- Email: [email protected]
The Effects of Target and Distractor Familiarity on Visual Search in Anxious Children
Pergamon
Journal of Anxiety Disorders, Vol. 14, No. 1, pp. 41–56, 2000 Copyright 2000 Elsevier Science Ltd Printed in the USA. All rights reserved 0887-6185/00 $–see front matter
PII S0887-6185(99)00038-9
The Effects of Target and Distractor Familiarity on Visual Search in Anxious Children: Latent Inhibition and Novel Pop-Out R. E. Lubow, Ph.D. Tel Aviv University, Tel Aviv, Israel
Paz Toren, M.D., and N. Laor, M.D., Ph.D. Tel Aviv Community Mental Health Center, Tel Aviv, Israel
Oren Kaplan, Ph.D. Tel Aviv University, Tel Aviv, Israel
Abstract—Children and adolescents (ages 6–17 years) diagnosed as having an anxiety disorder were compared to matched controls on a two-stage serial visual search task in which they identified presence or absence of a unique shape presented with homogeneous distractors. Response time was examined as a function of prior experience with either target, distractor, or both, allowing for a within-subject assessment of latent inhibition (LI: slower responding to a target that was formerly a distractor against a background of distractors that were formerly targets as compared to a novel target with distractors that were formerly targets) and novel pop-out effects (NPO: faster responding to a novel target against a background of familiar former targets as compared to the condition in which both the target and distractors were novel). There were robust LI and NPO effects for both anxious and control children. However, the predicted interaction between diagnosis and LI condition was not obtained. In general, the results suggest that children with diagnosed anxiety disorder do not differ from controls on basic information processing as assessed by this visual search task. 2000 Elsevier Science Ltd. All rights reserved.
The authors wish to thank Michal Sadeh and Osnat Shor for assistance in data collection, and Armonit Roter for helpful suggestions on an earlier draft of this paper. Requests for reprints should be sent to R. E. Lubow, Department of Psychology, Tel Aviv University, Ramat Aviv 69978, Israel. E-mail: [email protected]
41
42
R. E. LUBOW ET AL.
Keywords: Latent inhibition; Anxiety; Visual search
The relationships between behavioral/physiological pathologies and cognitive processes have been the subject of much research in a variety of areas (e.g., Parkinson’s disease, schizophrenia). Such research adds to an understanding of causes and/or consequences of particular disorders, as well as to knowledge of the cognitive process used in the specific research paradigm. With some exceptions, to be discussed below, research on cognitive processing in anxiety-disordered children (ADC) has been surprisingly limited, perhaps because ADC is not a clearly defined nosological category, but rather consists of a number of subclassifications (e.g., panic, separation anxiety, posttraumatic stress, and obsessive-compulsive disorders; see Kendall & Brady, 1995; Ollendick, King, & Yule, 1994), each with a high level of comorbidity (Bernstein & Borchardt, 1991; Werry, 1991). Nevertheless, childhood anxiety disorders, irrespective of their subclassifications, share at least one common denominator, namely anxiety itself. Consequently, such children provide a convenient sample for examining the behavioral components of anxiety. Against this background, it is somewhat surprising that, “. . . there are few studies that have assessed cognition in anxious children and there are no studies of diagnosed samples” (Kendall & Chansky, 1991, p. 168). In the 9 years since Kendall and Chansky’s assessment, the situation has improved, at least partially. One topic that has been the subject of research concerns the effects of anxiety on selective attention to threat-related as compared to neutral words. Studies with children, using either the emotional-Stroop paradigm (Kindt, Brosschot, & Everaerd, 1997; Martin, Horder, & Jones, 1992) or the probedetection task (e.g., Vasey, Daleiden, Williams, & Brown, 1995; Vasey, ElHag, & Daleiden, 1996), indicate that high-anxious children selectively attend to threat-related words. Although these data conform to those obtained from adults (e.g., Eysenck, MacLeod, & Mathews, 1987; Mogg, Bradley, Williams, & Mathews, 1993; Mogg, Bradley, & Williams, 1995), there are some qualifications. In the children’s literature, for example, Doost, Taghavi, Moradi, Yule, and Dalgleish (1997) failed to find biased processing of threatrelated words in a mixed anxiety-depression group as compared to normal controls. On the other hand, Kindt et al. (1997) did report a processing bias for threatening words, but only for girls, both low and high anxious. A gender effect was also found by Vasey et al. (1995, 1996). Only low-anxious males exhibited an attentional bias away from threatening word stimuli. All of the studies cited above used word stimuli. The lack of consistency may derive, in part, from the use of such verbal materials. Words may not be equally threatening or neutral for male and female participants. In addition, Stroop and dot-probe tasks may be less sensitive to the effects of anxiety than other tests of attention. Thus, with adults, Fox (1994) found that high-trait
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
43
anxious subjects had difficulty inhibiting distracting information in a visual search task but not in a Stroop task in which the positions of targets and distractors were constant. To help resolve these inconsistencies, we tested low and high anxious male and female children on a visual search task with completely neutral targets and distractors (nonsense figures). The specific procedure, based on a study by Lubow and Kaplan (1997, Exp. 2), has been used to explore the relationship between pathology and attention. It has been applied to schizophrenics (Lubow, Kaplan, Abramovich, Rudnick, & Laor, 2000) and Parkinson’s disease patients (Lubow, Dressler, & Kaplan, 1999), as well as normal adults (Lubow & Kaplan, 1997). In addition to evaluating standard visual search behavior, the procedure allows for the assessment of several attentionally related phenomena, including latent inhibition (LI: slower learning to a previously irrelevant stimulus than to a novel one) and novel pop-out (NPO: more rapid detection of a novel target on a background of familiar distractors as compared to a novel target with novel distractors). The basic procedure consists of two phases, preexposure and test. The same visual search task is used in both phases. Subjects are required to report whether or not there is a unique target amongst a set of homogeneous distractors. In the relatively simple preexposure phase, the distractor shapes are the same from trial to trial. Similarly, the unique target shape, when present, is the same on all trials. On the other hand, in the more complex test phase, the target can be congruent with its status during preexposure (a target in both phases), or incongruent (a distractor during preexposure becoming a target in test), or the test target can be novel. Similarly, the test distractors can be congruent, incongruent, or novel relative to their status in the preexposure phase. The two-phase procedure, then, allows for a within-subject comparison of visual search response times under easy and difficult conditions. Furthermore, it permits the assessment of LI, which has been found to differentiate between schizophrenics and normals, and between low and high schizotypal normals (see Gray, Feldon, Rawlins, Hemsley, & Smith, 1991 and Lubow & Gewirtz, 1995, for reviews). LI is acquired when a stimulus is repeatedly presented as irrelevant. Under such circumstances, in a subsequent test phase, it becomes difficult for that stimulus to enter into new associations. Typically, the human-LI procedure, like the present visual search procedure, consists of two phases, preexposure and test. In the preexposure phase, the stimulus preexposed group (PE) is presented with a number of trials on each of which appears the same meaningless, irrelevant stimulus. Concurrent with such presentations, the subject engages in a different activity, the masking task. The masking task serves to divert attention from the stimulus which will later be used as the target in the test phase. A second group, non-preexposed (NPE), also engages in the masking task, but in the absence of the to-be-target stimuli. In the test, the masking
44
R. E. LUBOW ET AL.
stimuli continue to be present on each trial, but on any given trial either the preexposed stimulus or a novel stimulus might appear. The subject’s test-task is to associate the presence of the previously irrelevant, preexposed stimulus with some consequence. LI is demonstrated when the PE group reaches a learning criterion more slowly than the NPE group. One explanation of LI holds that, during the preexposure phase, attention to the irrelevant stimulus is reduced (Lubow, 1989; Mackintosh, 1975; Pearce & Hall, 1980; Wagner, 1976). As a result, at the time of test, less attention is allocated to that stimulus than if it had not been preexposed, or it had been preexposed but relevant. The Lubow and Kaplan (1997) procedure captures the basic manipulation of the LI paradigm, but replaces the associative learning test with a more direct measure of attentional involvement, namely visual search performance. In the visual search test phase, the PE condition is created when the previous target becomes the distractor and the previous distractor becomes the target. The NPE condition is achieved when the previous target becomes the distractor and the target is new (having been experienced neither as a target nor a distractor in the preexposure phase). With such conditions, Lubow and Kaplan (1997) found that response time (RT) was faster in the NPE than the PE condition, thereby providing support for an attentional explanation of LI. They also reported a NPO effect (Johnston & Hawley, 1994), in which a new target presented in an array of familiar figures was detected faster than a new target in an all-novel array. The NPO effect, which reflects an interaction between controlled and automatic processing (Johnston & Hawley, 1994), has not yet been investigated as a function of anxiety. The present study, then, compares male and female ADC and control children on a visual search task. On the basis of Lubow and Kaplan’s (1997) results, it was expected that normal controls would exhibit both LI and NPO effects. However, it also was predicted that ADCs would have less LI than normals. This follows from Braunstein-Bercovitz’s (2000) data and her suggestion that attenuated LI in high schizotypal normals (e.g., Braunstein-Bercovitz & Lubow, 1998a, 1998b) is a result of a large anxiety component, a position supported by Lubow and Josman’s (1993) finding that ADHD children also have impaired LI.
METHOD Overview The experiment was divided into two phases. In phase 1, participants performed a visual search task in which the target and distractors retained their identity from trial to trial. Thus, if the target was shape-A, the target was
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
45
shape-A on every trial in which a target appeared. Similarly, if the distractors were composed of B-shapes, all trials contained the B-shape distractors, although sometimes with the A-target and sometimes with no target. Phase 1, then, contained a simple set of targets and distractors. In contrast, phase 2 contained a complex set of targets and distractors. The target could be either the same as that in phase 1, the same as the distractors of phase 1, or novel. Similarly, the distractors could be the same as the distractors of phase 1, the same as the target of phase 1, or novel. All viable combinations of target and distractors yield seven different target-distractor conditions in the test phase. In addition, the test phase had three conditions in which only the distractors were present. Subjects Participants consisted of 46 children and adolescents, 23 diagnosed as anxiety-disordered and 23 as healthy controls. All participants were diagnosed using the Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS; Puig-Antich, Orvaschel, Tabrizi, & Chambers, 1978; Apter, Orvaschel, Laseg, Moses, & Tyano, 1989). In addition, they completed two self-evaluation assessment questionnaires: the Revised Children’s Manifest Anxiety Scale (RCMAS; Reynolds & Richmond, 1978, 1985) and the Children’s Depression Inventory (CDI; Kovacs, 1985). Twenty-three subjects (12 girls and 11 boys: age range, 6–17 years, M ⫽ 11.2, SD ⫽ 2.71) were from the anxiety-disorder outpatient clinic, Tel Aviv Community Mental Health Center. These 23 consecutive-intakes were diagnosed with panic disorder, separation anxiety disorder, overanxious disorder, simple phobia, and/or avoidant disorder. Comorbidity included obsessivecompulsive disorder, tic disorder, depression, enuresis, and learning disabilities. Participants with neurological disorders or with mental retardation were excluded. All subjects were drug naive. The control group consisted of 23 healthy volunteers (11 girls and 12 boys; age range, 6–17 years, M ⫽ 12, SD ⫽ 3.53), without any history of psychopathology. The research protocol was approved by the Institutional Human Investigation committee. All parents signed an informed consent form, and all participants gave their agreement to participate in the study. Apparatus and Stimuli Interactions with the subject were conducted through an IBM-PC with a VGA color monitor. The left and right arrow keys were labeled “0” and “1,” respectively.
46
R. E. LUBOW ET AL.
Fig. 1. A sample screen showing a field of distractors with a unique target.
All stimuli, targets and distractors, were constructed from five randomly connected straight lines generated from a 3 ⫻ 3 matrix, measuring 1.5 ⫻ 1.5 cm (Musen & Treisman, 1990). A given display could either contain 20 identical figures (target-absent) or 19 identical figures plus one unique one (target-present). Figure 1 presents an example of a display frame in which the target was present. Self-Report Measures Revised Children’s Manifest Anxiety Scales (RCMAS). RCMAS consists of 37 items and measures chronic (trait) anxiety. In addition to an 11-item Lie scale (Reynolds & Richmond, 1985), the scale is composed of three anxiety factors, Physiological Symptoms, Worry and Oversensitivity, and Concentration. Each question requires a “yes” or “no” response. The score for each factor is composed of the number of “yes” responses. Reynolds and Paget (1983) describe normative data, as well as reliability and validity scores. Children’s Depression Inventory (CDI). CDI consists of 27 sets of items. Within each set, the subject is required to choose one of three items which best describes his/her situation. The three items are on a scale such that one item indicates a severe symptom, a second moderate, and a third, no symptom. Each set was scored 0, 1, or 2, depending on the item that the subject selected. Total test scores may vary from 0 to 54, with the higher score indicating more severe symptomology. Procedure Psychological tests. The self-report questionnaires for anxiety and depression (RCMAS and CDI) were individually administered within 1 month of participation in the visual search task. For three ADC subjects, scores on the CDI
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
47
were not obtained; for one Control, scores on both the CDI and RCMAS were not obtained. Visual search tasks. Participants, run individually, were seated at normal viewing distance from the computer screen. They were informed that they would see a series of computer displays with many figures on them. On some occasions the figures would all be the same, and on some there would be a unique figure. They were told to press “0” on the keyboard if all of the items were the same, and to press “1” if there was a unique item on the screen. They were asked to work as rapidly and accurately as possible. Thirty practice trials, constructed in a similar manner as those of the test phases, but with different target and distractor stimuli, were administered. In addition, unlike in the subsequent phases, subjects were given feedback for correct and incorrect responses. Preexposure phase. This phase, consisting of 100 trials, immediately followed the practice session. On each trial, display presentation was terminated when the subject responded. A short warning tone was sounded 2.7 seconds after the response. After an additional 300 milliseconds from the tone onset, the next display was presented. All frames contained 20 white figures. The 20 figures were either all the same or one of the 20 was different. On each trial the figures were randomly positioned on the screen within an imaginary 8 ⫻ 12 matrix. There were equal numbers of target-present and target-absent trials. Order of presentation was randomized with the constraint that there not be more than three consecutive trials of the same type. Figure shapes were completely counterbalanced across subjects and status as target or distractor. Test phase. The test phase began immediately after preexposure. Subjects were told that they would have the same task as before, but that now the targets and distractors would change from trial to trial. All subjects were treated identically during the test, with each one receiving 140 trials. Trials were organized into 7 blocks of 20 trials, with each of seven conditions represented twice in each block. The remaining six presentations did not contain a target. They were composed, two each, of the preexposed distractors, preexposed targets, or a novel shape. Within each block, the order of presentation was randomized, with the restriction that three no-target frames could not appear successively. Half the subjects received one random order and half the same order but in reverse. Response time (RT), as measured from the onset of the display to key press, and errors were recorded on every trial. In order to reduce the weights of extreme scores, data analyses were performed on median RTs.
48
R. E. LUBOW ET AL.
TABLE 1 Mean Test Scores (and Standard Deviations) for Scores on the Revised Children’s Manifest Anxiety Scale (RCMAS) and the Children’s Depression Inventory (CDI) for Anxiety Disorder (ADC) and Control Subjects RCMAS factors
ADC M (SD) Control M (SD) p
CDI
1
2
3
Total RCMAS
Lie
8.30 (3.44)
3.87 (1.79)
4.78 (2.04)
1.44 (1.35)
10.56 (1.42)
3.39 (2.33)
4.55 (4.07) ⬍.003
1.41 (2.14) ⬍.001
2.02 (2.24) ⬍.001
0.73 (1.38) .089
4.37 (1.68) ⬍.001
2.82 (2.11) .39
RESULTS Psychological Tests Table 1 presents the mean scores and standard deviations for the CDI and the three factors and lie scale of the RCMAS. On the RCMAS, the groups were significantly different on factors 1 and 2 (physiological symptoms of anxiety and worry/oversensitivity, respectively), but only marginally on factor 3 (concentration). The two groups did not differ on the lie scale. The ADC group scored significantly higher than the control group on the depression inventory. Table 2 presents the intercorrelations (Pearson’s r) between the scores on the subscales of RCMAS and CDI. With the exceptions of the RCMAS Lie scale, all of the scores were significantly correlated with each other. As would be expected, the correlations were stronger within the ADC group than within the control group, .52, .84, .83, and .15, .21, .50, respectively. TABLE 2 Correlation Matrix for Scores on Lie Scale and Three Factors of Revised Children’s Manifest Anxiety Scale (RCMAS) and Children’s Depression Inventory (CDI)
Factor 1 Factor 2 Factor 3 Lie CDI
RCMAS
Factor 1
Factor 2
Factor 3
Lie
.858 .898 .598 .040 .763
.645 .334 ⫺.044 .513
.360 ⫺.017 .679
⫺.046 .701
⫺.101
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
49
Fig. 2. Mean median response times (RT) for the detection of a unique target (T) among distractors and the detection of absence of target (NT) among distractors for Anxiety Disorder (ADC) and Control subjects in the preexposure phase (A) and in the test phase (B).
Visual Search Tasks Preexposure phase. Figure 2 (panel A) displays the mean of the median RTs for ADC and control groups for correct responses on target-trials (T) and notarget-trials (NT). RT scores were assessed with a 2 ⫻ 2 mixed analysis of variance (ANOVA), with the first factor being Diagnosis (ADC vs. control), and the second factor, Trial-type (T vs. NT). As can be seen in Figure 2, both subject groups responded more quickly on T than on NT trials, F(1, 44) ⫽ 134.66, p ⬍ .0001. Although the ADC group responded more slowly than the control group, the difference was not significant, nor was the Diagnosis ⫻ Trial-Type interaction (p values ⬎ .10). Mean percent of errors on T-trials was 7.0; on NT-trials, 2.3. Errors were analyzed with a 2 ⫻ 2 mixed ANOVA with the same factors as in the previous analysis. Again, the only significant main effect was that of Trial-Type, F(1, 44) ⫽ 22.83, p ⬍ .0001, with more errors on T- than NT-trials. Neither Diagnosis main effect nor Diagnosis ⫻ Trial-Type interaction was significant (p values ⬎ .10). Pearson correlations between errors and RTs were calculated separately for T- and NT-trials. On T-trials, the correlation was r ⫽ .45 (p ⫽ .002), indicating that as RTs got longer the errors increased. On NT-trials the correlation was r ⫽ ⫺.02. Test phase. RT and error data were collapsed across all T-trials and also all NT-trials. Mean median RTs (Figure 2, panel B) display the same pattern as in
50
R. E. LUBOW ET AL.
the preexposure phase. A 2 ⫻ 2 mixed ANOVA (T vs. NT; Diagnosis) indicated that RTs for NT-trials were longer than for T-trials, F(1, 44) ⫽ 127.78, p ⬍ .0001. Again, although the ADC group responded more slowly than the control group, the difference was not significant, nor was the Diagnosis ⫻ Trial-Type interaction (p values ⬎ .10). Mean percent errors on T-trials was 10.5; on NT-trials, 4.3. A mixed ANOVA indicated that the only significant main effect was that of Trial-Type, F(1, 44) ⫽ 32.08, p ⬍ .0001. Neither the Diagnosis main effect nor the Diagnosis ⫻ Trial interaction was significant (p values ⬎ .10). Pearson correlations between errors and RTs were calculated separately for T- and NT-trials (r ⫽ .02 and r ⫽ ⫺.24, respectively), and were not significant, indicating the absence of a trade off between speed and errors. Comparing RTs in preexposure and test phase. Two mixed 2 ⫻ 2 ANOVAs, one for T-trials and one for NT-trials, compared performance in preexposure and test phases. The first factor was Phase (PE vs. Test), the second factor was Diagnosis (ADC vs. control). For T-trials, Phase was significant, F(1, 44) ⫽ 56.74, p ⬍ .0001), indicating that responses on T-trials were faster in preexposure than in test. Diagnosis was not significant (p ⬎ .10). The Phase ⫻ Diagnosis interaction, shown in Figure 2, was marginally significant, F(1, 44) ⫽ 3.09, p ⫽ .09); RTs on T-trials were proportionately faster in preexposure than test for the ADC group as compared to the control group. For NT-trials, Phase also was significant, F(1, 44) ⫽ 7.89, p ⫽ .007); RTs on T-trials were faster in preexposure than test. Neither Diagnosis nor the Phase ⫻ Diagnosis interaction was significant (p values ⬎ .10). Selected test phase conditions. As noted, of the seven conditions that contained targets, only three relate to LI and NPO effects (PE, NPE, and NOV). Figure 3 presents the mean median RTs for correct responses for these conditions for ADC and control groups. The LI and NPO effects were analyzed separately, each with a 2 ⫻ 2 ⫻ 2 mixed ANOVA; the first factor was Diagnosis (ADC vs. control), the second factor, gender, and the third, Condition (PE and NPE for LI; NPE and NOV for NPO). LI effects. As can be seen in Figure 3, RTs were longer for the ADC than control group. This difference was significant, F(1, 42) ⫽ 4.20, p⫽.05. RTs were also longer in PE than NPE conditions (LI). The LI effect was significant, F(1, 42) ⫽ 16.97, p ⬍ .001. The same analysis was conducted on groups that were constructed from a median split of RCMAS scores. (In these new groups, six subjects who were diagnosed as ADC scored below the median RCMAS, and four subjects who were in the original control group scored above the median). As would be expected from the previous analysis, there was a main effect of Condition. Although there was a larger difference between PE and NPE conditions (more
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
51
Fig. 3. Mean median response times (RT) for the detection of a unique target among distractors for three different display conditions for Anxiety Disorder (ADC) and Control subjects. PE ⫽ the preexposed target is the test distractor, and the preexposed distractor is the test target; NPE ⫽ the preexposed target is the test distractor, and the test target is novel; NOV ⫽ both the test target and the test distractors are novel). LI is the difference between PE and NPE; Novel popout is the difference between NPE and NOV.
LI) in the low-anxious than in the high-anxious group, the Condition ⫻ Anxiety level interaction was not significant, F(1, 42) ⫽ 2.48, p⫽.12. NPO effects. The NPO effect, faster responding in the NPE than NOV condition, can be seen in Figure 3. The mixed ANOVA indicated that only the main effect of Condition (pop-out) was significant, F(1, 42) ⫽ 8.10, p ⫽ .007, although Diagnosis approached significance, F(1, 42) ⫽ 3.01, p ⫽ .09. None of the interactions was significant (Fs ⬍ 1), except for Condition ⫻ Gender, F(1, 42) ⫽ 5.40, p ⫽ .03, reflecting a large NPO effect in males (NOVNPE ⫽ .48 seconds) and the absence of such an effect in females (NOVNPE ⫽ .04 seconds). As with LI, the NPO data were analyzed after dividing the subjects on the basis of a median split of RCMAS scores. In addition to the main pop-out effect, only the Condition ⫻ Gender interaction was significant, F(1, 42) ⫽ 5.64, p ⫽ .02, again reflecting a larger NPO effect in males than females.
DISCUSSION Self-Report Questionnaires The anxiety-disordered children and adolescents, in addition to being diagnosed as such with the semi-structured K-SADS interview, had significantly more self-reported anxiety than normal controls on two of the three subscales of the RCMAS. They had higher scores on the factor describing physiological
52
R. E. LUBOW ET AL.
symptoms of anxiety and on the factor associated with worry and oversensitivity. The third factor, concentration, was in the same direction, but not statistically significant. Together with the higher self-rated anxiety scores, this group also had elevated scores on the CDI depression questionnaire as compared to the control subjects. Depression and anxiety scores were highly correlated (see Table 2), as would be expected (e.g., Cole, Truglio, & Peeke, 1997; Dalgleish et al., 1997; Delbarrio, Moreno-Rosset, Lopez-Martinez, & Olmedo, 1997). Indeed, the magnitude of the relationship suggests that self-reported anxiety and depression may reflect a single factor (Tannenbaum, Forehand, & Thomas, 1992). Visual Search Tasks Preexposure phase. In the visual search tasks, there were two types of displays, target-present and target-absent. In the preexposure phase, the target was the same shape on all target-present trials. Similarly, the distractors did not vary across preexposure trials. In this relatively simple situation, ADC and control subjects did not differ from each other, neither in RTs nor errors. The only reliable effect was that of target presence versus absence. The longer RTs in the latter compared to the former indicates that, for both groups, the task was performed as a serial, self-terminating search. Test phase. The test phase, consisting of displays constructed from a variety of different target and distractors, was more difficult than the preexposure phase. This was reflected in the generally elevated RTs in test as compared to preexposure (see panels A and B in Figure 2). Nevertheless, the pattern of results in the two phases was the same. Although ADC subjects responded somewhat slower than controls, they were not significantly different from them. As in the preexposure phase, the only reliable difference was between target-present and target-absent conditions. However, a comparison of RTs in preexposure and test phases suggests that ADCs responded relatively more slowly than controls when switching from the simple to more complex task. The Phase ⫻ Diagnosis interaction (p ⫽ .086) is displayed in Figure 2, where RTs on T-trials are proportionately slower in test than in preexposure for the ADC as compared to the control group. This finding, although not statistically significant, is consistent with reports of larger anxiety-control differences on difficult as opposed to simple RT tasks (e.g., Fox, 1994). LI effects. Target search time was slower in the PE than in the NPE test condition. This finding with children is the same as that reported for adults (Lubow & Kaplan, 1997), and it provides additional support to the claim that traditional LI, as indexed in a learning task, is modulated by attentional processes (Lubow, 1989; Lubow & Gewirtz, 1995).
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
53
In spite of the robust LI effect, the predicted difference between ADCs and controls, namely reduced LI in the former compared to the latter, was not obtained. However, when diagnostic classification was disregarded and the subjects were reclassified into low and high anxiety on the basis of RCMAS scores, there was some suggestion, albeit not statistically significant, of the expected pattern. The failure to obtain the predicted effect may be the result of low power from small sample size. Nevertheless, if proven to be reliable, then the finding that high anxiety level attenuates LI would be congruent with data from studies using standard LI tests. Thus, Lubow and Josman (1993) found a loss of LI in hyperactive children, and Braunstein-Bercovitz (2000) reported an attenuation of LI in high- compared to low-anxious adults, as assessed on the Trait-Anxiety Scale (Spielberger, Gorsuch, & Lushene, 1970). The failure to even approach a significant Anxiety ⫻ Condition interaction when anxiety was based on the clinical diagnosis may relate to the construct validity of anxiety as used in K-SADS and RCMAS. As frequently noted (e.g., Cole et al., 1997; Dalgleish et al., 1997; Delbarrio et al., 1997; Tannenbaum et al., 1992), and as indicated by the high correlations between the RCMAS and CDI scores in the present study, clinically diagnosed anxiety and depression may be inseparable. Consequently, the prediction of attenuated LI in the ADC group, which was based on anxiety level, may have been ill-founded. Depression may limit the effects of anxiety on LI. In spite of these qualifications, the similar levels of LI for clinically diagnosed ADCs and controls, suggest that, in the absence of emotionally related stimuli, the two groups do not differ in their processing of irrelevant stimuli. NPO effects. As in the Lubow and Kaplan (1997) study with adults, we obtained an NPO effect; both ADC and control groups found the novel target more quickly in displays containing familiar distractors than in displays in which the distractors were novel. The absence of an interaction between NPO conditions and anxiety level suggests that any differences in LI between low and high anxiety levels would be independent of NPO effects. In addition, if one accepts the position of Johnston and Hawley (1994) that NPO is a function of automatic and controlled processes, these results would suggest that these processes are unimpaired in ADCs. Summary and conclusions. Although the ADC group responded more slowly than controls in phase-two compared to the simpler phase-one search session, these differences were not significant. Similarly, the two groups did not differ in the magnitudes of either LI or NPO. The absence of the predicted differences in LI may be due to the confounding of anxiety and depression within the diagnostic classifications that comprise anxiety disorders, as seen in the self-report questionnaires, where RCMAS anxiety scores and CDI depression scores were highly correlated. In spite of the high correlation with CDI,
54
R. E. LUBOW ET AL.
RCMAS scores may represent a LI-modulating anxiety variable that is less contaminated by other factors than are the anxiety disorder diagnostic categories. In spite of these qualifications, overall, the pattern of results indicates that ADCs and normals do not differ in visual search behavior, nor does such behavior interact with gender. These negative findings reinforce the position that differences between the two groups obtained in selective attention tasks using threatening and neutral words relate to the specific nature of the stimuli and not to general differences in information processing.
REFERENCES Apter, A., Orvaschel, H., Laseg, M., Moses, T., & Tyano, S. (1989). Psychometric properties of the K-SADS-P in an Israeli adolescent in-patient population. Journal of the American Academy of Child and Adolescent Psychiatry, 28, 61–65. Bernstein, G. A., & Borchardt, C. M. (1991). Anxiety disorders of childhood and adolescence. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 519–532. Braunstein-Bercovitz, H. (2000). Is the attentional dysfunction in schizotypy related to anxiety? Schizophrenia Research, in press. Braunstein-Bercovitz, H., & Lubow, R. E. (1998a). Latent inhibition as a function of attentional decline to the irrelevant stimulus during preexposure. Learning and Motivation, 29, 261–279. Braunstein-Bercovitz, H., & Lubow, R. E. (1998b). Are high-schizotypal normal participants distractible or limited in attentional resources? A study of latent inhibition as a function of masking task load and schizotypy level. Journal of Abnormal Psychology, 107, 659–670. Cole, D. A., Truglio, R., & Peeke, L. (1997). Relation between symptoms of anxiety and depression in children: A multitrait-multimethod-multigroup assessment. Journal of Consulting and Clinical Psychology, 65, 110–119. Dalgleish, T., Taghavi, R., Neshat-Doost, H., Moradi, A., Yule, W., & Canterbury, R. (1997). Information processing in clinically depressed and anxious children and adolescents. Journal of Child Psychology and Psychiatry, 38, 535–541. Delbarrio, V., Moreno-Rosset, C., Lopez-Martinez, R., & Olmedo, M. (1997). Anxiety, depression, and personality structure. Personality and Individual Differences, 23, 327–335. Doost, H. T. N., Taghavi, M. R., Moradi, A. R., Yule, W., & Dalgleish, T. (1997). The performance of clinically depressed children and adolescents on the modified Stroop paradigm. Personality and Individual Differences, 23, 753–759. Eysenck, M. W., MacLeod, C., & Mathews, A. (1987). Cognitive functioning and anxiety. Psychological Research, 49, 189–195. Fox, E. (1994). Attentional bias in anxiety: A defective inhibition hypothesis. Cognition and Emotion, 8, 165–195. Gray, J. A., Feldon, J., Rawlins, J. N. P., Hemsley, D. R., & Smith, A. D. (1991). The neuropsychology of schizophrenia. Behavioral and Brain Sciences, 14, 1–20. Johnston, W. A., & Hawley, K. J. (1994). Perceptual inhibition of expected inputs: The key that opens closed minds. Psychonomic Bulletin and Review, 1, 56–72. Kendall, P. C., & Chansky, T. E. (1991). Considering cognition in anxiety-disordered children. Journal of Anxiety Disorders, 5, 167–185. Kendall, P. C., & Brady, E. U. (1995). Comorbidity in the anxiety disorders of childhood. In K. Craig & K. Dobson (Eds.), Anxiety and depression in adults and children (pp. 3–36). Thousand Oaks, CA: Sage.
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
55
Kindt, M., Brosschot, J. F., & Everaerd, W. (1997). Cognitive processing bias of children in a real life stress situation and a neutral situation. Journal of Experimental Child Psychology, 64, 79–97. Kovacs, M. (1985). The natural history and course of depressive disorders in childhood. Psychiatric Annals, 15, 387–389. Lubow, R. E. (1989). Latent inhibition and conditioned attention theory. New York: Cambridge University Press. Lubow, R. E., Dressler R., & Kaplan, O. (1999). The effects of target and distractor familiarity on visual search in de novo Parkinson’s disease patients: Latent inhibition and novel pop-out. Neuropsychology, 13, 415–423. Lubow, R. E., & Gewirtz, J. C. (1995). Latent inhibition in humans: Data, theory, and implications for schizophrenia. Psychological Bulletin, 117, 87–103. Lubow, R. E., & Josman, Z. E. (1993). Latent inhibition deficits in hyperactive children. Journal of Child Psychiatry and Psychology, 34, 959–973. Lubow, R. E., & Kaplan, O. (1997). Visual search as a function of the type of prior experience with target and distractors. Journal of Experimental Psychology: Human Perception and Performance, 23, 14–24. Lubow, R. E., Kaplan, O., Abramovich, P., Rudnick, A., & Laor, N. (2000). Visual search in schizophrenics: Latent inhibition and novel pop-out effects. Schizophrenia Research, in press. Mackintosh, N. J. (1975). A theory of attention: Variations in the associability of stimuli with reinforcement. Psychological Review, 82, 276–298. Martin, M., Horder, P., & Jones, G. V. (1992). Integral bias in naming of phobia-related words. Cognition and Emotion, 6, 479–486. Mogg, K., Bradley, B. P., & Williams, R. (1995). Attentional bias in anxiety and depression: The role of awareness. British Journal of Clinical Psychology, 34, 17–36. Mogg, K., Bradley, B. P., Williams, R., & Mathews, A. (1993). Subliminal processing of emotional information in anxiety and depression. Journal of Abnormal Psychology, 102, 304–311. Musen, G., & Treisman, A. (1990). Implicit and explicit memory for visual patterns. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 127–137. Ollendick, T. H., King, N. J., & Yule, W. (Eds.). (1994). International handbook of phobic and anxiety disorders in children and adolescents. New York: Plenum Press. Pearce, J. M., & Hall, G. (1980). A model of Pavlovian learning: Variations in the effectiveness of conditioned but not unconditioned stimuli. Psychological Review, 87, 532–552. Puig-Antich, J., Orvaschel, H., Tabrizi, R. N., & Chambers, W. J. (1978). The Schedule for Affective Disorders and Schizophrenia for School-Aged Children. New York: New York State Psychiatric Institute. Reynolds, C. R., & Paget, K. D. (1983). National normative and reliability data for the Revised Children’s Manifest Anxiety Scale. School Psychology Review, 12, 324–336. Reynolds, C. R., & Richmond, B.O. (1978). What I think and feel: A revised measure of children’s manifest anxiety. Journal of Abnormal Child Psychology, 6, 271–280. Reynolds, C. R., & Richmond, B. O. (1985). Revised children’s manifest anxiety scale. Los Angeles: Western Psychological Service. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety Inventory (self-evaluation questionnaire). Palo Alto, CA: Consulting Psychologists Press. Tannenbaum, L. E., Forehand, R., & Thomas, A. M. (1992). Adolescent self-reported anxiety and depression: Separate constructs or a single entity. Child Study Journal, 22, 61–72. Vasey, M. W., Daleiden, E. L., Williams, L. L., & Brown, L. M. (1995). Biased attention in childhood anxiety disorders: A preliminary study. Journal of Abnormal Child Psychology, 23, 267– 279. Vasey, M. W., El-Hag, N., & Daleiden, E. L. (1996). Anxiety and the processing of emotionally threatening stimuli: Distinctive patterns of selective attention among high- and low-test-anxious children. Child Development, 67, 1173–1185.
56
R. E. LUBOW ET AL.
Wagner, A. R. (1976). Priming in STM: An information processing mechanism for self-generated or retrieval-generated depression in performance. In T. Tighe & R. N. Leaton (Eds.), Habituation: Perspectives from child development, animal behavior, and neurophysiology (pp. 95–128). Hillsdale, NJ: Lawrence Erlbaum Associates. Werry, J. S. (1991). Overanxious disorder: A review of its taxonomic properties. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 533–544.
Journal of Anxiety Disorders, Vol. 14, No. 1, pp. 41–56, 2000 Copyright 2000 Elsevier Science Ltd Printed in the USA. All rights reserved 0887-6185/00 $–see front matter
PII S0887-6185(99)00038-9
The Effects of Target and Distractor Familiarity on Visual Search in Anxious Children: Latent Inhibition and Novel Pop-Out R. E. Lubow, Ph.D. Tel Aviv University, Tel Aviv, Israel
Paz Toren, M.D., and N. Laor, M.D., Ph.D. Tel Aviv Community Mental Health Center, Tel Aviv, Israel
Oren Kaplan, Ph.D. Tel Aviv University, Tel Aviv, Israel
Abstract—Children and adolescents (ages 6–17 years) diagnosed as having an anxiety disorder were compared to matched controls on a two-stage serial visual search task in which they identified presence or absence of a unique shape presented with homogeneous distractors. Response time was examined as a function of prior experience with either target, distractor, or both, allowing for a within-subject assessment of latent inhibition (LI: slower responding to a target that was formerly a distractor against a background of distractors that were formerly targets as compared to a novel target with distractors that were formerly targets) and novel pop-out effects (NPO: faster responding to a novel target against a background of familiar former targets as compared to the condition in which both the target and distractors were novel). There were robust LI and NPO effects for both anxious and control children. However, the predicted interaction between diagnosis and LI condition was not obtained. In general, the results suggest that children with diagnosed anxiety disorder do not differ from controls on basic information processing as assessed by this visual search task. 2000 Elsevier Science Ltd. All rights reserved.
The authors wish to thank Michal Sadeh and Osnat Shor for assistance in data collection, and Armonit Roter for helpful suggestions on an earlier draft of this paper. Requests for reprints should be sent to R. E. Lubow, Department of Psychology, Tel Aviv University, Ramat Aviv 69978, Israel. E-mail: [email protected]
41
42
R. E. LUBOW ET AL.
Keywords: Latent inhibition; Anxiety; Visual search
The relationships between behavioral/physiological pathologies and cognitive processes have been the subject of much research in a variety of areas (e.g., Parkinson’s disease, schizophrenia). Such research adds to an understanding of causes and/or consequences of particular disorders, as well as to knowledge of the cognitive process used in the specific research paradigm. With some exceptions, to be discussed below, research on cognitive processing in anxiety-disordered children (ADC) has been surprisingly limited, perhaps because ADC is not a clearly defined nosological category, but rather consists of a number of subclassifications (e.g., panic, separation anxiety, posttraumatic stress, and obsessive-compulsive disorders; see Kendall & Brady, 1995; Ollendick, King, & Yule, 1994), each with a high level of comorbidity (Bernstein & Borchardt, 1991; Werry, 1991). Nevertheless, childhood anxiety disorders, irrespective of their subclassifications, share at least one common denominator, namely anxiety itself. Consequently, such children provide a convenient sample for examining the behavioral components of anxiety. Against this background, it is somewhat surprising that, “. . . there are few studies that have assessed cognition in anxious children and there are no studies of diagnosed samples” (Kendall & Chansky, 1991, p. 168). In the 9 years since Kendall and Chansky’s assessment, the situation has improved, at least partially. One topic that has been the subject of research concerns the effects of anxiety on selective attention to threat-related as compared to neutral words. Studies with children, using either the emotional-Stroop paradigm (Kindt, Brosschot, & Everaerd, 1997; Martin, Horder, & Jones, 1992) or the probedetection task (e.g., Vasey, Daleiden, Williams, & Brown, 1995; Vasey, ElHag, & Daleiden, 1996), indicate that high-anxious children selectively attend to threat-related words. Although these data conform to those obtained from adults (e.g., Eysenck, MacLeod, & Mathews, 1987; Mogg, Bradley, Williams, & Mathews, 1993; Mogg, Bradley, & Williams, 1995), there are some qualifications. In the children’s literature, for example, Doost, Taghavi, Moradi, Yule, and Dalgleish (1997) failed to find biased processing of threatrelated words in a mixed anxiety-depression group as compared to normal controls. On the other hand, Kindt et al. (1997) did report a processing bias for threatening words, but only for girls, both low and high anxious. A gender effect was also found by Vasey et al. (1995, 1996). Only low-anxious males exhibited an attentional bias away from threatening word stimuli. All of the studies cited above used word stimuli. The lack of consistency may derive, in part, from the use of such verbal materials. Words may not be equally threatening or neutral for male and female participants. In addition, Stroop and dot-probe tasks may be less sensitive to the effects of anxiety than other tests of attention. Thus, with adults, Fox (1994) found that high-trait
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
43
anxious subjects had difficulty inhibiting distracting information in a visual search task but not in a Stroop task in which the positions of targets and distractors were constant. To help resolve these inconsistencies, we tested low and high anxious male and female children on a visual search task with completely neutral targets and distractors (nonsense figures). The specific procedure, based on a study by Lubow and Kaplan (1997, Exp. 2), has been used to explore the relationship between pathology and attention. It has been applied to schizophrenics (Lubow, Kaplan, Abramovich, Rudnick, & Laor, 2000) and Parkinson’s disease patients (Lubow, Dressler, & Kaplan, 1999), as well as normal adults (Lubow & Kaplan, 1997). In addition to evaluating standard visual search behavior, the procedure allows for the assessment of several attentionally related phenomena, including latent inhibition (LI: slower learning to a previously irrelevant stimulus than to a novel one) and novel pop-out (NPO: more rapid detection of a novel target on a background of familiar distractors as compared to a novel target with novel distractors). The basic procedure consists of two phases, preexposure and test. The same visual search task is used in both phases. Subjects are required to report whether or not there is a unique target amongst a set of homogeneous distractors. In the relatively simple preexposure phase, the distractor shapes are the same from trial to trial. Similarly, the unique target shape, when present, is the same on all trials. On the other hand, in the more complex test phase, the target can be congruent with its status during preexposure (a target in both phases), or incongruent (a distractor during preexposure becoming a target in test), or the test target can be novel. Similarly, the test distractors can be congruent, incongruent, or novel relative to their status in the preexposure phase. The two-phase procedure, then, allows for a within-subject comparison of visual search response times under easy and difficult conditions. Furthermore, it permits the assessment of LI, which has been found to differentiate between schizophrenics and normals, and between low and high schizotypal normals (see Gray, Feldon, Rawlins, Hemsley, & Smith, 1991 and Lubow & Gewirtz, 1995, for reviews). LI is acquired when a stimulus is repeatedly presented as irrelevant. Under such circumstances, in a subsequent test phase, it becomes difficult for that stimulus to enter into new associations. Typically, the human-LI procedure, like the present visual search procedure, consists of two phases, preexposure and test. In the preexposure phase, the stimulus preexposed group (PE) is presented with a number of trials on each of which appears the same meaningless, irrelevant stimulus. Concurrent with such presentations, the subject engages in a different activity, the masking task. The masking task serves to divert attention from the stimulus which will later be used as the target in the test phase. A second group, non-preexposed (NPE), also engages in the masking task, but in the absence of the to-be-target stimuli. In the test, the masking
44
R. E. LUBOW ET AL.
stimuli continue to be present on each trial, but on any given trial either the preexposed stimulus or a novel stimulus might appear. The subject’s test-task is to associate the presence of the previously irrelevant, preexposed stimulus with some consequence. LI is demonstrated when the PE group reaches a learning criterion more slowly than the NPE group. One explanation of LI holds that, during the preexposure phase, attention to the irrelevant stimulus is reduced (Lubow, 1989; Mackintosh, 1975; Pearce & Hall, 1980; Wagner, 1976). As a result, at the time of test, less attention is allocated to that stimulus than if it had not been preexposed, or it had been preexposed but relevant. The Lubow and Kaplan (1997) procedure captures the basic manipulation of the LI paradigm, but replaces the associative learning test with a more direct measure of attentional involvement, namely visual search performance. In the visual search test phase, the PE condition is created when the previous target becomes the distractor and the previous distractor becomes the target. The NPE condition is achieved when the previous target becomes the distractor and the target is new (having been experienced neither as a target nor a distractor in the preexposure phase). With such conditions, Lubow and Kaplan (1997) found that response time (RT) was faster in the NPE than the PE condition, thereby providing support for an attentional explanation of LI. They also reported a NPO effect (Johnston & Hawley, 1994), in which a new target presented in an array of familiar figures was detected faster than a new target in an all-novel array. The NPO effect, which reflects an interaction between controlled and automatic processing (Johnston & Hawley, 1994), has not yet been investigated as a function of anxiety. The present study, then, compares male and female ADC and control children on a visual search task. On the basis of Lubow and Kaplan’s (1997) results, it was expected that normal controls would exhibit both LI and NPO effects. However, it also was predicted that ADCs would have less LI than normals. This follows from Braunstein-Bercovitz’s (2000) data and her suggestion that attenuated LI in high schizotypal normals (e.g., Braunstein-Bercovitz & Lubow, 1998a, 1998b) is a result of a large anxiety component, a position supported by Lubow and Josman’s (1993) finding that ADHD children also have impaired LI.
METHOD Overview The experiment was divided into two phases. In phase 1, participants performed a visual search task in which the target and distractors retained their identity from trial to trial. Thus, if the target was shape-A, the target was
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
45
shape-A on every trial in which a target appeared. Similarly, if the distractors were composed of B-shapes, all trials contained the B-shape distractors, although sometimes with the A-target and sometimes with no target. Phase 1, then, contained a simple set of targets and distractors. In contrast, phase 2 contained a complex set of targets and distractors. The target could be either the same as that in phase 1, the same as the distractors of phase 1, or novel. Similarly, the distractors could be the same as the distractors of phase 1, the same as the target of phase 1, or novel. All viable combinations of target and distractors yield seven different target-distractor conditions in the test phase. In addition, the test phase had three conditions in which only the distractors were present. Subjects Participants consisted of 46 children and adolescents, 23 diagnosed as anxiety-disordered and 23 as healthy controls. All participants were diagnosed using the Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS; Puig-Antich, Orvaschel, Tabrizi, & Chambers, 1978; Apter, Orvaschel, Laseg, Moses, & Tyano, 1989). In addition, they completed two self-evaluation assessment questionnaires: the Revised Children’s Manifest Anxiety Scale (RCMAS; Reynolds & Richmond, 1978, 1985) and the Children’s Depression Inventory (CDI; Kovacs, 1985). Twenty-three subjects (12 girls and 11 boys: age range, 6–17 years, M ⫽ 11.2, SD ⫽ 2.71) were from the anxiety-disorder outpatient clinic, Tel Aviv Community Mental Health Center. These 23 consecutive-intakes were diagnosed with panic disorder, separation anxiety disorder, overanxious disorder, simple phobia, and/or avoidant disorder. Comorbidity included obsessivecompulsive disorder, tic disorder, depression, enuresis, and learning disabilities. Participants with neurological disorders or with mental retardation were excluded. All subjects were drug naive. The control group consisted of 23 healthy volunteers (11 girls and 12 boys; age range, 6–17 years, M ⫽ 12, SD ⫽ 3.53), without any history of psychopathology. The research protocol was approved by the Institutional Human Investigation committee. All parents signed an informed consent form, and all participants gave their agreement to participate in the study. Apparatus and Stimuli Interactions with the subject were conducted through an IBM-PC with a VGA color monitor. The left and right arrow keys were labeled “0” and “1,” respectively.
46
R. E. LUBOW ET AL.
Fig. 1. A sample screen showing a field of distractors with a unique target.
All stimuli, targets and distractors, were constructed from five randomly connected straight lines generated from a 3 ⫻ 3 matrix, measuring 1.5 ⫻ 1.5 cm (Musen & Treisman, 1990). A given display could either contain 20 identical figures (target-absent) or 19 identical figures plus one unique one (target-present). Figure 1 presents an example of a display frame in which the target was present. Self-Report Measures Revised Children’s Manifest Anxiety Scales (RCMAS). RCMAS consists of 37 items and measures chronic (trait) anxiety. In addition to an 11-item Lie scale (Reynolds & Richmond, 1985), the scale is composed of three anxiety factors, Physiological Symptoms, Worry and Oversensitivity, and Concentration. Each question requires a “yes” or “no” response. The score for each factor is composed of the number of “yes” responses. Reynolds and Paget (1983) describe normative data, as well as reliability and validity scores. Children’s Depression Inventory (CDI). CDI consists of 27 sets of items. Within each set, the subject is required to choose one of three items which best describes his/her situation. The three items are on a scale such that one item indicates a severe symptom, a second moderate, and a third, no symptom. Each set was scored 0, 1, or 2, depending on the item that the subject selected. Total test scores may vary from 0 to 54, with the higher score indicating more severe symptomology. Procedure Psychological tests. The self-report questionnaires for anxiety and depression (RCMAS and CDI) were individually administered within 1 month of participation in the visual search task. For three ADC subjects, scores on the CDI
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
47
were not obtained; for one Control, scores on both the CDI and RCMAS were not obtained. Visual search tasks. Participants, run individually, were seated at normal viewing distance from the computer screen. They were informed that they would see a series of computer displays with many figures on them. On some occasions the figures would all be the same, and on some there would be a unique figure. They were told to press “0” on the keyboard if all of the items were the same, and to press “1” if there was a unique item on the screen. They were asked to work as rapidly and accurately as possible. Thirty practice trials, constructed in a similar manner as those of the test phases, but with different target and distractor stimuli, were administered. In addition, unlike in the subsequent phases, subjects were given feedback for correct and incorrect responses. Preexposure phase. This phase, consisting of 100 trials, immediately followed the practice session. On each trial, display presentation was terminated when the subject responded. A short warning tone was sounded 2.7 seconds after the response. After an additional 300 milliseconds from the tone onset, the next display was presented. All frames contained 20 white figures. The 20 figures were either all the same or one of the 20 was different. On each trial the figures were randomly positioned on the screen within an imaginary 8 ⫻ 12 matrix. There were equal numbers of target-present and target-absent trials. Order of presentation was randomized with the constraint that there not be more than three consecutive trials of the same type. Figure shapes were completely counterbalanced across subjects and status as target or distractor. Test phase. The test phase began immediately after preexposure. Subjects were told that they would have the same task as before, but that now the targets and distractors would change from trial to trial. All subjects were treated identically during the test, with each one receiving 140 trials. Trials were organized into 7 blocks of 20 trials, with each of seven conditions represented twice in each block. The remaining six presentations did not contain a target. They were composed, two each, of the preexposed distractors, preexposed targets, or a novel shape. Within each block, the order of presentation was randomized, with the restriction that three no-target frames could not appear successively. Half the subjects received one random order and half the same order but in reverse. Response time (RT), as measured from the onset of the display to key press, and errors were recorded on every trial. In order to reduce the weights of extreme scores, data analyses were performed on median RTs.
48
R. E. LUBOW ET AL.
TABLE 1 Mean Test Scores (and Standard Deviations) for Scores on the Revised Children’s Manifest Anxiety Scale (RCMAS) and the Children’s Depression Inventory (CDI) for Anxiety Disorder (ADC) and Control Subjects RCMAS factors
ADC M (SD) Control M (SD) p
CDI
1
2
3
Total RCMAS
Lie
8.30 (3.44)
3.87 (1.79)
4.78 (2.04)
1.44 (1.35)
10.56 (1.42)
3.39 (2.33)
4.55 (4.07) ⬍.003
1.41 (2.14) ⬍.001
2.02 (2.24) ⬍.001
0.73 (1.38) .089
4.37 (1.68) ⬍.001
2.82 (2.11) .39
RESULTS Psychological Tests Table 1 presents the mean scores and standard deviations for the CDI and the three factors and lie scale of the RCMAS. On the RCMAS, the groups were significantly different on factors 1 and 2 (physiological symptoms of anxiety and worry/oversensitivity, respectively), but only marginally on factor 3 (concentration). The two groups did not differ on the lie scale. The ADC group scored significantly higher than the control group on the depression inventory. Table 2 presents the intercorrelations (Pearson’s r) between the scores on the subscales of RCMAS and CDI. With the exceptions of the RCMAS Lie scale, all of the scores were significantly correlated with each other. As would be expected, the correlations were stronger within the ADC group than within the control group, .52, .84, .83, and .15, .21, .50, respectively. TABLE 2 Correlation Matrix for Scores on Lie Scale and Three Factors of Revised Children’s Manifest Anxiety Scale (RCMAS) and Children’s Depression Inventory (CDI)
Factor 1 Factor 2 Factor 3 Lie CDI
RCMAS
Factor 1
Factor 2
Factor 3
Lie
.858 .898 .598 .040 .763
.645 .334 ⫺.044 .513
.360 ⫺.017 .679
⫺.046 .701
⫺.101
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
49
Fig. 2. Mean median response times (RT) for the detection of a unique target (T) among distractors and the detection of absence of target (NT) among distractors for Anxiety Disorder (ADC) and Control subjects in the preexposure phase (A) and in the test phase (B).
Visual Search Tasks Preexposure phase. Figure 2 (panel A) displays the mean of the median RTs for ADC and control groups for correct responses on target-trials (T) and notarget-trials (NT). RT scores were assessed with a 2 ⫻ 2 mixed analysis of variance (ANOVA), with the first factor being Diagnosis (ADC vs. control), and the second factor, Trial-type (T vs. NT). As can be seen in Figure 2, both subject groups responded more quickly on T than on NT trials, F(1, 44) ⫽ 134.66, p ⬍ .0001. Although the ADC group responded more slowly than the control group, the difference was not significant, nor was the Diagnosis ⫻ Trial-Type interaction (p values ⬎ .10). Mean percent of errors on T-trials was 7.0; on NT-trials, 2.3. Errors were analyzed with a 2 ⫻ 2 mixed ANOVA with the same factors as in the previous analysis. Again, the only significant main effect was that of Trial-Type, F(1, 44) ⫽ 22.83, p ⬍ .0001, with more errors on T- than NT-trials. Neither Diagnosis main effect nor Diagnosis ⫻ Trial-Type interaction was significant (p values ⬎ .10). Pearson correlations between errors and RTs were calculated separately for T- and NT-trials. On T-trials, the correlation was r ⫽ .45 (p ⫽ .002), indicating that as RTs got longer the errors increased. On NT-trials the correlation was r ⫽ ⫺.02. Test phase. RT and error data were collapsed across all T-trials and also all NT-trials. Mean median RTs (Figure 2, panel B) display the same pattern as in
50
R. E. LUBOW ET AL.
the preexposure phase. A 2 ⫻ 2 mixed ANOVA (T vs. NT; Diagnosis) indicated that RTs for NT-trials were longer than for T-trials, F(1, 44) ⫽ 127.78, p ⬍ .0001. Again, although the ADC group responded more slowly than the control group, the difference was not significant, nor was the Diagnosis ⫻ Trial-Type interaction (p values ⬎ .10). Mean percent errors on T-trials was 10.5; on NT-trials, 4.3. A mixed ANOVA indicated that the only significant main effect was that of Trial-Type, F(1, 44) ⫽ 32.08, p ⬍ .0001. Neither the Diagnosis main effect nor the Diagnosis ⫻ Trial interaction was significant (p values ⬎ .10). Pearson correlations between errors and RTs were calculated separately for T- and NT-trials (r ⫽ .02 and r ⫽ ⫺.24, respectively), and were not significant, indicating the absence of a trade off between speed and errors. Comparing RTs in preexposure and test phase. Two mixed 2 ⫻ 2 ANOVAs, one for T-trials and one for NT-trials, compared performance in preexposure and test phases. The first factor was Phase (PE vs. Test), the second factor was Diagnosis (ADC vs. control). For T-trials, Phase was significant, F(1, 44) ⫽ 56.74, p ⬍ .0001), indicating that responses on T-trials were faster in preexposure than in test. Diagnosis was not significant (p ⬎ .10). The Phase ⫻ Diagnosis interaction, shown in Figure 2, was marginally significant, F(1, 44) ⫽ 3.09, p ⫽ .09); RTs on T-trials were proportionately faster in preexposure than test for the ADC group as compared to the control group. For NT-trials, Phase also was significant, F(1, 44) ⫽ 7.89, p ⫽ .007); RTs on T-trials were faster in preexposure than test. Neither Diagnosis nor the Phase ⫻ Diagnosis interaction was significant (p values ⬎ .10). Selected test phase conditions. As noted, of the seven conditions that contained targets, only three relate to LI and NPO effects (PE, NPE, and NOV). Figure 3 presents the mean median RTs for correct responses for these conditions for ADC and control groups. The LI and NPO effects were analyzed separately, each with a 2 ⫻ 2 ⫻ 2 mixed ANOVA; the first factor was Diagnosis (ADC vs. control), the second factor, gender, and the third, Condition (PE and NPE for LI; NPE and NOV for NPO). LI effects. As can be seen in Figure 3, RTs were longer for the ADC than control group. This difference was significant, F(1, 42) ⫽ 4.20, p⫽.05. RTs were also longer in PE than NPE conditions (LI). The LI effect was significant, F(1, 42) ⫽ 16.97, p ⬍ .001. The same analysis was conducted on groups that were constructed from a median split of RCMAS scores. (In these new groups, six subjects who were diagnosed as ADC scored below the median RCMAS, and four subjects who were in the original control group scored above the median). As would be expected from the previous analysis, there was a main effect of Condition. Although there was a larger difference between PE and NPE conditions (more
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
51
Fig. 3. Mean median response times (RT) for the detection of a unique target among distractors for three different display conditions for Anxiety Disorder (ADC) and Control subjects. PE ⫽ the preexposed target is the test distractor, and the preexposed distractor is the test target; NPE ⫽ the preexposed target is the test distractor, and the test target is novel; NOV ⫽ both the test target and the test distractors are novel). LI is the difference between PE and NPE; Novel popout is the difference between NPE and NOV.
LI) in the low-anxious than in the high-anxious group, the Condition ⫻ Anxiety level interaction was not significant, F(1, 42) ⫽ 2.48, p⫽.12. NPO effects. The NPO effect, faster responding in the NPE than NOV condition, can be seen in Figure 3. The mixed ANOVA indicated that only the main effect of Condition (pop-out) was significant, F(1, 42) ⫽ 8.10, p ⫽ .007, although Diagnosis approached significance, F(1, 42) ⫽ 3.01, p ⫽ .09. None of the interactions was significant (Fs ⬍ 1), except for Condition ⫻ Gender, F(1, 42) ⫽ 5.40, p ⫽ .03, reflecting a large NPO effect in males (NOVNPE ⫽ .48 seconds) and the absence of such an effect in females (NOVNPE ⫽ .04 seconds). As with LI, the NPO data were analyzed after dividing the subjects on the basis of a median split of RCMAS scores. In addition to the main pop-out effect, only the Condition ⫻ Gender interaction was significant, F(1, 42) ⫽ 5.64, p ⫽ .02, again reflecting a larger NPO effect in males than females.
DISCUSSION Self-Report Questionnaires The anxiety-disordered children and adolescents, in addition to being diagnosed as such with the semi-structured K-SADS interview, had significantly more self-reported anxiety than normal controls on two of the three subscales of the RCMAS. They had higher scores on the factor describing physiological
52
R. E. LUBOW ET AL.
symptoms of anxiety and on the factor associated with worry and oversensitivity. The third factor, concentration, was in the same direction, but not statistically significant. Together with the higher self-rated anxiety scores, this group also had elevated scores on the CDI depression questionnaire as compared to the control subjects. Depression and anxiety scores were highly correlated (see Table 2), as would be expected (e.g., Cole, Truglio, & Peeke, 1997; Dalgleish et al., 1997; Delbarrio, Moreno-Rosset, Lopez-Martinez, & Olmedo, 1997). Indeed, the magnitude of the relationship suggests that self-reported anxiety and depression may reflect a single factor (Tannenbaum, Forehand, & Thomas, 1992). Visual Search Tasks Preexposure phase. In the visual search tasks, there were two types of displays, target-present and target-absent. In the preexposure phase, the target was the same shape on all target-present trials. Similarly, the distractors did not vary across preexposure trials. In this relatively simple situation, ADC and control subjects did not differ from each other, neither in RTs nor errors. The only reliable effect was that of target presence versus absence. The longer RTs in the latter compared to the former indicates that, for both groups, the task was performed as a serial, self-terminating search. Test phase. The test phase, consisting of displays constructed from a variety of different target and distractors, was more difficult than the preexposure phase. This was reflected in the generally elevated RTs in test as compared to preexposure (see panels A and B in Figure 2). Nevertheless, the pattern of results in the two phases was the same. Although ADC subjects responded somewhat slower than controls, they were not significantly different from them. As in the preexposure phase, the only reliable difference was between target-present and target-absent conditions. However, a comparison of RTs in preexposure and test phases suggests that ADCs responded relatively more slowly than controls when switching from the simple to more complex task. The Phase ⫻ Diagnosis interaction (p ⫽ .086) is displayed in Figure 2, where RTs on T-trials are proportionately slower in test than in preexposure for the ADC as compared to the control group. This finding, although not statistically significant, is consistent with reports of larger anxiety-control differences on difficult as opposed to simple RT tasks (e.g., Fox, 1994). LI effects. Target search time was slower in the PE than in the NPE test condition. This finding with children is the same as that reported for adults (Lubow & Kaplan, 1997), and it provides additional support to the claim that traditional LI, as indexed in a learning task, is modulated by attentional processes (Lubow, 1989; Lubow & Gewirtz, 1995).
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
53
In spite of the robust LI effect, the predicted difference between ADCs and controls, namely reduced LI in the former compared to the latter, was not obtained. However, when diagnostic classification was disregarded and the subjects were reclassified into low and high anxiety on the basis of RCMAS scores, there was some suggestion, albeit not statistically significant, of the expected pattern. The failure to obtain the predicted effect may be the result of low power from small sample size. Nevertheless, if proven to be reliable, then the finding that high anxiety level attenuates LI would be congruent with data from studies using standard LI tests. Thus, Lubow and Josman (1993) found a loss of LI in hyperactive children, and Braunstein-Bercovitz (2000) reported an attenuation of LI in high- compared to low-anxious adults, as assessed on the Trait-Anxiety Scale (Spielberger, Gorsuch, & Lushene, 1970). The failure to even approach a significant Anxiety ⫻ Condition interaction when anxiety was based on the clinical diagnosis may relate to the construct validity of anxiety as used in K-SADS and RCMAS. As frequently noted (e.g., Cole et al., 1997; Dalgleish et al., 1997; Delbarrio et al., 1997; Tannenbaum et al., 1992), and as indicated by the high correlations between the RCMAS and CDI scores in the present study, clinically diagnosed anxiety and depression may be inseparable. Consequently, the prediction of attenuated LI in the ADC group, which was based on anxiety level, may have been ill-founded. Depression may limit the effects of anxiety on LI. In spite of these qualifications, the similar levels of LI for clinically diagnosed ADCs and controls, suggest that, in the absence of emotionally related stimuli, the two groups do not differ in their processing of irrelevant stimuli. NPO effects. As in the Lubow and Kaplan (1997) study with adults, we obtained an NPO effect; both ADC and control groups found the novel target more quickly in displays containing familiar distractors than in displays in which the distractors were novel. The absence of an interaction between NPO conditions and anxiety level suggests that any differences in LI between low and high anxiety levels would be independent of NPO effects. In addition, if one accepts the position of Johnston and Hawley (1994) that NPO is a function of automatic and controlled processes, these results would suggest that these processes are unimpaired in ADCs. Summary and conclusions. Although the ADC group responded more slowly than controls in phase-two compared to the simpler phase-one search session, these differences were not significant. Similarly, the two groups did not differ in the magnitudes of either LI or NPO. The absence of the predicted differences in LI may be due to the confounding of anxiety and depression within the diagnostic classifications that comprise anxiety disorders, as seen in the self-report questionnaires, where RCMAS anxiety scores and CDI depression scores were highly correlated. In spite of the high correlation with CDI,
54
R. E. LUBOW ET AL.
RCMAS scores may represent a LI-modulating anxiety variable that is less contaminated by other factors than are the anxiety disorder diagnostic categories. In spite of these qualifications, overall, the pattern of results indicates that ADCs and normals do not differ in visual search behavior, nor does such behavior interact with gender. These negative findings reinforce the position that differences between the two groups obtained in selective attention tasks using threatening and neutral words relate to the specific nature of the stimuli and not to general differences in information processing.
REFERENCES Apter, A., Orvaschel, H., Laseg, M., Moses, T., & Tyano, S. (1989). Psychometric properties of the K-SADS-P in an Israeli adolescent in-patient population. Journal of the American Academy of Child and Adolescent Psychiatry, 28, 61–65. Bernstein, G. A., & Borchardt, C. M. (1991). Anxiety disorders of childhood and adolescence. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 519–532. Braunstein-Bercovitz, H. (2000). Is the attentional dysfunction in schizotypy related to anxiety? Schizophrenia Research, in press. Braunstein-Bercovitz, H., & Lubow, R. E. (1998a). Latent inhibition as a function of attentional decline to the irrelevant stimulus during preexposure. Learning and Motivation, 29, 261–279. Braunstein-Bercovitz, H., & Lubow, R. E. (1998b). Are high-schizotypal normal participants distractible or limited in attentional resources? A study of latent inhibition as a function of masking task load and schizotypy level. Journal of Abnormal Psychology, 107, 659–670. Cole, D. A., Truglio, R., & Peeke, L. (1997). Relation between symptoms of anxiety and depression in children: A multitrait-multimethod-multigroup assessment. Journal of Consulting and Clinical Psychology, 65, 110–119. Dalgleish, T., Taghavi, R., Neshat-Doost, H., Moradi, A., Yule, W., & Canterbury, R. (1997). Information processing in clinically depressed and anxious children and adolescents. Journal of Child Psychology and Psychiatry, 38, 535–541. Delbarrio, V., Moreno-Rosset, C., Lopez-Martinez, R., & Olmedo, M. (1997). Anxiety, depression, and personality structure. Personality and Individual Differences, 23, 327–335. Doost, H. T. N., Taghavi, M. R., Moradi, A. R., Yule, W., & Dalgleish, T. (1997). The performance of clinically depressed children and adolescents on the modified Stroop paradigm. Personality and Individual Differences, 23, 753–759. Eysenck, M. W., MacLeod, C., & Mathews, A. (1987). Cognitive functioning and anxiety. Psychological Research, 49, 189–195. Fox, E. (1994). Attentional bias in anxiety: A defective inhibition hypothesis. Cognition and Emotion, 8, 165–195. Gray, J. A., Feldon, J., Rawlins, J. N. P., Hemsley, D. R., & Smith, A. D. (1991). The neuropsychology of schizophrenia. Behavioral and Brain Sciences, 14, 1–20. Johnston, W. A., & Hawley, K. J. (1994). Perceptual inhibition of expected inputs: The key that opens closed minds. Psychonomic Bulletin and Review, 1, 56–72. Kendall, P. C., & Chansky, T. E. (1991). Considering cognition in anxiety-disordered children. Journal of Anxiety Disorders, 5, 167–185. Kendall, P. C., & Brady, E. U. (1995). Comorbidity in the anxiety disorders of childhood. In K. Craig & K. Dobson (Eds.), Anxiety and depression in adults and children (pp. 3–36). Thousand Oaks, CA: Sage.
TARGET AND DISTRACTOR FAMILIARITY EFFECTS
55
Kindt, M., Brosschot, J. F., & Everaerd, W. (1997). Cognitive processing bias of children in a real life stress situation and a neutral situation. Journal of Experimental Child Psychology, 64, 79–97. Kovacs, M. (1985). The natural history and course of depressive disorders in childhood. Psychiatric Annals, 15, 387–389. Lubow, R. E. (1989). Latent inhibition and conditioned attention theory. New York: Cambridge University Press. Lubow, R. E., Dressler R., & Kaplan, O. (1999). The effects of target and distractor familiarity on visual search in de novo Parkinson’s disease patients: Latent inhibition and novel pop-out. Neuropsychology, 13, 415–423. Lubow, R. E., & Gewirtz, J. C. (1995). Latent inhibition in humans: Data, theory, and implications for schizophrenia. Psychological Bulletin, 117, 87–103. Lubow, R. E., & Josman, Z. E. (1993). Latent inhibition deficits in hyperactive children. Journal of Child Psychiatry and Psychology, 34, 959–973. Lubow, R. E., & Kaplan, O. (1997). Visual search as a function of the type of prior experience with target and distractors. Journal of Experimental Psychology: Human Perception and Performance, 23, 14–24. Lubow, R. E., Kaplan, O., Abramovich, P., Rudnick, A., & Laor, N. (2000). Visual search in schizophrenics: Latent inhibition and novel pop-out effects. Schizophrenia Research, in press. Mackintosh, N. J. (1975). A theory of attention: Variations in the associability of stimuli with reinforcement. Psychological Review, 82, 276–298. Martin, M., Horder, P., & Jones, G. V. (1992). Integral bias in naming of phobia-related words. Cognition and Emotion, 6, 479–486. Mogg, K., Bradley, B. P., & Williams, R. (1995). Attentional bias in anxiety and depression: The role of awareness. British Journal of Clinical Psychology, 34, 17–36. Mogg, K., Bradley, B. P., Williams, R., & Mathews, A. (1993). Subliminal processing of emotional information in anxiety and depression. Journal of Abnormal Psychology, 102, 304–311. Musen, G., & Treisman, A. (1990). Implicit and explicit memory for visual patterns. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 127–137. Ollendick, T. H., King, N. J., & Yule, W. (Eds.). (1994). International handbook of phobic and anxiety disorders in children and adolescents. New York: Plenum Press. Pearce, J. M., & Hall, G. (1980). A model of Pavlovian learning: Variations in the effectiveness of conditioned but not unconditioned stimuli. Psychological Review, 87, 532–552. Puig-Antich, J., Orvaschel, H., Tabrizi, R. N., & Chambers, W. J. (1978). The Schedule for Affective Disorders and Schizophrenia for School-Aged Children. New York: New York State Psychiatric Institute. Reynolds, C. R., & Paget, K. D. (1983). National normative and reliability data for the Revised Children’s Manifest Anxiety Scale. School Psychology Review, 12, 324–336. Reynolds, C. R., & Richmond, B.O. (1978). What I think and feel: A revised measure of children’s manifest anxiety. Journal of Abnormal Child Psychology, 6, 271–280. Reynolds, C. R., & Richmond, B. O. (1985). Revised children’s manifest anxiety scale. Los Angeles: Western Psychological Service. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety Inventory (self-evaluation questionnaire). Palo Alto, CA: Consulting Psychologists Press. Tannenbaum, L. E., Forehand, R., & Thomas, A. M. (1992). Adolescent self-reported anxiety and depression: Separate constructs or a single entity. Child Study Journal, 22, 61–72. Vasey, M. W., Daleiden, E. L., Williams, L. L., & Brown, L. M. (1995). Biased attention in childhood anxiety disorders: A preliminary study. Journal of Abnormal Child Psychology, 23, 267– 279. Vasey, M. W., El-Hag, N., & Daleiden, E. L. (1996). Anxiety and the processing of emotionally threatening stimuli: Distinctive patterns of selective attention among high- and low-test-anxious children. Child Development, 67, 1173–1185.
56
R. E. LUBOW ET AL.
Wagner, A. R. (1976). Priming in STM: An information processing mechanism for self-generated or retrieval-generated depression in performance. In T. Tighe & R. N. Leaton (Eds.), Habituation: Perspectives from child development, animal behavior, and neurophysiology (pp. 95–128). Hillsdale, NJ: Lawrence Erlbaum Associates. Werry, J. S. (1991). Overanxious disorder: A review of its taxonomic properties. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 533–544.