

772
(127)
747
(105)

Positive

806
(142)
748
(108)

Stimulus origin

Neutral

786
(135)
763
(110)

796
(130)
759
(119)

Participant

Negative

812
(150)
756
(122)

496

materials is that these two sets of stimuli might differ on several lexical dimensions,
such as word length, word frequency, and so on. To assess whether these differences
in lexical properties influenced RT, we conducted several regression analyses. For
each valence category (i.e., positive, negative, and neutral), we regressed the mean
RT to participant-generated words (predicted variable) on three predictors: the RT
to experimenter-generated words, the mean frequency ratings of this participant-gen-
erated category (computed using Francis & Kucera, 1982, norms), and the mean word
length measures of this participant-generated category. For example, the RT to neu-
tral participant-generated words was regressed on the RT to neutral experimenter-
provided words, mean frequency ratings of the neutral key words provided by this
participant, and mean word length of these neutral key words. None of the partial
correlations between RT to participant-generated words and word frequency ratings
or word length measures approached significance (all ps 0.3). These analyses sug-
gest that differences in lexical properties between stimulus sets did not exert signifi-
cant influences on reaction time. It is important to note that this conclusion is in line
with the results of all past investigations regarding the effects of lexical properties on
emotion Stroop RTs: neither word length nor word frequency accounted for Stroop
interference effect with personally relevant words (e.g., Riemann, Amir, & Luoro,
1994; Riemann & McNally, 1995; Williams et al., 1996).
To examine concern relevance, emotionality, and mood-congruence hypothe-
ses, mean RTs were analyzed using a four-way ANOVA, with order of MIP (PN
vs. NP) as a between-subjects variable, and mood-induction valence (negative vs.
positive), word origin (participant vs. experimenter), and word valence (neutral,
mood congruent, or mood incongruent) as within-subject variables. There was no
main effect of order, F(1,79)
1.18, n.s. As expected, there was a time effect as
indicated in this analysis by a significant MIP Order interaction, F(1,79) 20.8,
p .001. This effect indicated that participants’ RTs were longer in the first mood-
induction condition than they were in the second mood-induction condition (mean
RTs for the first and second MIPs were 800.1 and 751.2, respectively). No other
significant interactions involving Order emerged from these analyses (all ps 0.1).
A significant effect of MIP was revealed, such that participants’ RTs in the
negative condition were significantly longer than were their RTs in the positive
condition (mean RTs in the negative condition
793.9 and positive condi-
tion
759.2, F(1,79)
9.6, p
.01).
A main effect of Origin indicated that participant-generated words elicited
significantly more interference than did words provided by the experimenter,
F(1,79)
19.9, p
.001. There was a significant effect of word Valence,
F(2,158)
3.14, p
.05. Two planned comparisons were conducted to clarify
this effect. The first indicated that mood-incongruent words did not elicit more
interference than did neutral words, F 1. The second planned comparison indi-
cated that mood-congruent words elicited more interference than did neutral words,
F(1,79)
11.0, p
.001. There was a significant MIP
Valence interaction,
F(2,158)
8.4, p
.001. A planned comparison indicated that, in the negative
MIP, participants’ RTs to negative words were longer than they were to positive
words, whereas this pattern was reversed in the positive MIP, F(1,79)
12.1,
p .01. There were no other significant interactions.

Gilboa-Schechtman, Revelle, and Gotlib

Stroop Following Mood Induction

To examine whether the main effects of Valence and Origin were significant
in the negative and positive mood-induction conditions, we conducted separate
three-way ANOVAs for each of these conditions. In each ANOVA, order of MIP
(PN vs. NP) was a between-subjects variable, and word origin (participant vs.
experimenter) and word valence (neutral, mood congruent, or mood incongruent)
were within-subject variables.
In the analysis of RTs in the negative condition no main effect of Order was
identified, F
1. There were no other interactions involving Order (all ps
.2).
There was a significant effect of word Valence, F(2,158)
7.7, p
.05. Planned
comparisons indicated that negative words elicited more interference than did
neutral or positive words among experimenter-provided words (t(79)
3.9; p
.01; t(79)
2.8, p
.05, respectively). Identical pattern was observed among
participant-generated words (t(79) 2.1, p .05; p(80) 1.95; p .05, respectively).
A main effect of Origin F(1,79)
15.6, p
.05 as found. RTs to participant
generated words were longer than to experimenter provided words. This effect was
modified by a significant Origin
Valence interaction, F(2,158)
4.7, p
.05.
Post-hoc comparisons indicated that the effect of word origin was greater for the
neutral words than for negative or positive words (F(1,79)
4.75, p
.05; and
F(1,79) 9.4, respectively).
In the analysis of RTs in the positive condition a main effect of Order was
identified, F(1,79) 9.8, p .001. RTs in the NP condition were faster than RTs
in the PN condition. There were no other interactions involving Order (all ps
.2). A main effect of Origin F(1,79) 8.5, p .05 was found. RTs to participant-
generated words were longer than to experimenter-provided words. There was a
significant effect of word Valence, F(2,158)
3.5, p
.05. Planned comparisons
indicated that positive words elicited more interference than neutral or negative
words among experimenter-provided words (t(79) 2.8; p .05; t(79) 2.0, p
.05, respectively). No such differences were observed among participant-generated
words (all ps .2). There was no significant Origin Valence interaction, F 1.

DISCUSSION

We investigated the effects of mood congruence, concern relevance, and emo-
tionality on Stroop interference following experimentally induced positive and nega-
tive moods in a nonclinical population. Our results support the mood-congruence
and the concern-relevance hypotheses, but not the emotionality hypothesis. First,
consistent with the mood-congruence hypothesis, we found that interference effects
in the negative condition were specific to negative-emotion words, and that interfer-
ence effects in the positive condition were specific to the positive-emotion words.
Our results are consistent with those reported by Richards et al. (1992) in demonstra-
ting that both positive and negative affective states increase the selective processing
of mood-congruent materials. Second, words generated by the participants elicited
more interference than did experimenter-provided materials, supporting the con-
cern-relevance hypothesis. It is important to note that this effect was not due to
lexical differences between the two stimulus sets. We concur, therefore, with

498

Mathews and Klug (1993), who state that judgments of personal emotional relevance
are stronger predictors of interference than are judgments of emotionality per se.
Finally, inconsistent with the emotionality hypothesis, mood-incongruent words,
although as ‘‘emotional’’ as mood-congruent ones, did not elicit interference com-
pared to neutral words. In contrast, mood-congruent words elicited significantly
more interference than did neutral words. It is possible that previous research that
identified emotionality effects (e.g., Martin, Williams, & Clark, 1991) confounded
the effects of emotionality with the effects of concern relevance.
Nonclinical individuals’ pattern of selective attention seems to parallel that of
high-trait anxious, clinically anxious, and clinically depressed individuals (Williams
et al., 1996). Combined, these data suggest that, for all populations, concern rele-
vance and mood congruence contribute independently to Stroop interference,
whereas emotionality does not. If so, what distinguishes clinical from nonclinical
processing of valenced stimuli? One possibility is that the difference is merely
quantitative. Indeed, the effect sizes in studies with clinical participants are greater
than the effect sizes in our study (see Williams et al., 1996). Alternatively, it is
possible that the differences between clinical and nonclinical individuals manifest
themselves in the strategies they use to cope with their cognitive interference
rather than in the interference itself. Individuals with clinically significant emotional
problems may not be able to inhibit such interference, whereas nonclinical individu-
als may be able to override it (Gotlib, Roberts, & Gilboa, 1996).
Our findings suggest several explanations for the inconsistent pattern of results
regarding the mood-congruence hypothesis using the emotional Stroop task. First,
the specifics of the mood-induction procedures might affect the strength of Stroop
interference. Bower (1987) suggests that the induction of ‘‘weak, temporary and
nonspecific moods of happiness and sadness in the laboratory have not primed
perceptual processing of the general class of positive and negative words, respec-
tively’’ (p. 448), and argues further that to obtain a mood-congruence effect, the
priming may have to point to a more specific set of words and emotional themes.
Consistent with Bower’s hypothesis, ‘‘focused’’ affective states (e.g., those induced
by autobiographic MIPs) may be more likely to be associated with Stroop interfer-
ence than ‘‘diffused’’ inductions (e.g., Velten MIP). Indeed, Niedenthal, Haberstadt,
and Setterlund (1997) found that focused emotional states affect lexical decisions
and word-naming latencies of emotion-congruent words.
Mood congruence is a well-recognized phenomenon with respect to memory and
judgment processes. Researchers have consistently found enhanced memory for
mood-congruent materials in nonclinical individuals (Blaney, 1986; Matt, Vazquez, &
Campbell, 1992). Similarly, research on judgment processes suggests that experimen-
tally induced moods affect participants’ judgments of a variety of real and hypothetical
events (for a review, see Gotlib, Gilboa, & Sommerfield, in press). However, mood
congruence is less well established with respect to attention. Thus far, the association
between attention and mood has been limited to negative affective states, especially
anxiety (e.g., Mathews & MacLeod, 1994). Moreover, studies attempting to obtain
mood-congruence effects in attentional tasks with a wider range of affective states
(typically sadness and happiness) yield an inconsistent pattern of findings (e.g., Chal-
lis & Krane, 1988; Clark, Teasdale, Broadbent, & Martin, 1983).

Gilboa-Schechtman, Revelle, and Gotlib

Stroop Following Mood Induction

Our findings using the emotion Stroop task suggest that mood congruence may
extend to attentional processes as well. However, it is frequently argued that this
task is not a pure measure of attention (e.g., MacLeod & Mathews, 1988; Mathews &
MacLeod, 1985). In particular, a Stroop task in which the stimuli are presented in
a ‘‘blocked’’ format (i.e., the stimuli are grouped by word type, such that, for
example, all negative words are presented in succession) may not provide a ‘‘pure’’
measure of attention because the post-attentional rumination about a negatively
valenced word might delay the color naming of the next word (e.g., McNally,
Riemann, & Kim, 1990). It is important to note that our Stroop task involved a
random rather than a blocked presentation of the Stroop stimuli, making the Stroop
interference index less likely to be contaminated by post-attentional processing.
Thus, our results, combined with those of Richards et al. (1992) and Niedenthal et
al. (1997), suggest that mood affects attentional processes in negative (e.g., anxious
and sad) as well as positive affective states (see also Williams et al., 1996). Therefore,
consistent with Beck’s (1976) and Bower’s (1981) theories of the relation between
affect and cognition, the effects of mood on cognitive processes extend beyond
memory and judgment to attentional processes.

LIMITATIONS AND FUTURE DIRECTIONS

The present research examined parameters affecting selective processing of
valenced information, using laboratory-induced positive and negative affective
states. Obviously, experimental mood induction enlists participants’ cooperation in
modifying their affective states and therefore creates demand characteristics. Might
the participants be consciously attending to critical words in order to please the
experimenter, rather than responding ‘‘naturally’’? We think this is unlikely for
two reasons. First, both musical MIPs and autobiographic MIPs were found to be
effective in influencing objective (e.g., reaction time) as well as subjective (e.g.,
mood rating) measures (e.g., Niedenthal et al., 1997; Salovey, 1992). In addition,
reviews of the mood-induction literature suggest that MIPs affect a variety of
cognitive, physiological, and psychomotor measures (e.g., Larsen & Sinnett, 1991;
Gerrards-Hesse, Spies, & Hesse, 1994). For example, Neidenthal et al. (1997, Experi-
ment 3) found that emotional state influenced attentional processes using lexical
decision and word-naming tasks. Second, although demand characteristics for clini-
cal subjects with their disorder-relevant words should be no less obvious than the
demand characteristics after MIPs, studies using subliminal Stroop found that the
pattern of interference with subliminally presented words resembled the pattern
of interference with superliminal stimuli (MacLeod & Rutherford, 1992; Williams
et al., 1996). Therefore, it seems unlikely that emotional Stroop interference in
our study depends on participants’ conscious decisions to attend more to mood-
congruent stimuli. However, it will be important to replicate these findings using
techniques in which the demand characteristics are less salient (e.g., film clips,
as in Gross & Levenson, 1995; or naturally occurring affective states) and other
cognitive measures.
In our discussion, we emphasize the effects of mood on the selective processing

500

of valenced information. Yet it is also possible that the observed selective processing
can be partially accounted for by the effects of cognitive priming independent of
mood. According to the cognitive-priming hypothesis, mood congruence is associ-
ated with the cognitive activation of affective concepts, rather than with the effect
of mood per se (e.g., Rholes, Riskind, & Lane, 1987). Although in our experiment
cognitive priming does contribute to selective processing of emotion-related materi-
als, affect-free cognition is unlikely to be the sole cause of this selectivity. First,
we found that participants selectively processed mood-congruent experimenter-
provided words, which were not part of the participants’ recollected events. Experi-
menter-provided words, not being part of the autobiographic memories retrieved
by the participants, are related to these memories primarily through affective associ-
ations. Second, the effects of cognitive priming typically dissipate within seconds
(e.g., Fischler & Goodman, 1978), whereas the selective-processing effects observed
in our studies persisted for minutes. These arguments notwithstanding, our results
are unlikely to resolve the debate about cognitive vs. affective priming, partly
because emotional effects and cognitive priming of emotion-relevant materials are
not easily separable. To fully disentangle the effects of cognitive and emotional
priming, future research might attempt to examine the effects of recollection per
se (without the emotional reexperience) on selective processing.
In interpreting our findings, we assumed that interference with participant-
generated stimuli is due to the relevance of these stimuli to participants’ personal
concerns. However, the process of generating these materials, rather than their
concern-relevant content, may have made those stimuli more salient to the partici-
pants. Indeed, words that are generated by participants tend to be better remem-
bered than words that are provided by the experimenter (e.g., Gardiner & Hampton,
1985). Although it is possible that the process of stimulus generation facilitates
selective attention, we believe it is unlikely that our effect is due solely to this
process rather than to the content of these stimuli. First, we have found differentially
selective processing of self-generated stimuli in positive and negative mood-
induction conditions. If only the process of generating these materials, rather than
their content, affected selective attention, such differences would not be found.
Second, the repeated activation of memory traces, which results in generation effects
in memory tasks, need not necessarily result in corresponding effects in attentional
processing. Future research, employing other procedures to elicit idiographic materi-
als, might examine whether our results regarding Stroop interference depend on
the self-generation effect. Enhanced salience of idiographic materials is not unique
to the self-generation procedure. For example, rating of materials for the degree
of ‘‘disturbance’’ before performing the Stroop task is also likely to increase the
salience of those materials (e.g., McNally et al., 1994). However, replicating our
results with other elicitation procedures (e.g., rating) will reinforce the present
findings.
Finally, future research should explore mood congruence in clinical and subclin-
ical populations. So far, research has concentrated on examining clinical individuals
in their natural (i.e., anxious or dysphoric) mood. This research established that
naturally occurring negative states are associated with selective processing of nega-
tive materials. However, a comprehensive test of the mood-congruence hypothesis

Gilboa-Schechtman, Revelle, and Gotlib

Stroop Following Mood Induction

also requires a demonstration that selective attention to positive emotional material
occurs in positive affective states. Unfortunately, positive mood inductions with
clinical populations have not been conducted. Such studies would enable us to
further explore the similarities and differences between clinical and nonclinical
populations and, ultimately, to understand the nature of emotional breakdown.

ACKNOWLEDGMENTS

This research has been supported in part by grant IRI-8812699 awarded to
Andrew Ortony by the National Science Foundation, in part by Anderson Con-
sulting through Northwestern University’s Institute for the Learning Sciences, and
in part by contracts MDA903-90-C-0108 and MDA903-93-K-0008 from the U.S.
Army Research Institute awarded to William Revelle and Kristen Anderson.
We are indebted to Andrew Ortony, Susan Mineka, and Colin MacLeod for
many thoughtful comments and suggestions that greatly influenced this work.
Correspondence concerning this article should be addressed to Eva Gilboa-
Schechtman, Department of Psychology, Bar-Ilan University, Ramat Gan, 52900,
Israel. E-mail: gilboae@mail.biu.ac.il.

REFERENCES

Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York: International Universi-
ties Press.
Blaney, P. H. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229–246.
Bower, G. H. (1981). Mood and memory. American Psychologist, 36(2), 129–148.
Bower, G. H. (1987). Commentary on mood and memory. Behaviour Research and Therapy, 25(6),
443–455.
Challis, B. H., & Krane, R. V. (1988). Mood induction and the priming of semantic memory in lexical
decision task: Asymmetric effects of elation and depression. Bulletin of the Psychonomic Society,
26(4), 309–312.
Clark, D. M., Teasdale, J. D., Broadbent, D. E., & Martin, M. (1983). Effect of mood on lexical decisions.
Bulletin of the Psychonomic Society, 21(3), 175–178.
Fischler, I., & Goodman, G. O. (1978). Latency of associative activation in memory. Journal of Experimen-
tal Psychology: Human Perception and Performance, 4, 455–470.
Francis, W. N., & Kucera, H. (1982). Frequency analysis of English usage. Lexicon and grammar. Boston,
MA: Houghon-Mifflin.
Gardiner, J. M., & Hampton, J. A. (1985). Semantic memory and the generation effect: Some tests of
the lexical activation hypothesis. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 11, 732–741.
Gerrards-Hesse, A., Speies, K., & Hesse, F. W. (1994). Experimental inductions of emotional states
and their effectiveness: A review. British Journal of Psychology, 85, 55–76.
Giles, M., & Cairns, E. (1989). Colour naming of violence-related words in Northern Ireland. British
Journal of Clinical Psychology, 28, 87–88.
Gotlib, I., Roberts, J., & Gilboa, E. (1996). Cognitive interference in depression. In I. G. Sarason, B.
R. Sarason, & G. R. Pierce (Eds.), Cognitive interference: Theories, methods, findings. (pp. 347–378).
Hillsdale, NJ: Lawrence Erlbaum.
Gotlib, I. H., Gilboa, E., & Sommerfield, B. L. (in press). Cognitive functioning in depression: Nature
and origins. In R. J. Davidson (Ed.), Wisconsin Symposium on Emotion, Vol. 1. New York: Oxford
University Press.
Gotlib, I. H., & McCann, C. D. (1984). Construct accessibility and depression: An examination of
cognitive and affective factors. Journal of Personality and Social Psychology, 47(2), 427–439.

502

Gross, J. J., & Levenson, R. W. (1995). Emotion elicitation using films. Cognition and Emotion, 9(1),
87–108.
John, C. H. (1988). Emotionality ratings and free association norms of 240 emotional and non-emotional
words. Cognition and Emotion, 2, 49–70.
Larsen, R. J., & Sinnett, L. M. (1991). Meta-analysis of experimental manipulations: Some factors
affecting the Velten mood induction procedure. Review of Personality and Social Psychology,
17(3), 323–334.
Logan, A., & Goetsch, V. (1993). Attention to external cues in anxiety states. Clinical Psychology
Review, 13, 541–559.
MacLeod, C., & Mathews, A. (1988). Anxiety and the allocation of attention to threat. Quarterly Journal
of Experimental Psychology: Human Experimental Psychology, 38, 659–670.
MacLeod, C., & Rutherford, E. M. (1992). Anxiety and the selective processing of emotional information:
Mediating roles of awerness, trait and state variables, and personal relevance of stimulus materials.
Behaviour Research and Therapy, 30, 479–491.
Martin, M., Williams, R., & Clark, D. M. (1991). Does anxiety lead to selective processing of threat-
related information? Behavior Research and Therapy, 29, 147–160.
Mathews, A., & Klug, F. (1993). Emotionality and interference with color naming in anxiety. Behavioral
Research and Therapy, 31, 57–62.
Mathews, A., & MacLeod, C. (1985). Selective processing of threat cues in anxiety states. Behaviour
Research and Therapy, 23(5), 563–569.
Mathews, A., & MacLeod, C. (1994). Cognitive approaches to emotions and emotional disorders. Annual
Review of Psychology, 45, 25–50.
Matt, G. E., Vazquez, C., &, Campbell, W. K. (1992). Mood congruent recall of affectively toned stimuli:
A meta-analytic review. Clinical Psychology Review, 12, 227–255.
McNally, R. J., Amir, N., Louro, C. E., Lukach, B. M., Riemann, B. C., & Calamari, J. E. (1994).
Cognitive processing of idiographic emotional information in panic disorder. Behavioral Research
and Therapy, 32, 119–122.
McNally, R. J., Riemann, B. C., & Kim, E. (1990). Selective processing of threat cues in panic disorder.
Behaviour Research and Therapy, 28(5), 407–412.
Mogg, K., Mathews, A., Bird, C., & Macgregor-Morris, R. (1990). Effects of stress and anxiety on the
processing of threat stimuli. Journal of Personality and Social Psychology, 59, 1230–1237.
Niedenthal, P., Halbestadt, J. B., & Setterlund, M. (1997). Being happy and seeing ‘‘happy’’: Emotional
state mediates visual word recognition. Cognition and Emotion, 11, 383–403.
Rholes, W. S., Riskind, J. H., & Lane, J. W. (1987). Emotional states and memory biases: Effects of
cognitive priming on mood. Journal of Personality and Social Psychology, 52(1), 91–99.
Richards, A., French, C. C., Johnson, W., Naparstek, J., & Williams, J. (1992). Effects of mood manipula-
tion and anxiety on performance of an emotional Stroop task. British Journal of Psychology,
83, 479–491.
Riemann, B. C., Amir, N., & Louro, C. E. (1994). Cognitive processing of personally relevant information
in panic disorder. Behavioral Therapy and Research, 32(1), 119–122.
Riemann, B., & McNally, R. J. (1995). Cognitive processing of personally relevant information. Cognition
and Emotion, 9, 325–340.
Salovey, P. (1992). Mood-induced self-focused attention. Journal of Personality and Social Psychology,
62(4), 699–707.
Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology.
Psychological Bulletin, 120, 3–24.

Gilboa-Schechtman, Revelle, and Gotlib