Defensive copers show a deficit in passive avoidance learning on newmans gono go task implications for self deception and socialization

Avoidance Learning on Newman’s Go/No-Go Task: Implications for Self-Deception andin response to contingent positive and negative feedback.. Defensive copers showed preferential reflection

Avoidance Learning on Newman’s Go/No-Go Task: Implications for Self-Deception and

in response to contingent positive and negative feedback The duration that feedback remained onscreen was self-regulated Defensive copers showed preferential reflection away from negative feedback, committed more passive-avoidance errors, and were characterized by impaired learn- ing, overall Further, the ratio of reflection on negative feedback to re- flection on positive feedback directly mediated both passive-avoidance errors and overall learning Defensive coping strategies, therefore, appear

to interfere with passive avoidance learning, thereby fostering severative, dysfunctional action patterns by reducing knowledge gained from previous mistakes Implications for the learning of effective social- ization strategies, and for psychopathy—which is commonly character- ized by similar passive-avoidance deficits—are subsequently considered.

per-This research was supported by the Social Sciences and Humanities Research Council

of Canada, and by a Connaught Grant from the University of Toronto We gratefully acknowledge Jessica Aurora’s help in the data collection and Phil Zelazo’s willingness

to read and offer advice on previous versions of the manuscript Correspondence cerning this article should be addressed to Jordan B Peterson, Department of Psy- chology, University of Toronto, 100 St George St., Toronto, Ontario, Canada, M5S 3G3.

con-Journal of Personality 72:5, October 2004.

Blackwell Publishing 2004

Research conducted over the last two decades has suggested thatindividuals who self-report low levels of anxiety may not constitute ahomogeneous group Rather, it appears meaningful to distinguishbetween those who are genuinely low-anxious, and those who utilizedefensive coping strategies to minimize their level of subjectively ex-perienced or reported anxiety (Weinberger, Schwartz & Davidson,1979).1 Individuals in both hypothetically low-anxious groups con-sistently score below the median on self-report anxiety measures, butdiverge dramatically on measures of conformity, defensiveness, orrestraint That is, whereas genuinely low-anxious individuals scorequite low on measures of defensiveness, defensive copers manifestelevated levels of defensiveness despite showing similarly low anxietylevels This combination of traits seems somewhat counterintuitive:

it is unlikely that individuals who are clearly defensive and sensitive

to the opinions of others could be genuinely free of anxiety In sequence, researchers have suggested that the defensive coper’ssubjective level of anxiety may not accurately reflect his or her ob-jective sensitivity to anxiety-provoking or threatening environmentalstimuli

con-Recent investigations have generally supported such suggestions.Defensive-copers manifest a unique pattern of threat processing,characterized by increased physiological responses to aversive orthreatening stimuli (e.g heart rate: Fuller, 1992; cortisol: Brown

et al., 1996; b-endorphins: Jamner & Hoyle, 1999) and, cally, to reduced attention to such stimuli Mogg and colleagues(2000), for example, have systematically demonstrated that individ-uals who utilize repressive coping styles allocate significantly fewerattentional resources toward threatening than toward neutral orpositive stimuli This complex avoidant bias has a clearly defensivequality and stands in marked contrast to the unbiased attentional

paradoxi-1 A number of different personality constructions have, in the past, fallen under the umbrella of individuals who utilize defensive coping mechanisms, including self-deceptive individuals and individuals with repressive coping styles Although there are important differences between each of these personality constructs, a core characteristic of each is the use of defensive strategies to reduce subjective levels of negative emotionality (be it fear, anxiety, sadness, guilt, or otherwise) Following the sound advice of an anonymous reviewer, rather than choose one of these particular personality constructs to refer to the participants in the present study, we have instead used the more global and straightforward term ‘‘defensive coper’’ throughout.

processing of genuinely low-anxious individuals (Mogg & Bradley,1998) Furthermore, experimental demonstrations of such defensiveprocessing are in accordance with a large body of evidence that hasless systematically demonstrated defensive copers’ reluctance toprocess aversive information Mischel, Ebbeson, and Zeiss (1973),for instance, demonstrated that individuals high in the use of self-deception were less likely to consider negative self-relevant informa-tion, while Davis and Schwartz (1987) found these individuals char-acterized by reduced ability to recall negative memories In line withthese findings, Eysenck (2000) has suggested, more generally,that defensive coping may occur in part through minimization

of threatening information across four broad categories: externalstimuli, internal physiological stimuli, personal behavior, and per-sonal cognitions The fact that defensive copers can accomplishthis avoidance, however, suggests that they must be as threat-sensi-tive as high-anxious individuals, despite their low self-reported levels

of anxiety In the absence of such threat sensitivity, after all, theywould not be capable of easily determining which stimuli should beavoided

Research suggesting that defensive copers actually show initialautomatic orientation toward threatening stimuli directly supportsthis latter notion Calvo and Eysenck (2000) measured the timecourse of threat-processing biases in individuals with repressive cop-ing styles by measuring their ability to name target words presented

at various delay intervals after being primed with threatening ornonthreatening, context-framing sentences Repressors were slower

at naming the target words after threat exposure when the delay terval was long (1050 ms), but were faster at naming the same targetwords when the word appeared after a shorter delay (550 ms) This

in-‘‘vigilance-avoidance’’ pattern of processing (Mogg & Bradley, 1998)indicates that defensive copers rapidly divert their attention else-where following initial orientation to threat The early facilitation ofthreat processing theoretically reflects the underlying anxiety sensi-tivity of defensive individuals The subsequent inhibition of thethreat-related word processing theoretically indexes or representstheir later onset avoidant coping strategies

What purpose might such avoidant strategies serve? The ing of aversive or threatening stimuli necessarily exposes the indi-vidual to a variety of psychological dangers, and has been shown tocorrelate positively with levels of arousal and anxiety (Carver &

process-Scheier, 1998; Dollard & Miller, 1950; Mogg & Bradley, 1998) thermore, evidence exists to suggest that defensive copers may, infact, be even more susceptible to such increased levels of arousal andanxiety in response to negative feedback than the general population(Johnson, Vincent & Ross, 1997) Presumably, therefore, avoidingprolonged processing of negatively valenced, aversive stimuli couldserve to minimize the heightened anxiety that more sustainedprocessing might cause Thus, defensive individuals may avoid thor-ough cognitive processing of negatively valenced information in or-der to minimize their level of short-term psychological risk Thisthreat minimization would allow defensive copers to remain subjec-tively stress-free, despite the inconsistencies between their subjectiveexperience and their objective body-state.

Fur-In keeping with such ideas, Taylor and Brown (1988) have posed that ‘‘positive illusions’’—cognitive filters that preferentiallyscreen out negative information—may protect against anxiety anddepression, enhance creativity (through increases in positive emo-tion), and promote successful life adjustment Such a hypothesisstands in direct contradiction to classic clinical wisdom, however,which is predicated on the idea that accurate contact with reality andself constitutes a necessary precondition for healthy psychologicaladjustment (Binswanger, 1963; Freud, 1963) According to suchclassic notions, individuals who habitually refuse to thoroughlyprocess aversive information should find it difficult to successfullyadapt their behavior to the ever-changing environment From thisperspective, effective adaptation requires constant self-regulation,based, in theory, on the operation of negative feedback systems thatcompare present states to ideal or goal states (Ansbacher & Ansbac-her, 1956; Carver & Scheier, 1998; Peterson, 1999; Rolls, 1999).When goal-directed behavior is disrupted, discrepancy between thedesired and actual states produces a rapidly elicited, aversive ‘‘errortag,’’ which indicates that a particular plan has failed (Damasio,

pro-1994, 1998; Gray, 1982; Peterson & Flanders, in press) The merefact of such error, however—even in combination with the initialaffective response—does not provide enough information to deter-mine the precise nature of the inadequacy (Peterson, 1999; Peterson

& Flanders, in press) Rather, the generation of such detailed mation requires further, effortful exploration or reconstrual of thesituation (Ohman, 1979, 1987; Sokolov, 1969) Thus, decreasedattention devoted towards stimuli signifying disruptions in goal-

infor-directed behavior should logically reduce the opportunity to correct the problems that caused the disruption and should thereby increase the chances of encountering similar disruptions in the future In other words, the ability to learn from mistakes requires paying attention to the reasons for such mistakes when they occur Without such reflection, it may be difficult to obtain the informa-tion necessary to avoid committing similar perseverative errors in the future

Such notions have received empirical support from a number of sources Newman and colleagues (Newman & Schmitt, 1998; New-man, Patterson, Howland & Nichols, 1990; Newman, Widom, & Nathan, 1985) were the first to propose that reflection on negative feedback may play an important role in the ability to learn from punishment These researchers demonstrated that undersocialized, disinhibited, and psychopathic individuals—predominantly charac-terized by an abundance of perseverative, dysfunctional behaviors— commonly reflect less on negative feedback and show marked deficits in passive avoidance learning (learning to inhibit a dysfunc-tional behavior to avoid punishment)

In the common version of Newman’s paradigm, participants must learn to press or to not press a button as a consequence of positive or negative feedback received during a trial-and-error learning task Correct presses are rewarded with positive feedback (and, often, a small monetary gain) while incorrect presses are punished with negative feedback (and a small monetary loss) While engaged in this task, participants are able to voluntarily regulate the length of time that the contingent positive or negative feedback remains available for on-screen reflection and consi-deration Newman and colleagues (e.g Howland, Patterson, Kosson & Newman, 1987; Newman, 1987; Newman et al., 1990) have consistently reported that disinhibited and psychopathic individuals exhibit a marked reduction in reflection on the negative feedback and a marked increase in the number of passive-avoidance errors committed on the task Newman has therefore proposed that psychopaths are victim to deficits in response modulation, at the level of automatic processing, and that these deficits render them unable to take sufficient time to pause and reflect on their errors Such reduced reflection on negative feedback then decreases the opportunity for evaluative and corrective measures, which, in turn, increases the likelihood of future perseverative, dysfunctional behavior

Further evidence of the link between reflection and learning comesfrom the education literature, where Shafrir and Pascual-Leone(1990) have shown, for example, that level of reflective preference(RP) predicts ability to learn through trial and error Shafrir andPascual-Leone (1990) operationally defined RP as length of timespent pausing after negatively valenced performance feedback/length

of time spent pausing after positively valenced performance feedback.2These authors have reported positive correlations between RP meas-ures and various academic and nonacademic learning tasks, infer-ence problems, and intelligence measures Shafrir and Pascual-Leone’s (1990) research thus demonstrates that degree of reflection

on information about past failure can have direct effects on theability to learn Furthermore—and of critical importance to theories

of self-deception—Shafrir’s work suggests that it is differential tention paid to negatively valenced information versus positivelyvalenced information that is most important with regard to the ac-quisition of knowledge Individuals who selectively refuse to engage

at-in higher-order threat/failure processat-ing therefore appear to fail totransform the information implicit in such failure into novel, pro-ductive, implementable behaviors and structures of abstract knowl-edge (Peterson, 1999) Individuals who deeply process such aversivestimuli, by contrast, have been shown capable of successfully adapt-ing, even to traumatic stress (Foa & Kozak, 1985, 1986; Pennebaker,Mayne & Francis, 1997)

The Present Study

We therefore hypothesized that defensive copers would show parative deficits in the ability to learn through trial and error whenprovided with contingent positive and negative feedback Further-more, we hypothesized that this learning deficit would occur inconjunction with a reduced level of RP (or, alternatively stated,

com-2 Shafrir and Pascual-Leone (1990) originally referred to this reflection measure

as ‘‘Post-failure Reflectivity.’’ We will instead use the term ‘‘Reflective Preference’’ throughout this manuscript, because the original label seems to imply, somewhat inaccurately, that the key measure is the length of time the negative feedback remains on-screen Shafrir and Pascual-Leone’s (1990) measure was, in fact, the ratio of reflection on negative feedback to reflection on positive feedback, which seems more aptly described by the term ‘‘Reflective Preference.’’

given the opportunity to reflect on both positive and negative back, defensive copers would pay comparatively less attention to thenegative feedback than high- or low-anxious individuals and wouldtherefore learn more slowly).

feed-We operationalized our hypothesis by identifying high-anxious,low-anxious, and defensive copers, using standard self-report per-sonality indices and tested it by running classified participantsthrough a modified version of the Newman et al (1990) computer-ized go/no-go task This task is the same as that described above,requiring participants to learn when to press and when not to press abutton on a computer keyboard, based on contingent positive andnegative feedback As in Newman et al (1990), correct presses re-ceived positive feedback, and incorrect presses received negativefeedback And, as with Newman et al (1990), participants were able

to regulate the length of time that feedback remained on-screen forconsideration

This task was specifically chosen for a number of reasons: First, itprovides a direct behavioral measure of the amount of time partic-ipants choose to spend attending to both positive and negative feed-back This measure allows the experimenter to derive both differenceand ratio scores of reflection on negative versus positive task-relatedfeedback (the ratio measure being Shafrir and Pascual-Leone’s[1990] RP measure), and allows for the comparison of the alloca-tion of reflective resources between groups We hypothesized that thedefensive copers would manifest lower RP than either high- or low-anxious individuals, and we felt that this would indicate their inferiorreflection strategy, skewed towards lesser reflection on negative in-formation

Second, the task provides a variety of easy and accurate measures

of the ability to learn through experience (number of omission rors, number of passive-avoidance errors, and overall task success),enabling multiple evaluations of performance Third, the task allowsparticipants’ learning to be directly correlated with the length of timethey leave the positive and negative feedback on the screen The no-tion that the (hypothetically) reduced RP scores of defensive coperscan, in fact, be linked to learning difficulties on the task can therefore

er-be directly assessed

Finally, the use of Newman et al.’s (1990) go/no-go task allowed

us to compare the performance of defensive copers with that ofundersocialized and disinhibited (including psychopathic) popula-

tions As discussed above, Newman and his colleagues have monly used this paradigm within these selected populations and havedemonstrated that such individuals exhibit chronic deficits inpassive-avoidance learning Furthermore, such undersocializedpopulations often exhibit what may be considered an extreme ina-bility to learn from their previous mistakes throughout their life-course, and thus they should be particularly interesting to investi-gators researching the processes involved in learning and knowledgeacquisition An additional aim of this research was, therefore, toexamine the possibility that use of defensive coping mechanismscould create deficits in learning similar to those of undersocializedpopulations Psychopathic individuals have generally been thoughtunable to generate enough anxiety to produce conditioning or tomediate appropriate avoidance learning (Fowles, 1980; Raine, 1989).This paper is part of a larger research program investigating thepossibility that some individuals may, through the use of defensivestrategies, reduce their subjective level of anxiety to such an extentthat they similarly disrupt their ability to learn effectively throughexperience.

com-METHODMeasures

Balanced Inventory of Desirable Responding The BIDR (Paulhus, 1991)

is a 40-item inventory consisting of two 20-item subscales: Self-Deceptive Enhancement (SDE) and Impression Management (IM) SDE assesses defensiveness towards personal weakness (e.g ‘‘I have never doubted

my ability as a lover’’) and a general egoistic or overconfident ponse bias) (e.g ‘‘I am fully in control of my own fate’’) (Paulhus & John, 1998) IM measures the tendency to make oneself look better by denying socially undesirable behavior (e.g ‘‘I never take things that don’t belong to me’’).

res-Taylor Manifest Anxiety Scale The 20-item short form of the Taylor Manifest Anxiety Scale (TMAS; Bendig, 1956) was used to measure trait anxiety The TMAS is highly correlated with other measures of trait anxiety and negative affectivity (Watson & Clark, 1984).

Group Classifications The categorical selection criteria utilized, similar

to those suggested by Weinberger (1990), were based on patterns of scores

on both the TMAS and the SDE.3Participants were classified as sive copers (n 5 18) if they scored above the median on the SDE and below the median on the TMAS Defensive copers were therefore those participants who rated themselves as high on self-deceptive enhancement and low on anxiety—a pattern of self-description suggest- ing suppression of negative affect In contrast, the high-anxious (n 5 21) and low-anxious (n 5 21) participants obtained scores below the median on the SDE, and above or below the median on the TMAS, respectively.

defen-Eight participants scored above the median on the TMAS and the SDE This group has sometimes been referred to as a ‘‘defensive high- anxious’’ group, but, historically, low numbers of individuals with this classification have consistently hampered their investigation Because only eight participants were categorized in this group in the present ex- periment, they were not included in any of the categorical analyses dis- cussed below.

Participant Selection

Participants initially completed a shortened version (the first 10 items) of the SDE during a mass testing session at the beginning of the school term (N 5 1472) A total of 85 participants were invited to attend experimental sessions: twenty-nine individuals who scored in the top 10% of the dis- tribution on the partial SDE and 56 individuals who scored in the bottom 10% of the distribution on the partial SDE We invited roughly twice as many participants from the bottom 10% of the distribution because we required a sufficient number of high- and low-anxious participants, both

of whom would need to be selected from this low SDE group (please refer

to the section above on group classifications for further details) The liability of the partial SDE scale was moderate, a 5 56, but quite ade- quate as an initial screening device.

re-During the experimental sessions, participants were readministered the full SDE, as well as the short form of the TMAS Based on these full SDE scores, eight individuals were classified as ‘‘defensive high-anxious,’’ and thus were not included in the categorical analyses Of the remaining 77

3 Weinberger (1990) utilized the Marlowe-Crown Social Desirability Index (MCSD) rather than the SDE for participant classification Furnham, Petrides,

& Spencer-Bowdage (2002) have demonstrated that use of either the MCSD or the SDE are valid, and similar, identifiers of individuals who utilize repressive coping styles We believe, however, that the SDE is a more appropriate measure, due to its ability to separate socially desirable responding from more internally self- enhancing tendencies (Paulhus, 1984, 1986).

participants, only those participants whose partial and full SDE scores were consistently within the same half of the distribution, using a median split analysis, were included in the final analysis of the data (N 5 60) Although the loss of 17 participants is somewhat unfortunate, this double classification procedure best ensures the accuracy of each participant’s group membership As evidence to this effect, the correlation between partial and full SDE scores for all included participants was r 5 81.

Data Analytic Strategies

Following Wienberger et al (1979), most recent research on defensive coping has classified participants as defensive copers in the categorical fashion described above Both the TMAS and the SDE evaluate contin- uous personality traits, however, and thus, by rights, should be analyzed

in a continuous manner (Wright, 2003) With these two considerations in mind, we decided to supplement the categorical analyses with continuous analyses, using regression with TMAS and SDE as individual predictors

of each dependent variable The categorical analysis is most easily compared to previous literature in the area However, the continuous analyses should provide a more in-depth investigation into the individual contributions of anxiety and defensiveness on attentional and learning processes.4

Apparatus and Task

The experimental task was conducted on a Pentium computer with a 14" monitor in a small quiet room The experimental task was a modified version of the go/no-go discrimination task used by Newman et al (1990) Participants were instructed that they would have to learn through trial and error when to respond (by pressing the spacebar) and when not to respond Stimuli consisted of 10 two-digit numbers (e.g.

15, 24, 38, 47) presented nine times each in pseudo-random order No more than three consecutive ‘‘go’’ or ‘‘no-go’’ stimuli were presented In total, there were 90 experimental trials in the task (plus 10 practice trials and a 5 trial ‘‘go’’ pretreatment, as explained below) Two different sets of numbers were used on the task to ensure there were no unintentional patterns to the digits that could prove helpful for the participants In addition, stimuli that were ‘‘go’’ stimuli for one half of the participants

4 Note that the 8 ‘‘defensive high-anxious’’ participants, not included in the egorical analyses due to low numbers, are included in the regression analyses re- ported throughout Thus, 60 participants are included in each of the categorical analyses, whereas 68 participants are included in the continuous analyses.

cat-were ‘‘no-go’’ stimuli for the other half, making a total of four different stimulus sets.

Following Newman et al (1990), participants received a five-trial ward pretreatment, during which each of the ‘‘go’’ stimuli was presented,

re-as in the test trials The purpose of the pretreatment wre-as to establish a dominant response set, by providing a high probability of reward for re- sponding at the beginning of the task (see also Siegel, 1978; Newman, Patterson, & Kosson, 1987) Test trials began immediately after the five- trial pretreatment Following a correct response, the stimulus number was immediately replaced by the message ‘‘Correct You win 20 cents!’’ ac- companied by a high-pitched tone (625 Hz), played through the computer speaker If the response was incorrect, the message ‘‘Wrong You lose 20 cents!’’ appeared, and a low-pitched tone (125 Hz) was played No feed- back was provided in the absence of a response.

All participants were informed that they would have to re-press the spacebar, after receiving feedback, in order to proceed to the next trial In this way, participants controlled the duration that the negative or positive contingent feedback was displayed If they did not respond within five seconds, the computer prompted them to do so Participants were not informed that their response times were being recorded No feedback was provided, and, therefore, no second response was required when a participant did not respond to the stimulus In the case of a nonresponse, the stimulus remained on the screen for 3 seconds; the intertrial interval was 1 second.

Procedure

Eligible participants were provided with a brief description of the entire project Those providing consent for contact were invited into the laboratory Upon arrival, an additional consent form describing the task and the personality measures were completed Task order was counterbalanced so that half of the participants performed the go/no-go task before filling out the personality measures and half per- formed the task subsequent to the personality measures Participants received 10 practice trials before the onset of the go/no-go task, using the numerals 01 and 02 as stimuli The five-trial ‘‘go’’ pretreatment was then given, followed immediately by the 90 test trials Upon completion of the study, participants were awarded course credit and paid the money earned on the computer task They were then fully de- briefed and allowed to leave The entire study was reviewed for compli- ance with ethical requirements by the University of Toronto Institutional Review Board.

RESULTSTwo low-anxious participants were initially removed from the anal-yses, bringing the total number of participants to 58.5 Preliminaryanalyses revealed no significant differences between the four differ-ent stimulus sets administered Similarly, no differences were foundbetween participants who completed the go/no-go task before or af-ter administration of the questionnaires None of these variableswere therefore considered further in any analyses Means and stand-ard deviations of group SDE and TMAS scores are displayed inrows 1 and 2 of Table 1.

Task SuccessParticipants received 20 cents for every correct response and lost 20cents for every incorrect response In consequence, the total amount

of money awarded to each participant could be used as a measure ofoverall success on the task Row 3 of Table 1 shows the means andstandard deviations for the amount of money awarded to partici-pants in each group A one-way Analysis of Variance (ANOVA)with group as the between-subjects variable revealed a significantmain effect for number of tokens gained during the task,

F(2,55) 5 5.63, Z 5 170, p 5 006 Planned comparisons indicatedthat defensive copers acquired significantly less money during thetask than both high-anxious, t(55) 5 2.34, d 5 77, p 5 023, and low-anxious participants t(55) 5 3.278, d 5 1.09, p 5 002, who did notdiffer from each other

Regression analyses, with SDE and TMAS as individual tors of task success, were also conducted to evaluate the individualcontribution of each personality construct SDE and TMAS wereentered in Block 1, and the SDE  TMAS interaction term was en-tered in Block 2 Initial collinearity between the measures and theinteraction term was reduced by centering the predictors before en-tering them in the regression model SDE emerged as the only sig-nificant predictor of task performance, b 5 383, p 5 003; other

predic-5 One was caught cheating on the task (she had taken out a pen and was writing down the stimulus contingencies) The other clearly did not understand the task at hand (or decided to be purposefully noncompliant), pressing a key on only 21 of the 90 trials in an apparently random fashion By comparison, the next lowest number of button press responses was 30, while the mean number of responses across all participants was 51.1.

p’s4.25, suggesting that defensiveness may have played a larger rolethan anxiety level in influencing performance on the task.

Error TypeRows 4 and 5 of Table 1 display the means and standard deviationsfor the number of omission and passive-avoidance errors made byeach group during the task, respectively

Passive-avoidance errors involved pressing the button on no-gotrials, while omission errors involved not pressing the button on gotrials A 3  2 repeated measures ANOVA was conducted withGroup as the between subjects variable and Error type (passive-avoidance, omission) as the repeated measure A significant maineffect for Error type emerged, F(2,55) 5 33.35, Z 5 377, po.001, in-dicating that all groups committed more passive-avoidance thanomission errors on the task The predicted main effect of Group alsoemerged, F(1,55) 5 5.67, Z 5 171, p 5 006, indicating more errors

Table 1 Means and Standard Deviations of All Personality, Performance, and

Reflection Measures High Anxious Low Anxious Defensive Coper

committed by defensive copers, as was demonstrated in the previousanalysis of overall money earned on the task The Group  ErrorType interaction approached significance, F(2,55) 5 2.33, Z 5 078,

p 5 107, and exploratory planned comparisons indicated thatdefensive copers committed significantly more passive-avoidanceerrors than both high-anxious t(55) 5 2.77, d 5 75, p 5 008,and low-anxious participants, t(55) 5 2.98, d 5 80, p 5 004.Groups did not differ on the mean number of omission errorscommitted, p4.1

Once again, regression analyses were performed to investigate theindividual contributions of SDE and TMAS on committed passive-avoidance and omission errors Individual regression analyses wereperformed to predict each error type, entering the predictors in thesame fashion as described above As with overall task performance,only SDE emerged as a significant predictor of commission errors,

b 5 401, p 5 002, however, both TMAS, b 5 203, p 5 109, andSDE  TMAS, b 5 218, p 5 100, trended toward significant pre-dictions, suggesting the possibility that an interactive effect may ex-ist Neither SDE nor TMAS emerged as significant predictors ofomission errors; however, SDE  TMAS emerged as highly signif-icant, b 5 311, p 5 027

ReflectivityTwo low-anxious participants were not included in the reflectivitydata because they did not make any commission errors on thego/no-go task and, therefore, had no negative reflectivity score.One additional low-anxious participant was dropped from theanalysis because she waited the entire 5000 ms duration on each

of the 90 trials, suggesting that she was unaware that she couldproceed to the next trial by pressing the space bar again Follow-upquestions during the debriefing session verified that this was, in fact,the case

Rows 6 and 7 of Table 1 display the mean length of time eachgroup left the positive feedback (positive feedback reflectivity, PFR)and negative feedback (negative feedback reflectivity, NFR) onscreen, respectively Row 8 displays the mean RPs for the threegroups One-way ANOVAs indicated that the only group main effectwas for RP, F(2,52) 5 2.63, Z 5 107, p 5 040, one-tailed Subse-quent planned comparisons indicated that although all three groups