18. COGNITION AND EMOTION 1 18 - psypress.co.ukpsypress.co.uk/ek5/resources/pdf/chap18.pdf · 18....

18. COGNITION AND EMOTION 1

1

18

Cognition and Emotion

INTRODUCTION

Much of cognitive psychology is still influencedto some extent by the computer analogy or meta-phor, although much less so than used to be thecase. For example, this can be seen in the emphasison information-processing models. One of thelimitations of this approach is that it does not lenditself readily to an examination of the relation-ship between cognition and emotion (especiallythe effects of emotion on cognition). The reasonis that it is difficult to think of computers as hav-ing emotional states.

Most cognitive psychologists ignore the issueof the effects of emotion on cognition by tryingto ensure that all their participants are in a relat-ively neutral emotional state. However, there hasbeen a rapid increase in the number of cognitivepsychologists carrying out research in the areaof cognition and emotion. Some of that research(e.g., the role of emotional states in eyewitnesstestimony; flashbulb memories; autobiographicalmemory) is discussed in Chapter 8.

The most common approach adopted by cog-nitive psychologists wishing to study the effects ofemotion on cognition has involved manipulating

participants’ emotional states in a systematic way.In contrast, some researchers (e.g., Lazarus, 1991;Smith & Lazarus, 1993) have studied the effectsof cognition on emotion. As there are almostconstant interactions between cognition and emo-tion in everyday life, any attempt to provide anadequate theory of cognition that ignores emotionis likely to prove inadequate.

Before proceeding, it is worth considering somedefinitions. The term “affect” is very broad, andhas been used to cover a wide variety of experi-ences such as emotions, moods, and preferences.In contrast, the term “emotion” tends to be usedto refer to fairly brief but intense experiences,although it is also used in a broader sense.Finally, “mood” or “state” are terms describinglow-intensity but more prolonged experiences.

COGNITIVE DETERMINANTS OF EMOTION

What factors determine our emotional experience?There is reasonable agreement concerning therelevant factors, but much controversy about therelative importance of these factors. Accordingto Parkinson (e.g., 1994), emotional experiencedepends on four separate factors:

CP_18_C01 9/28/06, 6:55 PM1

2 COGNITIVE PSYCHOLOGY: A STUDENT ’S HANDBOOK

(1) Appraisal of some external stimulus or situ-ation. This is generally accepted as the mostimportant factor, and was the one emphasisedby Lazarus (e.g., 1991).

(2) Reactions of the body (e.g., arousal). Accord-ing to the James–Lange theory, our emotionalexperience depends on our perception ofour own bodily symptoms. As James (1890,p. 451) argued, “If we fancy some strongemotion, and then try to abstract from ourconsciousness of it all the feelings of itsbodily symptoms, we find we have nothingleft behind.” The finding that patients withspinal cord injuries (who have little or no dir-ect experience of their bodily symptoms) haveno reported reduction in emotional experience(e.g., Cobos, Sanchez, Perez, & Vila, 2004) isinconsistent with this theoretical position.

(3) Facial expression. The importance of thisfactor was shown in a study by Strack, Martin,and Stepper (1988). Participants were moreamused by cartoons when adopting a facialexpression close to a smile than when pro-ducing an expression resembling a frown.

(4) Action tendencies. For example, preparingto advance in a threatening way is associatedwith anger, whereas preparing to retreat isassociated with fear (Frijda, Kuipers, & terSchure, 1989).

It is very important to bear in mind that thesefour factors are not independent of each other(see Figure 18.1). More specifically, cognitive

appraisal of the situation affects bodily reactions,facial expression, and action tendencies, as wellas having a direct effect on emotional experience.These considerations inspired the bold notion thatcognitive factors (especially appraisals) are alwaysof fundamental importance in determining emo-tional experience. This was precisely what wasargued by Lazarus (1982, 1991) in his appraisaltheory, which led to the development of severalother appraisal theories (see Barrett, in press, fora review). The essence of this theoretical approachis as follows: “Whether emotion is generated inresponse to perceived, remembered, or imaginedevents, and by automatic or controlled process-ing, appraisal theories claim that appraisals startthe emotion process, initiating the physiological,expressive, behavioural and other changes thatcomprise the resultant emotional state” (Roseman& Smith, 2001, p. 7).

Appraisal theory

The original version of appraisal theory was putforward by Lazarus (1982). According to thistheory, cognitive appraisal can be subdivided intothree more specific forms of appraisal:

• Primary appraisal: an environmental situationis regarded as being positive, stressful, orirrelevant to well-being.

• Secondary appraisal: account is taken of theresources that the individual has available tocope with the situation.

Appraisalof externalstimulus orsituation

Facialexpression

EMOTIONALEXPERIENCE

Bodilyreactions

Actiontendencies

FIGURE 18.1

Factors influencing emotionalexperience. Based on

Parkinson (1994).

CP_18_C01 9/28/06, 6:55 PM2


• Re-appraisal: the stimulus situation and thecoping strategies are monitored, with the prim-ary and secondary appraisals being modifiedif necessary.

The descriptions of these forms of appraisalseem to imply that they involve deliberate consci-ous processing. However, that is not necessarilythe case. For example, Lazarus (1991, p. 169)referred to “two kinds of appraisal processes—one that operates automatically without aware-ness or volitional control, and another that isconscious, deliberate, and volitional”.

There have been two major developments inappraisal theory since the theoretical formulationof Lazarus (1982). First, it is increasingly claimedthat each distinct emotion is elicited by a specificand distinctive pattern of appraisal. For example,Smith and Lazarus (1993) argued that there aresix appraisal components, two involving primaryappraisal and four involving secondary appraisal:

• Primary: motivational relevance (related topersonal commitments?).

• Primary: motivational congruence (consistentwith the individual’s goals?).

• Secondary: accountability (who deserves thecredit or blame?).

• Secondary: problem-focused coping potential(can the situation be resolved?).

• Secondary: emotion-focused coping potential(can the situation be handled psychologically?).

• Secondary: future expectancy (how likely is itthat the situation will change?).

According to Smith and Lazarus (1993), differ-ent emotional states can be distinguished on thebasis of which appraisal components are involvedand how they are involved. For example, anger,guilt, anxiety, and sadness all possess the primaryappraisal components of motivational relevanceand motivational incongruence (these emotionsonly occur when goals are blocked). However,they differ in terms of secondary appraisal com-ponents. Guilt involves self-accountability, anxietyinvolves low or uncertain emotion-focused cop-ing potential, and sadness involves low futureexpectancy for change.

Second, most early appraisal theories (includ-ing that of Smith & Lazarus, 1993) focused on thestructure of appraisal rather than the processesinvolved. In other words, they emphasised thecontents of any given appraisal but largely ignoredthe underlying mechanisms involved in produ-cing appraisals. This issue has been addressed inseveral recent theories (see Barrett, in press). Ofparticular importance is the theoretical approach ofSmith and Kirby (2001). According to their theory,various appraisal processes occur in parallel. Thereare three basic mechanisms (see Figure 18.2). First,there is associative processing, which involvespriming and activation of memories. This form ofprocessing occurs rapidly and automatically, andlacks flexibility. Second, there is reasoning. Thisprocess involves deliberate thinking, and is slowerand more flexible than associative processing.Third, there are appraisal detectors that continu-ously monitor appraisal information coming fromthe associative and reasoning processes. The emo-tional state an individual experiences at any givenmoment is determined by the total informationregistered by the appraisal detectors.

Evidence

There are numerous studies showing that emo-tional experience is influenced by cognitive ap-praisal. For example, consider an early experimentreported by Speisman, Lazarus, Mordkoff, andDavison (1964). Participants saw various anxiety-evoking films. One film showed a Stone Ageritual in which adolescent boys had their penisesdeeply cut, and another film showed various work-shop accidents. The most horrific of these accid-ents involved a board caught in a circular sawthat rammed with tremendous force through themid-section of a worker, who died writhing onthe floor.

In the study by Speisman et al. (1964), cogni-tive appraisal was manipulated by varying theaccompanying soundtrack. For example, denialwas produced by indicating that the incision filmdid not show a painful operation, or that thoseinvolved in the workshop film were actors. Intel-lectualisation was produced in the incision filmby considering matters from the perspective of

CP_18_C01 9/28/06, 6:55 PM3


Affectivepriming

Contents offocal awareness Perceived

stimuli

Associativelyactivated

representations

Appraisaldetectors

Reasoning

Appraisalintegration

Emotional response• Appraisal outcome• Physiological activity• Action tendencies

Subjectiveaffect

FIGURE 18.2

an anthropologist viewing strange native customs.It was produced in the workshop film by tellingthe viewer to consider the situation in an objectiveway. Various psychophysiological measures ofarousal or stress (e.g., heart rate; galvanic skinresponse) were taken continuously during theviewing of each film.

Denial and intellectualisation both producedsubstantial reductions in stress as indexed bythe psychophysiological measures compared to acontrol condition in which there was no sound-track. Thus, manipulating an individual’s cognitiveappraisal can have a significant impact on physio-logical stress reactions.

As discussed earlier, Smith and Lazarus (1993)argued that which emotion is experienced by anindividual in a given situation depends on the spe-cific appraisal components activated by the situ-ation. They presented scenarios to their participantsand instructed them to identify with the centralcharacter. In one scenario, the central characterhad performed poorly in an important course and

he appraised the situation. Other-accountabilitywas produced by having him put the blame on theunhelpful teaching assistants. Self-accountabilitywas produced by having him argue that he madea lot of mistakes (e.g., doing work at the lastminute). Low emotion-focused coping potentialwas produced by having him think that there was agreat danger he would finish with a poor academicrecord. Low future expectancy was produced byhaving him think it was impossible to succeedwith his chosen academic path.

Smith and Lazarus (1993) found that theappraisal manipulations generally had the pre-dicted effects on the emotional states reported byparticipants. For example, anger was more com-mon when there was other-accountability ratherthan self-accountability. In contrast, guilt was morecommon when there was self-accountability ratherthan other-accountability. In spite of the apparentlypositive nature of these findings, Parkinson (2001)argued that they were relatively unimpressive.He pointed out that under 30% of the variance

Mechanisms involved in theappraisal process. From Smith

and Kirby (2001), toward deliveringon the promise of appraised theory.

In K. R. Scherer, A. Schoor, &T. Johnson (Eds.), Appraisal

processes in emotion:Theory, methods, research.Reprinted by permission of

Oxford University Press.

CP_18_C01 9/28/06, 6:55 PM4


in emotion ratings was accounted for by theappraisal manipulations. In other words, most ofthe differences in emotional states across situationscould not be explained directly by the appraisalmanipulations.

Bennett, Lowe, and Honey (2003) suggestedthat more impressive findings might be obtainedif real-life emotional events were studied ratherthan the hypothetical scenarios used by Smith andLazarus (1993). Bennett et al. asked participantsto think of the most stressful event they had experi-enced over the previous 4 weeks. The emotionalstates experienced by the participants were pre-dicted reasonably well by the cognitive appraisalsthey had used. However, the relationship betweenappraisals and emotional experience was nostronger in this study than in the one by Smithand Lazarus (1993).

The relationship between cognitive appraisalsand specific emotional experience may some-times be weak because any given emotion can beproduced by various combinations of appraisals.This possibility was explored by Kuppens, vanMechelen, Smits, and de Boeck (2003). Theystudied four appraisals (goal obstacle; other-accountability; unfairness; control) relevant tothe experience of anger. Participants described re-cently experienced unpleasant situations in whichone of the four appraisals was present or absent.What Kuppens et al., p. 266, found was that thedeterminants of anger are relatively flexible andchangeable: “None of the selected components[of appraisal] can be considered as a truly singlynecessary or sufficient condition for anger.” Thus,for example, we can feel angry without theappraisal of unfairness or the presence of a goalobstacle.

According to Smith and Kirby (2001), appraisalcan involve very rapid associative processesoccurring below the level of conscious awareness.There is much evidence to support this viewpoint(discussed in the next section). For example,Chartrand, van Baaren, and Bargh (2006) con-firmed that automatic appraisal processes caninfluence emotional state. Positive (e.g., music;friends), negative (e.g., war; cancer), or neutral(e.g., building; plant) words were presentedrepeatedly but so briefly they could not be identi-

fied at the conscious level. Participants receivingthe negative words reported a more negative moodstate than those receiving the positive words.

Evaluation

There is a consensus that appraisal processesare of fundamental importance in influencingemotional experience. Such processes not onlydetermine whether or not we experience emotion,but also strongly influence the precise emotionexperienced. Smith and Kirby (2001), with theirdistinction between associative processes andreasoning, have clarified the processes involvedin appraisal, which were left rather vague in pre-vious versions of appraisal theory.

Various criticisms can be made of the appraisalapproach to emotion. First, the assumption thatappraisal always plays a crucial role in determin-ing emotional experience is probably too strong.In fact, it is likely that appraisal is sometimesa consequence of emotional responding rather thana cause. For example, there is much evidencethat depression can be caused by neurotransmitterabnormalities and can be alleviated by the useof anti-depressant drugs (see Eysenck, 2004).Second, the rapid and automatic associative pro-cesses involved in appraisal are difficult to studyin detail, and there has been relatively little pro-gress in understanding them. Third, it is assumedthat appraisal causes emotional experience, butappraisal and emotional experience often seem toblur into each other. As Parkinson (2001, p. 181)pointed out, “It seems likely that a willingness toendorse items describing one’s helplessness andfeelings of loss [appraisal] implies a tendency toagree that one is also sad and sorrowful [emo-tional experience].” Fourth, “Appraisal theory hastaken the paradigm [model] of emotional experi-ence as an individual passive subject confrontinga survival-threatening stimulus” (Parkinson &Manstead, 1992, p. 146). Thus, appraisal theoryde-emphasises the social context in which mostemotion is experienced. Fifth, appraisal theoriesfocus on “a relatively limited number of emo-tions, leaving large gaps in our understanding ofthe full spectrum of emotional phenomena in needof explanation” (Barrett, in press).

CP_18_C01 9/28/06, 6:55 PM5


MULTI-LEVEL THEORIES

As we have seen throughout the book, thecognitive system is complex and multi-faceted.For example, Baddeley’s working memory modelnow consists of four different components (seeChapter 6). Accordingly, it is probable that severaldifferent cognitive processes underlie emotionalexperience. For example, as discussed above, Smithand Kirby (2001) argued that emotional experi-ence is influenced by associative processes andby reasoning.

The complexity of the cognitive system is oneimportant reason why theorists are increasinglyputting forward multi-level theories to identifythe key cognitive processes underlying emotion.Another reason is that such theories seem to beneeded to account for the emotional conflicts mostof us experience from time to time. For example,individuals with spider phobia become very fright-ened when they see a spider even though theymay “know” that most spiders are harmless. Theeasiest way of explaining such emotional con-flicts is to assume that one cognitive processproduces fear in response to the sight of a spider,whereas a second cognitive process provides con-flicting knowledge that it is probably harmless.

There are many multi-level theories (seereviews by Robinson, 1998, and Teasdale, 1999),and these theories differ in important ways. How-ever, there is fairly general agreement that oneway (but not the only one) in which emotionalreactions can be triggered is by preattentive orautomatic processes. We start by considering someof the evidence on this issue before discussingtwo multi-level theories.

Automatic or preattentive processes

In the previous section, we discussed findings byChartrand et al. (2006) suggesting that processesbelow the level of conscious awareness can pro-duce emotional reactions. Some of the most con-vincing evidence that emotional responses can beactivated automatically was reported by Öhmanand Soares (1994). They presented snake- andspider-phobic participants with pictures of snakes,

spiders, flowers, and mushrooms. These pictureswere presented very rapidly so that they couldnot be identified. In spite of this, the spider-phobic participants reacted emotionally to thespider pictures, as did the snake-phobic particip-ants to the snake pictures. More specifically, therewere greater physiological responses (in the formof skin conductance responses) to the phobia-relevant pictures. In addition, the participantsexperienced more arousal and felt more negativewhen exposed to those pictures than to the otherones.

A related phenomenon is worth mentioning atthis point. There have been numerous studies inwhich stimuli (e.g., pictures; Chinese ideographs)were presented very briefly below the level ofconscious awareness. Even though these stimulicould not subsequently be recognised, participantswere still more likely to choose previously pre-sented stimuli than comparable new ones whenasked to select the ones they preferred. Thus, therewas a positive affective reaction to the previouslypresented stimuli (as assessed by their preferencejudgements) in spite of the fact that these stimulihad not been perceived consciously. This phenom-enon is known as the mere exposure effect (seeBornstein, 1989, for a review).

Monahan, Murphy, and Zajonc (2000) pointedout that most studies of the mere exposure effecthave focused only on positive affective reactionsto the specific stimuli previously presented belowthe level of conscious awareness. They wonderedwhether the positive affect produced by repeatedexposure to stimuli might generalise to novelstimuli. Monahan et al. explored this issue asfollows. Participants were initially presented with25 different polygons or Chinese ideographs onceeach (single exposure) or with five polygons orideographs presented five times each (repeatedexposure). Each stimulus was presented for only5 ms and was followed by a mask so that it couldnot be identified consciously. After that, all par-ticipants rated their liking for various stimuli.Some of these stimuli had been presented pre-viously, some were novel but from the samecategory (i.e., polygons or ideographs), and somewere novel and from the other category. Whatwas found was that all three types of stimulus

CP_18_C01 9/28/06, 6:55 PM6


(1) A slow-acting thalamus-to-cortex-to-amygdalacircuit involving detailed analysis of sensoryinformation.

(2) A fast-acting thalamus–amygdala circuit basedon simple stimulus features (e.g., intensity);this circuit bypasses the cortex.

Why do we have two emotion circuits? Thethalamus–amygdala circuit allows us to respondrapidly in threatening situations, and thus can bevaluable in ensuring our survival. In contrast, thecortical circuit produces a detailed evaluation ofthe emotional significance of the situation. Assuch, it allows us to respond to situations in themost appropriate way.

The findings of Öhman and Soares (1994; dis-cussed previously) are consistent with the notionthat there is a fast-acting system for threat detec-tion that involves only minimal cortical process-ing. Additional support was reported by Morris,DeGelder, Weiskrantz, and Dolan (2001). Theystudied a patient who had no conscious visualperception of stimuli presented to damaged areas

3.4

3.2

3.0

2.8

2.6

2.4

2.2

2.0

Liki

ng

rat

ing

s

Oldpreviously

shown

Novelsimilar

Noveldifferent

Control

a

bc

a

bcb

dcd

Stimuli rated

Repeated exposure

Single exposure

Control

FIGURE 18.3

were liked significantly better by participantswho had initially been presented with repeatedexposures of stimuli they could not consciouslyperceive (see Figure 18.3).

What do these findings mean? First, preat-tentive or automatic processing based on repeatedexposures to the same stimuli can produce posi-tive affect. Second, the positive affect thus createdcan influence the emotional reactions to a sur-prisingly wide range of novel stimuli.

LeDoux: Fear

LeDoux (1992, 1996) has focused mainly on fearin his research. He has emphasised the role of theamygdala, which he regards as the brain’s “emo-tional computer” for working out the emotionalsignificance of stimuli. According to LeDoux,sensory information about emotional stimuli isrelayed from the thalamus simultaneously to theamygdala and to the cortex. Of key relevance here,LeDoux (1992, 1996) argues that there are twodifferent emotion circuits in fear:

Liking ratings as a function ofexposure conditions and stimulustype. Bars marked with the sameletter do not differ significantly.From Monahan et al. (2000).Reprinted with permission ofthe Association for PsychologicalScience.

CP_18_C01 9/28/06, 6:55 PM7


of his primary visual cortex. In spite of this, hehad significant activation of the amygdala to fear-ful faces presented to the damaged area. The over-all pattern of brain activation indicated that thefast-acting system was involved and that therewas little or no cortical processing. Other studiesconsistent with LeDoux’s theoretical approach arediscussed in a review by Öhman (2005).

Interacting Cognitive Subsystems framework

Teasdale and Barnard (1993) developed oneof the most influential multi-level theoreticalapproaches, the Interacting Cognitive Subsystemsframework. In essence, it is assumed that thereare several kinds of information codes (see Fig-ure 18.4):

(1) Sensory codes for basic features of visual,acoustic, and proprioceptive [concerned withone’s own body state] sensory input.

(2) Intermediate codes for speech-level andobject information.

(3) Propositional codes representing specificmeanings based on conceptual processing.

These meanings can be expressed verbally andcan be tested for accuracy with evidence.

(4) Implicational codes representing a more gen-eral level of implicit, non-verbal meaningbased on schemas and expressing feelings withimplicit content.

Three points need to be emphasised about theInteractive Cognitive Subsystems approach. First,what Teasdale and Barnard (1993) proposed is aframework rather than a fully developed theory.As a consequence, it does not directly generatetestable predictions. Second, it is assumed thatthe various subsystems interact strongly with eachother. Third, it is assumed that emotional experi-ence depends much more on implicational mean-ings than propositional ones. In the words ofBarnard and Teasdale (1991, p. 24), the essenceof the distinction is between “knowing with thehead (propositional) and knowing with the heart(implicational)”.

Walz and Rapee (2003) carried out a study totest the notion that there are separate propositionaland implicational levels of meaning. They arguedthat the content of spoken words determines their

Acousticsubsystem

Visualsubsystem

Body statesubsystem

Morphonolexicalsubsystem

Objectsubsystem

Propositionalsubsystem

Implicationalsubsystem

Sensory

Intermediatestructuraldescription

Meaning

Co-occurringregularities inoriginating codesabstracted intoderived codes

FIGURE 18.4

The hierarchical relationshipbetween sensory, structural

description and meaning codesin the ICS framework. From

Teasdale and Barnard (1993).

CP_18_C01 9/28/06, 6:55 PM8


propositional meaning, whereas the way they areexpressed (e.g., tone; volume) determines theirimplicational meaning. The two forms of meaningcan be distinguished by saying emotional wordsin a neutral tone (e.g., “fury” said neutrally) or bysaying neutral words with an emotional expres-sion (e.g., “jury” said with anger). They had twofurther conditions in which emotional words weresaid with an emotional expression (e.g., “fury”said with anger) and neutral words were saidunemotionally (e.g., “jury” said neutrally). Therewere two tasks. The content task required parti-cipants to decide as rapidly as possible whetherthe word meaning was emotional or neutral whileignoring the expression. The expression task in-volved deciding whether the expression was emo-tional or neutral while ignoring the word content.

What did Walz and Rapee (2003) find? Ofcrucial importance, implicational meaning inthe form of expression played a key role in bothtasks. Unsurprisingly, performance speed on theexpression task was determined almost exclusivelyby whether the expression of the stimulus wordwas angry or neutral. In addition, however, wordexpression influenced performance speed on thecontent task (see Figure 18.5). Word content wasimportant on the content task but had a relativelymodest impact on the expression task.

The above findings provide general support forthe Interacting Cognitive Subsystems framework.

First, they suggest that implicational and proposi-tional meaning can have separate influences onperformance of emotion-based tasks. Second, itis assumed within the framework that implicationalmeaning is of central importance in emotion, andthat form of meaning influenced performancemore across the two tasks than did propositionalmeaning. Third, the finding that angrily expressedanger words produced the fastest decisions on thecontent task (see Figure 18.5) is consistent withthe framework on the basis that information fromthe implicational and propositional subsystemswas congruent and mutually supportive.

Overall evaluation

Multi-level theories have much to recommendthem. They are typically more consistent than pre-vious theories with the accumulating evidenceconcerning the nature of the cognitive system. Inaddition, there is convincing evidence that emo-tional experience can be triggered by preattent-ive or automatic processes as well as by moreconscious, deliberate processes. Such evidenceis relevant to the common experience of beingfrightened of some object or situation in spite of“knowing” that it is actually harmless.

There are various limitations with multi-leveltheories. First, they differ significantly from eachother (see Robinson, 1998), but the evidence does

Angry content

Neutral content

975

950

925

900

875

850

825

800

775

Mea

n R

T (m

sec)

Angryexpression

Neutralexpression

FIGURE 18.5

Mean reaction times on the contenttask as a function of word content(angry vs neutral) and wordexpression (angry vs neutral).From Walz and Rapee 2003.

CP_18_C01 9/28/06, 6:55 PM9


not clearly indicate the superiority of one multi-level theory over the others. Second, some theoret-ical approaches (e.g., the Interacting CognitiveSubsystems approach) lack sufficient detail topermit direct testing of their assumptions. Third,more generally, “If we can ‘explain’ all aspectsof cognition–emotion relations simply by anunconstrained re-description of those relations inthe language of a complex framework which doesnot yield verifiable predictions, then we mayhave really explained very little” (Teasdale, 1999,p. 677). In spite of such pessimism, however, it ishighly probable that future theories will be multi-level rather than single-level.

MOOD AND COGNITION

Suppose you are in a depressed mood. What arethe likely effects of your mood on your cog-nitive processes? Most people find that unhappymemories spring to mind when they are depressed.

They also tend to think more negatively aboutthemselves and the world around them. More gen-erally, any given mood state (negative or positive)seems to influence cognitive processing so thatwhat we think and remember matches (or is con-gruent with) that mood state.

Bower’s network theory

Bower (1981) and Gilligan and Bower (1984) putforward a semantic network theory designed toaccount for phenomena such as those mentionedin the previous paragraph (see Figure 18.6). Thetheory as developed by Gilligan and Bower (1984)makes six assumptions:

(1) Emotions are units or nodes in a semanticnetwork, with numerous connections to relatedideas, to physiological systems, to events, andto muscular and expressive patterns.

(2) Emotional material is stored in the semanticnetwork in the form of propositions orassertions.

Expressivebehaviours

AutonomicpatternsUnworthiness

Loss

Sadness(emotion node)

Hopelessness

Despair

FIGURE 18.6

Bower’s semantic network theory.The ovals represent nodes or units

within the network. Adapted fromBower (1981).

CP_18_C01 9/28/06, 6:55 PM10


(3) Thought occurs via the activation of nodeswithin the semantic network.

(4) Nodes can be activated by external or byinternal stimuli.

(5) Activation from an activated node spreads torelated nodes. This assumption is crucial—itmeans that activation of an emotion node (e.g.,sadness) leads to activation of emotion-relatednodes or concepts (e.g., loss; despair) in thesemantic network.

(6) “Consciousness” consists of a network ofnodes activated above some threshold value.

The above assumptions lead to several testablehypotheses:

• Mood-state-dependent memory: memory isbest when the mood at retrieval matches thatat the time of learning.

• Mood congruity: emotionally toned informa-tion is learned and retrieved best when thereis correspondence between its affective valueand the learner’s (or rememberer’s) currentmood state.

• Thought congruity: an individual’s free asso-ciations, interpretations, thoughts, and judge-ments are thematically congruent with his/hermood state.

• Mood intensity: increases in intensity of moodcause increases in the activation of associatednodes in the associative network.

How do the four hypotheses relate to the sixtheoretical assumptions? So far as mood-state-dependent memory (typically assessed by recall)is concerned, associations are formed at the timeof learning between the activated nodes repres-enting the to-be-remembered items and theemotion node or nodes activated because of theparticipant’s mood state. At recall, the mood stateat that time activates the appropriate emotion node.Activation then spreads from that emotion nodeto the various nodes associated with it. If themood state at learning matches that at recall,this increases activation of the nodes of to-be-remembered items, and produces enhanced recall.However, the associative links between the to-be-remembered stimulus material and the relevant

emotion nodes are likely to be relatively weak. Asa result, mood-state-dependent effects are likelyto be greater when the memory test is a difficultone offering few retrieval cues (e.g., free recall)than when it provides strong retrieval cues (e.g.,recognition memory). When the memory testprovides powerful retrieval cues, weak activationof list words via emotion nodes is unable to boostperformance.

Mood-state-dependent effects are also pre-dicted by other theories. According to Tulving’sencoding specificity principle (see Chapter 6), thesuccess of recall or recognition depends on theextent to which the information available at re-trieval matches the information stored in memory.If information about the mood state at the time oflearning is stored in memory, then being in thesame mood state at the time of retrieval increasesthis information matching. Theoretically, thisshould increase both recall and recognition.

Mood congruity occurs when people in a goodmood learn (and remember) emotionally positivematerial better than those in a bad mood, whereasthe opposite is the case for emotionally negativematerial. According to Gilligan and Bower (1984),mood congruity depends on the fact that emo-tionally loaded information is associated morestrongly with its congruent emotion node thanwith any other emotion node. For example, thosenodes containing information about sadness-provoking events and experiences are associativelylinked to the emotion node for sadness (see Fig-ure 18.6). To-be-remembered material congruentwith the current mood state links up with thisassociative network of similar information. Thisleads to extensive or elaborative encoding of theto-be-remembered material, and thus to superiorlong-term memory. A similar process is involvedduring retrieval. Information congruent with anindividual’s mood state will be more activated thaninformation that is incongruent, and so can beretrieved more easily.

Thought congruity occurs for two reasons.First, the current mood state activates the corres-ponding emotion node. Second, activation spreadsfrom that emotion node to other, associated nodes,which will tend to contain information emotion-ally congruent with the activated emotion node.

CP_18_C01 9/28/06, 6:55 PM11


Mood-state-dependent memory

Mood-state-dependent memory has typically beenstudied by having the participants initially learnone or two lists of words. Learning occurs inone mood state (e.g., happy or sad), and recalloccurs in the same mood state or in a differentone (see Figure 18.7). When two lists are pre-sented (e.g., Bower, Monteiro, & Gilligan, 1978;Schare, Lisman, & Spear, 1984), one list is learnedin one mood and the other list is learned in adifferent mood. After that, participants are putback into one of these two moods and they areinstructed to recall only the list learned first.Recall should be higher when the mood state atthe time of recall is the same as that at the timeof learning.

Schare et al. (1984) and Bower et al. (1978)obtained mood-state-dependent recall with thetwo-list design but not with the one-list design. Itis possible that participants trying to recall thefirst list with the mood appropriate to the secondlist thought of some of the words from the secondlist, and this interfered with the task of recallingfirst-list words.

Ucros (1989) reviewed 40 published studiesof mood-state-dependent memory. There was amoderate tendency for people to remember mater-ial better when there was a match between themood at learning and that at retrieval. However,the effects were generally stronger when parti-cipants were in a positive mood than a negativemood. This may well be because individuals in anegative mood are motivated to change their moodstate (discussed in more detail later).

Ucros (1989) also found that mood-state-dependent effects were stronger when the learning

material had personal relevance. According toBower’s (1992) causal belongingness hypothesis,memory is mainly affected by mood state whenparticipants believe that their emotional state atlearning is caused by the to-be-learned material.Causal attribution leads to an effective associ-ation between the material and the emotional state.This is much more likely to occur with personalevents (e.g., feeling delighted after succeedingin an important examination) than when anemotional state is induced before presenting thelearning task.

Kenealy (1997) noted various problems withprevious research. First, the level of learning wasnot assessed in most studies. As a result, it is notclear whether poor performance reflected deficientmemory or deficient learning. Second, there wasno check in some studies that the mood manipula-tions had been successful. Third, only one memorytest was generally used. However, the extent ofany mood-state-dependent effects on memoryoften depends on the nature of the memory test.For example, Kihlstrom (1991) suggested thatthe effects of mood state will be weaker whenrich and informative cues are provided within theretrieval environment (e.g., cues on a test of cuedrecall).

Kenealy (1997) addressed all these issues in aseries of experiments. In one study, the participantslooked at a map and learned a set of instructionsconcerning a particular route until their learn-ing performance exceeded 80%. The followingday they were given tests of free recall and cuedrecall (the visual outline of the map). Therewere strong mood-state-dependent effects in freerecall but not in cued recall (see Figure 18.8).Thus, mood state can affect memory even when

Mood state at learning

Happy

Happy

Sad

Sad

Mood state at recall

Happy

Sad

Happy

Sad

Predicted level of recall

High

Low

Low

High

FIGURE 18.7

Design for a study of mood-state-dependent memory together withthe predicted results on Bower’s

(1981) theory.

CP_18_C01 9/28/06, 6:55 PM12


old and an aggressive heavy drinker. Another per-sonality was Bonnie, 36, who was very social andwhose main interests were in the theatre. In onestudy, the same story was read to five of the per-sonalities in turn, with each personality providingalmost immediate recall. There was no systematicimprovement in recall across personalities, sug-gesting the existence of mood-state-dependentforgetting or inter-identity amnesia in explicitmemory. On another task, memory for words wastested by means of an implicit memory test (wordcompletion) and an explicit memory test (recall).Memory performance was much worse when thepersonality at the time of test differed from thatat learning. Thus, the findings on explicit memorysupported Bower’s (1994) theory but those onimplicit memory did not.

There are two major limitations in most studiesof memory in dissociative identity disorder. First,very few studies have included more than onepatient. Second, and more important, there arevarious reasons why a patient with dissociativeidentity disorder might apparently fail to remem-ber information previously learned by a differentpersonality. The information may be genuinelyinaccessible or it may be more-or-less deliberatelywithheld. It is very useful to include a controlgroup of healthy individuals instructed to simulatethe effects of inter-identity amnesia.

Huntjens et al. (2003) compared memoryperformance in patients with dissociative identity

100

90

80

70

60

(a) Free recall

Happy Sad

Mood at learning

Lear

ned

item

s re

calle

d (

%) 100

90

80

70

60

(b) Cued recall

Happy Sad

Mood at learningLe

arn

ed it

ems

reca

lled

(%

)

Sad at recallHappy at recall

FIGURE 18.8

Free and cued recall as a functionof mood state (happy or sad) atlearning and at recall. Based ondata in Kenealy (1997).

learning is controlled, but does so mainly whenno other powerful retrieval cues are available.

Mood-state-dependent memory:Dissociative identity disorder

Bower (1994) argued that research on patientswith dissociative identity disorder (formerlycalled multiple personality disorder) is relevantto his network theory. Such patients have two ormore separate personalities, each having its owncharacteristic mood state. According to Bower(1994), information learned when one personalityis in control should be relatively inaccessibleto explicit memory (involving conscious recol-lection) when control passes to a different per-sonality. This is sometimes termed inter-identityamnesia. In essence, the prediction is that mood-state-dependent memory effects extend to disso-ciative identity disorder.

Bower (1994) argued that there would be lessevidence for mood-state-dependent effects withtests of implicit memory (not involving consciousrecollection). The reason is that the stimuli relev-ant to implicit memory tests are “uncontrollable”or “obligatory” (Bower, 1994, p. 230), and so notsubject to mood-dependent states.

Nissen et al. (1988) studied explicit and implicitmemory in a 45-year-old woman with 22 differ-ent personalities ranging in age from 5 to 45. Oneof her personalities was Charles, who was 45 years

CP_18_C01 9/28/06, 6:55 PM13


sociative identity disorder patients denying anymemory of having learned List A at all!

The findings of Huntjens et al. (2003) suggeststrongly that patients with dissociative identitydisorder do not show mood-state-dependent effectsin explicit memory. As Huntjens et al. (2003,p. 296) concluded, “Reports of inter-identityamnesia, although possibly reflecting the patient’ssubjective experience, should not be taken as evid-ence for objective episodic memory impairmentfor neutral material.”

Bower (1994) argued that patients with dis-sociative identity disorder should not exhibit mood-state-dependent effects with implicit memory.Thus, the prediction is that patients should showclear evidence of implicit memory when tested ina different identity from the one in control at thetime of learning. The relevant studies have pro-duced inconsistent findings (see Huntjens et al.,2005, for a review). In their own study, Huntjenset al. used participants with dissociative identitydisorder, healthy controls told to simulate dis-sociative identity disorder, and healthy controlsgiven no special instructions. All groups performeda serial reaction time task. The same repeatingsequence of stimulus locations was used across

6.0

5.0

4.0

3.0

2.0

0

Non-shared categories

Shared categories

Healthycontrols

Healthysimulating

controls

Dissociativeidentitydisorderpatients

Mea

n r

ecal

l

FIGURE 18.9

Mean recall of words from sharedand unshared categories by

healthy controls, healthy simulatingcontrols, and dissociative identity

disorder patients. Based on data inHutjens et al. (2003).

disorder, healthy controls instructed to simulateinter-identity amnesia, and healthy controls. Allparticipants learned two lists (List A and List B).List A contained the names of vegetables, animals,and flowers, and List B contained the names ofdifferent vegetables, different animals, and piecesof furniture. The first two groups learned List Ain one identity and then learned List B in a differ-ent identity or a feigned identity. Finally, recallfor List B was tested by free recall.

What findings would we expect? List B recallfor the categories shared by both lists (i.e., veget-ables and animals) should be subject to proactiveinterference (in which memory is disrupted bysimilar previous learning; see Chapter 6). How-ever, this should not be the case for the categoryunique to List B (pieces of furniture). If patientswith dissociative identity disorder have inter-identity amnesia, they should not show proactiveinterference because the information from List Awould be inaccessible to them. In fact, all threegroups showed comparable amounts of proactiveinterference (see Figure 18.9), indicating thatmemory of List A words disrupted List B recall.What is especially striking is that Huntjens et al.(2003) only analysed the data from those dis-

CP_18_C01 9/28/06, 6:55 PM14


a story containing emotionally toned material.There is then a memory test for the list or thestory after the participant’s mood has returned tonormal. Mood congruity is shown by recall beinggreatest when the affective value of the to-be-learned material matches the participant’s moodstate at learning. Alternatively, emotionally tonedmaterial can be learned when the participant’smood state is fairly neutral. Recall is subsequentlytested when the participant is in a positive ornegative mood state. Mood congruity is shown ifhe/she recalls more information congruent thanincongruent with his/her mood state at recall.

Bower, Gilligan, and Monteiro (1981) studiedmood congruity. Participants hypnotised to feelhappy or sad read a story about two college men,Jack and André. Jack is very depressed and glum,because he is having problems with his academicwork, with his girlfriend, and with his tennis. Incontrast, André is very happy, because things aregoing very well for him in all three areas. Parti-cipants identified more with the story characterwhose mood resembled their own while readingthe story, and they recalled more information abouthim.

Mood states produce changes in physiologicalarousal as well as in cognitive activity. Bower

1000

900

800

700

600

500

400

3001 2 3 4 5 6 7 8

Block

Mea

n R

T (m

sec)

PatientsControlsSimulators

FIGURE 18.10

Mean response times on a serialreaction time task across blocksfor dissociative identity disorderpatients, healthy controls andhealthy simulators. From Huntjenset al. (2005) with permission fromElsevier.

blocks 2–7, but the participants were unaware ofthis. The first two groups were instructed to switchidentities for blocks 5–8.

According to Bower (1994), the performanceof the patients with dissociative identity disordershould not have suffered when they switched ident-ities. In contrast, their performance should havedeteriorated after switching identities if implicitmemory is influenced by mood-state-dependenteffects. The findings were much more in line withthe latter prediction (see Figure 18.10). However,the healthy controls simulating dissociative ident-ity disorder showed the same pattern of results asthe patients, and it is unlikely that this was attrib-utable to a change in mood state. As Huntjenset al. (2005, p. 387) concluded, “Because ofthe ability of simulators to mimic inter-identityamnesia, the results of patients cannot be inter-preted unambiguously. Their pattern of perform-ance can both indicate inter-identity amnesia orsimulation of inter-identity amnesia.”

Mood congruity

One common procedure to test for mood con-gruity is as follows. First, a mood is induced,followed by the learning of a list or the reading of

CP_18_C01 9/28/06, 6:55 PM15


Induction condition

6

5

4

3

2

1

0Depressed Schema

(essay writing)Neutral Arousal

Mea

n w

ord

s re

calle

d

Depression-related words

Essay-writing-related words

FIGURE 18.11

Free recall of depression-relatedand essay-writing-related words in

four conditions: depressed moodinduction; arousal induction; neutral

mood induction; and schemainduction. Adapted from Varner

and Ellis (1998).

(1981) assumed that the cognitive changes wereresponsible for mood congruity. However, moodcongruity may actually depend on the arousalchanges. Varner and Ellis (1998) compared thesepossibilities in two experiments in which parti-cipants were presented with word lists. Somewords were associated with being depressed andthe rest were related to the organisation of skillswhen writing an essay. There were four condi-tions: depressed mood induction, arousal induction(stepping on and off a wooden platform), neutralmood induction, and schema induction (readingstatements relevant to writing an essay). Therewas a test of free recall after list presentation.

What were the findings? Mood congruity wasfound in the depressed mood induction conditionbut not in the arousal induction condition (seeFigure 18.11). Varner and Ellis (1998) obtainedsimilar findings when the various induction pro-cedures were used after learning but before recall.Thus, mood congruity can affect retrieval as wellas learning, and depends on mood state ratherthan arousal.

Rusting and DeHart (2000) pointed out thatthere is more evidence of mood-congruent retrievalwith positive than with negative affect. How canwe explain this? People in a negative mood aremuch more likely to be motivated to change theirmood. As Rusting and DeHart (2000, p. 738)expressed it, “When faced with an unpleasantemotional state, individuals may regulate their

emotional states by retrieving pleasant thoughtsand memories, thus reducing or reversing a negat-ive mood-congruency effect.”

Rusting and DeHart (2000) tested the abovehypothesis. Participants were presented with 20positive words, 20 negative words, and 20 neutralwords, and wrote 60 sentences each containingone of the words. Then there was a negative moodinduction in which participants imagined experi-encing distressing events. After that, some parti-cipants were told to engage in positive re-appraisalof the distressing events (“List some good thingsthat could occur as a result of any of the negativeevents in the stories”). Others were told to con-tinue to focus on negative thoughts, and the controlparticipants were left free to choose the directionof their thoughts. Finally, all participants weregiven an unexpected test of free recall for the60 words.

The findings depended very much on the mood-regulation strategies used by the participants (seeFigure 18.12). Those in the continued focus con-dition showed the typical mood-congruity effect,whereas those in the positive re-appraisal condi-tion showed mood incongruity. These effects weremuch stronger among participants who had pre-viously indicated that they were generally success-ful at regulating negative moods. Many failuresto find mood-congruent effects probably occurbecause individuals in a negative mood are motiv-ated to improve their mood.

CP_18_C01 9/28/06, 6:55 PM16


Thought congruity

Thought congruity has been studied in variousways. One method is to put participants into apositive or negative mood state before asking themto make various judgements. Thought congruityis shown if the judgements are positive or lenientamong participants in a positive mood state butnegative or harsh among those in a negative moodstate.

Forgas and Locke (2005) provided good evid-ence for thought congruity. Experienced teacherswere first of all given a mood induction to putthem into a happy or sad mood state. After that,they were given four vignettes describing work-place situations (e.g., a colleague cutting in frontof you in a photocopier queue). For each vignette,the teachers had to make judgements based onimagining themselves in the situation described.Mood state influenced the participants’ judge-ments. More specifically, “Happy mood producedmore optimistic and lenient causal attributionswhile those in a negative mood were more critical”(Forgas & Locke, 2005, p. 1071).

Several studies designed to test the affect infu-sion model (discussed shortly) have failed to find

evidence of thought congruity. In essence, judge-ments that require extensive processing are ofteninfluenced by mood state, but those that can bemade easily are less likely to be (e.g., Sedikides,1994).

Mood intensity

There has been relatively little work on the moodintensity hypothesis. However, Rinck, Glowalla,and Schneider (1992) considered a related issue—the emotional intensity of the stimulus material.Participants put into a happy or sad mood ratedwords for pleasantness–unpleasantness. There wasa mood-congruency effect for the intensely emo-tional words (but not the weakly emotional words)on a later unexpected recall test.

More direct evidence in favour of the moodintensity hypothesis was discussed by Eich (1995)in a review of mood-state-dependent memory. Thepredicted effects of mood state on memory wereobtained more often when free recall was usedrather than recognition memory. Of relevance here,mood-state-dependent memory was obtained mostconsistently when the induced mood states werestrong rather than weak.

50

45

40

35

30

0Continued

focusControl Positive

re-appraisal

Perc

ent

wo

rd r

ecal

lPositive words

Negative words

FIGURE 18.12

Mean percent recall of positiveand negative words as a function ofcondition (continued focus; control;positive re-appraisal). Based ondata in Rusting and DeHart (2000).

CP_18_C01 9/28/06, 6:55 PM17


Evaluation

Bower’s network theory has been extremely influ-ential, and has played a key role in opening up anentire research area. There is reasonable empir-ical support for most of the predicted phenomenaflowing from the theory, including mood-state-dependent recall, mood congruity, and thoughtcongruity. However, there have been many fail-ures to obtain the predicted effects, which aresometimes eliminated by the motivation of indi-viduals in a negative mood state to improve theirmood.

There are several limitations with Bower’stheoretical approach. First, it predicts that moodwill influence cognitive processing more generallythan is actually the case. This issue is discussedin more detail shortly. Second, as Forgas (1999,p. 597) pointed out, the theory “is notoriouslydifficult to falsify . . . The problem of falsifiabilitymainly arises because in practice it is difficult toprovide a complete a priori specification of thekind of cognitive contents likely to be activatedin any particular cognitive task.” Third, the the-ory is clearly oversimplified. Emotions or moodsand cognitive concepts are both represented asnodes within a semantic network. In reality, how-ever, moods typically change slowly in intensity,whereas cognitions tend to be all-or-none and thereis rapid change from one cognition to another. AsPower and Dalgleish (1997, p. 74) pertinentlyremarked, “A theory that gives emotion the samestatus as individual words or concepts is theoretic-ally confused.”

Affect infusion model

A theoretical approach that is more general inscope than Bower’s network theory is the affectinfusion model (e.g., Bower & Forgas, 2000;Forgas, 1995). The starting point for this modelis the notion of affect infusion, which occurswhen affective information selectively influencesattention, learning, and memory. At the heart ofthe model is the assumption that there are fourprocessing strategies varying in the extent to whichthey involve affect infusion. Here are the fourstrategies:

(1) Direct access: This involves the strongly cuedretrieval of stored cognitive contents and isnot influenced by affect infusion. For example,our retrieval of strongly held political attitudeswould not be altered by changes in mood.

(2) Motivated processing strategy: This involvesinformation processing being influenced bysome strong, pre-existing objective (e.g., en-hancing our current mood state). There is littleaffect infusion with this processing strategy.

(3) Heuristic processing: This strategy (whichrequires the least amount of effort) involvesusing our current feelings as informationinfluencing our attitudes. For example, if askedhow satisfied we are with our lives, we mightcasually respond in a more positive way on asunny day than an overcast one (Schwarz &Clore, 1983).

(4) Substantive processing: This strategy involvesextensive and prolonged processing. Peopleusing this strategy select, learn, and interpretinformation, and then relate it to pre-existingknowledge. Affect often plays a major rolewhen this strategy is used, because it influ-ences the information used during cognitiveprocessing. Use of this strategy typically pro-duces findings in line with those predicted byBower’s network theory.

What are the crucial differences between theaffect infusion model and Bower’s (1981) networktheory? Bower assumed that mood or affect typic-ally has various different effects on processing. Incontrast, it is assumed within the affect infusionmodel that mood often has little or no effect oncognition (e.g., when people use direct accessor motivated processing). Mood or affect doesinfluence processing and performance when theheuristic or substantive processing strategies areused. However, the way heuristic processing causesmood or affect to influence performance is clearlydifferent from that envisaged by Bower (1981).In sum, one of the four processing strategiesidentified within the affect infusion model (i.e.,substantive processing) has the effects predictedby Bower (1981). The other three processing strat-egies represent a considerable development andextension of network theory.

CP_18_C01 9/28/06, 6:55 PM18


Evidence

We will briefly consider a few studies showingthe use of each of the four processing strategies.Direct access is used mainly when judgementsare made about familiar objects or events and sothe relevant information is readily accessible inlong-term memory. For example, Salovey andBirnbaum (1989) found that mood had no influ-ence on estimates of familiar positive healthyevents (e.g., those involved in maintaining health),suggesting the use of direct access. In contrast,mood did influence estimates of unfamiliar negat-ive health events (e.g., chances of contracting agiven illness) about which participants possessedlittle relevant knowledge.

Bower and Forgas (2000) argued that peoplein a sad mood often engage in motivated process-ing to try to improve their mood, and we sawsome evidence of this in a study by Rusting andDeHart (2000). Forgas (1991) asked happy or sadparticipants to select a work partner for them-selves. Descriptions about potential partners wereprovided either on information cards or on a com-puter file. Sad participants selectively searchedfor rewarding partners, whereas happy participantsfocused more on selecting task-competent part-ners. Thus, rather than showing mood-congruentprocessing, sad individuals actively attempted toreduce their sad mood by selecting a partner whowould be supportive.

The strategy of heuristic processing (whichrequires the least amount of effort) involves usingour current feelings as information influencingour attitudes. For example, Schwarz and Clore(1983) found that people expressed more positiveviews about their happiness and life satisfac-tion when questioned on a sunny rather than arainy, overcast day. This happened because theywere feeling better when the weather was fine,and this influenced their views about overall lifesatisfaction. However, when participants’ atten-tion was directed to the probable source of theirmood (i.e., the weather) by a previous question,then their mood state no longer influenced theirviews about life satisfaction. In these circum-stances, superficial heuristic processing was nolonger feasible.

Sedikides (1994) compared the effects ofsubstantive processing and direct access. Particip-ants had to make judgements about themselvesinvolving central or peripheral self-conceptions.Sedikides argued that judgements about centralself-conceptions are easily made and shouldinvolve direct access. In contrast, judgementsabout peripheral self-conceptions require moreextensive processing and so involve substantiveprocessing. As predicted from the affect infusionmodel, mood state influenced judgements ofperipheral self-conceptions but not those of centralself-conceptions.

Evaluation

The theoretical assumption that affect or moodwill influence cognitive processing and judge-ments when certain processes are used (heuristicor substantive processing) has received muchsupport. Some of this support comes from studiesdesigned to test Bower’s (1981) network theory.There is also reasonable support for the additionalassumption that affect or mood will not influencecognitive processing when other processes (directaccess or motivated processing) are used. Theaffect infusion model has more general applic-ability than the network theory, and is much betterequipped to explain non-significant effects ofaffect or mood on performance.

The central limitation of most of the researchis that the precise cognitive processes used byparticipants performing the various tasks are notclearly specified. For example, it is plausible thatmore extensive processing is required to makeunfamiliar judgements than familiar ones. How-ever, it would be useful to know which additionalprocesses are involved in the former case. Themodel is also limited in not focusing enough ondifferences in processing associated with differ-ent mood states. For example, individuals in apositive mood tend to use heuristic processing,whereas those in a negative mood use substantiveprocessing. In one study, Chartrand et al. (2006)found that participants in a positive mood reliedon superficial stereotypical information whereasthose in a negative mood engaged in more detailednon-stereotypical processing.

CP_18_C01 9/28/06, 6:55 PM19


ANXIETY, DEPRESSION, ANDCOGNITIVE BIASES

Much of the research discussed in the previoussection was concerned with the effects of moodmanipulations on cognitive processing and per-formance. However, it is also possible to focus oncognitive processing in individuals who are in agiven mood state most of the time. For example,we can study patients suffering from an anxietydisorder or from depression. Alternatively, we canstudy healthy individuals having anxious or de-pressive personalities. This may seem like an easyresearch strategy to adopt. However, a significantproblem is that individuals high in anxiety tendto be high in depression, and vice versa. This isthe case in both normal and clinical populations.

One important difference between anxiety anddepression concerns the negative events associatedwith each emotion. More specifically, past lossesare associated mainly with depression, whereasfuture threats are associated mainly with anxiety.For example, Eysenck, Payne, and Santos (2006)presented participants with scenarios referringto severe negative events (e.g., the potential oractual diagnosis of a serious illness). There werethree versions of each scenario, depending onwhether it referred to a past event, a future pos-

sible event, or a future probable event. Participantsindicated how anxious or depressed each eventwould make them. As can be seen in Figure 18.13,anxiety was associated more with future than withpast events, whereas the opposite was the case fordepression.

Why is it important to study cognitive proc-esses in anxious and depressed individuals? Oneof the key assumptions made by many researchersin this area (e.g., Beck & Clark, 1997; MacLeod,Campbell, Rutherford, & Wilson, 2004; Williams,Watts, MacLeod, & Mathews, 1997) is that vulner-ability to clinical anxiety and depression dependsat least in part on various cognitive biases. Anotherkey assumption is that cognitive therapy shouldfocus on reducing or eliminating these cognitivebiases as a major goal of treatment. The mostimportant cognitive biases are as follows:

• Attentional bias: selective attention to threat-related stimuli presented at the same time asneutral stimuli.

• Interpretive bias: the tendency to interpretambiguous stimuli and situations in a threaten-ing fashion.

• Explicit memory bias: the tendency to retrievemostly negative or unpleasant rather than posit-ive or neutral information on a test of memoryinvolving conscious recollection.

120

115

110

105

100

Anxiety

Depression

Past Futureuncertain

Futureprobable

Scenario type

Mea

n a

nxi

ety

and

dep

ress

ion

sco

res

FIGURE 18.13

Mean anxiety and depressionscores (max. = 140) as afunction of scenario type

(past; future uncertain;future probable) (N = 120).

From Eysenck et al. (2006).

CP_18_C01 9/28/06, 6:55 PM20


• Implicit memory bias: the tendency to exhibitbetter performance for negative or threaten-ing than for neutral or positive informationon a memory test not involving consciousrecollection.

It seems reasonable to assume that someonepossessing most (or all) of these cognitive biaseswould be more likely to develop an anxiety dis-order than someone lacking these bias. However,there are two issues that need to be addressedhere. First, do patients with an anxiety disorderor depression typically possess all of these cognit-ive biases or only certain ones? As we will see,theorists differ among themselves concerning theanswer to that question. Second, there is the cau-sality issue. Suppose we find that most depressedpatients possess various cognitive biases. Doesthat mean that the cognitive biases played a rolein triggering the depression, or did the depres-sion enhance the cognitive biases? Much of whatfollows represents various attempts to addressthese two issues.

Beck’s schema theory

Beck (e.g., 1976) has been the most influentialtheorist in this area. The essence of his approachis that some individuals have greater vulnerabilitythan others to developing depressive or anxietydisorders. Such vulnerability depends on the forma-tion in early life of certain schemas or organisedknowledge structures. According to Beck andClark (1988, p. 20):

The schematic organisation of the clinicallydepressed individual is dominated by anoverwhelming negativity. A negative cog-nitive trait is evident in the depressed per-son’s view of the self, world and future . . . Incontrast, the maladaptive schemas in theanxious patient involve perceived physicalor psychological threat to one’s personaldomain as well as an exaggerated sense ofvulnerability.

Beck and Clark (1988) assumed that schemasinfluence most cognitive processes such as atten-

tion, perception, learning, and retrieval of informa-tion. Schemas produce processing biases in whichthe processing of schema-consistent or emotion-ally congruent information is favoured. Thus,individuals with anxiety-related schemas shouldselectively process threatening information, andthose with depressive schemas should selectivelyprocess emotionally negative information. WhileBeck and Clark (1988) emphasised the role ofschemas in producing processing biases, theyclaimed that schemas would only become activeand influence processing when the individual wasin an anxious or depressed state.

It follows from Beck’s schema theory that indi-viduals with clinical anxiety or depression shouldtypically possess all four of the cognitive biasesdescribed above. The reason is that the schemasallegedly have pervasive influences on cognitiveprocessing. It is worth noting that essentially thesame predictions follow from Bower’s (1981) net-work theory with its emphasis on mood-congruityeffects.

Evidence relevant to Beck’s theoreticalapproach will be discussed in detail shortly. How-ever, we will note two very general limitations ofit at this juncture:

First, the central theoretical construct of‘schema’ is amorphous [vague], and oftenseems to mean little more than ‘belief’.Second, the evidence for the existence ofspecific schemas is often based on a cir-cular argument. Behavioural evidence of acognitive bias in anxious patients is used toinfer the presence of a schema, and thenthat schema is used to ‘explain’ the observedcognitive bias. In other words, there is gen-erally no direct or independent evidence ofthe existence of a schema. (Eysenck, 1997,pp. 95–96).

Williams et al. (1997)

An important theory in the area of anxiety, de-pression, and cognitive biases is that of Williamset al. (1997). They argued that anxiety anddepression fulfil different functions, and this hasconsequences for information processing. Anxiety

CP_18_C01 9/28/06, 6:55 PM21


has the function of anticipating danger or futurethreat. As a result, it is “associated with a tendencyto give priority to processing threatening stimuli;the encoding involved is predominantly percep-tual rather than conceptual in nature” (Williamset al., 1997, p. 307). In contrast, if depressioninvolves the replacement of failed goals, “thenthe conceptual processing of internally generatedmaterial related to failure or loss may be morerelevant to this function than perceptual vigilance”(Williams et al., 1997, p. 315).

It should be noted that the approach of Williamset al. (1997) is not only applicable to patientssuffering from an anxiety disorder or from depres-sion. It is also of relevance to healthy individualshaving an anxious or a depressive personality. Wewill be focusing mainly on findings from clinicalsamples, but the general pattern of results is sim-ilar within healthy populations (see Eysenck, 1997,for a review).

Williams et al. (1997) made use of Roediger’s(1990) distinction between perceptual and concep-tual processes. Perceptual processes are essentiallydata-driven processes, and are typically involvedin basic attentional processes and in implicitmemory. In contrast, conceptual processes are top-down processes, and are typically involved inexplicit memory (but can be involved in attentionalprocesses and implicit memory). Williams et al.assumed that anxiety facilitates the perceptualprocessing of threat-related stimuli, whereasdepression facilitates the conceptual processingof threatening information. This leads to variouspredictions:

(1) Anxious patients should have an attentionalbias for threatening stimuli, but depressedpatients should not when only perceptual pro-cesses are involved.

(2) Anxious and depressed patients should havean interpretive bias for ambiguous stimuli andsituations.

(3) Depressed patients should have an explicitmemory bias but anxious patients should not.

(4) Anxious patients should have an implicitmemory bias, but depressed patients shouldnot provided that only perceptual processesare involved.

Experimental evidence: Cognitive biases

We have seen that four main cognitive biaseshave been identified: attentional bias, interpretivebias, explicit memory bias, and implicit memorybias. We will consider the evidence concerningthe existence of each of these biases in clinicallyanxious and depressed groups. There are severalanxiety disorders including generalised anxietydisorder (chronic worry and anxiety about severallife domains), panic disorder (frequent occurrenceof panic attacks), post-traumatic stress disorder(anxiety and flashbacks associated with a previ-ous traumatic event), and social phobia (extremefear and avoidance of social situations). Themost common form of clinical depression by faris major depressive disorder, characterised bysadness, depressed mood, tiredness, and loss ofinterest in various activities. The reader shouldbe forewarned that some of the findings aresomewhat inconsistent and difficult to interpret.

Attentional bias

We start by considering attentional bias. Two maintasks have been used to study attentional bias (seeEysenck, 1997, for a review). First, there is thedot-probe task. In the original version of this task,two words are presented at the same time, one toan upper and the other to a lower location on acomputer screen. On critical trials, one of thesewords is threat-related and the other is neutral.The allocation of attention is measured by record-ing speed of detection of a dot that can replaceeither word. It is assumed that detection latenciesare shorter in attended areas. Therefore, attentionalbias is indicated by a consistent tendency fordetection latencies to be shorter when the dotreplaces the threat-related word rather than theneutral one.

The first experiment to study attentional biassystematically was carried out by Halkiopoulos(unpublished). In his research, carried out at thestart of the 1980s, he used the auditory modality.Participants were given a task in which pairsof words were presented at the same time, oneto each ear. They were instructed to attend toone ear and to repeat back aloud all the words

CP_18_C01 9/28/06, 6:55 PM22


presented to that ear while ignoring words on theunattended ear. A tone was sometimes presentedto one ear very shortly after a pair of words hadbeen presented, and the participants were in-structed to respond to this tone as rapidly as pos-sible. Anxious individuals high in trait anxiety(but not those low in trait anxiety) responded fasterto the tone when it followed a threatening word tothe same ear than when it followed a threateningword to the other ear. This provided the first clearevidence of attentional bias in anxiety.

Attentional bias has also been studied byusing the emotional Stroop task. The participantsname the colours in which words are printed asrapidly as possible. Some of the words are emo-tional (e.g., stupid; inadequate) whereas othersare neutral. Attentional bias is shown if particip-ants take longer to name the colours of emotion-congruent words than neutral words, on theassumption that the increased naming time occursbecause emotion-congruent words have beenattended to more than neutral words. However,there has been some dispute about the appropri-ate interpretation of findings with the emotionalStroop. According to De Ruiter and Brosschot(1994, p. 317), “The increased Stroop interfer-ence might . . . be the result of an attempt to avoidprocessing the stimulus because it contains emo-tionally valenced [loaded] information.”

There is reasonably convincing evidence thatanxious individuals typically show an attentionalbias (see Heinrichs & Hofmann, 2001, for areview). For example, Calvo and Avero (2005)reviewed studies on the effects of trait anxietyon attention to threat-related pictures. Attentionalbias associated with high anxiety was found in58% of the studies and the findings were non-significant in the remaining 42%. The main quali-fication that needs to be placed on that statementis that there is more convincing evidence foran attentional bias in attention at early than atlater stages of attentional allocation (e.g., Calvo& Avero, 2005; Mogg, Millar, & Bradley, 2000).For example, Calvo and Avero (2005) found atten-tional bias towards harm pictures over the first500 ms after stimulus presentation, but this be-came attentional avoidance 1500–3000 ms afterstimulus presentation.

Relatively little is known about attentionalbias and depression. As Rinck and Becker (2005,p. 63) pointed out, “The empirical evidenceregarding attentional biases in depression is scarceand heterogeneous, making it very difficult todraw firm conclusions.” Williams et al. (1997)reviewed seven studies, six of which found evid-ence for an attentional bias in depression. How-ever, all of these studies used the emotional Strooptask, and findings from that task are difficult tointerpret.

In spite of the inconsistencies in the literatureon depression and attentional bias, we can makesome sense of the findings. Williams et al. (1997)argued that depression is associated with biasesin conceptual processes rather than the more auto-matic perceptual ones. Accordingly, there mightbe an attentional bias in depression provided thatconceptual or elaborative attentional processesare required. Support for this position comesfrom studies by Matthews and Antes (1992) andby Rinck and Becker (2005) in which depressedindividuals showed an attentional bias on tasksinvolving controlled attentional processes. Forexample, Matthews and Antes measured the eyemovements of depressed and non-depressed indi-viduals presented with slides containing “sad” and“happy” regions. Depressed participants focusedrelatively more on the sad regions and less on thehappy ones. Mogg and Bradley (2005, p. 29)reviewed the relevant literature, and concluded that“Where an attentional bias has been found indepressed patients, it seems to occur mainly forself-relevant negative information which is pre-sented under conditions that allow or encourageelaborative [conceptual] processing.”

Interpretive bias

There is general agreement that anxious anddepressed individuals possess interpretive biases.So far as anxiety is concerned, Eysenck, MacLeod,and Mathews (1987) asked healthy participantsof varying levels of trait anxiety to write downthe spellings of auditorily presented words. Someof the words were homophones having separatethreat-related and neutral interpretations (e.g., die,dye; pain, pane). There was a correlation of +0.60

CP_18_C01 9/28/06, 6:55 PM23


between trait anxiety and the number of threaten-ing homophone interpretations.

A potential problem with the homophonetask is that the participants may think of bothspellings. In that case, their decision as to whichword to write down may involve response bias(e.g., which spelling is more socially desirable?).Eysenck, Mogg, May, Richards, and Mathews(1991) assessed response bias using ambiguoussentences (e.g., “The doctor examined littleEmily’s growth”). Patients with generalised anxietydisorder were more likely than healthy controls tointerpret such sentences in a threatening fashion,and there were no group differences in responsebias.

Research on depressed individuals consistentlyindicates that they have an interpretive bias. Vari-ous studies (discussed by Rusting, 1998) havemade use of the Cognitive Bias Questionnaire.Events are described briefly, with participantshaving to select one out of four possible inter-pretations of each event. Depressed patientstypically select more negative interpretations thanhealthy controls.

Pyszczyski, Holt, and Greenberg (1987) haddepressed and non-depressed students rate the like-lihood of various possible future events. Depressedparticipants rated negative future events as morelikely to happen than did non-depressed ones,whereas the opposite was the case for positivefuture events.

Memory biases

It will be remembered that Williams et al. (1997)predicted that explicit memory bias would befound more often in depressed than in anxiouspatients, but the opposite would be the case forimplicit memory bias. In their review of therelevant studies, Williams et al. (1997, pp. 285–288) concluded that the evidence supported theirposition:

Out of nine studies using indirect [implicit]tests of memory in anxious subjects orpatients, seven have found significant biastowards negative material . . . no study hasyet found word congruent bias in implicit

memory in depression . . . all publishedstudies appear to find explicit memorybiases in depression, yet only a third of thestudies on trait anxiety or GAD [gener-alised anxiety disorder] find explicit memorybiases.

Rinck and Becker (2005) have provided a morerecent review of the findings, concluding thatthey are less consistent than claimed by Williamset al. (1997). We will consider memory biases indepression first, followed by anxiety. It remainsthe case that depressed patients typically exhibitan explicit memory bias. However, Murray,Whitehouse, and Alloy (1999) raised an issueregarding the interpretation of this finding. Theyasked healthy participants low and high in depres-sion as assessed by the Beck Depression Inventoryto perform a self-referential task (“Describesyou?”) on a series of positive and negative words.Then the participants provided free recall or forcedrecall, in which they were required to write downa large number of words. There was the typicalexplicit memory bias in depression with freerecall, but no bias at all with forced recall (seeFigure 18.14).

What do these findings mean? According toMurray et al. (1999, p. 175), they “implicate animportant contribution of diminished motivationand/or conservative report criterion in the mani-festation of depression-related biases and deficitsin recall.”

Williams et al. (1997) concluded that no pub-lished studies showed implicit memory bias indepressed individuals. Since then, however, therehave been several reports of implicit memorybias in depression (e.g., Rinck & Becker, 2005;Watkins, Martin, & Stern, 2000). The precisereasons for these mixed findings are not known.However, Rinck and Becker argued that implicitmemory bias is more likely to be found in indi-viduals with a clinical level of depression, andtheir demonstration of implicit memory biasinvolved patients with major depressive disorder.

Williams et al. (1997) concluded that there wasmuch more evidence for implicit memory biasthan for explicit memory bias in anxious indi-viduals. More recent research has mostly indicated

CP_18_C01 9/28/06, 6:55 PM24


3

2

1

0Positivewords

Negativewords

Positivewords

Negativewords

Free recall Forced recall

Mea

n w

ord

s re

calle

d

Depressed participants

Non-depressed participants

FIGURE 18.14

Free and forced recall for positiveand negative words in individualshigh and low in depression. Basedon data in Murray et al. (1999).

that anxious patients have an implicit memorybias, but suggests they also have an explicitmemory bias. In a review, Coles and Heimberg(2002) concluded that patients suffering with eachof the main anxiety disorders exhibit implicitmemory bias. However, the evidence was weakestfor social phobia, and Rinck and Becker (2005)failed to find an implicit memory bias in social-phobic participants. Coles and Heimberg foundclear evidence in the literature for an explicit mem-ory bias in panic disorder, post-traumatic stressdisorder, and obsessive-compulsive disorder. How-ever, there was little support for this memory biasin generalised anxiety disorder or social phobia,and Rinck and Becker (2005) subsequently foundno evidence for explicit memory bias in social-phobic participants.

Evaluation

There is indisputable evidence that anxious anddepressed patients show various cognitive biases.Such evidence indicates the value of consideringanxiety and depression in terms of cognitive pro-cesses. The evidence does not seem to supportpredictions from Beck’s schema-based theorythat clinically anxious and depressed individualsshould show the complete range of attentional,interpretive, and memory biases. More specifically,

there are relatively few convincing findings show-ing that depressed individuals have an attentionalbias or an implicit memory bias.

How successful is the theoretical approachof Williams et al. (1997)? Their assumption thatthe pattern of cognitive biases differs betweenanxious and depressed individuals seems to becorrect. On balance, anxious individuals showmore evidence than depressed ones of attentionalbias. The exception is that depressed individualssometimes have an attentional bias when concep-tual or elaborative processing is involved (Mogg& Bradley, 2005), but this can be explained byWilliams et al. (1997). As predicted, anxious indi-viduals show more evidence than depressed onesof attentional bias and implicit memory bias,and depressed individuals show more evidencethan anxious ones of explicit memory bias. How-ever, some of the findings are less supportive ofthis approach. This is especially the case withrespect to explicit memory bias, which shouldtheoretically not be present in anxious patients.In fact, patients with panic disorder, obsessive-compulsive disorder, and post-traumatic stressdisorder all show explicit memory bias (Coles &Heimberg, 2002). However, and more in line withthe theory, patients with generalised anxiety dis-order and social phobia generally do not. Thus,the theoretical approach of Williams et al. (1997)

CP_18_C01 9/28/06, 6:55 PM25


can account for most (but not all) of the relevantfindings.

Experimental evidence: Causality

As mentioned earlier, evidence that patients withan anxiety disorder or major depressive disorderpossess various cognitive biases can be interpretedin various ways. Of crucial importance is the needto establish the direction of causality—do cognit-ive biases make individuals vulnerable to develop-ing anxiety or depression, or does having clinicalanxiety or depression lead to the development ofcognitive biases? Of course, it is also possible thatthe causality runs in both directions. This issue issignificant. Cognitive biases have much greaterpractical and theoretical importance if they helpto cause clinical anxiety and depression than ifthey are merely the by-products of a pre-existingdisorder.

Attentional bias

MacLeod et al. (2002) trained some of theirhealthy participants to develop an attentional bias.This was done by altering the dot-probe task sothat the target dot always appeared in the locationin which the threatening word had been presented.In another group of participants, the target dotalways appeared in the non-threat location. Whenboth groups were exposed to a moderately stress-ful anagram task, those who had developed anattentional bias exhibited more anxiety than thosein the other group.

Mathews and MacLeod (2002) discussed vari-ous studies producing results similar to those ofMacLeod et al. (2002). They also investigated theeffects of training healthy participants to developan opposite attentional bias, i.e., selectively avoid-ing attending to threat-related stimuli. In one oftheir studies, they used only healthy participantshigh in the personality dimension of trait anxiety.There were two groups, both receiving 7500 train-ing trials. The training for one group was designedto produce an opposite attentional bias, but thiswas not the case for the other group. WhatMathews and MacLeod discovered was that only

the group that developed an opposite attentionalbias showed a substantial reduction in their levelof trait anxiety.

Interpretive bias

Mathews and Mackintosh (2000) made use of avariety of methods to induce healthy individualsto develop interpretive biases for ambiguousinformation. Their key finding was that particip-ants who actively generated personally relevantmeanings developed an interpretive bias leadingto a significant elevation in experienced anxiety.Mathews and Mackintosh (2000, p. 602) con-cluded that their findings “provide evidence con-sistent with a causal link between the deploymentof interpretive bias and anxiety.”

Wilson, MacLeod, and Mathews (2006) usedhomographs having a threatening and a neutralmeaning (e.g., stroke; fit; sink). Extensive train-ing was provided so that one group of healthyparticipants would focus on the threatening inter-pretations of these homographs whereas the othergroup would focus on the neutral interpretations.After training, both groups were exposed to avideo showing near-fatal accidents. The group thathad been trained to produce threatening inter-pretations of homographs became more anxiousthan the other group during the showing of thevideo.

Reasonable evidence that interpretive bias inthe form of dysfunctional attitudes plays a role inthe development of major depressive disorder wasreported by Lewinsohn, Joiner, and Rohde (2001).They measured dysfunctional attitudes (e.g., “Mylife is wasted unless I am a success”; “I should behappy all the time”) in adolescents not having amajor depressive disorder at the outset of the study.One year later, Lewinsohn et al. assessed the negat-ive life events experienced by the participantsover the 12-month period. They also assessedwhether the participants were suffering from majordepressive disorder. Those who experienced manynegative life events had an increased likelihoodof developing major depressive disorder only ifthey were initially high in dysfunctional attitudesor interpretive bias (see Figure 18.15). Since

CP_18_C01 9/28/06, 6:55 PM26


0.30

0.25

0.20

0.15

0.10

0.05

0Low Medium High

Dysfunctional attitudes

MD

D in

cid

ence

High negativelife events

Low negativelife events

FIGURE 18.15

Probablility of developing MajorDepressive Disorder (MDD) as afunction of number of life events(high vs. low) and dysfunctionalattitudes (low, medium, or high).Adapted with permission fromLewinsohn, P.M., Joiner, T.E., &Rohde, P. (2001). Evaluation ofcognitive diathesis-stress modelsin predicting major depressivedisorder in adolescents. Journalof Abnormal Psychology, 110,203–215.

CHAPTER SUMMARY

• Cognitive determinants of emotionAccording to Lazarus’s appraisal theory, cognitive appraisal is always of fundamental importancein determining emotional experience. According to appraisal theory, each distinct emotion iselicited by a specific pattern of appraisal. Smith and Kirby (2001) have extended appraisal theoryby distinguishing between rapid and automatic associative processes and a much slower reason-ing process, with both types of process occurring in parallel. The output of these processes ismonitored by appraisal detectors. It is likely that cognitive appraisal is sometimes a consequencerather than a cause of emotional experience. In addition, appraisal theory de-emphasises thesocial context in which emotion is typically experienced.

dysfunctional attitudes were assessed before theonset of major depressive disorder, it appears thatdysfunctional attitudes or interpretive bias is a riskfactor for developing that disorder when exposedto stressful life events.

Evaluation

Research to date has produced promising findingssuggesting that training attentional and interpret-ive biases can influence individuals’ subsequentmood state. As Mathews (2004, p. 1033) argued,“Not only can training lead to persisting alterationsin encoding bias, but consequent mood changeshave provided the most convincing evidence to

date that such biases play a causal role in emo-tional vulnerability.”

There are two limitations with the researchdiscussed in this section. First, the emphasis hasbeen much more on anxiety than on depression,and it is not altogether clear that the possessionof cognitive biases triggers vulnerability to depres-sion. Second, there is some artificiality aboutthe experiments that have been carried out. AsMathews and MacLeod (2002, p. 349) concluded,“It remains uncertain whether induced biases arereally the same as those occurring naturally inclinically anxious patients. The biases induced inmost of the studies . . . are likely to be very tran-sient compared with those occurring naturally.”

CP_18_C01 9/28/06, 6:55 PM27


• Multi-level theoriesMulti-level theories designed to identify the key cognitive processes underlying emotion have theadvantage of recognising the complexity of the cognitive system. It is assumed in many multi-level theories that emotional reactions can be triggered by preattentive or automatic processes.There is much support for this assumption. LeDoux has identified separate slow-acting and fast-acting circuits for processing threat, with only the former circuit involving extensive cognitiveprocessing. It is assumed within the Interactive Cognitive Subsystems framework that know-ing with the heart involves implicational meaning, whereas knowing with the head involvespropositional meaning. Multi-level theories have considerable promise but at present are ratherdifficult to test.

• Mood and cognitionAccording to Bower’s semantic network theory, activation of an emotion node leads to activationof many related nodes in the semantic network. Such patterns of activation produce severaleffects including mood-state-dependent memory, mood congruity, and thought congruity. Thereis evidence for all of these effects. However, they are found more often with positive than withnegative mood states, because individuals in a negative mood state are motivated to change theirmood state. There is some evidence that patients with dissociative identity disorder show inter-identity amnesia, which resembles mood-state-dependent forgetting. However, recent researchhas cast doubt on the reality of this phenomenon. According to the affect infusion model, affectiveinformation influences heuristic and substantive processing but not direct access or motivatedprocessing. There is much support for these assumptions, and this model has more generalapplicability than Bower’s network theory. However, more remains to be discovered about thefactors determining which of the four processes will be used by a given individual on a given task.

• Anxiety, depression, and cognitive biasesThere is evidence for several cognitive biases including attentional bias, interpretive bias, explicitmemory bias, and implicit memory bias. According to Beck, anxious and depressed individualsshould exhibit all four biases. According to Williams et al. (1997), anxiety and depression shouldbe associated with different patterns of biases. Anxious individuals should have attentional bias,interpretive bias, and implicit memory bias, whereas depressed individuals should have inter-pretive bias and explicit memory bias. Research is more supportive of the latter theory, but itappears that patients with most anxiety disorders have an explicit memory bias. There is increas-ing evidence that cognitive biases can play a causal role in triggering anxiety and depression, butfurther research is needed.

FURTHER READING

• Barrett, L.F. (Ed.) (2005). Emotion and consciousness. New York: Guilford Press. This editedbook contains various chapters focusing on the relationship between emotion and cognition.

• Barrett, L.F. (in press). On the automaticity of emotion. In J. Bargh. (Ed.), Automatic processes insocial thinking and behaviour. New York: Psychology Press. Appraisal theories of emotion areevaluated critically with an emphasis on the main limitations of such theories.

• Hertel, P.T. (Ed.) (2002). Cognitive biases in anxiety and depression: Special issue of Cognitionand Emotion. Hove, UK: Psychology Press. This book by leading experts provides detailedinformation on the crucial issue of the similarities and differences in cognitive biases betweenanxious and depressed individuals.

CP_18_C01 9/28/06, 6:55 PM28


• Martin, L.L., & Clore, G.L. (Eds.) (2001). Theories of mood and cognition: A user’s guidebook.Mahwah, NJ: Lawrence Erlbaum Associates Inc. The main theoretical approaches to the area ofmood and cognition are discussed at length in this edited book.

• Scherer, K.R., Schorr, A., & Johnstone, T. (Eds.) (2001). Appraisal processes in emotion: Theory,methods, research. Oxford, UK: Oxford University Press. This edited book contains a range ofchapters by experts varying in their degree of enthusiasm for the appraisal-based approach toemotion.

• Yiend, J. (Ed.) (2004). Cognition, emotion and psychopathology. Cambridge, UK: CambridgeUniversity Press. Several of the chapters in this edited book are concerned with cognitive biasesin anxiety.

CP_18_C01 9/28/06, 6:55 PM29

30 GLOSSARY

30

affect infusion: the influence of affect or mood onprocesses such as attention, learning, memory,and judgement.

attentional bias: selective attention to threat-related stimuli when presented at the same timeas neutral ones.

dissociative identity disorder: a mental disorderin which the individual claims to have two ormore personalities that appear to know littleabout each other; it is claimed that there isinter-identity amnesia.

explicit memory bias: the tendency to retrieverelatively more negative or unpleasant informa-tion than positive or neutral information on atest of explicit memory dependent on consci-ous recollection.

implicit memory bias: the tendency to haverelatively better memory performance for

Glossary

negative than for neutral information on a testof implicit memory not involving consciousrecollection.

interpretive bias: the tendency to interpret am-biguous stimuli and situations in a threateningway.

mere exposure effect: an effect in which stimulipresented previously (even when not perceivedconsciously) are preferred to new stimuli.

mood congruity: learning and memory are betterwhen the affective value of what is learned orremembered matches the individual’s moodstate.

mood-state-dependent memory: the findingthat memory is better when the mood state atretrieval is the same as that at learning thanwhen there is a discrepancy between the twomood states.

CP_18_D01 9/28/06, 6:56 PM30

REFERENCES 31

31

Barnard, P.J., & Teasdale, J.D. (1991). Interactingcognitive subsystems: A systemic approach tocognitive–affective interaction and change. Cogni-tion and Emotion, 5, 1–39.

Barrett, L.F. (in press). On the automaticity of emotion.In J. Bargh (Ed.), Automatic processes in socialthinking and behaviour. New York: Psychology Press.

Beck, A.T., & Clark, D.A. (1997). An information-processing model of anxiety: Automatic and strategicprocesses. Behaviour Research and Therapy, 35, 49–58.

Beck, A.T. (1976). Cognitive therapy and the emotionaldisorders. New York: International Universities Press.

Beck, A.T., & Clark, D.A. (1988). Anxiety and depres-sion: An information processing perspective. AnxietyResearch, 1, 23–36.

Bennett, P., Lowe, R., & Honey, K.L. (2003).Appraisals, core relational themes, and emotions:A test of the consistency of reporting and theirassociations. Cognition and Emotion, 17, 511–520.

Bornstein, R.F. (1989). Exposure and affect: Overviewand meta-analysis of research, 1968–1987. Psycho-logical Bulletin, 106, 265–289.

Bower, G.H. (1981). Mood and memory. AmericanPsychologist, 36, 129–148.

Bower, G.H. (1992). How might emotions affectlearning? In S-A. Christianson (Ed.), The handbookof emotion and memory: Research and theory.Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

Bower, G.H. (1994). Temporary emotional states actlike multiple personality. In R.M. Klein & B.K.Doane (Eds.), Psychological concepts and dissocia-tive disorders. Hillsdale, NJ: Lawrence ErlbaumAssociates Inc.

Bower, G.H., & Forgas, J.P. (2000). Affect, memory,and social cognition. In E. Eich, J.F. Kihlstrom,G.H. Bower, J.P. Forgas, & P.M. Niedenthal (Eds.),

References

Cognition and emotion. Oxford, UK: Oxford Uni-versity Press.

Bower, G.H., Gilligan, S.G., & Monteiro, K.P. (1981).Selectivity of learning caused by affective states.Journal of Experimental Psychology: General, 110,451–473.

Bower, G.H., Monteiro, K.P., & Gilligan, S.G. (1978).Emotional mood as a context for learning and recall.Journal of Verbal Learning and Verbal Behavior,17, 573–585.

Calvo, M.G., & Avero, P. (2005). Time course ofattentional bias to emotional scenes in anxiety: Gazedirection and duration. Cognition and Emotion, 19,433–451.

Chartrand, T.L., van Baaren, R.B., & Bargh, J.A.(2006). Linking automatic evaluation to moodand information-processing style: Consequencesfor experienced affect, impression formation, andstereotyping. Journal of Experimental Psychology:General, 135, 7–77.

Cobos, P., Sanchez, M., Perez, N., & Vila, J. (2004).Effects of spinal cord injuries on the subjectivecomponent of emotions. Cognition and Emotion,18, 281–287.

Coles, M.E., & Heimberg, R.G. (2002). Memory biasesin the anxiety disorders: Current status. ClinicalPsychology Review, 22, 587–627.

De Ruiter, C., & Brosschot, J.F. (1994). The emotionalStroop interference effect in anxiety: Attentional biasor cognitive avoidance? Behaviour Research andTherapy, 32, 315–319.

Eich, E. (1995). Searching for mood dependent memory.Psychological Science, 6, 67–75.

Eysenck, M.W. (1997). Anxiety and cognition: A uni-fied theory. Hove, UK: Psychology Press.

Eysenck, M.W. (2004). Psychology: An internationalperspective. Hove, UK: Psychology Press.

CP_18_D02 9/28/06, 6:56 PM31

32 REFERENCES

Eysenck, M.W., MacLeod, C., & Mathews, A. (1987).Cognitive functioning and anxiety. PsychologicalResearch, 49, 189–195.

Eysenck, M.W., Mogg, K., May, J., Richards, A.,& Mathews, A. (1991). Bias in interpretation ofambiguous sentences related to threat in anxiety.Journal of Abnormal Psychology, 100, 144–150.

Eysenck, M.W., Payne, S., & Santos, R. (2006). Anxi-ety and depression: Past, present, and future events.Cognition and Emotion, 20, 274–294.

Forgas, J.P. (1991). Mood effects on partner choice:Role of affect in social decisions. Journal of Per-sonality and Social Psychology, 61, 708–720.

Forgas, J.P. (1995). Mood and judgement: The affectinfusion model (AIM). Psychological Bulletin, 117,39–66.

Forgas, J.P. (1999). Network theories and beyond. In T.Dalgleish & M.J. Power (Eds.), Handbook of cogni-tion and emotion. Chichester, UK: Wiley.

Forgas, J.P., & Locke, J. (2005). Affective influences oncausal inferences: The effects of mood on attribu-tions for positive and negative interpersonal episodes.Cognition and Emotion, 19, 1071–1081.

Fridja, N.H., Kuipers, P., & ter Schure, E. (1989). Rela-tions among emotion, appraisal, and emotionalaction readiness. Journal of Personality and SocialPsychology, 57, 212–228.

Gilligan, S.G., & Bower, G.H. (1984). Cognitiveconsequences of emotional arousal. In C. Izard,J. Kagen, & R. Zajonc (Eds.), Emotions, cognition,and behaviour. New York: Cambridge UniversityPress.

Heinrichs, N., & Hofmann, S.G. (2001). Informationprocessing in social phobia: A critical review. Clin-ical Psychology Review, 21, 751–770.

Hertel, P.T. (2002). Cognitive biases in anxiety anddepression: Introduction to the special issue. Cogni-tion and Emotion, 16, 321–330.

Huntjens, R.J.C., Postma, A., Peters, M.L., Woertman,L., & van der Hart, O. (2003). Inter-identity amnesiafor neutral, episodic information in dissociativeidentity disorder. Journal of Abnormal Psychology,112, 290–297.

Huntjens, R.J.C., Postma, A., Woertman, L., van derHart, O., & Peters, M.L. (2005). Procedural memoryin dissociative identity disorder: When can inter-identity amnesia be truly established? Consciousnessand Cognition, 14, 377–389.

James, W. (1890). Principles of psychology. New York:Holt.

Kenealy, P.M. (1997). Mood-state-dependent retrieval:The effects of induced mood on memory recon-sidered. Quarterly Journal of Experimental Psycho-logy, 50A, 290–317.

Kihlstrom, J.F. (1991). On what does mood-dependentmemory depend? In D. Kuiken (Ed.), Mood andmemory: Theory, research and applications. London:Sage.

Kuppens, P., van Mechelen, I., Smits, D.J.M., & deBoeck, P. (2003). The appraisal basis of anger:Specificity, necessity, and sufficiency of components.Emotion, 3, 254–269.

Lazarus, R.S. (1982). Thoughts on the relations betweenemotion and cognition. American Psychologist, 37,1019–1024.

Lazarus, R.S. (1991). Emotion and adaptation. Oxford,UK: Oxford University Press.

LeDoux, J.E. (1992). Emotion as memory: Anatomicalsystems underlying indelible neural traces. In S-A.Christianson (Ed.), The handbook of emotion andmemory: Research and theory. Hillsdale, NJ: Law-rence Erlbaum Associates Inc.

LeDoux, J.E. (1996). The emotional brain: Themysterious underpinnings of emotional life. NewYork: Simon & Schuster.

Lewinsohn, P.M., Joiner, T.E. Jr., & Rohde, P. (2001).Evaluation of cognitive diathesis–stress models inpredicting major depressive disorder in adolescents.Journal of Abnormal Psychology, 110, 203–215.

MacLeod, C., Campbell, L., Rutherford, E.M., &Wilson, E. (2004).

MacLeod, C., Rutherford, E.M., Campbell, L.,Ebsworthy, G., & Holker, L. (2002). Selectiveattention and emotional vulnerability: Assessingthe causal basis of their association through theexperimental manipulation of attentional bias.Journal of Abnormal Psychology, 111, 107–123.

MacLeod, C., Campbell, L., Rutherford, E., & Wilson,E. (2004). The causal status of anxiety-linkedattentional and interpretive bias. In J. Yiend (Ed.),Cognition, emotion and psychopathology: Theoret-ical, empirical and clinical directions. Cambridge:Cambridge University Press.

Martin, L.L., & Clore, G.L. (Eds.) (2001). Theories ofmood and cognition: A user’s guidebook. Mahwah,NJ: Lawrence Erlbaum Associates Inc.

Mathews, A. (2004). On the malleability of emotionalencoding. Behaviour Research and Therapy, 42,1019–1036.

Mathews, A., & Mackintosh, B. (2000). Inducedemotional interpretation bias and anxiety. Journalof Abnormal Psychology, 109, 602–615.

Mathews, A., & MacLeod, C. (2002). Induced process-ing biases have causal effects on anxiety. Cognitionand Emotion, 16, 331–354.

Matthews, G., & Antes, J.R. (1992). Visual attentionand depression: Cognitive biases in the eye fixationsof the dysphoric and the non-depressed. CognitiveTherapy & Research, 16, 359–371.

Mogg, K., & Bradley, B.P. (2005). Attentional bias ingeneralised anxiety disorder versus depressive dis-order. Cognitive Therapy and Research, 29, 29–45.

Mogg, K., Millar, N., & Bradley, B.P. (2000). Biases ineye movements to threatening facial expressions ingeneralised anxiety disorder and depressive disorder.Journal of Abnormal Psychology, 109, 695–704.

CP_18_D02 9/28/06, 6:56 PM32

REFERENCES 33

Monahan, J.L., Murphy, S.T., & Zajonc, R.B. (2000).Subliminal mere exposure: Specific, general, anddiffuse effects. Psychological Science, 11, 462–466.

Morris, J.S., DeGelder, B., Weiskrantz, L., & Dolan,R.J. (2001). Differential extrageniculostriate andamygdala responses to presentation of emotionalfaces in a cortically blind field. Brain, 124, 1241–1252.

Murray, L.A., Whitehouse, W.G., & Alloy, L.B. (1999).Mood congruence and depressive deficits in mem-ory: A forced-recall analysis. Memory, 7, 175–196.

Nissen, M., Ross, J., Willingham, D., Mackenzie, T., &Schacter, D. (1988). Memory and amnesia in apatient with multiple personality disorder. Brain &Cognition, 8, 117–134.

Öhman, A. (2005). The role of the amygdala inhuman fear: Automatic detection of threat. Psy-choneuroendocrinology, 30, 953–958.

Öhman, A., & Soares, J.J.F. (1994). “Unconsciousanxiety”: Phobic responses to masked stimuli.Journal of Abnormal Psychology, 103, 231–240.

Parkinson, B. (1994). Emotion. In A.M. Coleman (Ed.),Companion encyclopaedia of psychology, Vol. 2.London: Routledge.

Parkinson, B. (2001). Putting appraisal in context.In K.R. Scherer, A. Schorr, & T. Johnstone (Eds.),Appraisal processes in emotion: Theory, methods,research. Oxford, UK: Oxford University Press.

Parkinson, B., & Manstead, A.S.R. (1992). Appraisalas a cause of emotion. In M.S. Clark (Ed.), Reviewof personality and social psychology (Vol. 13). NewYork: Sage.

Power, M., & Dalgleish, T. (1997). Cognition and emo-tion: From order to disorder. Hove, UK: PsychologyPress.

Pyszczynski, T., Holt, K., & Greenberg, J. (1987).Depression, self-focused attention, and expectanciesfor positive and negative future life events for selfand others. Journal of Personality and Social Psy-chology, 52, 994–1001.

Rinck, M., & Becker, E.S. (2005). A comparison ofattentional biases and memory biases in women withsocial phobia and major depression. Journal ofAbnormal Psychology, 114, 62–74.

Rinck, M., Glowalla, U., & Schneider, K. (1992). Mood-congruent and mood-incongruent learning. Memory& Cognition, 20, 29–39.

Robinson, M.D. (1998). Running from William James’bear: A review of preattentive mechanisms and theircontributions to emotional experience. Cognition andEmotion, 12, 667–696.

Roediger, H.L. (1990). Implicit memory: Retentionwithout remembering. American Psychologist, 45,1043–1056.

Roseman, I.J., & Smith, C.A. (2001). Appraisal theory:Overview, assumptions, varieties, controversies. InK.R. Scherer, A. Schorr, & T. Johnstone (Eds.),

Appraisal processes in emotion: Theory, methods,research. Oxford, UK: Oxford University Press.

Rusting, C.L. (1998). Personality, mood, and cognitiveprocessing of emotional information: Three con-ceptual frameworks. Psychological Bulletin, 124,165–196.

Rusting, C.L., & DeHart, T. (2000). Retrieving positivememories to regulate negative mood: Consequencesfor mood-congruent memory. Journal of Personal-ity and Social Psychology, 78, 737–752.

Salovey, P., & Birnbaum, D. (1989). Influence of moodon health-related cognitions. Journal of Personalityand Social Psychology, 57, 539–551.

Schare, M.L., Lisman, S.A., & Spear, N.F. (1984).The effects of mood variation on state-dependentretention. Cognitive Therapy & Research, 8, 387–408.

Scherer, K.R., Schorr, A., & Johnstone, T. (2001).Appraisal processes in emotion: Theory, methods,research. Oxford, UK: Oxford University Press.

Schwarz, N., & Clore, G.L. (1983). Mood, misattribu-tion and judgements of well-being: Informativeand directive functions of affective states. Journalof Personality and Social Psychology, 45, 513–523.

Sedikides, C. (1994). Incongruent effects of sad moodon self-conception valence: It’s a matter of time.European Journal of Social Psychology, 24, 161–172.

Smith, C.A., & Kirby, L.D. (2001). Toward deliveringon the promise of appraisal theory. In K.R. Scherer,A. Schoor, & T. Johnstone (Eds.), Appraisal pro-cesses in emotion: Theory, methods, research.Oxford, UK: Oxford University Press.

Smith, C.A., & Lazarus, R.S. (1993). Appraisal com-ponents, core relational themes, and the emotions.Cognition and Emotion, 7, 233–269.

Speisman, J.C., Lazarus, R.S., Mordkoff, A.M., &Davison, L.A. (1964). Experimental reduction ofstress based on ego-defence theory. Journal ofAbnormal Psychology, 68, 367–380.

Strack, F., Martin, L.L., & Stepper, S. (1988). Inhibit-ing and facilitating conditions of the human smile: Anon-obtrusive test of the facial feedback hypothesis.Journal of Personality and Social Psychology, 54,768–777.

Teasdale, J.D. (1999). Multi-level theories of cognition–emotion relations. In T.D. Dalgleish & M.J. Power(Eds.), Handbook of cognition and emotion.Chichester, UK: Wiley.

Teasdale, J.D., & Barnard, P.J. (1993). Affect, cognitionand change: Re-modelling depressive thought. Hove,UK: Lawrence Erlbaum Associates Ltd.

Ucros, C.G. (1989). Mood state-dependent memory:A meta-analysis. Cognition and Emotion, 3, 139–167.

Varner, L.J., & Ellis, H.C. (1998). Cognitive activityand physiological arousal: Processes that mediate

CP_18_D02 9/28/06, 6:56 PM33

34 REFERENCES

mood-congruent memory. Memory & Cognition, 26,939–950.

Walz, P.G., & Rapee, R.M. (2003). Disentanglingschematic and conceptual processing: A test of theInteracting Cognitive Subsystems framework. Cogni-tion and Emotion, 17, 65–81.

Watkins, P.C., Martin, C.K., & Stern, L.D. (2000).Unconscious memory bias in depression: Percep-tual and conceptual processes. Journal of AbnormalPsychology, 109, 282–289.

Williams, J.M.G., Watts, F.N., Macleod, C., & Mathews,A. (1997). Cognitive psychology and emotional dis-orders (2nd. ed.). Chichester, UK: Wiley.

Wilson, E.J., MacLeod, C., Mathews, A., & Rutherford,E.M. (2006). The causal role of interpretive bias inanxiety reactivity. Journal of Abnormal Psychology,115, 103–111.

Yiend, J. (Ed.) (2004). Cognition, emotion and psycho-pathology. Cambridge, UK: Cambridge UniversityPress.

CP_18_D02 9/28/06, 6:56 PM34

18. COGNITION AND EMOTION 1 18 - psypress.co.ukpsypress.co.uk/ek5/resources/pdf/chap18.pdf · 18....

Documents

Transcript of 18. COGNITION AND EMOTION 1 18 - psypress.co.ukpsypress.co.uk/ek5/resources/pdf/chap18.pdf · 18....