Demonstrating emotional processing differences in psychopathyusing affective ERP modulation
NATHANIEL E. ANDERSONa and MATTHEW S. STANFORDb
aMind Research Network, Albuquerque, New Mexico, USAbDepartment of Psychology and Neuroscience, Baylor University, Waco, Texas, USA
Abstract
Psychopaths exhibit abnormalities processing emotional information, but there is less certainty regarding the roleattention plays in these processes. We present data from two affective picture-viewing tasks comparing event-relatedpotential (ERP) modulation effects when emotional information is present but not task relevant (Task 1) followed by acondition directing attention to the categorization of emotional content (Task 2). Controls show a robust, persistent ERPpositivity (200–900 ms) associated with emotional target photos compared to neutral targets in both tasks. Individualswith psychopathy only showed this differentiation when explicitly attending to the emotional content of the photos (Task2), and these effects remained smaller than the amplitude differences demonstrated by controls. Although abnormalallocation of attention may play a critical role, this cannot completely account for emotional processing deficitsassociated with psychopathy.
Descriptors: EEG/ERP, Emotion, Psycopathological
A prominent concern in psychopathy research is the deficit peoplewith psychopathy have in processing emotionally relevant infor-mation. Cleckley’s (1941) original case studies illustrated callous-unemotional traits as a crucial element of psychopathy, andsubsequent research has emphasized a fundamental deficit in thebrain’s response to emotionally salient events, which has beendemonstrated with peripheral autonomic measures (Hare, 1968;Lykken, 1957), startle reflex modulation (Benning, Patrick, &Iacono, 2005; Patrick, Bradley, & Lang, 1993) and functional brainimaging (Birbaumer et al., 2005; Glenn, Raine, & Schug, 2009;Kiehl et al., 2001). Still, there exist competing theories about thefundamental processes underlying psychopathic individuals’muted affective response. Some suggest a basic failure to encodeemotionally relevant information at the amygdala (Blair, 2006),whereas others have suggested a fundamental deficit diverting andallotting attentional resources to emotionally salient events(Newman & Lorenz, 2003). Clarifying the nature of psychopathicindividuals’ affective deficits is an essential component forprogress in this field, which has broader implications for our under-standing of human thought and emotion in general.
The healthy brain deals with emotional information in a uniqueway: by preferentially devoting processing resources to stimuli thathave significance for the safety and survival of the species (Bradley,Codispoti, Cuthbert, & Lang, 2001; Ohman, Hamm, & Hugdahl,2000). This automatic, bottom-up filtering process is a major com-ponent of attention (Knudsen, 2007) and has robust consequenceson physiological responses, including autonomic arousal, startle
modulation (Bradley, Cuthbert, & Lang, 1990), and electrocorticalmeasures (Hajcak & Olvet, 2008). Many previous investigationshave examined the sensitivity of event-related potentials (ERPs) toemotional information in healthy individuals. The most prominentamong these emotion-sensitive components is the late positivepotential (LPP), which occurs as a positive deflection in the ERPwaveform at and beyond around 500 ms poststimulus and appearsto be elicited specifically by the presence of emotional information,independent of stimulus rarity or task relevance (Diedrich,Naumann, Maier, & Becker, 1997; Naumann, Bartussek, Diedrich,& Laufer, 1992). Another emotion-sensitive component that hasreceived some recent attention is the early posterior negativity(EPN), an enhanced negativity around 200 ms poststimulus foremotional content (Schupp, Junghofer, Weike, & Hamm, 2003a,2003b). However, the effect of emotional salience on ERPamplitude is not limited to these emotion-specific components and,in fact, appears to be a persistent effect across a wide range ofcomponents.
Voluntary, selective attention also plays a key role in ERPmodulation throughout the information processing stream, includ-ing early stimulus feature analysis, having reported effects on theP1, N1, P2 complex. The relative gain (amplification) of theseneural signals that accompanies attended stimuli has been sug-gested as the actual mechanism by which selective attention oper-ates (see Hillyard, Vogel, & Luck, 1998). This set of components,prior to about 200 ms poststimulus, shares the quality that attentionis not required to elicit them; however, voluntary selective attentiondoes modulate them. Subsequent ERP components such as the N2and P3 have increasingly complex sensitivities to task parametersand accompany higher order cognitive processes, including selec-tion and working memory. The N2 is indicative of changing
Address correspondence to: Nathaniel E. Anderson, Mind ResearchNetwork, 1101 Yale Blvd. NE, Albuquerque, NM 87131, USA. E-mail:[email protected]
features in the stimulus environment and has been interpreted as anautomatic filtering stage for selective attention toward novelty(Luck & Hillyard, 1994a, 1994b). The P3 marks the engagementof working memory in the evaluation of the stimulus environmentwhen an ongoing task requires identification of salient information(Donchin & Coles, 1988). If emotional information, per se, cap-tures attention automatically (Ohman et al., 2000), then one wouldexpect to see amplitude differences in ERP components beginningat a stage when emotional information is accessible to cognitiveprocesses.
Investigations of the time course of affective response usingelectrocortical measures have generally agreed that several standardERP components are sensitive to changes in emotional valence(e.g., Foti, Hajcak, & Dien, 2009; Keil et al., 2002). Reports haveindicated significant emotion-related amplitude modulation as earlyas 100 ms after an eliciting stimulus (Carretie, Hinjosa, Martin-Loeches, Mercado, & Tapia, 2004; Smith, Cacioppo, Larsen, &Chartrand, 2003). In an integrative review, Olofsson, Nordin,Sequeira, and Polich (2008) determined that component amplitudesare reliably modulated for affective stimuli, most consistently fornegative valences in early stages (<300 ms) and for both positiveand negative relative to neutral stimuli beyond 300 ms. The inter-pretation generally applied in these reports is that the differencesin amplitude represent the enhanced, preferential processing ofemotionally salient information consistent with long-establishednotions of orienting and attention (see Ohman et al., 2000).
As psychopathic individuals have physiologically demonstrabledeficiencies in the processing of emotional information, it is rea-sonable to suspect differences between psychopaths and non-psychopaths in their patterns of emotion-related modulation ofERPs; however, this has rarely been investigated. In fact, the use ofERPs in studying psychopathy has been almost exclusively devotedto determining grand amplitude differences between groups (e.g.,Kiehl, Bates, Laurens, Hare, & Liddle, 2006). Preliminary evi-dence for these expected differences in emotion-related ERP modu-lation comes from Williamson, Harpur, and Hare (1991), whoreported that psychopaths demonstrated reduced ERP differentia-tion between affective words and neutral words compared to non-psychopaths in a lexical decision-making task. Similarly, Howardand McCullagh (2007) reported that psychopathic individualsdemonstrate deficits in affective ERP modulation, but only underconditions when affective stimuli are relevant to the task at hand.Because of their specific methodological designs, there remainuntested ambiguities in defining psychopathic individuals’ neu-roaffective processing deficits using ERPs. For instance, in Wil-liamson and colleagues’ report, the affective content of the lexicalstimuli was never a directed primary focus for any task condition;therefore, attention to affective content was not systematicallytested. In the report by Howard and McCullagh, although pictureswith affective content were among the target stimuli in one condi-tion, participants were only explicitly required to differentiatebetween pictures of man-made objects and living things; thereforeaffective content was never a primary focus of attention. Further-more, affective pictures were never the response targets; therefore,the attention to affective content was an unmanipulated variable inthis case as well. Although it has been a theoretically relevantvariable in psychopathy literature for quite some time (Patterson &Newman, 1993), there are currently no existing reports usingERPs to investigate the moderating role of directed attention inpsychopaths’processing of emotional information, as has beenrecently demonstrated using startle eyeblink modulation (Newman,Curtin, Bertsch, & Baskin-Sommers, 2010).
The primary goal of this investigation was to adapt the ERPmethods of previous reports that had examined the time course ofaffective processing in healthy individuals and apply these methodsmore specifically to examine affective processing differences inthose with high psychopathic traits. We also set out to determinewhether directed attention to emotional content would impact anydifferences in affective processing associated with psychopathyas has been suggested in recent work by Newman and colleagues(2010) examining effects of attention in psychopaths. It washypothesized that psychopathic individuals would show reduced orabsent emotion-dependent modulation of ERP components com-pared to nonpsychopathic controls. Furthermore, it was hypoth-esized that psychopathic individuals would demonstrate relativelyaugmented differentiation between affective and neutral stimuliwhen their attention was explicitly directed to identifying emo-tional content compared to passive processing of this information,suggesting a specific deficit in the automatic allocation of attentionto emotionally salient information.
Method
Participant Recruitment
Volunteers from the local community completed the PsychopathicPersonality Inventory–Revised (PPI-R; Lilienfeld & Widows,2005), a self-report measure designed to identify psychopathicpersonality traits in a noninstitutionalized population. With theintended goal of comparing extreme groups on this measure, adver-tisements were placed in newspapers and distributed on fliers thatdescribed psychopathic personality traits, in nonpejorative terms,as desirable qualities for research participants (e.g., cunning,charming, intelligent, impulsive), similar to Widom (1977). Tofacilitate recruitment of low scorers, advertisements describingopposite traits were also used (e.g., modest, respectful of others,careful). Ultimately, all initial volunteers completed the PPI-R,and those with the highest and lowest scores, falling at least inthe upper and lower quartiles of published norms for their respec-tive age groups, were actively recruited for participation in thesecond phase of the study, which included psychophysiologicalmeasures.
A short preliminary inventory, given concurrently with thePPI-R, screened for any major exclusionary criteria. These con-sisted of a history of any prior major head injury, treatment for AxisI thought or mood disorders (examples of schizophrenia and majordepression were given), posttraumatic stress disorder, recent treat-ment for drug or alcohol dependence (not within the past 12months), epilepsy, any hearing deficits or required use of a hearingaid, and any uncorrected visual deficits. Participants reporting anyexclusionary criteria were precluded from selection for psycho-physiological assessment. Additionally, demographic informationwas gathered, including age, sex, race, and level of education.Willing participants who met criteria were scheduled for laboratorysessions, during which all additional data were collected, includingan assessment of intelligence using the verbal subscale (VIQ score)of the Wechsler Abbreviated Scale of Intelligence (WASI;Wechsler, 1999).
Psychophysiological Assessment
Participants who were selected to undergo psychophysiologicalassessment were scheduled to complete this portion of the study ina single laboratory session. After administration of the WASI for
Emotional processing differences in psychopathy 793
assessment of VIQ, participants underwent two components ofpsychophysiological assessment: a standard startle eyeblink modu-lation task and a novel ERP protocol, designed as an affectivepicture-viewing oddball task. It was decidedly beneficial to includean elementary assessment of emotional processing on a predictablepsychophysiological measure, with well-established relationshipswith the psychopathy construct. This would provide a manner ofvalidating that selected groups indeed demonstrate physiologicaldifferences related to emotional processing, which are traditionallyexpected to accompany psychopathic traits in addition to evalua-tion on the novel ERP protocol. Participants therefore underwent astandard affective picture-viewing eyeblink startle paradigm inaddition to a novel task developed to assess the modulation ofvarious ERP components with emotionally evocative stimuli.
Startle paradigm. Physiological recordings took place in asound- and light-attenuated, radio frequency anechoic chamber(Raymond EMC Enclosures Ltd., Ottawa, Ontario, Canada). Par-ticipants sat directly in front of a computer monitor, and were fittedwith a pair of noise-canceling headphones, through which thestartle probes were delivered. Subjects were instructed simply towatch the picture presentation and that they would occasionallyhear noises that sound like bursts of static and that these noisescould simply be ignored.
The startle modulation task utilized pictures from the Interna-tional Affective Picture System (IAPS; Lang, Ohman, & Vaitl,1988), a set of images with standardized ratings of affectivevalence and arousal level (Lang & Greenwald, 1988). Forty-fivetotal pictures1 were chosen from the set, with 15 pictures fallinginto each of three valence categories: pleasant, neutral, andunpleasant. These pictures were chosen such that neutral picturesdid not favor positive or negative valences and hade very lowarousal ratings. Emotional images were chosen such that valenceand arousal levels were matched for both male and female raters.That is, valences for pleasant and unpleasant pictures were equi-distant from neutral ratings with arousal levels matched.
During the presentation, all pictures remained on the screen for6 s, separated by 10-s intertrial intervals (ITIs), during which smallcrosshairs were centered on the screen to indicate a center of focus.Ten of the 15 pictures from each valence were paired with a startleprobe, a 50-ms, 100-dB white-noise burst with near instant rise.Probes occurred randomly between 3 and 5 s after picture onset.Pictures were ordered pseudo-randomly, such that no two picturesof the same valence, paired with a startle probe, occurred sequen-tially. The presentation began with three startle probes, 3 s apart,during an initial screen with crosshairs (same as ITI screen). Thiswas intended to acclimate the subject to the startle probe and toavoid the initial large blinks associated with initial probes. Theseblinks were not included in the analysis (cf. Patrick & Berthot,1995).
The magnitude of the startle response was measured by elec-tromyogram (EMG) recordings from the orbicularis oculi muscleof each subject’s right eye. A pair of Ag-AgCl electrodes (BiopacSystems Inc., Goleta, CA) was placed 1 cm below the lower lid ofthe right eye, one directly below the pupil, and a second 1 cm to the
right of that electrode. A third electrode was placed in the center ofthe forehead as a ground. Skin was prepared with isopropyl alcoholand a mildly abrasive gel (NuPrep) to improve surface conduction;Signa gel brand saline gel was used as a conducting medium. EMGsignals were collected with a sampling rate of 2,000 Hz and a10–500-Hz bandpass filter, rectified and integrated with a timeconstant of 10 ms. Blink magnitudes were defined as a smoothedEMG signal, recorded as baseline to peak differences for eachstartle probe. The baseline was defined as the mean orbicularisoculi EMG reading during the 25 ms prior to the onset of the noise;peaks were defined as the maximum EMG amplitude between 40and 150 ms after the onset of the noise.
Affective oddball ERP task. The ERP task was designed to elicittarget-relevant ERP components, including N2, P3, and LPP aswell as earlier sensory components, allowing comparisons ofamplitude within subjects for ERP components elicited by affectiveversus neutral visual target conditions. The protocol was alsodesigned such that the task relevance of the emotional content oftarget pictures could be manipulated across two variations of thesame task. Each participant engaged in both versions of the task, sothat comparisons of task condition could be made both withinsubjects and across groups.
Participants were instructed to view a picture presentationconsisting of two types of images: a repeated abstract design andintermittent actual photographs (response targets). Their task wasto respond as quickly as possible (under the guise of a reaction-time test) by pressing a single response button whenever an actualphotograph was presented. The picture presentation was designedusing slides from the IAPS (Lang et al., 1988). A total of 40photographic targets were used2 and occurred with an overall prob-ability of .12. Pictures remained on the screen for 2 s, with a500-ms interstimulus interval. Target stimuli were comprised oftwo varieties: negatively valenced emotional photos and neutralphotos, each with equal probability (20 of each target variety). Thedecision to exclude positively valenced targets was made to elimi-nate superfluous evaluation of a possible negativity bias in earlystage processing (see Cacioppo & Gardner, 1999), which was irrel-evant to the research questions of primary interest. In this versionof the task (Task 1), the participants were unaware that the photo-graphic targets would vary in emotional content, and, in fact, thisinformation is irrelevant to their performance on the task. Anydifferences between ERPs elicited by these categories of targetswould represent implicit differentiation between emotional andneutral target photos.
A second version of the task (Task 2) was implemented as amanipulation of the task relevance of the emotional content. ForTask 2, participants were informed that they would see a presenta-tion similar to that in the first task (in fact, it was identical), butinstead of simply responding to all target photos, they were tocategorize each photo as either emotionally evocative or neutral bymeans of pressing one of two response buttons. They were encour-aged to do this as quickly as possible, but not to sacrifice accuracyfor speed. In contrast to Task 1, the emotional content of thephotographic targets was task relevant, and, in fact, their attentionto this content was explicitly required. All participants performedthis task after Task 1. The decision was made not to counterbalance
these task conditions, as changing the task order would compro-mise the intended purpose of Task 1, which was to minimizevoluntary attention to variation in emotional content. Furthermore,the main effects of interest were Group ¥ Target Category interac-tions for each task condition, which minimizes the relevance of anytask order effects across groups.
Aside from response targets, both tasks included a no-go con-dition that was intended to assess potential differences in affectivemodulation between two varieties of the P3: the P3a and the P3b.The P3a is similar in form to the standard target-detection P3b, butis more frontally distributed and is traditionally elicited by novelstimuli that do not match target response conditions or by distractorstimuli that indicate the need to inhibit a prepotent response con-dition (see Polich, 2007). The no-go stimuli were similar to pho-tographic targets. They were unique neutral photographs from theIAPS but were superimposed with words from the Affective Normsfor English Words list (ANEW; Bradley & Lang, 1999). ANEWwords were either neutral or negative and were matched for wordlength, commonality of usage, and arousal levels per gender (i.e.,words are equally arousing for both men and women). A total of 40no-go stimuli were used (20 emotional, 20 neutral), and this cat-egory also occurred with an overall probability of .12. Participantswere instructed that when they encountered photographic targetssuperimposed with a word, they were to inhibit their response andsimply read the word silently. Their instructions for the no-gostimuli remained the same for both tasks.
Electroencephalogram (EEG) data were recorded from scalpsites using fitted, elastic caps (Electro-Cap International, Inc.,Eaton, OH) consisting of 64 tin electrodes arranged in the inter-national 10-20 system, with standard and intermediate positions,and Electro-Gel brand saline gel was used as a recording medium.Participants’ scalps were prepared with isopropyl alcohol andNuPrep abrasive gel to improve surface conduction. Impedancesof electrodes were kept below 10 kW. During recording, scalpelectrodes were referenced to a single electrode, Cz, and re-referenced off-line to electrodes affixed to the mastoids in post-processing. Four electrodes placed around the subjects eyes wereused to record blinks and eye movements, for off-line removal ofocular artifacts via spatial filter. The EEG was recorded using aNeuroscan Synamps2 system, digitized at a sampling rate of1,000 Hz and amplified with a gain of 1,000, and stored for off-line analysis. Postprocessing consisted of a digital bandpass filter(0.1 Hz to 35 Hz, 12 dB/oct slope) followed by removal of ocularartifacts using a spatial filter (Neuroscan) created for each subjectusing the electrooculogram (EOG), which was applied across thecontinuous data to reduce contamination of the EEG by eye move-ments. This process was followed by visual inspection and manualdeletion of incidental epochs contaminated by other sources ofnoise; however, manual deletions were minimal, and most partici-pants required no further artifact rejection beyond the spatial filterfor eyeblink removal. Groups were nearly equal in the number ofretained epochs per subject, which remained very near 20 trialsper condition. The EEG was re-referenced to averaged mastoids.Epochs were isolated from -100 ms to 900 ms after stimulus, anda correction to baseline was implemented per epoch. Trials wereaveraged within subjects to obtain relevant ERPs, time-locked toeach target category described above. ERP amplitudes weredefined as the maximum positive or minimum negative peakswithin relevant intervals relative to baseline, 100 ms prior tostimulus onset with intervals as follows: N1 (50–150 ms), P2(125–225 ms), N2 (225–350 ms), P3 (250–500 ms), and LPP(500–900 ms).
Hypotheses and Data Analysis
Analyses were carried out with the goal of exhibiting differences inboth affective startle modulation and affective ERP modulationbetween high and low scorers on the PPI-R. Startle data havewell-established expectations to reveal smaller or absent startlepotentiation effects in those with high psychopathic traits; however,the ERP analysis was more exploratory. It was expected thatincreased amplitudes for several ERP components would accom-pany emotionally evocative pictures for nonpsychopathic controlsubjects, but those with high psychopathic traits would fail todemonstrate modulation of at least one ERP component, but pos-sibly several. Furthermore, it was expected that conditional direc-tion of attention toward emotional information as a task-relevantproperty of the photographic targets would promote ERP differen-tiation between stimulus categories for psychopathic individuals.For all analyses, group divisions were determined by PPI-R totalscore, and it should be noted that any potential divisions by PPI-Rfactor scores did not change the composition of the groups, as theywere selected for extreme scores on this measure.
For analysis of the startle data, standardized blink amplitudeswere analyzed with a (2 ¥ 3) mixed-model anlysis of variance(ANOVA) comparing two groups (High Psychopathy ¥ Low Psy-chopathy) and three picture valences (pleasant, neutral, andunpleasant). A significant Psychopathy Group ¥ Picture Valenceinteraction would be the main effect of interest and would indicatedifferent patterns of blink modulation between groups, whichwould be further analyzed by comparing scores at individualpicture valences.
For analysis of ERP data, each component peak was analyzedseparately and submitted to a (2 ¥ 2 ¥ 9) mixed model ANOVAcomparing the two groups (High Psychopathy ¥ Low Psychopa-thy) on two repeated measures variables representing emotionalcontent (emotional vs. nonemotional) and electrode site repre-sented by nine regions (F3, Fz, F4, C3, Cz, C4, P3, Pz, P4) toaccount for potential topographical variations between groups.Analyses were also carried out dividing these regions into two(3 ¥ 3) repeated measures accounting for lateral scalp region (left,middle, right) and anteroposterior scalp region (frontal, central,parietal) in order to distinguish any differences in scalp distributionof component amplitude differences between groups; however,none were expected. The main effects of interest were any Psycho-pathy Group ¥ Emotional Content interaction, which, similar to thestartle analysis, would indicate differences in emotional modula-tion of ERP amplitude and could be explored in more detail withfollow-up tests. This same set of analyses was carried out for bothERP tasks. For all repeated measure analyses Greenhouse–Geysercorrections were implemented where the sphericity assumptionwas violated, and these corrections are reflected in reported values;these corrections did not impact significance for any analysesreported below.
Results
Participants
Analyses were carried out on data from 40 total participants (20 pergroup). Of these 40, 21 were female, 32 were Caucasian/Whitewith 7 Hispanic and 1 Asian/Pacific Islander. The mean age was 25years, with a range of 18 to 57. Examining possible differencesbetween groups on variables of gender, age, race, and WASI VIQrevealed no significant effects. Level of education was divided intofive categories and also revealed no significant differences between
Emotional processing differences in psychopathy 795
groups. The only significant differences between groups on subjectvariables were the intended differences on PPI-R scores. Becauseindividual percentile ranks differ based on gender and age, averagepercentile ranks (which take these variables into account) are usedto represent group characteristics. These results are summarized inTable 1.
Startle Data
When we examined raw blink amplitudes, collapsed across valencecategory, groups did not differ overall, t(38) = .189, p > .80. Exam-ining blinks standardized within subjects, the (2 ¥ 3) mixed modelANOVA revealed a main effect of valence, F(2,76) = 22.43,p < .001, h2 = .371, indicating that the amplitude of blinks variedsignificantly across the three affective valences of pictures. Thiseffect was superseded by a Psychopathy Group ¥ Valence interac-tion, F(2,76) = 4.66, p < .02, h2 = .109, indicating that the modu-lation effects due to picture valence vary significantly between thetwo groups.
Follow-up repeated measures ANOVAs for each group revealedthe nature of the significant modulation differences. Nonpsycho-pathic individuals demonstrated a strong effect, F(2,38) = 17.28,p < .001, h2 = .476, of the typical modulation patterns for stand-ardized blink magnitudes such that blinks during negative pictureswere larger (mean z = .311) and blinks during positive pictureswere smaller (mean z = -.254) relative to those during neutralpictures (mean z = -.058). Paired t tests confirmed that nonpsycho-pathic people demonstrated significant blink potentiation duringnegative pictures, t(19) = 3.10, p < .01, and significant blink attenu-ation during positive pictures, t(19) = 2.10, p < .05. Those withhigh psychopathic traits, however, exhibited a different modulationpattern, while still showing a large valence-dependent modulationeffect, F(2,38) = 10.09, p < .001, h2 = .347. This group exhibitedsmaller blink amplitudes during both positive (mean z = -.261) andnegative (mean z = .100) pictures relative to neutral ones (meanz = .161). Paired t tests revealed that the attenuation effect during
positive pictures was driving the overall amplitude effect for thehigh psychopathic trait group, t(19) = 3.981, p < .001, whereas thesmall difference in amplitude between negative and neutral pictureswas not significant, t(19) = .646, p > .50. A comparison of blinkmodulation patterns for each group is represented in Figure 1.
ERP Behavioral Performance Data
Although no formal hypotheses were made regarding behavioralperformance data, these outcomes are relevant for interpretation of
Note. SCI = Self Centered Impulsivity factor of PPI. FD = Fearless Dominance factor of PPI. There were no significant differences between groups on anydemographic data; the only differences were those defined by group division on PPI-R total and factor scores.*p < .001.
Figure 1. Nonpsychopathic individuals show the typical startlemodulation patterns with significantly potentiated eyeblinks viewingnegative photos and significantly attenuated responses during positivephotos. Psychopathic individuals’ attenuation during positive photos wasintact, but they lacked the effect of blink potentiation during negativephotos.
796 N.E. Anderson and M.S. Stanford
ERP modulation effects described below. Performance data wereanalyzed using a mixed model ANOVA with psychopathy catego-rization as a between-groups factor and task variables as repeatedmeasures. Behavioral performance was expectedly different acrossthe two versions of the ERP modulation task, F(1,38) = 120.19,p < .001, h2 = .760. Response latencies were much longer on Task2, 95% confidence interval (CI) [737.69 ms, 866.59 ms], comparedto Task 1, 95% CI [510.50 ms, 611.87 ms], indicating participants’cognitive appraisal of the emotional content of the photographictargets in Task 2. These differences in response latency across tasksdid not vary significantly between groups. There was a significantmain effect for target category (emotional vs. neutral) on reactiontimes, F(1,38) = 34.80, p < .001, h2 = .478, such that, collapsedacross psychopathy groups, reaction times were faster for emo-tional targets than for neutral targets. For Task 1 emotional targets95% CI was [517.82 ms, 592.63 ms]; for Task 1 neutral targets95% CI was [537.97 ms, 616.83 ms]. This difference was simplyexaggerated for Task 2, as participants took longer to categorizestimuli: Task 2 emotional targets, 95% CI [681.75 ms, 763.75 ms],Task 2 neutral targets, 95% CI [754.95 ms, 865.87 ms]. Theseexaggerated differences for Task 2 led to a significantTask ¥ Category interaction effect, F(1,38) = 13.43, p < .001,h2 = .261, but the main effect for emotional category remainedsignificant for both tasks analyzed independently; that is, responseswere faster for emotional content compared to neutral content forboth Task 1, F(1,38) = 6.17, p < .02, h2 = .140, and Task 2,F(1,38) = 30.35, p < .001, h2 = .444. Responses were simply muchfaster for emotional content in Task 2 compared to Task 1. Again,there were no significant differences in reaction time betweengroups, nor were there significant interactions involving groups.
The number of task errors for each subject was also examined,including errors of commission and omission on Task 1 and exam-ining emotional content categorization on Task 2. Groups did notdiffer on the number of errors; in fact, error rates were very smallover all. Each group averaged between one and two errors ofcommission and less than one error of omission. For Task 2, eachgroup averaged less than one categorization (emotional vs. neu-tral) discrepancy per individual, an expectedly low rate due tothe obvious differences in emotional content between targetconditions.
ERP Affective Modulation Data
Task 1: Implicit differentiation. Mixed model ANOVAs wereperformed for each of five ERP components (N1, P2, N2, P3, andLPP), using electrode sites as repeated measures, to compareamplitudes across emotional valence categories (within groups)and to examine overall amplitude differences between groups aswell as interactions between these effects that would indicate dif-ferences between groups in the patterns of emotional ERP modu-lation. Because of the extensive nature of these data and numerouseffects of interest, pertinent effects will be described below, and adetailed account of F values and probabilities will be given inaccompanying tables.
For Task 1, in which the emotional content of photographs wastask irrelevant, a significant main effect for target picture categoryindicated that amplitudes were significantly increased in the posi-tive direction for emotional targets compared to neutral targets atthe P2, N2, P3, and LPP components, but not the N1 component.This main effect was superseded by a significant TargetCategory ¥ Psychopathy Group interaction at components P2, N2,P3, and LPP, but again not at the N1 component. Between-groups
effects were not significant; therefore neither group demonstratedlarger amplitudes for any of the components when the data werecollapsed across target category. These results are given in moredetail at the top of Table 2 and in Figure 2. The trends evident inthis set of analyses suggest that the Group ¥ Target Content inter-action, then, represents different emotional modulation patternsbetween groups.
To confirm this, follow-up ANOVAs were conducted for eachpsychopathy group separately. These analyses revealed thatnonpsychopathic individuals indeed exhibited significant positiveincreases in amplitude for all ERP components excluding theN1 during photographic targets with emotional content. F statis-tics ranged from 14.62 to 48.27, all p values < .001; N1:F(1,19) = 0.002, p > .90, h2 = .000; P2: F = 14.62, p < .001,h2 = .435; N2: F = 27.94, p < .001, h2 = .595; P3: F = 22.86,p < .001, h2 = .546; LPP: F = 48.27, p < .001, h2 = .718. For thosewith high psychopathic traits, however, there were no significantdifferences between amplitudes for emotional and nonemotionaltargets for components prior to the LPP, and the valence effect atLPP was only marginally significant, F(1,19) = 4.33, p = .051,h2 = .186.
None of these effects were superseded by three-way interac-tions involving electrode site, which would have indicated unex-pected regional scalp differences in ERP modulation patterns.Differences in scalp distributions for ERP components acrossgroups were neither evident nor expected; therefore, we opted notto normalize amplitude measures as has been suggested for suchanalyses (McCarthy & Wood, 1985). The ANOVA on each com-ponent revealed an (expected) main effect for electrode site, indi-cating larger amplitudes near specific topographical sites. Forexample, the P3 is largest around the centro-parietal midline, andthe N2 is largest around fronto-central regions. These component-specific topographical distributions were typical and did not varyby group.
ERPs were also examined for the no-go stimuli that includedwords (emotional or neutral) superimposed over neutral pictures.Again, these stimuli were intended to elicit a no-go or inhibitory setof components, including what might be categorized as a P3a. Thesame analyses were done for these components, and these resultsare given at the top of Table 3. Significant main effects for wordcategory (emotional vs. neutral) revealed that ERP components hadlarger positive amplitudes for emotional words compared to neutralwords at the N1, N2, P3, and LPP components, but not the P2component. For these no-go evoked responses, there were no sig-nificant between-groups effects, nor were there any significantinteractions involving the grouping variable, although main effectsizes were generally smaller for this modulation effect acrossgroups, as can be seen in Figure 3. The only other significanteffects for these stimuli were the main effects for electrode siteindicating the topographical distribution of the particular compo-nent. Again, these topographical distributions were typical and didnot vary by group.
Follow-up ANOVAs were carried out to examine groups sepa-rately on the no-go stimuli as well. Psychopathic individuals exhi-bited no significant main effects for emotional category of thewords at any of the ERP components. For nonpsychopathic sub-jects, the emotional valence main effects were relatively small,with the exception of the LPP, and were nonsignificant in somecases; N1: F(1,19) = 5.25, p < .04, h2 = .217; P2: F = 1.17, p > .20,h2 = .058; N2: F = 2.95, p > .10, h2 = .134; P3: F = 5.80, p < .04,h2 = .217; LPP: F = 26.07, p < .001, h2 = .578. The absence of aGroup ¥ Emotional Category interaction effect in the preceding
Emotional processing differences in psychopathy 797
analysis appears, then, to be a result of smaller emotional modula-tion overall among nonpsychopathic individuals for these lexicalstimuli, whereas emotional modulation remained effectively absentfor psychopathic individuals, similar to results for photographictargets.
Task 2: Explicit categorization. For Task 2, in which participantswere required to attend to the emotional content of the photo-graphs, ANOVAs revealed that the emotional category of thepicture (emotional vs. neutral) again had a significant effect oncomponent amplitude, this time at all components (N1, P2, N2, P3,LPP), demonstrating that emotional modulation was preservedacross the two tasks. As in Task 1, these effects were superseded bysignificant Group ¥ Target Category interactions at N1, P2, P3, andLPP, but not N2. There were no significant between-groups differ-ences in amplitude collapsed across valence category; that is,neither group exhibited larger overall component amplitudes forTask 2. These results are summarized in the lower half of Table 2and in Figure 4.
Follow-up ANOVAs run on each group separately revealed thenature of these interaction effects. For nonpsychopathic individu-als, just as in Task 1, robust emotional modulation was apparentbeginning about 200 ms onward; N1: F(1,19) = 0.08, p > .70,h2 = .004; P2: F = 22.15, p < .001, h2 = .538; N2: F = 13.77,p < .001, h2 = .420; P3: F = 32.69, p < .001, h2 = .632; LPP:F = 38.50, p < .001, h2 = .670. Psychopathic individuals, in
contrast to Task 1, demonstrated significant modulation effectsbetween target categories at most components; however, the effectsremained relatively smaller than those for nonpsychopathic sub-jects, thereby driving the significant interaction effects apparent inthe mixed-model analysis; N1: F(1,19) = 9.56, p < .01, h2 = .335;P2: F = 3.59, p = .07, h2 = .159; N2: F = 8.08, p < .010, h2 = .298;P3: F = 19.19, p < .001, h2 = .502; LPP: F = 11.49, p < .01,h2 = .377.
Another means of analyzing the effect of task requirements is toexamine groups individually using task condition as a within-subjects repeated measure. A significant Task ¥ Target Categoryinteraction would indicate a change in emotion-dependent ERPamplitude from Task 1 to Task 2. Nonpsychopathic subjects exhib-ited no significant Task ¥ Target Category interactions at any ERPcomponent (all Fs < 1.00), which suggests that task requirementshad no further impact on the affective modulation patterns alreadyevident between target conditions during Task 1. Those withhigh psychopathic traits, however, exhibited two significantTask ¥ Target Category interactions: N1 showed a significant nega-tive amplitude increase, F(1,19) = 13.34, p < .01, h2 = .412, and P3showed a significant positive increase, F(1,19) = 7.25, p < .02,h2 = .276, all other Fs < 1.00. These effects can be seen by com-paring this group’s modulation patterns in Figures 2 and 3.
The negative amplitude increase for N1 is a curious deviationfrom the other patterns reported here, as it is the only negative-going modulation trend evident in the entire study. This effect is
Table 2. ERP ANOVA Results for Photographic Target Stimuli
Component Effect F value (1,38) p value Effect size (h2)
Task 1N1 Stimulus category 0.164 .688 .004
Stimulus ¥ Group 0.116 .736 .003Between groups 2.168 .149 .054
P2 Stimulus category 12.191 .001 .243Stimulus ¥ Group 4.222 .047 .100Between groups 0.332 .568 .009
N2 Stimulus category 23.006 <.001 .377Stimulus ¥ Group 4.458 .041 .105Between groups 0.600 .443 .016
P3 Stimulus category 21.902 <.001 .366Stimulus ¥ Group 13.006 .001 .255Between groups 0.061 .806 .002
LPP Stimulus category 37.309 <.001 .495Stimulus ¥ Group 8.725 .005 .187Between groups 1.338 .255 .034
Task 2N1 Stimulus category 6.072 .018 .138
Stimulus ¥ Group 4.270 .046 .101Between groups 0.172 .681 .004
P2 Stimulus category 23.571 .000 .383Stimulus ¥ Group 3.495 .069 .084Between groups 1.522 .225 .039
N2 Stimulus category 17.411 <.001 .314Stimulus ¥ Group 1.762 .192 .044Between groups 0.172 .681 .005
P3 Stimulus category 50.728 <.001 .572Stimulus ¥ Group 5.789 .021 .132Between groups 1.233 .274 .039
LPP Stimulus category 47.715 <.001 .557Stimulus ¥ Group 6.653 .014 .149Between groups 0.083 .775 .002
Notes. For photographic targets, trends show emotional modulation differences in ERP amplitudes between people with psychopathy and controls as earlyas about 200 ms when affective content is processed passively. When affective content is processed actively, differences appear earlier, but less consistentlyand perhaps with smaller effect sizes.
798 N.E. Anderson and M.S. Stanford
discussed in more detail below. Considering this and the posi-tive amplitude increase at P3 in Task 2, these data suggest that,where emotional processing deficits are present, potentially inter-pretable cognitive changes accompany changes in task parametersinfluencing attention. As in Task 1, significant main effects forelectrode site indicate a typical scalp distribution of these ERPcomponents, and there were no significant Group ¥ Electrode Siteinteractions.
ERPs were also examined for the lexical no-go stimuli in Task 2,although the instructions for these stimuli remained the same(simply read the words silently). Interpretation of this set of ERP
components is theoretically ambiguous, as no button pressing wasinvolved for these stimuli, only the context for these stimuli changedas responses for categorizing target stimuli were required in thistask. As shown in the lower half of Table 3, these data produced nosignificant differences at any ERPcomponent for emotional content,between-groups differences, or Group ¥ Target interactions; there-fore, their spatial averages have been omitted here. Standard ERPcomponents were still present, and significant main effects forelectrode site were still present for each ERP component indicatingtypical scalp distributions, but these components did not varysignificantly based on emotional content or group for Task 2.
Figure 2. Significant positive amplitude increases of ERP components elicited by photographic target stimuli for nonpsychopathic individuals is apparentbeginning at approximately 200 ms (P2) and continues through LPP. Psychopathic individuals only demonstrate marginal positive enhancement of the LPPand no apparent differentiation at earlier processing stages. Waveforms are referenced to mastoids.
Emotional processing differences in psychopathy 799
Discussion
Although many have examined the time course of affectiveprocessing using ERPs in healthy individuals, very few haveapplied these techniques to examine individual differences in per-sonality and pathology despite obvious applications to this realm ofinvestigation (Davidson, 1998). The outcomes presented heredemonstrate that affective ERP modulation is a useful tool forexhibiting emotional processing abnormalities associated withpsychopathy and provide preliminary evidence that mechanisms ofvoluntary attention are capable of moderating these differences inpsychopathic individuals, but only to a limited degree. Further-more, the methodology described here may prove to be a versatileand adaptable protocol effective for illuminating specific featuresof emotional and cognitive processing.
Startle Modulation
Lack of startle potentiation has been a psychophysiological hall-mark of psychopathy since Patrick and colleagues (1993) demon-strated this in an incarcerated sample. It was utilized here as ameans of verifying group differences on a psychophysiologicalmeasure commonly attributed to psychopaths, which adds credenceto the construct validity of the presently defined groups based onpsychopathic traits. The patterns exhibited here reflect those com-monly reported previously. Nonpsychopathic individuals demon-strated significant blink potentiation during aversive photos andblink attenuation during pleasant photos. Those with high psycho-
pathic traits showed divergent blink modulation patterns, lackingthe potentiation effect ordinarily seen under aversive conditions.These effects demonstrate that the selected groups exhibit both thepersonality traits (as measured by the PPI-R) and physiologicalcharacteristics consistent with psychopathy classified by variousmethods and in various populations.
ERP Behavioral Performance
The large differences in response latencies between task conditionswere an expected consequence of the deliberation required to cat-egorize the stimuli as emotional or nonemotional. Participantsrequired on average approximately 200 ms longer to respond inTask 2 compared to Task 1, regardless of group membership. Therewere no group differences in accuracy of categorization; the over-whelming equivalence of participants’ individual categorization ofthe photos in Task 2 attests to both the clarity of emotional contentand to sustained attention to the affective content across bothgroups. The absence of between-groups differences in responselatency or accuracy of target categorization suggests that the dif-ferences observed between groups in ERP waveforms are notdirectly related to the behavioral outcome of explicit emotivecategorization. That is to say, psychopathic individuals were noworse at identifying emotional content in photographs based onspeed or accuracy of categorization.
There have been some suggestions that psychopathic indivi-duals have an impaired ability to recognize emotional information,especially judging subtle emotive facial expressions (e.g., Kosson,
Table 3. ERP ANOVA Results for Lexical, No-Go Stimuli
Component Effect F value (1,38) p value Effect size (h2)
Task 1N1 Stimulus category 8.400 .006 .181
Stimulus ¥ Group 0.529 .472 .014Between groups 0.238 .629 .006
P2 Stimulus category 2.290 .138 .057Stimulus ¥ Group 0.066 .799 .002Between groups 2.833 .101 .069
N2 Stimulus category 4.673 .037 .109Stimulus ¥ Group 0.464 .500 .012Between groups 0.132 .718 .000
P3 Stimulus category 9.610 .004 .202Stimulus ¥ Group 0.448 .507 .012Between groups 0.070 .793 .002
LPP Stimulus category 13.983 .001 .269Stimulus ¥ Group 3.143 .084 .076Between groups 0.037 .849 .001
Task 2N1 Stimulus category 0.085 .773 .002
Stimulus ¥ Group 3.198 .082 .078Between groups 0.077 .783 .002
P2 Stimulus category 1.657 .206 .042Stimulus ¥ Group 2.741 .106 .067Between groups 0.017 .898 .000
N2 Stimulus category 0.812 .373 .021Stimulus ¥ Group 0.446 .508 .012Between groups 0.165 .687 .004
P3 Stimulus category 0.789 .380 .020Stimulus ¥ Group 3.671 .063 .088Between groups 0.363 .550 .09
LPP Stimulus category 1.416 .241 .036Stimulus ¥ Group 0.288 .594 .008Between groups 0.949 .336 .024
Notes. For lexical stimuli compared to photographic stimuli, there appear to be smaller overall effects for passive emotional modulation, making differencesin modulation between groups inconsequential. In Task 2 the emotional modulation effects are no longer apparent in either group.
800 N.E. Anderson and M.S. Stanford
Suchy, Mayer, & Libby, 2002). The present data suggest that theemotional content was conspicuous enough for this not to be adetermining factor in the observed neural processing differences.The observed differences in affective modulation of ERP wave-forms must therefore translate into neural processing abnormalitiesthat do not impair overt identification of emotionally relevantvisual information in late-stage processing (at or around 700 ms),concordant with behavioral responses.
Response latencies were shorter for emotional targets comparedto nonemotional targets for both versions of the task for both
groups. This difference was simply exaggerated under conditionsof target categorization; that is, both groups took longer to catego-rize neutral photos than to categorize emotional photos as such.This main effect for emotional content is consistent with a largebody of research suggesting that emotionally relevant informationis capable of facilitating reflexes and behavioral responses (Ohman,Flykt, & Esteves, 2001; Strauss, 1983).
It may seem dubious that psychopathic individuals would dem-onstrate facilitated responses. However, it has been suggested thatpsychopaths’ basic orienting response may be intact (Patrick,
Figure 3. Emotional lexical no-go stimuli only modestly modulated ERP components for nonpsychopathic subjects. Psychopathic subjects demonstrated nosignificant emotion modulation by these stimuli at any stage of component analysis. Waveforms are referenced to the mastoids.
Emotional processing differences in psychopathy 801
1994), and others have suggested that psychopaths’ orientingresponse may only be impaired when this information is peripheralto ongoing behavior (Hare, 1968; Patterson & Newman, 1993). Infact, certain reports indicate that psychopaths are capable of normalfacilitation of reflexes when a task requires direct focus on emo-tional stimuli (Newman et al., 2010), when peripheral informationis spatially contiguous with information required for a primary task(Hiatt, Schmitt, & Newman, 2004), and when there is no conflict inpunishment–reward contingency (Newman & Kosson, 1986). Animportant conclusion here is that differences in ERP modulation
across groups cannot be accounted for by any immediate behavio-ral performance differences.
ERP Modulation
Task 1 interpretation. Results from target stimuli in Task 1 inthe ERP protocol clearly indicate differences between groups inprocessing emotional stimuli. Nonpsychopathic individuals exhi-bited a global, persistent, enhanced positivity in waveforms elic-ited by affective targets as compared to neutral targets. In contrast
Figure 4. When actively categorized as emotional or non-emotional, photographic targets produced similar modulation patterns as in Task 1 fornonpsychopaths, but elicited significantly more pronounced modulation patterns in psychopathic individuals, an effect absent for this group in Task 1.Waveforms are referenced to the mastoids.
802 N.E. Anderson and M.S. Stanford
to an overall increase in amplitude across the average ERP wave-form, which would magnify values in both the positive and nega-tive directions, this effect represents an exaggerated positive shiftaround 200 ms, which is sustained across the remaining epoch.This enhanced positivity was absent in psychopathic individuals,who presented with a relatively late positive shift, which remainedmodest compared to the emotion-dependent differences observedin nonpsychopathic individuals. Further attention must be given tothe presence of this persistent emotion-related positivity innonpsychopathic individuals before a broader interpretation canbe made regarding its absence or diminishment in people withpsychopathy.
Previous investigations of affective ERP modulation in healthyindividuals have sometimes reported early (<300 ms) enhancednegativities (e.g., Junghofer, Bradley, Elbert, & Lang, 2001; Keilet al., 2002), and others have demonstrated early enhanced posi-tivities (e.g., Cuthbert, Schupp, Bradley, Birbaumer, & Lang,2000; Palomba, Angrilli, & Mini, 1997). A great deal of this vari-ability may be accounted for by stimulus presentation rates,which vary widely in these studies (Bradley & Lang, 2007), andsome may be accounted for by specific task parameters; forinstance, many of these investigations have implemented simplepassive picture-viewing protocols. The persistent, positiveincreases demonstrated here are likely an effect related to rela-tively slow picture presentations and the requirements of a targetdetection task.
Most resources agree, at or near 300 ms and beyond, affectivecontent contributes to enhanced positive amplitude in ERP compo-nents (Olofsson et al., 2008). This is consistent with literature inmemory encoding, which has demonstrated strong relationshipsbetween middle-late ERP positivity and subsequent memory foritems, an effect that is apparently tied to the distinctiveness of itemsrecalled (Fabiani, Karis, & Donchin, 1990; Rugg, 1995). The per-sistent ERP positivity in nonpsychopathic individuals likelyreflects recognition of a distinctive quality in affective photos.These data suggest that people with psychopathy fail to engage inearly discriminatory processes that demarcate emotionally salientevents, but are apparently able to effectively categorize thesestimuli as such in a relatively simple behavioral task. The preser-vation of this behavioral operation may therefore be due to somealternative or compensatory neural mechanisms, which are perhapsrepresented by the modest, late-onset ERP modulation apparentaround 500 ms.
ERPs elicited by lexical no-go stimuli were also examined.These stimuli were intended to elicit an inhibitory, no-go P3, orwhat might be categorized as a P3a. Topographical scalp distribu-tions for this P3 were not substantially different from those gener-ated by the affective photographic targets; that is, peaks werelargest in the parietal regions for both. In reviewing features of bothP3a and P3b, Polich (2007) discussed distinctions between the“novelty P3a,” and the “no-go P3a,” suggesting they are likelyvariants of the same potential but with slightly different topo-graphical scalp distributions, with the no-go P3 having a morecentral/parietal distribution. They are both distinguishable fromthe canonical P3b in that prefrontal cortex is clearly integral as aneural generator for both P3a variants (Knight, 1984; Knight,Grabowecky, & Scabini, 1995), whereas neural generators of theP3b are more widely distributed (Soltani & Knight, 2000). Thedistribution of potentials in the present data may cast some doubton whether these stimuli successfully generated a P3a or whetherthey simply produce a lexically induced P3b, and therefore inter-pretations should be made cautiously.
P3s generated by the lexical no-go stimuli in Task 1 producedessentially the same pattern of effects as the visual target stimuli;however, these effects were much smaller than those elicited byphotographs. The no-go stimuli containing emotional words (com-pared to neutral words) produced larger, more positive potentialsacross several components in the processing stream for nonpsy-chopathic individuals (excluding P2), but there were no significantdifferences between emotional and nonemotional conditions forthose with high psychopathic traits at any component stage duringTask 1. These results and their potential interpretations are con-gruent with those described above. The smaller effect sizes arepotentially due to the slightly more challenging cognitivedemands. In general, divided attention and increases in taskdemands produce reliably smaller overall potentials (Kok, 2001),which may limit the size of differences between differentiablestimulus categories.
Task 2 interpretation. The aim of Task 2 was to provide a com-parison condition under which automatic and implicit differentia-tion of emotional content could be contrasted with active andvoluntary attention to emotional content within subjects. It washypothesized that psychopathic individuals would show increasedERP differentiation between affective and neutral stimuli whentask requirements demanded attention to the emotional content ofthe photos, as is suggested by recent work by Newman and col-leagues (2010). Results indicated that this group indeed demon-strated increased differentiation between emotional and neutraltargets for this task; however, the effect sizes for these differencesremained modest compared to the affect-related modulation exhib-ited by nonpsychopathic individuals. Follow-up tests revealed thatnonpsychopathic individuals showed virtually identical modulationpatterns for the two tasks, which may indicate a kind of ceilingeffect for affective differentiation of content. Psychopathic indi-viduals demonstrated significant changes in amplitude betweentasks represented by an enhanced negativity at the N1 andenhanced positivity at the P3 component.
The enhanced negativity at N1 during Task 2 for psychopathicindividuals is an unexpected outcome that deserves additionalattention here. This exaggerated negative deflection falls outsideall the regular patterns recognized in this study as a whole, as itis the only instance in this investigation of an increased negativityfor any stimulus with emotional content. It is tempting to dis-count this as a possible singularity in the data; however, afterreexamining individual subject amplitudes on this component,no outliers were found to be driving this effect. The direction ofmodulation and the latency and distribution of this effect isreminiscent of the EPN described by Schupp et al. (2003a,2003b). Why this effect would only appear for those with highpsychopathic traits and, further, only under this attentive condi-tion (Task 2) is beyond what the data collected here can explain.A very conservative interpretation is simply to recognize it as aunique marker for differentiation between neutral and emotionaltargets in psychopathic individuals, when attention is effortfullydirected toward categorizing emotional content, and it maysuggest alternative cognitive processes being engaged to accom-plish this task. The robust positive amplitude increase of the P3for psychopathic individuals in Task 2 is qualitatively similar towhat nonpsychopathic individuals display for both versions ofthe task, which suggests that whatever cognitive processescontribute to this pattern occur effortlessly in nonpsychopathicindividuals, but only through directed, effortful attention in psy-chopathic individuals.
Emotional processing differences in psychopathy 803
Results for lexical stimuli during Task 2 are difficult to inter-pret. There was no evidence across groups for any significantdifferentiation between emotional and neutral words during Task2. When comparing these results to those of Task 1, it is importantto recognize that the differences between the two task conditionsfor photographic targets do not generalize to the lexical no-gostimuli, being that the lexical stimuli were cues to inhibitresponses to targets. The relevance of emotional content for thewords was incidental for both tasks, as the instructions remainedthe same for lexical no-go stimuli in both tasks (i.e., simplyread the words silently). Any interpretation made regarding thedifference between Task 1 and 2 for these no-go stimuli arespeculative at best.
It may be that the increased attention to picture categorizationrequired for Task 2 made participants less attentive to the wordmeanings. Their only behavioral requirement was to refrain fromany response during these stimuli; therefore, motivation for readingthese words may have decreased considerably in Task 2, as the taskdemands increased for the target stimuli. Despite the problematicfeatures of interpreting these particular results, the most pertinentinformation for these lexical no-go stimuli was gathered in Task 1.The behavioral demands of Task 2 were more purposefullydesigned to allow for a multifaceted analysis of target identifica-tion, and no additional theoretical perspectives were hypothesizedregarding features of no-go stimuli in this task. It is a curiousdiversion in the data, however, which may at least serve as a cue forfuture investigations of this nature to take better care to ensuremotivated processing of these stimuli.
The Effect of Attention
Overall, the outcomes of this study are particularly intriguingwhen considered in the context of hypotheses regarding possibleabnormalities in attention in people with psychopathy, such asNewman’s response modulation hypothesis (RMH; Newman &Lorenz, 2003; Patterson & Newman, 1993). To reiterate the keyfeatures of RMH, it proposes that apparent deficits in emotionalprocessing stem from a failure to process peripheral cues notimmediately relevant to ongoing, goal-directed behavior. Much ofthe foregoing support underlying this notion specifically examinesthe impact of information that is peripheral and distracting, appro-priating resources from a primary task, and these outcomes havedemonstrated that people with psychopathy are less prone to inter-ference from external distracters in this context (Jutai & Hare,1983; Mitchell, Richell, Leonard, & Blair, 2006; Newman,Schmitt, & Voss, 1997). RMH specifically predicts that peoplewith psychopathy are capable of processing emotional informationat appropriate levels when it is an element of immediate interest, inother words, when it is task relevant (Newman et al., 2010). In thepresent study, emotional information is never used as a distractingelement; it is simply superfluous information in the first task andtask relevant in the second. The data here support RMH insomuchas those with high psychopathic traits do show increased ERPdifferentiation between emotional and neutral cues when theirattention is explicitly directed toward these features (Task 2), yetto say that psychopathic individuals process this information atnormal levels may be incorrect. Data here suggest that psycho-pathic individuals fail to achieve the same form or level of differ-entiation recognized in nonpsychopathic individuals, despite thesesignificant increases. Attention clearly seems to play a moderatingrole in the information processing abnormalities evident in peoplewith psychopathy, but care must be taken in the interpretation of
the physical characteristics of ERP waveform, which may suggestsome alternative neural processes at work.
Summary and Conclusion
The outcomes of this investigation exhibit multifaceted demonstra-tions of how those with high psychopathic traits process and useemotional information differently. Psychopathic individuals exhibitunencumbered performance categorizing emotional stimuli anddemonstrate the typical facilitation of physical responses commen-surate with an intact orienting response. Consistent with this, startlereactivity is appropriately attenuated under pleasant conditions,similar to responses in healthy, normal individuals. Where psycho-pathic people’s divergent physiology has observable behavioralconsequences may be limited to the utilization of affective contentto prime defensive networks, as evidenced here and elsewhere bylack of affective startle potentiation along with other reports ofdeficient aversive conditioning and poor autonomic response topunishment cues. So, although ample evidence in ERP and func-tional magnetic resonance imaging literature suggests divergentpatterns of neural processing of affective information, care must betaken in identifying the functional circuits responsible for theserather specific behavioral consequences.
The current experimental protocol was intended to exploreabnormalities in emotional processing engaged in by those withhigh psychopathic traits as revealed by affective ERP modulationand the potential moderating effects that goal-directed attention onthese processes. The evidence presented here indicates clear defi-cits in the implicit differentiation between emotionally salientand neutral stimuli such that early discriminatory processes thatsupport facilitated attention and memory for this distinctive contentare absent or substantially delayed. However, when effortful, vol-untary attention to emotional information is explicitly required fortask performance, ERP waveforms suggest compensatory modifi-cation of this level of processing apparent at both the N1 and P3components. However, the electrocortical representation of thesemodifications remains quantitatively diminished and qualitativelyunusual compared to the consistent patterns exhibited in nonpsy-chopathic individuals across both tasks. These persistent deviationslikely represent neural processing differences that may account forspecific abnormalities in the means by which psychopaths incor-porate and utilize emotional information in the governance ofongoing cognitive and behavioral processes.
Investigations into the physiological basis of deficits associatedwith psychopathy have already revealed important ways that emo-tional processing contributes to healthy cognitive functioning. Thecurrent investigation was designed to expand this line of researchby incorporating ERP methodologies to better account for proc-esses of automated and directed attention in a formula that mayeventually help us to understand how affective content is distin-guished as significant and worthy of facilitated processing. Thenotion that automated attention is a fundamental process by whichaffective content becomes salient in our environment is wellaccepted, and the hypothesis that abnormalities in attentionundergird psychopaths’ deficits in emotional processing hasgained noteworthy support. The prospect of alternative, divergent,or compensatory means for incorporating emotional informationinto cognitive and ultimately behavioral processes deserves furtherinvestigation. These efforts will not only broaden our understan-ding of the physiological basis of psychopathy, but will contri-bute to a broader understanding of pathology and individualdifferences.
804 N.E. Anderson and M.S. Stanford
References
Benning, S. D., Patrick, C. J., & Iacono, W. G. (2005). Psychopathy,startle blink modulation, and electrodermal reactivity in twinmen. Psychophysiology, 42, 753–762. doi: 10.1111/j.1469-8986.2005.00353.x
Birbaumer, N., Veit, R., Lotze, M., Erb, M., Hermann, C., Grodd, & Flor,H. (2005). Deficient fear conditioning in psychopathy: A functionalmagnetic resonance imaging study. Archives of General Psychiatry, 62,799–805. doi: 10.1001/archpsyc.62.7.799
Blair, R. J. R. (2006). The emergence of psychopathy: Implications for theneuropsychological approach to developmental disorders. Cognition,101, 414–442. doi: 10.1016/j.cognition.2006.04.005
Bradley, M. M., Codispoti, M., Cuthbert, B. N., & Lang, P. J. (2001).Emotion and motivation: I. Defensive and appetitive reactions in pictureprocessing. Emotion, 1, 276–298. doi: 10.1037/1528-3542.1.3.276
Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1990). Startle reflexmodification: Emotion or attention? Psychophysiology, 27, 513–522.doi: 10.1111/j.1469-8986.1990.tb01966.x
Bradley, M. M., & Lang, P. J. (1999). Affective norms for English words(ANEW): Instruction manual and affective ratings. Technical ReportC-1, The Center for Research in Psychophysiology, University ofFlorida.
Bradley, M. M., & Lang, P. J. (2007). Emotion and motivation. In J. T.Cacioppo, L. G. Tassinary, & G. G. Berntson (Eds.), Handbook ofpsychophysiology (3rd ed., pp. 581–607). New York: Cambridge Uni-versity Press.
Cacioppo, J. T., & Gardner, W. L. (1999). Emotion. Annual Review ofPsychology, 50, 191–214. doi: 10.1146/annurev.psych.50.1.191
Carretie, L., Hinojosa, J. A., Martin-Loeches, M., Mercado, F., & Tapia, M.(2004). Automatic attention to emotional stimuli: Neural correlates.Human Brain Mapping, 22, 290–299. doi: 10.1002/hbm.20037
Cleckley, H. (1941). The mask of sanity: An attempt to reinterpret theso-called psychopathic personality. Oxford, England: Mosby.
Cuthbert, B. N., Schupp, H. T., Bradely, M. M., Birbaumer, N., & Lang, P.J. (2000). Brain potentials in affective picture processing: Covariationwith autonomic arousal and affective report. Biological Psychology, 62,95–111. doi: 10.1016/j.clinph.2003.11.021
Davidson, R. J. (1998). Affective style and affective disorders: Perspectivesfrom affective neuroscience. Cognition & Emotion, 12, 307–330. doi:10.1080/026999398379628
Diedrich, O., Naumann, E., Maier, S., & Becker, G. (1997). A frontalpositive slow wave in the ERP associated with emotional slides. Journalof Psychophysiology, 11, 71–84.
Donchin, E. & Coles, M. G. H. (1988). Is the P300 component a manifes-tation of context updating? Behavioral and Brain Sciences, 11, 357–374. doi: 10.1017/S0140525X00058027
Fabiani, M., Karis, D., & Donchin, E. (1990). Effects of mnemonic strategymanipulation in a Von Restorff paradigm. Electroencephalographyand Clinical Neurophysiology, 75, 22–35. doi: 10.1016/0013-4694(90)90149-E
Foti, D., Hajcak, G., & Dien, J. (2009). Differentiating neural responses toemotional pictures: Evidence from temporo-spatial PCA. Psychophysi-ology, 46, 521–530. doi: 10.1111/j.1469-8986.2009.00796.x
Glenn, A. L., Raine, A., & Schug, R. A. (2009). The neural correlates ofmoral decision-making in psychopathy. Molecular Psychiatry, 14, 5–6.doi: 10.1038/mp.2008.104
Hare, R. D. (1968). Psychopathy, autonomic functioning, and the orientingresponse. Journal of Abnormal Psychology, 73, 1–24. doi: 10.1037/h0025873
Hajcak, G., & Olvet, D. M. (2008). The persistence of attention to emotion:Brain potentials during and after picture presentation. Emotion, 8, 250–255. doi: 10.1037/1528-3542.8.2.250
Hiatt, K. D., Schmitt, W. A., & Newman, J. P. (2004). Stroop tasks revealabnormal selective attention among psychopathic offenders. Neuropsy-chology, 18, 50–59. doi: 10.1037/0894-4105.18.1.50
Hillyard, S. A., Vogel, E. K., & Luck, S. J. (1998). Sensory gaincontrol (amplification) as a mechanism of selective attention:Electrophysiological and neuroimaging evidence. PhilosophicalTransactions of the Royal Society B, 353, 1257–1270. doi: 10.1098/rstb.1998.0281
Howard, R., & McCullagh, P. (2007). Neuroaffective processing in criminalpsychopaths: Brain event-related potentials reveal task-specific anoma-lies. Journal of Personality Disorders, 21, 322–339. doi: 10.1521/pedi.2007.21.3.322
Junghofer, M., Bradley, M. M., Elbert, T. R., & Lang, P. J. (2001). Fleetingimages: A new look at early emotion discrimination. Psychophysiology,22, 545–560. doi: 10.1111/1469-8986.3820175
Jutai, J. W., & Hare, R. D. (1983). Psychopathy and selective attentionduring performance of a complex perceptual-motor task. Psychophysi-ology, 20, 146–151. doi: 10.1111/j.1469-8986.1983.tb03280.x
Keil, A., Bradley, M. M., Hauk, D., Rockstroh, B., Elbert T., & Lang, P. J.(2002). Large scale neural correlates of affective picture processing.Psychophysiology, 39, 641–649. doi: 10.1017/S0048577202394162
Kiehl, K. A., Bates, A. T., Laurens, K. R., Hare, R. D., & Liddle, P. F.(2006). Brain potentials implicate temporal lobe abnormalities in crimi-nal psychopaths. Journal of Abnormal Psychology, 115, 443–453. doi:10.1037/0021-843X.115.3.443
Kiehl, K. A., Smith, A. M., Hare, R. D., Mendrek, A., Forster, B. B., Brink,J., & Liddle, P. F. (2001). Limbic abnormalities in affective processingby criminal psychopaths as revealed by functional magnetic resonanceimaging. Biological Psychiatry, 50, 677–684. doi: 10.1016/S0006-3223(01)01222-7
Knight, R. T. (1984). Decreased response to novel stimuli after prefrontallesions in man. Electroencephalography and Clinical Neurophysiology,59, 9–20. doi: 10.1016/0168-5597(84)90016-9
Knight, R. T., Grabowecky, M., & Scabini, D. (1995). Role of humanprefrontal cortex in attention control. In H. H. Herbert, R. Silvana, &P. S. Goldman-Rakic (Eds.), Epilepsy and the functional anatomyof the frontal lobe, advances in neurology (pp. 21–34) New York: RavenPress.
Knudsen, E. I. (2007). Fundamental components of attention. AnnualReview of Neuroscience, 30, 57–78. doi: 10.1146/annurev.neuro.30.051606.094256
Kok, A. (2001). On the utility of P3 amplitude as a measure ofprocessing capacity. Psychophysiology, 38, 557–577. doi: 10.1017/S0048577201990559
Kosson, D. S., Suchy, Y., Mayer, A. R., & Libby, J. (2002). Facial affectrecognition in criminal psychopaths. Emotion, 2, 398–411. doi:10.1037/1528-3542.2.4.398
Lang, P. J., & Greenwald, M. K. (1988). The international affective picturesystem standardization procedure and initial group results for affectivejudgments: Technical reports 1A & 1B. Gainsville, FL: Center forResearch in Psychophysiology, University of Florida.
Lang, P. J., Ohman, A., & Vaitl, D. (1988). The international affectivepicture system [Photographic slides]. Gainesville, FL: Center forResearch in Psychophysiology, University of Florida.
Lilienfeld, S. O., & Widows, M. (2005). Professional manual for thePsychopathic Personality Inventory–Revised (PPI-R). Lutz, FL: Psy-chological Assessment.
Luck, S. J., & Hillyard, S. A. (1994a). Electrophysiological correlates offeature analysis during visual search. Psychophysiology, 31, 291–308.doi: 10.1111/j.1469-8986.1994.tb02218.x
Luck, S. J., & Hillyard, S. A. (1994b). Spatial filtering during visual search:Evidence from human electrophysiology. Journal of Experimental Psy-chology: Human Perception and Performance, 20, 1000–1014. doi:10.1037/0096-1523.20.5.1000
Lykken, D. T. (1957). A study of anxiety in sociopathic personality. Journalof Abnormal and Social Psychology, 55, 6–10. doi: 10.1037/h0047232
McCarthy, G., & Wood, C. C. (1985). Scalp distributions of event-relatedpotentials: An ambiguity associated with analysis of variance models.Electroencephalography and Clinical Neurophysiology, 62, 203–208.doi: 10.1016/0168-5597(85)90015-2
Mitchell, D. G. V., Richell, R. A., Leonard, A., & Blair, R. J. R. (2006).Emotion at the expense of cognition: Psychopathic individualsoutperform controls on an operant response task. Journal of AbnormalPsychology, 115, 559–566. doi: 10.1037/0021-843X.115.3.559
Naumann, E., Bartussek, D., Diedrich, O., & Laufer, M. E. (1992). Assess-ing cognitive and affective information processing functions of the brainby means of the late positive complex of the event-related potential.Journal of Psychophysiology, 6, 285–298.
Newman, J. P., Curtin, J. J., Bertsch, J. D., & Baskin-Sommers,A. R. (2010). Attention moderates the fearlessness of psychopathicoffenders. Biological Psychiatry, 67, 66–70. doi: 10.1016/j.biopsych.2009.07.035
Newman, J. P., & Kosson, D. S. (1986). Passive avoidance learning inpsychopathic and nonpsychopathic offenders. Journal of AbnormalPsychology, 95, 257–263. doi: 10.1037/0021-843X.95.3.252
Emotional processing differences in psychopathy 805
Newman, J. P., & Lorenz, A. R. (2003). Response modulation and emotionprocessing: Implications for psychopathy and other dysregulatorypsychopathology. In R. J. Davidson, K. Scherer, & H. H. Goldsmith(Eds.), Handbook of affective sciences (pp. 904–929). New York:Oxford University Press.
Newman, J. P., Schmitt, W. A., & Voss, W. D. (1997). The impactof motivationally neutral cues on psychopathic individuals: Assessingthe generality of the response modulation hypothesis. Journal ofAbnormal Psychology, 106, 563–575. doi: 10.1037/0021-843X.106.4.563
Ohman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention:Detecting the snake in the grass. Journal of Experimental Psychology:General, 130, 466–478. doi: 10.1037/0096-3445.130.3.466
Ohman, A., Hamm, A., & Hugdahl, K. (2000). Cognition and theautonomic nervous system: Orienting, anticipation, and conditioning. InJ. T. Cacioppo, L. G. Tassinary, G. G. Berntson (Eds.), Handbook ofpsychophysiology (2nd ed., pp. 533–575). New York: CambridgeUniversity Press.
Olofsson, J. K., Nordin, S., Sequeira, H., & Polich, J. (2008). Affec-tive picture processing: An integrative review of ERP findings.Biological Psychology, 77, 247–265. doi: 10.1016/j.biopsycho.2007.11.006
Palomba, D., Angrilli, A., & Mini, A. (1997). Visual evoked potentials,heart rate responses, and memory to emotional pictorial stimuli. Inter-national Journal of Psychophysiology, 27, 55–67. doi: 10.1016/S0167-8760(97)00751-4
Patrick, C. J. (1994). Emotion and psychopathy: Startling newinsights. Psychophysiology, 3, 319–330. doi: 10.1111/j.1469-8986.1994.tb02440.x
Patrick, C. J., & Berthot, B. D. (1995). Startle potentiation duringanticipation of a noxious stimulus: Active versus passive responsesets. Psychophysiology, 32, 72–80. doi: 10.1111/j.1469-8986.1995.tb03408.x
Patrick, C. J., Bradley, M. M., & Lang, P. J. (1993). Emotion in the criminalpsychopath: Startle reflex modulation. Journal of Abnormal Psychol-ogy, 102, 82–92. doi: 10.1037/0021-843X.102.1.82
Patterson, C. M., & Newman, J. P. (1993). Reflectivity and learning fromaversive events: Toward a psychological mechanism for the syndromesof disinhibition. Psychological Review, 100, 716–736. doi: 10.1037/0033-295X.100.4.716
Polich, J. (2007). Updating P300: An integrative theory of P3a and P3b.Clinical Neurophysiology, 118, 2128–2148. doi: 10.1016/j.clinph.2007.04.019
Rugg, M. (1995). ERP studies of memory. In M. D. Rugg & M. G. H. Coles(Eds.), Electrophysiology of mind (pp. 132–166). New York: OxfordUniversity Press.
Schupp, H. T., Junghofer, M., Weike, A. I., & Hamm, A. O. (2003a).Attention and emotion: An ERP analysis of facilitated emotional stimu-lus processing. NeuroReport, 14, 1107–1110. doi: 10.1097/01.wnr.0000075416.59944.49
Schupp, H. T., Junghofer, M., Weike, A. I., & Hamm, A. O. (2003b).Emotional facilitation of sensory processing in the visual cortex.Psychological Science, 14, 7–14. doi: 10.1111/1467-9280.01411
Smith, N. K., Cacioppo, J. T., Larsen, J. T., & Chartrand, T. L. (2003). MayI have your attention please: Electrocortical responses to positive andnegative stimuli. Neuropsychologia, 41, 171–183. doi: 10.1016/S0028-3932(02)00147-1
Soltani, M., & Knight, R. T. (2000). Neural origins of the P300. CriticalReviews in Neurobiology, 14, 199–224.
Strauss, E. (1983). Perception of emotional words. Neuropsychologia,21, 99–103. doi: 10.1016/0028-3932(83)90104-5.
Wechsler, D. (1999). Wechsler Abbreviated Scale of Intelligence (WASI).San Antonio, TX: Harcourt Assessment.
Widom, C. S. (1977). A methodology for studying non-institutionalizedpsychopaths. Journal of Consulting and Clinical Psychology, 45, 674–683. doi: 10.1037/0022-006X.45.4.674
Williamson, S., Harpur, T. J., & Hare, R. D. (1991). Abnormal processingof affective words by psychopaths. Psychophysiology, 28, 260–273. doi:10.1111/j.1469-8986.1991.tb02192.x
(Received October 5, 2011; Accepted February 8, 2012)
