Deficient Response Modulation and Emotion Processingin Low-Anxious Caucasian Psychopathic Offenders:

Results From a Lexical Decision Task

Amanda R. Lorenz and Joseph P. NewmanUniversity of WisconsinMadison

The clinical and research literatures on psychopathy have identified an emotionparadox: Psychopaths display normal appraisal but impaired use of emotion cues.Using R. D. Hares (1991) Psychopathy ChecklistRevised and the G. S. WelshAnxiety Scale (1956), the authors identified low-anxious psychopaths and controlsand examined predictions concerning their performance on a lexical-decision task.Results supported all the predictions: (a) low-anxious psychopaths appraised emo-tion cues as well as controls; (b) their lexical decisions were relatively unaffectedby emotion cues; (c) their lexical decisions were relatively unaffected by affectivelyneutral word-frequency cues; and (d) their performance deficits were specific toconditions involving right-handed responses. The authors propose that deficientresponse modulation may underlie both the emotional and cognitive deficits asso-ciated with low-anxious psychopaths.

On the basis of Cleckleys (1976) classic portrayalof psychopathy, investigators have attributed psycho-paths disinhibited and antisocial lifestyle to an emo-tion deficit (Blair, 1995; Hare, 1998; Lykken, 1957,1995; McCord & McCord, 1964; Patrick, 1994). Con-sistent with this interpretation, psychopaths displayless reactivity to threat cues (Lykken, 1957), anoma-lous performance in an emotion-modulated startleparadigm (Patrick, Bradley, & Lang, 1993), less au-tonomic arousal in response to interpersonal distresscues (Blair, Jones, Clark, & Smith, 1997), and less

pronounced attention and memory effects associatedwith emotional cues relative to nonpsychopathic con-trols (Christiansen et al., 1996; Day & Wong, 1996;Kiehl, Hare, McDonald, & Brink, 1999). Together,these findings provide compelling evidence that emo-tion cues are less likely to influence the behavior ofpsychopaths.

Though investigators have focused their attentionon psychopaths emotion deficit, Cleckleys (1976)writings also highlighted an emotion paradox betweenthe psychopaths ability to appraise emotion cues andtheir inability to use emotion cues. That is, psycho-paths demonstrate normal appraisal of emotional cuesand situations in the abstract (i.e., verbal discussion),but they are deficient in using emotion cues to guidetheir judgments and behavior in the process of living.

Williamson, Harpur, and Hare (1991) found em-pirical support for the emotion paradox in psychopa-thy by using a lexical-decision task. The lexical-decision task requires participants to determine if abriefly presented letter string is a word or a nonword.Paralleling results for controls (Challis & Krane,1988; Graves, Landis, & Goodglass, 1981; Strauss,1983), nonpsychopathic offenders responded faster toemotion words than to neutral words. Psychopaths,however, did not demonstrate this emotion facilitationin response to positive or negative words although,similar to controls, they were more accurate at iden-tifying the emotional words. Thus, in contrast to the

Amanda R. Lorenz and Joseph P. Newman, Departmentof Psychology, University of WisconsinMadison.

This research was supported by a research grant from theNational Institute of Mental Health. We gratefully acknowl-edge the assistance of Cathy Farrey, Jeff Endicott, JimBoorman, Janet Walsh, Jeffrey Wydeven, Bernice Conner,Bev Mars, Deb March and the staff at the Oakhill Correc-tional Institution, the staff at the Columbia CorrectionalInstitution, and the cooperation of the Wisconsin Depart-ment of Corrections. We thank Bill Schmitt, Chad Brinkley,and Jennifer Vitale for interviewing and diagnosing partici-pants.

Correspondence concerning this article should be ad-dressed to Joseph P. Newman or Amanda R. Lorenz,Department of Psychology, University of WisconsinMadison, 1202 West Johnson Street, Madison, Wisconsin53706. E-mail: jpnewman@facstaff.wisc.edu

Emotion Copyright 2002 by the American Psychological Association, Inc.2002, Vol. 2, No. 2, 91104 1528-3542/02/$5.00 DOI: 10.1037//1528-3542.2.2.091

91

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

observed group differences in using emotion cues tofacilitate task performance, Williamson and col-leagues found no group differences in participantsword ratings. Similarly, Patrick et al. (1993) assessedstartle responses while participants viewed slides thatvaried in affective valence, and Patrick, Cuthbert, andLang (1994) assessed electrodermal activity in re-sponse to tones that were cues for threat-related orneutral situations. In both studies, Patrick and col-leagues observed significant differences in the physi-ological indices of emotion processing, but psycho-paths and nonpsychopaths performed similarly inrating the pleasantness, arousal, and dominance of thefear-related situations and slide stimuli. Furthermore,in an investigation of children with psychopathic ten-dencies, this group demonstrated smaller electroder-mal responses to distress cues than nonpsychopathicchildren, but the two groups did not differ when ratingthe valence of the cues (Blair, 1999; see also Blair etal., 1997). Taken together, these findings providegood empirical support for the emotion paradox: psy-chopaths can appraise emotion cues but such cueshave little measurable influence on their behavior.

Although existing theories address emotion deficitsin psychopaths, it is unclear how they account for theparadox. One potential explanation for these phenom-ena is the response modulation hypothesis (RMH; Go-renstein & Newman, 1980; Newman, 1998; Newman& Lorenz, in press; Patterson & Newman, 1993). TheRMH relates to Cleckleys (1976) concept of psy-chopathy, which holds that genuine psychopaths arecharacterized by low levels of neurotic anxiety, ab-sence of psychotic thinking, and good intelligence.Indeed, response modulation deficits have been foundto be relatively specific to low-anxious psychopaths(see Newman & Brinkley, 1997). Response modula-tion is defined as a brief and highly automatic shift ofattention that enables individuals to monitor and, ifrelevant, use information that is peripheral to theirdominant response set (i.e., deliberate focus of atten-tion). According to the RMH, the impulsivity, poorpassive avoidance, and emotion-processing deficits ofpsychopathic individuals may all be understood as afailure to process the meaning of information that isperipheral or incidental to their deliberate focus ofattention (Newman & Lorenz, in press). Consistentwith the RMH, psychopathic individuals display ad-equate passive avoidance when avoiding punishmentis their primary (i.e., sole) task but manifest perfor-mance deficits when avoidance requires them to pro-cess nondominant (i.e., secondary or latent) informa-tion (e.g., Lykken, 1957; Newman & Kosson, 1986).

Moreover, there is evidence that psychopaths are in-sensitive to secondary neutral as well as secondaryemotional cues (Newman, Schmitt, & Voss, 1997).

Given the similarities between their emotional andmore general information-processing deficits, it ispossible that psychopaths emotion deficits, like theirself-regulatory deficits, involve a deficiency in pro-cessing incidental cues. Indeed, Newman and Lorenz(in press) recently proposed that psychopaths deficitsin emotion processing may be understood as a failureto process the associative networks primed by second-ary emotion cues. Although emotion processing mayoperate effortlessly and automatically to influence thebehavior of most individuals (e.g., Bower, 1981), weproposed that emotions are less likely to influence thebehavior of low-anxious psychopaths because theiremotion processing is more dependent on deliberateprocessing (i.e., less automatic). Restated, psycho-paths may be less likely to activate and/or use theassociative networks primed by peripheral emotionalcues that other individuals use automatically.

With regard to the emotion paradox, then, the RMHis consistent with the hypothesis that psychopathicindividuals (a) display normal emotion processingwhen instructed to attend to emotion stimuli and (b)are less influenced by emotion cues that are peripheralto their dominant response set (i.e., directed atten-tion). Regarding emotion facilitation in a lexical-decision task, we presume that emotion words primeassociational networks on the basis of their emotionalvalence (Balota & Chumbley, 1984; Bower, 1981)and, then, activate emotional characteristics that fa-cilitate responding by rapidly indicating the presenceof the stimuli in the mental lexicon. For instance, theword sunset may prime positive emotions that activatenumerous words associated with happy, peaceful, andcontented feelings. This rapid activation of a positiveemotion makes it easier for the participant to identifythe stimulus as a word. However, to the extent thatpsychopaths are deficient in processing incidental in-formation as suggested by the RMH, they are lessinfluenced by a words affective connotations.

Conceptualizing facilitation on the lexical-decisiontask in terms of response modulation, we used a modi-fied version of the lexical-decision task employed byWilliamson and her colleagues (1991) to replicate andextend their findings. On the basis of Cleckleys(1976) conceptualization of psychopathy and past in-vestigations of the RMH (see Newman & Brinkley,1997; Schmitt & Newman, 1999b), planned compari-sons focus on the performance of low-anxious (i.e.,

LORENZ AND NEWMAN92

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

primary) psychopaths and controls.1 Nevertheless, weinclude the data for high-anxious groups in the overallanalyses so that investigators interested in Hares(1991) psychopathy construct (high PsychopathyChecklistRevised [PCLR] scorers without regardfor anxiety) can interpret out findings. Because sepa-rate planned comparisons (i.e., t tests) were conductedto test hypotheses using low-anxious groups only,they are unaffected by the inclusion of the high-anxious groups.

Given prior work demonstrating an emotion deficitin psychopaths, our first prediction was that low-anxious psychopaths would demonstrate less emotionfacilitation than low-anxious controls. This predictionis consistent with the RMH and Williamson et al.s(1991) findings that psychopaths demonstrated lessemotion facilitation than controls in a lexical-decisiontask. In addition, based on prior work demonstratingcomparable appraisal of emotion stimuli, we pre-dicted that low-anxious psychopaths would not differfrom low-anxious controls in rating the emotional va-lence of the words used in the lexical-decision task.Support for these two predictions would replicate ear-lier work demonstrating the emotion paradox in low-anxious psychopaths.

One interpretation of the psychopaths deficientemotion facilitation is that they fail to access or elabo-rate the semantic and affective associations or net-works primed by the stimulus words (Hare, 1998;Newman & Lorenz, in press). To the extent that thisfailure to process associative networks relates to re-sponse modulation as opposed to deficient emotionprocessing per se, psychopaths should be less influ-enced by emotionally neutral as well as emotionalaspects of a words meaning. To test this third pre-diction, we examined the ability of low-anxious psy-chopaths to use an emotionally neutral secondary cue(i.e., word frequency) to facilitate performance on theprimary lexical-decision task. Paralleling our predic-tion involving weaker emotion facilitation, we pre-dicted that low-anxious psychopaths would demon-strate less frequency facilitation than controls.

Recent research by Kosson (1996, 1998) and col-leagues (Howland, Kosson, Patterson, & Newman,1993; Kosson & Harpur, 1997) suggests that psycho-paths demonstrate performance anomalies on mea-sures of attention when using their right hand as op-posed to their left hand to respond. Although the moregeneral literature on cerebral asymmetries in psycho-paths is inconsistent (Hare, 1979; Hare & Forth, 1985;Hare & Jutai, 1988; Hare & McPherson, 1984; Hiatt,Lorenz, & Newman, in press; Jutai, Hare, & Con-

nolly, 1987; Smith, Arnett, & Newman, 1992), Kos-son speculated that psychopaths deficits on tasks re-quiring right-handed responses might be because ofinefficient processing in the left hemisphere. Consis-tent with this speculation, Bernstein, Newman, Wal-lace, and Luh (2000) found that psychopaths per-formed more poorly than did controls in usinginformation presented to their right-visual field. No-tably, however, this deficit was specific to the pro-cessing of incidental cues. Thus, evidence for ineffi-cient left hemisphere processing in psychopaths maybe specific to incidental cues and/or the integration ofprimary and secondary aspects of a stimulus as speci-fied by the RMH.

To the extent that lexical decisions normally in-volve integrating incidental information (i.e., the au-tomatic associations primed by a word) with onesdirected inspection of the word, it follows that psy-chopaths insensitivity to primed, secondary associa-tions, like their insensitivity to peripheral cues in theBernstein et al. (2000) study, may be specific to con-ditions that activate the left hemisphere primarily. Al-though such theorizing is still speculative, we de-signed the lexical-decision task to explore thishypothesis. In particular, we divided the task intoeight blocks of experimental trials and instructed par-ticipants to switch the hand used to respond after eachtrial block. If low-anxious psychopaths benefit lessthan controls from primed associations when the lefthemisphere is differentially activated, then the pre-dicted group differences in facilitation should be spe-cific to right-hand trial blocks.

MethodParticipants

The participants in this study were 100 Caucasianmale inmates residing at either a minimum or maxi-mum security prison in southern Wisconsin. Inmates

1 Schmitt and Newman (1999a) demonstrated that indi-viduals with high scores on the Psychopath ChecklistRevised (PCLR) are as likely to earn high scores on theWelsh Anxiety Scale (WAS; Welsh, 1956) and other mea-sures of anxiety-negative affect as they are to earn lowscores (i.e., psychopathy and anxiety are statistically inde-pendent). Thus, investigators wishing to exclude secondaryor neurotic psychopaths from analyses (Cleckley, 1976;Hare, 1970) must use a measure of anxiety-negative affectin conjunction with the PCLR. To this end, we use theWAS to assign participants to low- and high-anxiety groups.

DEFICIENT RESPONSE MODULATION 93

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

were excluded from participation if they were age 40or older, currently taking psychotropic medication, orleft-handed (total scores on Chapman HandednessScale > 21; Chapman & Chapman, 1987). In addition,inmates were excluded from analyses if they scoredbelow the 4th-grade level on prison achievement testsor had earned estimated Wechsler Adult IntelligenceScaleRevised (WAISR; Wechsler, 1981) scoresthat were less than 70 on the Shipley Institute of Liv-ing Scale (Zachary, 1986). Psychopathic and nonpsy-chopathic participants did not differ significantly onage or WAISR scores (low-anxious psychopaths: M4 100.82, SE 4 11.54; low-anxious controls: M 499.63, SE 4 11.87; high-anxious psychopaths: M 498.04, SE 4 11.01; high-anxious controls: M 496.49, SE 4 12.70). Inmates were paid $3.00 for theirparticipation in this study.

Psychopathy was assessed with the PCLR (Hare,1991), which has demonstrated good reliability andvalidity in Caucasian samples (Hare, 1996; Hare etal., 1990; Harpur, Hare, & Hakstian, 1989; Kosson,Smith, & Newman, 1990; Lorenz, Smith, Bolt,Schmitt, & Newman, 2001). In the current study, weevaluated the interrater reliability of PCLR ratingswith a subset of 31 participants. To provide a conser-vative estimate of the reliability, we calculated anintraclass correlation (ICC) using absolute rather thanconsistency agreement and treated both raters and par-ticipants as random effects (McGraw & Wong, 1996).This analysis yielded an ICC of .96.

Consistent with Hare (1991) and previous reportsfrom our laboratory, participants with PCLR scoresgreater than or equal to 30 were rated as psychopathsand those with PCLR scores less than or equal to 20were rated as nonpsychopathic controls. Anxiety wasmeasured by the Welsh Anxiety Scale (Welsh, 1956).Individuals scoring above the median (11) were ratedas high-anxious and individuals scoring below themedian were rated as low-anxious. This median splitresulted in 37 high-anxious (17 psychopaths and 20controls) and 37 low-anxious (11 psychopaths and 26controls) participants.2

Stimulus Material

The stimuli in the lexical-decision task consisted of48 wordnonword pairs for the experimental trials.These stimuli consisted of 12 positive words, 12 nega-tive words, 24 neutral words, and 48 nonwords andwere grouped into four experimental blocks: A, B, C,and D. Each experimental block consisted of 3 posi-tive, 3 negative, 6 neutral, and 12 nonwords. In addi-

tion, a practice block consisted of 12 neutral wordsand 12 nonwords that differed from the words usedfor the test trials.

The positive, negative, and neutral words for theexperimental trials were selected from Rubin andFriendlys (1986) word list. On scales that rangedfrom 1 (negative) to 7 (positive), the neutral words fellbetween 3.25 and 4.75 on the Goodness scale andunder 2.0 on the Emotionality scale. Examples of neu-tral words were bowl, event, and vacuum. Thepositive and negative words were rated above 4.0 onthe Emotionality scale and above 5.25 or below 2.75on the Goodness scale, respectively. Examples ofpositive words were kiss, sunset, and heaven.Examples of negative words were tomb, devil,and hostage. As expected, the positive and negativewords differed significantly from the neutral words onemotionality, F(2, 47) 4 135.70, p < .001. The posi-tive, negative, and neutral words differed significantlyfrom each other on goodness, F(2, 47) 4 333.06, p