Cognition & Emotion | My Assignment Tutor

Full Terms & Conditions of access and use can be found at & EmotionISSN: 0269-9931 (Print) 1464-0600 (Online) Journal homepage: anxiety, anxious mood, and threat detectionAngela Byrne & Michael W. EysenckTo cite this article: Angela Byrne & Michael W. Eysenck (1995) Trait anxiety, anxious mood, andthreat detection, Cognition & Emotion, 9:6, 549-562, DOI: 10.1080/02699939508408982To link to this article: online: 07 Jan 2008.Submit your article to this journalArticle views: 559View related articlesCiting articles: 123 View citing articlesCOGNITION AND EMOTION, 1995, 9 (6). 549-562Trait Anxiety, Anxious Mood, and Threat DetectionAngela Byrne and Michael W. EysenckRoyal Holloway, University of London, UKSubjects were required to detect either an angry or a happy target face in astimulus array of 12 photographs. It was found with neutral distractor facesthat those high in trait anxiety detected angry faces faster than did low traitanxious subjects, but the two groups did not differ in their speed of detectionof happy targets. In addition, high trait-anxious subjects detected happytarget faces slower than low trait-anxious subjects when the distractor faceswere angry. Comparable findings were obtained whether or not there wasanxious mood induction. It was concluded that high trait-anxious individualshave facilitated detection and processing of environmental threat relative tolow trait-anxious subjects, which enhance performance when the target isthreatening, but which impair performance when the distractors are threatening.INTRODUCTIONIt has often been proposed (e.g. Eysenck, 1992) that a primary function ofanxiety is to facilitate the rapid detection of threat. Much research isconsistent with this proposal (see Eysenck, 1992, for a review), but thereare no studies in which it has been tested directly. For example, there isevidence (Broadbent & Broadbent, 1988; Fox, 1993) that normals high intrait anxiety exhibit a selective attentional bias. This is typically assessed ina situation in which the concurrent presentation of threat-related andneutral stimuli is followed immediately by a probe requiring rapidresponse. A selective attentional bias is defined by relatively fasterresponding to the probe when it replaces the threat-related stimulus thanwhen it replaces the neutral stimulus. Such findings indicate that there isselective attention to threat-related stimuli. However, because probe onsettypically occurs 500msec after the onset of the threat-related and neutralstimuli, this paradigm is uninformative concerning the speed with whichthreat-related stimuli are detected.Requests for reprints should be sent to Professor Michael W. Eysenck, Department ofPsychology, Royal Holloway. University of London, Egham, Surrey TW20 OEX,UK0 1995 Erlbaum (UK) Taylor & Francis550 BYRNE AND EYSENCKA paradigm which is of more relevance to the present concerns was usedby MacLeod and Mathews (1991) and by Mogg, Mathews, Eysenck, andMay (1991). The paradigm consisted of a lexical decision task, with twoletter strings being presented concurrently in the key conditions. MacLeodand Mathews (1991) obtained evidence of reduced lexical decision latencies for threat-related words in anxious individuals in those conditions.This finding was partially replicated by Mogg et al. (1991), but only whenthere was attentional search. This paradigm provides only an indirectmeasure of speed of threat detection, because it cannot be assumed thatthe time taken to make a lexical decision about a threatening word is thesame as the time taken to identify threat.Another paradigm which is of partial relevance is the perceptual defenceparadigm, in which the presentation times required for identification ofemotionally loaded and neutral stimuli are assessed. If high trait-anxiousindividuals have facilitated processing of threatening stimuli, it might beexpected that they would have lower perceptual thresholds for emotionallythreatening stimuli than would low trait-anxious individuals. In fact, therewere no effects of trait anxiety in the studies of perceptual defence inwhich systematic attempts were made to distinguish between perceptualsensitivity and response bias (Van Egeren, 1968; Wagstaff, 1974). According to Eysenck, MacLeod, and Mathews (1987), these non-significantfindings may have occurred because only a single stimulus was presentedon each trial. High trait-anxious individuals may exhibit additional processing of threatening stimuli only when at least one threatening and onenonthreatening stimulus are presented concurrently, so that a selectiveprocessing mechanism can operate. In any case, it should be noted thatthe perceptual defence paradigm does not provide a direct measure ofspeed of threat detection.In addition to the prediction that threatening stimuli will be detectedmore rapidly by anxious individuals than by nonanxious ones, there isanother prediction that follows from the general position adopted here. Ifhigh trait-anxious individuals selectively attend to, and process, threatrelated stimuli, then it is likely that such individuals would be moreslowed down in their task performance by threatening distractors thanwould low trait-anxious individuals. Evidence consistent with that expectation was reported by Eysenck and Byrne (1992) when physical threatdistractors were used, but not when social threat distractors were used.One of the main objectives of the study to be reported was to use asituation resembling that found in everyday life. As the major concerns ofindividuals high in trait anxiety relate to social evaluation (e.g. Eysenck &van Berkum, 1992), it was decided to use as stimuli photographs of peopleexpressing different emotions rather than the word stimuli employed innearly all of the previous research. More specifically, faces expressingTRAIT ANXIETY, ANXIOUS MOOD, THREAT DETECTION 551happiness represented the positive stimuli, faces expressing anger represented the threatening faces, and emotionless faces formed the neutralcategory.A paradigm similar to the one used here was previously utilised byHansen and Hansen (1988). They presented their subjects with a seriesof photographs containing a number of faces all expressing one type ofemotion. On half of the trials there was a single discrepant face, i.e. oneexpressing a different emotion. They discovered that subjects were faster tolocate an angry face in a happy crowd than to locate a happy face in anangry crowd. The time taken to detect an angry face was not affected by thenumber of distractor faces, whereas detection of a happy face was influenced by the number of distractors. Hansen and Hansen (1988) argued thatprocessing was parallel for angry but not for happy faces. However, thisconclusion is suspect, because the error rates were extremely high in thevarious conditions (ranging from 30.3% to 79.3%).An important difference between the current study and that of Hansenand Hansen (1988) is that the effects of individual differences in traitanxiety on face detection were considered. Various predictions weremade based on the assumption that anxiety is associated with facilitateddetection of threat. First, it was predicted that high trait-anxious individualswould be relatively faster than low trait-anxious individuals to detect angryfaces. Secondly, it was predicted that the use of angry faces as distractorswould selectively disrupt the performance of high trait-anxious subjects.Thirdly, there was a condition in which subjects were told to detect anemotional face, and there were two relevant targets in the display (onehappy and one angry). It was predicted that high trait-anxious subjectswould be more likely than low trait-anxious subjects to select the angryface.It has often been found in studies of selective attentional bias (e.g.Broadbent & Broadbent, 1988; Fox, 1993) that the bias is determinedinteractively by trait and state anxiety. More specifically, the effects oftrait anxiety were greater when the level of state anxiety was high thanwhen it was low. A musical mood induction procedure was used to createanxiety rather than the Velten technique based on anxiety-related statements. The reason is that there is evidence (Albersnagel, 1988) that it hasmore intense and longer-lasting effects. In view of previous findings, itseemed likely that the various predicted effects of trait anxiety on facedetection would be greater when an anxious mood state had been induced.Finally, it was considered important to distinguish between the effects oftrait anxiety and of depression, because measures of trait anxiety anddepression typically correlate about +0.6 or +0.7 with each other (Watson& Clark, 1984). There are empirical and theoretical grounds for arguingthat it is trait anxiety rather than depression which is responsible for most552 BYRNE AND EYSENCKof the effects of attention on the processing of threat-related stimuli(Williams, Watts, MacLeod, & Mathews, 1988). Accordingly, it waspredicted that face-detection times would be determined by trait anxietyrather than by depression.METHODSubjectsTwenty-five subjects took part in the experiment. They were pre-selectedon the basis of their scores on the trait scale of the Spielberger State-TraitAnxiety Inventory. The high trait-anxious group consisted of 12 subjectsscoring above 48,with a mean of 50.0; their mean score on the BeckDepression Inventory was 10.8. The low trait-anxious group consisted of13 subjects scoring below 32, with a mean of 30.8; their mean score on theBeck Depression Inventory was 5.6. There were 11 males and 15 females,and ages ranged from 20 to 47 (mean = 25 years). All subjects were studentvolunteers. The sex ratio was almost equivalent across groups (six maleand seven female subjects in high-anxious group; five male and eightfemale subjects in the low-anxious group).MaterialsThere were 120 slides, each one consisting of 12 black-and-white photographs of faces arranged in four rows and three columns. Each slidecontained photographs of the same 12 individuals, of whom 7 werewomen and 5 were men. These faces were taken from the Ekman picturesof facial affect (Ekman & Friesen, 1975).From this set, the pictures chosenwere those with a neutral expression, those expressing happiness arrd thoseexpressing anger or anger mixed with disgust. These pictures were presented to six independent judges who rated them for emotionality andthreat value. Only those pictures rated as very positive, very threatening,and neutral were used. The positive and threatening faces were matched forrated emotionality.There were five blocks of 24 slides. In each block, half of the slidescontained a discrepant or target face, and half did not. Three of the blocksinvolved crowds with neutral expressions. In one of these blocks half of theslides contained a target face expressing anger; in another block, half of theslides contained a target face expressing happiness; and in the third block,half of the slides contained two possible targets: one happy face and oneangry face. In the fourth block, one consisted of slides in which the crowdsexpressed anger and half of these slides contained a discrepant happy face;and in the final block, the crowds expressed happiness and half of the slidesTRAIT ANXIETY, ANXIOUS MOOD, THREAT DETECTION 553contained a discrepant angry face. The trials may be summarised asfollows: angry target, neutral crowd (AN); happy target, neutral crowd(HN); both types of target, neutral crowd (BN); angry target, happy crowd(AH); and happy target, angry crowd (HA). The order of presentation ofblocks was randomised.The location of each face on each slide was randomised with theconstraint that the target should appear once in each location for thoseslides containing a target face. In those slides which contained two possibletargets, there was a constraint that there should be a separation of no morethan one position in any given direction between the two targets. Thepositions were also counterbalanced.Presentation was controlled by computer. The keypad was covered sothat only 12 keys were visible. The keys were numbered 1 to 12 corresponding to the arrangement of the faces on each slides. When the subjectpressed a response key the shutter closed, reaction time and key numberwere logged, and the next slide appeared. Subjects were instructed to pressthe space bar to indicate absence of a target and this response was logged asthe number 13.Two pieces of music were used for the mood induction procedure. Thesewere Stravinsky’s “Rite of Spring” and Faure’s “Ballad for Piano andOrchestra”. They were used to create the anxiety and non-anxiety conditions, respectively.ProcedureTesting took place on two occasions. On the first occasion, subjectscompleted the trait state scales of the Spielberger State-Trait AnxietyInventory and the Beck Depression Inventory. Subjects were theninstructed that their task was to locate the face expressing a differentemotion to the others, and they were told how to select the appropriateresponse key. Subjects were instructed to work as rapidly as possiblewithout making mistakes. Subjects were allowed to take a break betweenblocks. Before each block, subjects were instructed to look for the appropriate face, e.g. “Your task is to look for the angry face”. In the conditionin which the target could be a happy or an angry face, subjects wereinstructed to “look for the emotional face”.On the first testing session, subjects listened to one of the pieces ofmusic while completing the five blocks of trials. The subjects were notexplicitly instructed to work at getting into the desired mood. Kenealy(1988) found this procedure successful even in the absence of such instructions. When the blocks of trials were completed, subjects completed themeasure of state anxiety for a second time. Subjects were then requested toreturn at the same time the following day. On the next testing occasion,554 BYRNE AND EYSENCKsubjects performed the task as before while listening to the other piece ofmusic. At the end of the second testing session, subjects completed themeasure of state anxiety for the last time.RESULTSThe mean response times in all the single target conditions are shown inTable 1. As the hypotheses related to performance on the target trials only,analyses were conducted separately on the data from target and nontargettrials.Target TrialsThe first analysis compared the speed of detection of angry faces in neutralcrowds with that of happy faces in neutral crowds. Mood induction did notproduce a significant main effect and there were no significant interactionsinvolving this variable; accordingly, the data were pooled across thisvariable. There was a significant main effect of target type, [ F ( l , 23) =5.83,P < 0.025],with happy faces in neutral crowds being detected fasterTABLE 1Mean Response Interviews in msec as a Function of Trial Type, MoodInduction Condition, Target Presentvs. Absent, and Trait Anxiety (StandardDeviationsin Brackets) Trial TypesLow TAHigh TALow TAHigh TANeutral MoodAnxious Mood ~~ ~ – ~~HN TargetNo TargetAN TargetNo TargetHA TargetNo TargetAH TargetNo Target2875(1743)3583(1295)4352(1851)3844(1806)2454(563)4431(1023)2778( I 108)4078(1085)2471(686)4022(1302)2548(1198)4175(885)3226(808)4325(772)2532(77213790(887)2248 2592(1027) (1117)3626 3660(1310) (1030)3081 2415(1411) (1098)3926 3956(1528) (936)2640 3234(933) (920)3823 4680(1382) (1120)2931 2768(1884) (842)3877 4283(1249) (1038)TRAIT ANXIETY, ANXIOUS MOOD, THREAT DETECTION 555than angry faces in neutral crowds. There was a significant interactionbetween trait anxiety and target type, [F(1, 23) = 7.74, P < 0.0251 (seeFig. 1). In this interaction, the high trait-anxious group was significantlyfaster than the low trait-anxious group in detecting angry faces, P < 0.01,but the two groups did not differ in their response times to emotionallypositive faces.A bias score was calculated by subtracting mean detection latencies toangry targets from mean detection latencies to happy targets. This correlated significantlywith trait anxiety scores, r =+0.48,P < 0.025, but it didnot correlate significantly with depression scores, r = +0.34. The correlation of the bias score with trait anxiety remained significant when depression was partialled out, r = +0.38, P < 0.05.The next analysis addressed the second hypothesis. It compared thespeed of detection of happy faces in neutral crowds with that of happyfaces in angry crowds. The main effect of type of crowd was significant[F(1,23)=4.34, P


Leave a Reply

Your email address will not be published. Required fields are marked *