Processing expected and unexpected uncertainty is ...

0 downloads 0 Views 1MB Size Report
Oct 24, 2016 - fearless-dominance personality traits – An exploratory ERP study on feedback processing. Lydia Kogler a,⁎, Uta Sailer b,c, Birgit Derntl a, ...
Physiology & Behavior 168 (2017) 74–83

Contents lists available at ScienceDirect

Physiology & Behavior journal homepage: www.elsevier.com/locate/phb

Processing expected and unexpected uncertainty is modulated by fearless-dominance personality traits – An exploratory ERP study on feedback processing Lydia Kogler a,⁎, Uta Sailer b,c, Birgit Derntl a, Daniela M. Pfabigan c,d,⁎ a

Department of Psychiatry and Psychotherapy, Medical School, University of Tübingen, Calwerstrasse 14, 72074 Tübingen, Germany Faculty of Medicine, Institute of Basic Medical Sciences, Dept. of Behavioural Sciences in Medicine, University of Oslo, PO Box 1111, Blindern, 0317 Oslo, Norway c Social, Cognitive and Affective Neuroscience Unit, Department of Basic Psychological Research and Research Methods, Faculty of Psychology, University of Vienna, Liebiggasse 5, 1010 Vienna, Austria d Culture and Social Cognitive Neuroscience Laboratory, School of Psychological and Cognitive Sciences, Peking University, 52 Haidian Road, Beijing 100871, China b

H I G H L I G H T S • • • • •

Expected and unexpected uncertainty was introduced in a gambling task. FRN and P300 amplitudes were measured during the uncertainty manipulation. Their association with fearless-dominance personality traits was investigated. Associations found for expected-uncertainty/FRN and expected-certainty/P300. Indicating adaptive use of response-monitoring resources in high-trait individuals.

a r t i c l e

i n f o

Article history: Received 12 June 2016 Received in revised form 21 October 2016 Accepted 21 October 2016 Available online 24 October 2016 Keywords: Fearless-dominance Performance monitoring FRN P300 Expectancy Certainty

a b s t r a c t Expectancy and certainty regarding an outcome are important factors during performance monitoring. However, the separate contributions of expected and unexpected uncertainty on different measures of performance monitoring, including feedback-related negativity (FRN) and P300 components, are not well established. The current study investigated their relationship to fearless-dominance, a personality construct described by high social potency and low anxiety. Accurately predicting environmental outcomes in certain and uncertain situations might be a prerequisite of social potency, therefore it may be associated with increased performance monitoring and its ERP correlates. Consequently, expected-uncertain and unexpected-uncertain feedback (by violating previously learned certain and expected feedback) was introduced in addition to expected-certain feedback in healthy individuals during a probabilistic gambling task. In both FRN and P300 components, difference waves were more pronounced for unexpected-uncertain and expected-uncertain compared to expected-certain feedback. Moreover, more fearless-dominant individuals showed diminished feedback processing specifically in expected-uncertain trials, but concurrently enhanced attentional processing in expected-certain trials. These findings indicate adaptive and situation-appropriate utilization of performance monitoring resources in individuals with more pronounced fearless-dominance personality traits. The results indicate that a precise differentiation of expected and unexpected uncertainty in fearless-dominant individuals is mandatory in order to better understand the underlying personality construct and related behavior. © 2016 Published by Elsevier Inc.

1. Introduction ⁎ Correspondence to: L. Kogler, Department of Psychiatry and Psychotherapy, Medical School, University of Tübingen, Calwerstrasse 14, 72074 Tübingen, Germany; D. M. Pfabigan, Social, Cognitive and Affective Neuroscience Unit, Department of Basic Psychological Research and Research Methods, Faculty of Psychology, University of Vienna, Liebiggasse 5, 1010 Vienna, Austria. E-mail addresses: [email protected] (L. Kogler), [email protected] (U. Sailer), [email protected] (B. Derntl), [email protected] (D.M. Pfabigan).

http://dx.doi.org/10.1016/j.physbeh.2016.10.016 0031-9384/© 2016 Published by Elsevier Inc.

Successful performance monitoring requires us to adapt our behavior to a changing environment and to deal with uncertainty regarding the occurrence of specific events or behaviors. However, not all of us accomplish this endeavor in the same way and with the same success. In particular, individuals disregarding social norms, such as those with psychopathic traits [22], might monitor their performance and deal with uncertainty differently than those without these social inabilities.

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83

Adjacent to this assumption, the current study aimed to investigate performance monitoring during different uncertainty conditions in individuals with varying degrees of fearless-dominance traits, a set of personality traits often associated with psychopathy. 1.1. Fearless-dominance and performance monitoring Fearless-dominance reflects an extroversive personality characteristic that is currently a matter of scientific debate regarding its association with psychopathic traits [32,35,37]. It is thought to integrate traits such as social potency, lack of anxiety and high stress coping abilities [33], which may be helpful in manipulating and directing social situations. Cumulating research suggests dual-process models of psychopathy including an affective-interpersonal and an antisocial-impulsive dimension [59]. Thereby, these models refer to psychopathy as an extreme manifestation of these broad personality dimensions that are continuously distributed in the general population (i.e., in non-forensic samples) [16,38]. Fearlessness-dominance - a characteristic of the affective-interpersonal dimension of psychopathy - is described by emotional detachment [44] and possibly altered cognitive processes such as performance monitoring [36,55]. As part of performance monitoring, error and feedback processing are important functions to adapt to rapidly changing contexts. On the other hand, the antisocial-impulsive dimension of psychopathy is related to an impulsive and socially deviant lifestyle [44]. There is already evidence that this dimension is associated with diminished error processing in both adults and children [13,21,46,49,52,54,60]. Contrary to the rather consistent findings on antisocial-impulsive aspects of psychopathy, fewer studies investigated the association between the affective-interpersonal dimension and performance monitoring and revealed ambiguous findings: Heritage and Benning [25] failed to show any association between error processing and fearless-dominance scores, whereas others reported enhanced error processing [8] or diminished feedback processing in more fearless-dominant individuals [56]. Summarizing the available evidence, Schulreich [55] suggested cautiously that affective-interpersonal aspects might be related to diminished feedback processing. The current study aimed to address the important aspect of feedback processing under uncertainty and its relation to the affective-interpersonal dimension since accurately predicting environmental outcomes might be one hallmark to successfully manipulate and control situations. 1.2. Performance monitoring and feedback processing To investigate the neuronal correlates of feedback processing, two event-related potentials (ERPs) have been repeatedly addressed. The first ERP is the feedback-related negativity (FRN; [39]) which is associated with initial feedback evaluation. Larger FRN amplitudes have been reported in feedback- and gambling-paradigms after the presentation of unfavorable compared to favorable outcomes [17,27,39] in which unexpected compared to expected outcomes prompted the largest FRN deflections [20,45]. FRN amplitude variation is assumed to classify outcomes in a good/bad fashion [19,63] therewith indicating a negative reward prediction error signal (RPEs), i.e., outcomes “worse than expected” [26], also called signed RPEs [23,24]. RPEs reflect the discrepancy between expected and received outcomes and account for internal expectation updates of stimulus-outcome relations [50]. Important to note, FRN amplitude variation and associated RPEs often reflect variations in feedback expectancy, which depend on factors such as certainty regarding the outcome and the probability of its occurrence. Mushtaq, Bland, and Schaefer [40] define certainty as the ability to accurately predict future events. Referring to the opposite of certainty, Yu and Dayan [64] state two different forms of uncertainty: expected uncertainty reflects an observed and familiar unreliability of weak action-outcome associations (i.e., of probabilistic outcomes), whereas unexpected uncertainty reflects a not signaled change in strong actionoutcome associations (i.e., a change of previously deterministic

75

outcomes), which may also be seen as surprise. More precisely, the violation of strong action-outcome associations refers to unexpected uncertainty and the violation of weak action-outcome associations refers to expected uncertainty. To give an everyday life example, in a city known for unstable weather condition, estimating the weather in the afternoon is difficult while leaving the house during the morning. In contrast, in a city known for its daily sunshine a storm in the afternoon is unexpected. Thus, in the second city the sudden weather change indicates unexpected uncertain weather conditions while in the first one the weather is expected to be uncertain [64]. Yu, Zhou, and Zhou [65] tried disentangling uncertainty and probability in a decision-making task with feedback lacking behavioral consequences. The authors observed that FRN amplitudes were modulated by expected uncertainty only after positive feedback. A further study investigated FRN amplitude variation based on unexpected-uncertainty via the violation of previously learned strong stimulus-action association [45]. In contrast to Yu et al. [65], the authors reported the largest FRN amplitude variation for unexpected-uncertain feedback for both positive and negative outcomes. Comprehensive studies directly comparing effects of unexpected and expected uncertainty on FRN and also P300 amplitudes are still missing (but see [65]). The P300 is the second feedback-related ERP component which is primarily associated with memory processing and attention [48], but has been shown to be sensitive to expectancy in gambling tasks [63] and to motivational salience in feedback processing [41,62]. More positive P300 amplitudes were reported for unexpected compared to expected, for positive compared to negative, and for more salient than insignificant outcomes [5,47]. Thus, the aim of the current study was to extend the literature on feedback processing and RPEs in non-forensic individuals with different levels of fearless-dominance personality traits dealing with certain and uncertain environments. To directly investigate the effects of uncertainty on FRN and P300 amplitudes, we administered a gambling paradigm yielding positive and negative feedback with high or low feedback certainty and high or low feedback probability, combined in expectedcertain, expected-uncertain, and unexpected-uncertain feedback conditions. To focus on the current uncertainty manipulation, we calculated valence difference waves for both FRN and P300 components (via subtracting positive from negative conditions) to directly represent RPE signals. Since unexpected uncertainty is characterized by the violation of strong action-outcome-associations, we assumed that it induces the strongest RPEs, indicated by the most negative FRN difference waves [10,29,39,45]. We further assumed that individuals with higher fearless-dominance scores are highly attentive to errors and external feedback in order to adjust their behavior accordingly [8]; a pattern which they should also show during uncertain situations. Therewith, we expected fearless-dominance traits to be related with enhanced performance monitoring in gambling situations [8] and the RPEs reflecting unexpected as well as expected uncertainty to be positively associated with fearless-dominance personality traits (e.g., [8]). Additionally, we hypothesized that P300 amplitudes vary with feedback expectancy, with larger amplitudes for unexpected-uncertain and expected-uncertain feedback than for expected-certain feedback [6,14, 30,61]. We expected P300 amplitudes to be associated with fearlessdominance personality traits, but were not able to predict a direction of this effect due to inconsistent previous findings [7,9,11]. 2. Material and methods 2.1. Participants Twenty-five volunteers (all students; 14 women) participated in the experiment. All were right-handed [43] and reported no history of neurological or psychiatric disorders. Two male participants were excluded from further analysis due to data recording artifacts and inconsistent response behavior. The remaining 23 participants had a mean age of

76

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83

22.8 ± 3.0 years. The study was conducted in accordance with the Declaration of Helsinki (1983, revised 2013) and local ethical guidelines for experimentation with human participants (including approval by an institutional review board) at the Faculty of Psychology, University of Vienna. Written informed consent was obtained from each participant prior to the experimental procedure. Participants further filled in the German version of the Psychopathic Personality Inventory – Revised (PPI-R; [1]) to assess individual fearless-dominance personality traits. The PPI-R is a self-report questionnaire assessing psychopathic traits via eight subscales, which are subsumed under two higher-order factors: fearless-dominance (reflecting affective-interpersonal aspects) and self-centered impulsivity (reflecting antisocial-impulsive aspects). Internal consistency is satisfying with Cronbach alpha = 0.85. 2.2. Task Stimulus presentation and electroencephalogram (EEG) data collection were controlled by E-Prime 2.0 (Psychology Software Tools, Inc., Sharpsburg, PA). Participants were seated 70 cm in front of a 19″ cathode ray tube monitor (Sony GDM-F520, 75 Hz refresh rate). A modified version of the gambling paradigm by Pfabigan, Alexopoulos, Bauer and Sailer [45] was administered (see Fig. 1A for illustration of paradigm and time-line). The experiment started with a training phase of 60 trials during which participants learned specific cue-response-mappings. There, participants were instructed to figure out these cue-response

associations in the pre-experimental training phase. Each trial started with a fixation cross with a random duration of 650–1250 ms. Subsequently, a geometrical figure (circle, triangle, star; height and width: 2.5 cm) was presented for 450 ms as imperative cue. The subsequently presented question mark prompted participants to respond via button press on the response pad in front of them within the next 2000 ms. During the training phase the three geometrical figures were assigned to two buttons in the following manner: cue1 (either circle, triangle, or star) was associated with 100% reward probability only for button 1; cue2 was associated with 50% reward probability only for button 2; cue3 was not rewarded at all (0%), irrespective of button choice (see Table 1 and Fig. 1B). The button press triggered a 400 ms delay, then feedback was provided for 700 ms. The next trial started again with the presentation of the fixation cross. The number 50 in green color indicated a correct response and a gain of €0.5; the number 50 in red color indicated an incorrect response and a loss of €0.5. Thus, participants built up expectations regarding the association between the imperative cues and feedback outcome in a certain, deterministic manner for cue1 and cue3 (100% positive outcome for cue1 after pressing button 1; 100% negative outcome for cue3 irrespective of button press) and in an uncertain, probabilistic manner for cue2 (50% expectation for positive and 50% for negative feedback after pressing button 2) (see Table 1 and Fig. 1B). In the subsequently following experimental phase, participants were asked to accumulate as much money (starting with a deposit of €20) as possible by applying the previously learned

Fig. 1. A) Schematic time-line of the modified version of the paradigm [45]. B) Reward probability and button affiliation for each cue in training and experimental session (see also Table 1). Note: The assignment of the three visual cues (geometrical figures) to the experimental conditions was counterbalanced across participants.

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83 Table 1 Reward probability, number of trials, expectancy and valence of feedback per cue in training and experimental session. Note: The assignment of the three visual cues (geometrical figures) to the experimental conditions was counterbalanced across participants. Training Ratio

Experimental session

Trial number

Valence

Expectancy

Ratio

Trial number

Cue1 100% 0%

20 –

Positive Negative

Expected-certain Unexpected-uncertain

80% 20%

200 50

Cue2 50% 50%

10 10

Positive Negative

Expected-uncertain Expected-uncertain

50% 50%

50 50

Cue3 0% 100%

– 20

Positive Negative

Unexpected-uncertain Expected-certain

20% 80%

50 200

action-outcome-associations. However, reward probabilities changed compared to the training: cue1 was now associated with 80% reward probability for button 1, and cue3 was associated with 20% reward probability irrespective of button choice. Thus, the previously learned certain, deterministic expectation was partly violated (20% unexpecteduncertain negative feedback for cue1 and 20% unexpected-uncertain positive feedback for cue3). Nevertheless, in 80% of the trials participants still received the expected feedback (80% expected-certain positive feedback for cue1, and 80% expected-certain negative feedback for cue3). In contrast, cue2 contingencies (button 2, 50%) were not changed, thus, probabilistic feedback was not violated (50% expecteduncertain positive and negative feedback). Therewith, this experimental design introduced the following three expectancy levels: (I) expectedcertain, (II) expected-uncertain and (III) unexpected-uncertain feedback, each either with positive or negative valence. Importantly, certainty/uncertainty and probability of an outcome are highly associated [65]: certainty reaches its maximum with highest (100%) or lowest (0%) reward probability, whereas uncertainty reaches its maximum with medium reward probability (50%). Thus, action-outcome associations are driven by expectations regarding the probability of an outcome. Therefore, the current paradigm allows manipulating uncertainty as well as probability within one task. In the experimental phase, 600 trials were presented (see Table 1). After blocks of 50 trials, participants were provided with overall performance feedback and were given short rests. Participants were asked to estimate the reward probabilities for the three cues after the trainingphase and after the main experiment to assess whether the experimental manipulation was successful. Based on individual task performance, participants received a financial bonus (between €10 and €20) after task completion. EEG data collection lasted on average 43.35 min (SD = 6.49; range [35.22; 62.87]).

weighted actual EOG signals were subtracted from the EEG in the experimental trials [3]. Using EEGLAB 6.03b [12], EEG data were digitally low-pass filtered at 30 Hz (roll-off 6 dB/octave) and segmented from 200 ms prior to until 700 ms after feedback onset. The mean of the first 200 ms served as baseline interval. These epochs were screened for further eye-movement related artifacts applying additional extended (infomax) independent component analysis (ICA; [4,31]). On average, 1.52 independent components (SD = 0.90; range [0;4]) were removed per participant. Subsequently, a semi-automatic artifact detection procedure was employed to discard epochs containing voltage values exceeding ± 75 μV or voltage drifts of N50 μV in all EEG electrodes. Please refer to Table 2 for exact numbers of available trials after artifact correction. Artifact-free trials were grouped in six conditions: expected-certain positive feedback (cue1), expected-certain negative feedback (cue3), unexpected-uncertain positive feedback (cue3), unexpected-uncertain negative feedback (cue1), expected-uncertain positive feedback (cue2), and expected-uncertain negative feedback (cue2). Data were averaged subject- and condition-wise. To investigate a measure of neural activity specific to reward prediction errors (RPEs) for the three expectancy levels, we chose to apply a difference wave approach by subtracting averaged positive from averaged negative feedback trials per participant to remove processes common to both negative and positive feedback trials [69] – resulting in the conditions expected-certain, expected-uncertain, and unexpected-uncertain. To quantify FRN and P300 components, mean amplitude values were extracted from the difference waves at electrode site FCz (FRN, time interval 200–300 ms after feedback onset) and Pz (P300, time interval 300–500 ms after feedback onset).

Table 2 Means and standard deviation (SD) of behavioral and physiological variables of the current experiment.

Exp-certain

Negative Positive Exp-uncertain Negative Positive Unexp-uncertain Negative Positive

Exp-certain

Negative Positive Exp-uncertain Negative Positive Unexp-uncertain Negative Positive

2.3. Electroencephalographic recording The 64-multi-channel-EEG was recorded via equidistantly positioned Ag/AgCl-electrodes in an elastic electrode cap (EASYCAP GmbH, Herrsching, Germany). EEG was recorded with a balanced sterno-vertebral reference [57] from 0.016–125 Hz and sampled at 250 Hz. The ground electrode was placed centrally on the forehead. Eye-movement artifacts were discarded in a first step via channelwise subtraction of subject- and channel-specific weighted movement coefficients [3] which were acquired in pre-experimental calibration trials in which participants performed guided vertical and horizontal eye movements. The weighted coefficients for vertical and horizontal eye movement artifacts were calculated as the ratio of the covariance between each EEG channel and the electrooculogram (EOG), and the variance within the vertical and horizontal channels. Subsequently, the

77

Nr. artifact-corrected trials

Reaction times

M

SD

M

SD

187.26 173.65 34.04 31.70 42.78 44.65

10.76 17.42 8.24 8.91 5.44 3.66

531.32 428.61 526.38 535.45 412.24 547.02

184.05 99.70 206.66 205.78 86.34 182.24

FRN mean amplitudes

P300 mean amplitudes

M

SD

M

SD

10.54 11.14 11.52 16.97 8.86 15.97

4.38 4.39 4.95 7.96 4.25 7.96

9.69 7.60 11.27 16.36 12.87 17.59

6.34 4.87 6.86 8.34 6.39 7.09

M

SD

M

SD

276.87 286.09 271.48 275.30 277.74 267.65

43.53 41.62 38.66 37.09 27.34 40.61

371.30 354.09 378.96 369.22 390.43 368.35

71.52 68.73 74.96 53.74 56.91 54.80

Latency FRN

Exp-certain

Negative Positive Exp-uncertain Negative Positive Unexp-uncertain Negative Positive

Exp-certain Diff Exp-uncertain Diff Unexp-uncertain Diff

Latency P300

FRN mean diff amplitudes

P300 mean diff amplitudes

M

SD

M

SD

−0.84 −5.45 −6.47

2.95 4.75 4.18

2.32 −5.44 −4.73

4.94 4.39 3.40

78

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83

Time windows and electrode positions were chosen based on visual inspection of the original amplitude courses (Fig. 3), the difference wave amplitude courses (Fig. 4), and recent literature (e.g., [19,20,45]). To keep statistical models simple (as suggested by [68]), we refrained from adding more electrodes of interest into the current models. To account for potential outliers, a winsorisation procedure [70] was applied to guarantee normally distributed data. With this method all amplitudes higher than the value corresponding to the 75th percentile plus 1.5 times the interquartile range per condition were replaced with the maximum amplitude within this range in the corresponding condition. Accordingly, mean amplitudes lower than 25th percentiles minus 1.5 times the interquartile range per condition were replaced with the minimal amplitude within this range in the correspondent condition. Additionally, FRN and P300 latencies were extracted, starting from feedback onset until the respective minimum (FRN) or maximum (P300) of the original waveforms. 2.4. Statistical analysis Reaction times, accuracy rates, and subjective probability estimates were assessed as behavioral measures of the current paradigm. As physiological measures, the difference between negative and positive feedback conditions (i.e., the RPEs) was addressed first - we tested whether FRN (and also P300) mean difference wave amplitudes were significantly different from zero using t-tests. Subsequently, FRN and P300 mean difference wave amplitudes were investigated with separate linear mixed models consisting of the repeated fixed factor condition (expected-certain vs. expected-uncertain vs. unexpected-uncertain) and the standardized covariate fearless-dominance score. We did not include self-centered impulsivity scores in the current model since they were highly correlated with fearless-dominance scores and not in the scope of the current manuscript. Linear mixed models were chosen for analysis instead of classical analysis of covariance (ANCOVA) since the slopes between the covariate and the three conditions might differ from each other. Maximum likelihood estimations were applied to estimate the parameters for the mixed models, using an unstructured covariance matrix. Planned pairwise comparisons for the factor condition were corrected with the Bonferroni method; Cohen's d is reported to indicate effect sizes for significant comparisons. In addition, ERP latencies were explored via subjecting them to a two-way ANOVA model with the within-subject factors condition and valence (negative vs. positive). The level of significance was set at p b 0.05 for all tests. Statistical analyses were performed using PASW 18 (SPSS Inc., IBM Corporation, NY) and Statistica 6.0 (StatSoft Inc., Tulsa, OK). Moreover, we calculated Spearman correlations (rs) between the ERP difference wave amplitudes and fearless-dominance scores to demonstrate potential associations. Spearman correlations were chosen to meet concerns regarding outliers and not normally distributed data as is often the case with self-report questionnaire data [51]. Steiger's Z test was performed to explore differences between these correlations.

after the training; and 78.4% (cue1), 37.2% (cue2) and 14.3% (cue3) after the experimental manipulation. Participants chose the correct button for each cue on the majority of trials. One-tailed t-tests revealed that percentage of positive feedback (30.9%) after the training for cue2 was significantly higher than chance level's average percentage (t(22) = 13.85; p b 0.001) indicating that participants learned cue2’s reward rate and applied the correct button. This was further corroborated by an almost significant increase in percentage of correct answers from training to the experiment (t(22) = 1.99, p = 0.059, d = 0.46). Average positive feedback (i.e., accuracy rates) and subjective estimates are displayed in Fig. 2. Furthermore, we assessed whether the experimental manipulation was also reflected in response times (measured from question mark onset until button press; also winsorised). The ANOVA model resulted in significant main effects of condition (F(2,44) = 8.08, p(GG) = 0.007, η2p = 0.27) and valence (F(1,22) = 4.89, p = 0.038, η2p = 0.18), and a significant interaction (F(2,44) = 23.25, p(GG) b 0.001, η2p = 0.51). Tukey post-hoc tests showed that reaction times were comparable for expected-certain negative, expected-uncertain negative and positive, and unexpected-uncertain positive (all pvalues N 0.988). In contrast, reaction times were significantly faster for expected-certain positive and unexpected-uncertain negative conditions than the remaining ones (all p-values b0.003) – see Table 2. When testing associations between fearless-dominance scores and behavioral data, we observed a significant positive correlation between fearless-dominance scores and subjective estimates of cue3 during experimental trials (rs(23) = 0.446, p = 0.034) – participants with higher fearless-dominance scores rated subjective reward probabilities for cue3 higher than participants with lower scores. No other correlations reached significance (all p-values N 0.157). 3.2. ERP data Original waveforms of FRN and P300 amplitude courses are depicted in Fig. 3, while Fig. 4 demonstrates difference wave amplitude courses and scalp topographies of these. 3.2.1. FRN FRN mean difference wave amplitudes differed significantly from zero for expected-uncertain (t(22) = −5.50, p b 0.001) and unexpected-uncertain conditions (t(22) = −7.42, p b 0.001), but not for expected-certain outcomes (t(22) = − 1.36, p = 0.187). This indicates that

3. Results 3.1. Behavioral results The mean of the PPI-R total score was 291.22 (SD = 36.53; range 226–369; within the range of the German norm sample [1]); the mean fearless-dominance score was 103.35 (SD = 15.87; range 76– 132), the mean self-centered impulsivity score was 157.22 (SD = 19.94; range 122–206). In the current sample, fearless-dominance and self-centered impulsivity scores were positively correlated (rs(23) = 0.806, p b 0.001). Participants learned the cue-response associations successfully during the training: They received positive feedback on average in 85.9% after cue1, and in 30.9% after cue2. Subjective estimations of the reward probabilities were 77.9% for cue1, 35.3% for cue2 and 12.6% for cue3

Fig. 2. Average positive feedback (i.e., accuracy rate) and average subjective estimates for positive feedback for each cue for the training (TRAIN) and the experimental (EXP) session; in percent. Error bars denote SEM.

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83

79

Fig. 3. Original amplitude courses of FRN (left panel) and P300 (right panel) components, separately presented for negative and positive conditions. Feedback-onset at 0 ms; negative is drawn upwards per convention. Abb.: exp-certain: expected-certain; exp-uncertain: expected-uncertain; unexp-uncertain: unexpected-uncertain.

Fig. 4. Left side: Difference waves signal courses (negative minus positive) at electrodes FCz (FRN, upper panel), and Pz (P300, lower panel) for the three expectancy conditions. Please note: more negative values represent more pronounced FRN amplitudes for negative compared to positive feedback, but more pronounced P300 amplitudes for positive compared to negative feedback. Feedback-onset at 0 ms; negative is drawn upwards per convention. Right side: topographical scalp maps for mean FRN (200–300 ms) and P300 (300–500 ms) components of the three expectancy conditions; in μV.

80

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83

both uncertain conditions elicited reliable RPEs whereas expected-certain feedback did not. FRN mean difference wave amplitudes differed significantly for the three conditions (F(2,23) = 14.54, p b 0.001). Pairwise comparisons showed that FRN mean difference wave amplitudes were significantly smaller after expected-certain than after expected-uncertain (t(22) = 3.92, p = 0.001, d = 1.17) and unexpected-uncertain feedback (t(22) = 4.65, p b 0.001, d = 1.57). Amplitudes for expected-uncertain and unexpected-uncertain feedback conditions did not differ from each other (t(22) = 1.06, p = 303). Moreover, an interaction between condition and fearless-dominance scores was found (F(3,23) = 3.10, p = 0.047). Parameter estimates were significant for the interaction with expected-uncertain feedback (β = 2.366; t(23) = 2.81, p = 0.010), but not for the interaction with expected-certain (β = − 0.553; p = 0.363) and unexpected-uncertain (β = 1.229; p = 0.147) feedback. Fearless-dominance scores correlated significantly with FRN mean difference wave amplitudes of expected-uncertain feedback (r s(23) = 0.518, p = 0.011). Higher fearless-dominance scores were associated with less pronounced FRN mean difference wave amplitudes, i.e., less pronounced RPEs (the original waveforms showed only a significant association between fearlessdominance scores and expected-uncertain positive outcomes [(rs(23) = − 0.424, p = 0.044]). Correlations with expected-certain (rs(23) = − 0.211, p = 0.333) and unexpected-uncertain (rs(23) = 0.316, p = 0.142) feedback were not significant. Interestingly, the correlation coefficient of fearless-dominance scores and expectedcertain feedback differed significantly from the correlation coefficient with expected-uncertain feedback (Z = − 2.52, p = 0.012). FRN latencies in the six original conditions were not different from each other (all p-values N 0.308). 3.2.2. P300 P300 mean difference wave amplitudes differed significantly from zero for all three conditions: expected-certain (t(22) = 2.25, p = 0.035), expected-uncertain (t(22) = −5.94, p b 0.001), and unexpected-uncertain (t(22) = − 6.67, p b 0.001). This indicates that we observed reliable differences between negative and positive feedback in all conditions. However, the pattern was not uniform since negative feedback elicited larger P300 amplitudes than positive feedback for expected-certain outcomes, whereas this pattern was reversed for expected-uncertain and unexpected-uncertain outcomes. P300 mean difference wave amplitudes also differed significantly for the three conditions (F(2,23) = 31.90, p b 0.001). Pairwise comparisons showed that P300 mean difference wave amplitudes had significantly less pronounced values (i.e., more positive values) after expectedcertain than expected-uncertain (t(22) = 6.31, p b 0.001, d = 1.66) and unexpected-uncertain feedback (t(22) = 5.96, p b 0.001, d = 1.65), whereas expected-uncertain and unexpected-uncertain feedback did not differ from each other (t(22) = − 0.79, p = 0.441). As mentioned above, difference waves were calculated by subtracting positive from negative feedback trials, thus more negative values represent more pronounced P300 amplitudes for positive compared to negative feedback. Again, an interaction between condition and fearless-dominance scores was found (F(2,23) = 5.81, p = 0.004): Parameter estimates were significant for the interaction with expected-certain feedback (β = − 3.085; t(23) = − 3.92, p = 0.001), but not for the interaction with expected-uncertain (β = − 0.116; p = 0.899) and unexpecteduncertain (β = − 0.522; p = 0.456) feedback. A significant correlation was observed for fearless-dominance scores and P300 mean difference wave amplitudes of expected-certain (r s(23) = − 0.626, p = 0.001) feedback. Higher fearless-dominance scores were associated with less positive P300 mean difference wave amplitudes (i.e., with less differentiation between negative and positive feedback; the original waveforms showed only a significant association

between fearless-dominance scores and expected-certain negative outcomes [rs(23) = − 0.622, p = 0.002]). The correlations for fearless-dominance scores with P300 mean difference wave amplitude of expected-uncertain (r s(23) = 0.051, p = 0.816) and unexpected-uncertain feedback (rs(23) = − 0.196, p = 0.370) were not significant. Again, the correlation coefficients differed significantly between expected-certain and expected-uncertain conditions (Z = − 2.59, p = 0.010). P300 latencies in the six original conditions were not different from each other (all p-values N 0.083).1 Please see Fig. 5 for results of correlation analyses. 4. Discussion The current study aimed to investigate differences in reward prediction error signals (RPEs) in expected and unexpected uncertainty in association with fearless-dominance personality traits in a healthy student sample. Subjective estimates and button press behavior indicated that participants appropriately learned action-outcome associations. Furthermore, FRN and P300 amplitudes differed between expectancy conditions and were differentially associated with fearless-dominance personality traits. 4.1. Fearless-dominance and FRN amplitudes The present results are consistent with previous studies [29,45] revealing differences in FRN amplitudes related to expectancy mismatch. Unexpected-uncertain feedback elicited more negative FRN difference wave amplitudes than expected-certain but comparable amplitudes to expected-uncertain feedback. In the current sample, the two forms of uncertainty were comparably mapped onto FRN mean difference waves. However, considering fearless-dominance scores, we observed a significant positive association between expected-uncertain feedback and fearless-dominance: Individuals with higher fearless-dominance scores showed less negative FRN difference wave amplitudes, i.e., less RPEs, in the expected-uncertain condition. This observation contradicts our initial prediction because we assumed enhanced performance monitoring during both uncertainty conditions. Thus, in an environment with weak action-outcome contingencies, individuals with higher fearless-dominance scores showed diminished feedback monitoring. This is in contrast to expected-certain environments, in which the correlation between the FRN mean difference wave amplitude and fearless-dominance scores had a negative association (indicating enhanced monitoring), although not on a significant level. It is known that subjective involvement modulates FRN amplitudes [42,62]. Performance monitoring in an uncertain environment is an important feature to adapt behavior, but it seems that with increasing reward uncertainty the accuracy to improve performance diminishes and probably subjective involvement decreases as well. Thus, putting more cognitive resources into further increasing performance monitoring might not be the optimal strategy in these kind of uncertain situations. Our data suggest that individuals with higher fearless-dominance scores put less effort into monitoring situations in which they expect the “unpredictable”. Instead, they might cleverly shift monitoring processes to reduce performance 1 To control for potential FRN effects on P300 amplitude variation, we conducted an ANCOVA with the within-subject factor condition and standardized mean FRN difference wave values as covariate. We observed the expected main effect of condition (F(2,42) = 29.69, p b 0.001, η2p = 0.59). Planned comparisons showed that expected-certain outcomes elicited more positive P300 difference waves than the two other conditions (all p-values b 0.001). The covariate was not significant (F(1,21) = 1.43, p = 0.245), nor was the interaction between condition (i.e., expectation level) and the covariate (F(2,42) = 1.07, p = 0.353). We further conducted partial correlations to test whether the association between fearless-dominance scores and P300 amplitude variation changed when controlling for FRN variation in the respective condition. Corroborating the independence of our P300 results of FRN variation, the overall correlation pattern did not change (expected-certain: r(20) = − 0.703, p b 0.001; expected-uncertain: r(20) = −0.225, p = 0.313; unexpected-uncertain: r(20) = −0.260, p = 0.242).

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83

81

be true in unpredictable environments with low action-outcome contingencies, monitoring external feedback may be reduced in more fearless-dominant individuals to spare cognitive resources. This might be one possible mechanism by which more fearless-dominant individuals secure their ability to control and possibly manipulate external situations successfully. Significantly stronger correlations of fearless-dominance scores with RPEs in expected-uncertain compared to expectedcertain feedback conditions are supporting the importance of assessing the interaction of expectancy and uncertainty. 4.2. Fearless-dominance and P300 amplitudes

Fig. 5. Spearman correlations including regression lines between fearless-dominance scores and FRN (upper panel) and P300 difference wave amplitudes (lower panel) for the three feedback expectancy conditions.

evaluation, spare cognitive resources and avoid inefficient efforts in gambling situations where external feedback cannot be used to improve their own outcome. Thus, similar to a study in which the probability of positive outcomes was about 50% and thus not controllable [56], more fearless-dominant individuals of the current study showed diminished feedback processing, but this was specific for environments in which unpredictability was expected. This adaptive strategy might be useful to avoid wasting resources in unpredictable – and unchangeable – environments, whereas it might be a less efficient strategy in more predictable environments. Bresin et al. [8] showed that in individuals with a criminal history, higher fearless-dominance traits were associated with higher internal error monitoring in a flanker task and with adaptive behavior to internal error processing, indicated by post-error slowing. Contrarily, as already mentioned, an adaptive time-estimation paradigm including emotional stimuli resulted in reduced feedback processing in non-forensic individuals with higher fearless-dominance scores [56]. Our results further corroborate the notion that fearless-dominance scores seem to be indicative of reduced feedback monitoring, while error monitoring might be enhanced [55]. Importantly, this reduction in feedback monitoring could possibly represent a strategy for more efficient performance monitoring in unpredictable contexts, and should not solely be regarded as a deficit in performance monitoring. Thus, on the one hand, more fearless-dominant individuals may internally control their own behavior appropriately. On the other hand, and this may especially

Regarding P300 modulation, results revealed that P300 difference wave amplitudes were less pronounced in expected-certain compared to unexpected-uncertain or expected-uncertain feedback. Contrarily to expected-certain trials, in expected-uncertain and unexpected-uncertain ones, the differentiation between positive and negative feedback was more pronounced; with positive feedback yielding more positive P300 amplitudes. These findings are in line with previous studies reporting larger P300 amplitudes following unexpected feedback [6, 18,61,63]. According to Wu and Zhou [61], P300 variation due to valence requires the development and, subsequently, the violation of expectations concerning an outcome – thereby possibly explaining the less pronounced valence effect in expected-certain trials in the current sample. P300 amplitude variation is strongly associated with attention processing [48], which is definitely required in uncertain environments. For example, a match-mismatch control strategy might be useful for efficient performance monitoring [53]. Interestingly, in the current study, a general increase in attention processing was seen in higher fearless-dominant individuals in expected-certain situations due to less differentiation between negative and positive outcomes (driven mostly by negative outcomes). In line with the current results, an increase in attention processing was also shown specifically for more fearless-dominant individuals [9]. Our results corroborate these findings and additionally show that increased attention processing in more fearless-dominant individuals is related to expected-certain environments. Over-focused top-down attention in more fearless-dominant individuals in expected-certain situations may be associated with constantly monitoring reliability of the situation – less for negative, but more for positive outcomes [9,34]. Furthermore, there is evidence that attention processing modulates threat processing in more fearless-dominant individuals [15]. This is in accordance with the current results and indicates that guided attention processing in more fearless-dominant individuals may account for a lack of anxiety and high stress-coping abilities in these individuals [9]. Interestingly, the current expectancy manipulations led to different patterns as to whether positive or negative feedback elicits more pronounced P300 amplitudes. Previous inconsistent results in this regard might be attributable to different experimental designs, different manipulations of reward certainty and reward probability, and possibly even to the use of different feedback stimuli. For example, studies with high outcome uncertainty often report larger P300 amplitudes after positive/favorable than negative/unfavorable outcomes (e.g., [47, 71]). We speculate that positive outcomes might be perceived as more salient (i.e., attention focusing) and task-relevant in uncertain environments, while for negative outcomes this might be the case in certain environments. Future studies should address this possibility by designing experimental paradigms that specifically target the interplay of reward certainty and probability. 4.3. Limitations The experimental design was limited in the way that the expectedcertain conditions contained four times as many trials as the other conditions. Nevertheless, we are confident that particularly the uncertain conditions were comparable on the level of trial numbers. The pre-

82

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83

experimental learning phase and the subsequent modulation of expectancy in addition to different reward probabilities are critical assets of the current study, since previous research did not show differences in FRN amplitudes after uncertain and unexpected feedback [28], but largest FRN effects when a prediction was learned previously [28]. Our results are in accordance with previous theories indicating that the design was successfully applied. Future research could extend the current paradigm and add neutral trials to expected-certain, expected-uncertain, and unexpected-uncertain feedback conditions. We used both a regression-based approach and independent component analysis to detect and correct eye-movement related artifacts in this experiment. While we have successfully applied this approach in previous studies (e.g., [45,56]), we acknowledge that this might limit the comparability of our data with other studies applying solely one or the other method. For both feedback and attention processing in relation to fearlessdominance personality traits, sex might be a significant contributing factor [2,58]. Since we have to acknowledge that the current sample size was already rather small – rendering the current results as preliminary evidence - adding the factor sex to the statistical models was not an option. In this regard, we did not perform an a priori power analysis. While the small sample size is a limitation of the current study, effect sizes of the pairwise post-hoc comparisons can be classified as large for both FRN and P300 difference wave amplitudes [66]. We further conducted post-hoc power analyses on the pairwise comparisons (using GPower 3.1; [67]), which suggested that the power of our explorative study, despite the small sample size, was appropriate for the tested effects (π N 0.80). Thus, the current findings can be regarded as preliminary evidence and we strongly emphasize that future studies should strive for larger sample sizes allowing for example the possibility to assess sex differences concerning fearless-dominance personality traits and feedback processing or learning effects in relation to fearless-dominance and uncertainty. 5. Conclusions The current study aimed at specifically investigating the association between feedback processing in expected and unexpected environments and fearless-dominance personality traits. Both FRN and P300 results emphasize the importance of uncertainty in efficient feedback processing. More fearless-dominant individuals showed diminished feedback processing in expected-uncertain situations. Thus, individuals with higher fearless-dominance scores might put less effort into monitoring situations in which they expect the “unpredictable” and external feedback cannot be used to improve the outcome. They might reduce performance evaluation, spare cognitive resources and avoid inefficient efforts in these gambling situations. Furthermore, different correlation patterns of fearless-dominance scores with expected-certain and expected-uncertain feedback in FRN as well as P300 difference wave amplitudes demonstrated the significance of differentiating these situations when assessing RPEs. The probabilistic features of situations have to be taken into account when assessing performance monitoring in more fearless-dominant individuals. Assessing the interaction between fearless-dominance personality traits and performance monitoring complements the knowledge on personality aspects associated with psychopathy in non-forensic individuals. Acknowledgements The participants' financial remuneration was funded by a scholarship of the University of Vienna awarded to LK (Förderstipendium StudFG). Part of this work was undertaken while LK was funded by the IRTG1328 (DFG).

Preliminary results of parts of this study were presented at the OHBM Beijing (2012). All authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References [1] G.W. Alpers, H. Eisenbarth, Psychopathic Personality Inventory-Revised (PPI-R). Deutsche Version, Hogrefe Verlag, Göttingen, 2008. [2] M.E. Anton, A.R. Baskin-Sommers, J.E. Vitale, J.J. Curtin, J.P. Newman, Differential effects of psychopathy and antisocial personality disorder symptoms on cognitive and fear processing in female offenders, 2012. Cogn. Affect. Behav. Neurosci. 12 (4) 761–776, http://dx.doi.org/10.3758/s13415-012-0114-x.Differential. [3] H. Bauer, W. Lauber, Operant conditioning of brain steady potential shifts in man, Biofeedback Self Regul. 4 (2) (1979) 145–154. [4] A.J. Bell, T.J. Sejnowski, An information-maximization approach to blind separation and blind deconvolution, Neural Comput. 7 (6) (1995) 1129–1159. [5] C. Bellebaum, D. Polezzi, I. Daum, It is less than you expected: the feedback-related negativity reflects violations of reward magnitude expectations, 2010. Neuropsychologia 48 (11) 3343–3350 . http://www.scopus.com/inward/record.url? eid=2-s2.0-77955847129&partnerID=40&md5=d9d38304387b78a1302ac1562 b8d1edf. [6] A.R. Bland, A. Schaefer, Electrophysiological correlates of decision making under varying levels of uncertainty, 2011. Brain Res. 1417 55–66, http://dx.doi.org/10. 1016/j.brainres.2011.08.031. [7] M.H. Branchey, L. Buydens-Branchey, T.B. Horvath, Event-related potentials in substance-abusing individuals after long-term abstinence, Am. J. Addict. 2 (2) (1993) 141–148. [8] K. Bresin, M.S. Finy, J. Sprague, E. Verona, Response monitoring and adjustment: differential relations with psychopathic traits, J. Abnorm. Psychol. 123 (3) (2014) 634–649. [9] S.R. Carlson, S. Thái, ERPs on a continuous performance task and self-reported psychopathic traits: P3 and CNV augmentation are associated with fearless dominance, 2010. Biol. Psychol. 85 (2) 318–330, http://dx.doi.org/10.1016/j.biopsycho.2010.08. 002. [10] A.J. Cooper, É. Duke, A.D. Pickering, L.D. Smillie, Individual differences in reward prediction error: contrasting relations between feedback-related negativity and trait measures of reward sensitivity, impulsivity and extraversion, 2014. Front. Hum. Neurosci. 8 1–11, http://dx.doi.org/10.3389/fnhum.2014.00248. [11] L. Costa, L. Bauer, S. Kuperman, B. Porjesz, S. O'Connor, V. Hesselbrock, ... H. Begleiter, Frontal P300 decrements, alcohol dependence, and antisocial personality disorder, Biol. Psychiatry 47 (12) (2000) 1064–1071. [12] A. Delorme, S. Makeig, EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis, J. Neurosci. Methods 134 (1) (2004) 9–21. [13] Z.V. Dikman, J.J.B. Allen, Error monitoring during reward and avoidance learning in high- and low-socialized individuals, Psychophysiology 37 (1) (2000) 43–54. [14] C.C. Duncan-Johnson, E. Donchin, On quantifying surprise. The variation of event related potentials with subjective probability, Psychophysiology 14 (5) (1977) 456–467. [15] J.D. Dvorak-Bertscha, J.J. Curtin, T.J. Rubinstein, J.P. Newman, Psychopathic traits moderate the interaction between cognitive and affective processing, 2009. Psychophysiology 46 (5) 913–921, http://dx.doi.org/10.1111/j.1469-8986.2009.00833.x. Psychopathic. [16] J.F. Edens, D.K. Marcus, M.G. Vaughn, Exploring the taxometric status of psychopathy among youthful offenders: is there a juvenile psychopath taxon? 2011. Law Hum. Behav. 35 (1) 13–24, http://dx.doi.org/10.1007/s10979-010-9230-8. [17] W.J. Gehring, A.R. Willoughby, The medial frontal cortex and the rapid processing of monetary gains and losses, 2002. Science 295 (5563) 2279–2282, http://dx.doi.org/ 10.1126/science.1066893. [18] G. Hajcak, C.B. Holroyd, J.S. Moser, R.F. Simons, Brain potentials associated with expected and unexpected good and bad outcomes, 2005. Psychophysiology 42 (2) 161–170, http://dx.doi.org/10.1111/j.1469-8986.2005.00278.x. [19] G. Hajcak, J.S. Moser, C.B. Holroyd, R.F. Simons, The feedback-related negativity reflects the binary evaluation of good versus bad outcomes, 2006. Biol. Psychol. 71 (2) 148–154, http://dx.doi.org/10.1016/j.biopsycho.2005.04.001. [20] G. Hajcak, J.S. Moser, C.B. Holroyd, R.F. Simons, It's worse than you thought: the feedback negativity and violations of reward prediction in gambling tasks, 2007. Psychophysiology 44 (6) 905–912 . http://www.scopus.com/inward/record.url?eid=2-s2. 0-35348999040&partnerID=40&md5=0f71530459c35d847651ece2fae6e889. [21] J.R. Hall, E.M. Bernat, C.J. Patrick, Externalizing psychopathology and error-related negativity, 2007. Psychol. Sci. 18 (4) 326–333, http://dx.doi.org/10.1111/j.14679280.2007.01899.x.Externalizing. [22] R.D. Hare, The Hare Psychopathy Checklist-Revised. (Vol. Second Edi), Multi-Health Systems, Toronto, ON, 2003. [23] T.U. Hauser, R. Iannaccone, P. Stampfli, R. Drechsler, D. Brandeis, S. Walitza, S. Brem, The feedback-related negativity (FRN) revisited: new insights into the localization, meaning and network organization, 2014. NeuroImage 84 159–168, http://dx.doi. org/10.1016/j.neuroimage.2013.08.028. [24] B.Y. Hayden, S.R. Heilbronner, J.M. Pearson, M.L. Platt, Surprise signals in anterior cingulate cortex: neuronal encoding of unsigned reward prediction errors driving adjustment in behavior, 2011. J. Neurosci. 31 (11) 4178–4187, http://dx.doi.org/ 10.1523/jneurosci.4652-10.2011.

L. Kogler et al. / Physiology & Behavior 168 (2017) 74–83 [25] A.J. Heritage, S.D. Benning, Impulsivity and response modulation deficits in psychopathy: evidence from the ERN and N1, 2013. J. Abnorm. Psychol. 122 (1) 215–222, http://dx.doi.org/10.1037/a0030039. [26] C.B. Holroyd, M.G.H. Coles, The neural basis of human error processing: reinforcement learning, dopamine, and the error-related negativity, Psychol. Rev. 109 (4) (2002) 679–709. [27] C.B. Holroyd, G. Hajcak, J.T. Larsen, The good, the bad and the neutral: electrophysiological responses to feedback stimuli, 2006. Brain Res. 1105 (1) 93–101, http://dx. doi.org/10.1016/j.brainres.2005.12.015. [28] C.B. Holroyd, O.E. Krigolson, R. Baker, S. Lee, J. Gibson, When is an error not a prediction error? An electrophysiological investigation, 2009. Cogn. Affect. Behav. Neurosci. 9 (1) 59–70, http://dx.doi.org/10.3758/CABN.9.1.59. [29] C.B. Holroyd, S. Nieuwenhuis, N. Yeung, J.D. Cohen, Errors in reward prediction are reflected in the event-related brain potential, Neuroreport 14 (18) (2003) 2481–2484. [30] R.J. Johnson, E. Donchin, P300 and stimulus categorization: two plus one is not so different from one plus one, Psychophysiology 17 (2) (1980) 167–178. [31] T.W. Lee, M. Girolami, T.J. Sejnowski, Independent component analysis using an extended infomax algorithm for mixed subgaussian and supergaussian sources, Neural Comput. 11 (2) (1999) 417–441. [32] S.O. Lilienfeld, C.J. Patrick, S.D. Benning, J. Berg, M. Sellbom, J.F. Edens, The role of fearless dominance in psychopathy: confusions, controversies, and clarifications, 2012. Personal. Disord.: Theory, Res. Treat. 3 (3) 327–340, http://dx.doi.org/10. 1037/a0026987. [33] S.O. Lilienfeld, M.R. Widows, Psychopathic Personality Inventory-Revised: Professional Manual, Psychological Assessment Resources, Lutz, Florida, 2005. [34] Y. Long, X. Jiang, X. Zhou, To believe or not to believe: trust choice modulates brain response in outcome evaluation, 2012. Neuroscience 200 50–58, http://dx.doi.org/ 10.1016/j.neuroscience.2011.10.035. [35] D.R. Lynam, J.D. Miller, Fearless dominance and psychopathy: a response to Lilienfeld et al, 2012. Personal. Disord.: Theory, Res. Treat. 3 (3) 341–353, http:// dx.doi.org/10.1037/a0028296. [36] J.H.R. Maes, I.A. Brazil, No clear evidence for a positive association between the interpersonal-affective aspects of psychopathy and executive functioning, 2013. Psychiatry Res. 0, http://dx.doi.org/10.1016/j.psychres.2013.09.028. [37] J.D. Miller, D.R. Lynam, An examination of the psychopathic personality inventory's nomological network: a meta-analytic review, 2012. Personal. Disord.: Theory, Res. Treat. 3 (3) 305–326, http://dx.doi.org/10.1037/a0024567. [38] J.D. Miller, D.R. Lynam, T.a. Widiger, C. Leukefeld, Personality disorders as extreme variants of common personality dimensions: can the five-factor model adequately represent psychopathy? 2001. J. Pers. 69 (2) 253–276, http://dx.doi.org/10.1111/ 1467-6494.00144. [39] W.H.R. Miltner, C.H. Braun, M.G.H. Coles, Event-related brain potentials following incorrect feedback in a time-estimation task: evidence for a “generic” neural system for error detection, 1997. J. Cogn. Neurosci. 9 (6) 788–798 http://www.scopus. com/inward/record.url?eid=2-s2.0-0031436055&partnerID=40&md5= 23777dda239215a7398ce2674d74a241. [40] F. Mushtaq, A.R. Bland, A. Schaefer, Uncertainty and cognitive control, 2011. Front. Psychol. 2 249, http://dx.doi.org/10.3389/fpsyg.2011.00249. [41] S. Nieuwenhuis, G. Aston-Jones, J.D. Cohen, Decision making, the P3, and the locus coeruleus-norepinephrine system, 2005. Psychol. Bull. 131 (4) 510–532, http://dx. doi.org/10.1037/0033-2909.131.4.510. [42] S. Nieuwenhuis, N. Yeung, C.B. Holroyd, A. Schurger, J.D. Cohen, Sensitivity of electrophysiological activity from medial frontal cortex to utilitarian and performance feedback, 2004. Cereb. Cortex 14 (7) 741–747, http://dx.doi.org/10.1093/cercor/ bhh034. [43] R.C. Oldfield, The assessment and analysis of handedness: the Edinburgh inventory, Neuropsychologia 9 (1) (1971) 97–113. [44] C.J. Patrick, E.M. Bernat, Neurobiology of psychopathy: a two-process theory, in: G.G. Berntson, J.T. Cacioppo (Eds.), Handbook of Neuroscience for the Behavioral Sciences, John Wiley & Sons, New York 2009, pp. 1110–1131. [45] D.M. Pfabigan, J. Alexopoulos, H. Bauer, U. Sailer, Manipulation of feedback expectancy and valence induces negative and positive reward prediction error signals manifest in event-related brain potentials, 2011. Psychophysiology 48 (5) 656–664, http://dx.doi.org/10.1111/j.1469-8986.2010.01136.x. [46] D.M. Pfabigan, J. Alexopoulos, H. Bauer, C. Lamm, U. Sailer, All about the money – external performance monitoring is affected by monetary, but not by socially conveyed feedback cues in more antisocial individuals, 2011. Front. Hum. Neurosci. 5 100. http://dx.doi.org/10.3389/fnhum.2011.00100.

83

[47] D.M. Pfabigan, U. Sailer, C. Lamm, Size does matter! Perceptual stimulus properties affect event-related potentials during feedback processing, 2015. Psychophysiology 52 1238–1247, http://dx.doi.org/10.1111/psyp.12458. [48] J. Polich, Updating P300: an integrative theory of P3a and P3b, 2007. Clin. Neurophysiol. 118 (10) 2128–2148, http://dx.doi.org/10.1016/j.clinph.2007.04.019. [49] G.F. Potts, M.R.M. George, L.E. Martin, E.S. Barratt, Reduced punishment sensitivity in neural systems of behavior monitoring in impulsive individuals, Neurosci. Lett. 397 (1–2) (2006) 130–134. [50] R.A. Rescorla, A.R. Wagner, A Theory of Pavlovian Conditioning: Variations in the Effectiveness of Reinforcement and Nonreinforcement, Classical Conditioning II: Current Research and Theory, 1972, pp. 64–99. [51] G.A. Rousselet, C.R. Pernet, Improving standards in brain-behavior correlation analyses, 2012. Front. Hum. Neurosci. 6 119, http://dx.doi.org/10.3389/fnhum.2012. 00119. [52] M.T. Ruchsow, M. Spitzer, G. Grön, J. Grothe, M. Kiefer, Error processing and impulsiveness in normals: evidence from event-related potentials, 2005. Cogn. Brain Res. 24 317–325, http://dx.doi.org/10.1016/j.cogbrainres.2005.02.003. [53] U. Sailer, F.P.S. Fischmeister, H. Bauer, Effects of learning on feedback-related brain potentials in a decision-making task, 2010. Brain Res. 1342 85–93 http://www. scopus.com/inward/record.url?eid=2-s2.0-77953293507&partnerID=40&md5= e9c7855502e3ae436c76e5cb7450cd33. [54] D.L. Santesso, S.J. Segalowitz, L.A. Schmidt, ERP correlates of error monitoring in 10year olds are related to socialization, 2005. Biol. Psychol. 70 (2) 79–87, http://dx.doi. org/10.1016/j.biopsycho.2004.12.004. [55] S. Schulreich, Altered performance monitoring in psychopathy: a review of studies on action selection, error, and feedback processing, Curr. Behav. Neurosci. Rep. 3 (2016) 19–27. [56] S. Schulreich, D.M. Pfabigan, B. Derntl, U. Sailer, Fearless dominance and reduced feedback-related negativity amplitudes in a time-estimation task – further neuroscientific evidence for dual-process models of psychopathy, 2013. Biol. Psychol. 93 (3) 352–363, http://dx.doi.org/10.1016/j.biopsycho.2013.04.004. [57] W.A. Stephenson, F.A. Gibbs, A balanced non-cephalic reference electrode, Electroencephalogr. Clin. Neurophysiol. 3 (2) (1951) 237–240. [58] E. Verona, K. Bresin, C.J. Patrick, Revisiting psychopathy in women: Cleckley/Hare conceptions and affective response, 2013. J. Abnorm. Psychol. 122 (4) 1088–1093, http://dx.doi.org/10.1037/a0034062. [59] E. Verona, C.J. Patrick, T.E. Joiner, Psychopathy, antisocial personality, and suicide risk, J. Abnorm. Psychol. 110 (3) (2001) 462–470. [60] B.M. Wilkowski, M.D. Robinson, Putting the brakes on antisocial behavior: secondary psychopathy and post-error adjustments in reaction time, 2008. Personal. Individ. Differ. 44 (8) 1807–1818, http://dx.doi.org/10.1016/j.paid.2008.02.007. [61] Y. Wu, X. Zhou, The P300 and reward valence, magnitude, and expectancy in outcome evaluation, 2009. Brain Res. 1286 114–122 . http://www.scopus.com/inward/record.url?eid=2-s2.0-67849090370&partnerID=40&md5= 3994ff89a75d7215490d109032062d3d. [62] N. Yeung, C.B. Holroyd, J.D. Cohen, ERP correlates of feedback and reward processing in the presence and absence of response choice, Cereb. Cortex 15 (5) (2005) 535–544. [63] N. Yeung, A.G. Sanfey, Independent coding of reward magnitude and valence in the human brain, J. Neurosci. 24 (28) (2004) 6258–6264. [64] A.J. Yu, P. Dayan, Uncertainty, neuromodulation, and attention, 2005. Neuron 46 (4) 681–692, http://dx.doi.org/10.1016/j.neuron.2005.04.026. [65] R. Yu, W. Zhou, X. Zhou, Rapid processing of both reward probability and reward uncertainty in the human anterior cingulate cortex, 2011. PLoS One 6 (12), e29633, http://dx.doi.org/10.1371/journal.pone.0029633. [66] J. Cohen, Statistical Power Analysis for the Behavioral Sciencies, Routledge, 1977. [67] F. Faul, E. Erdfelder, A.-G. Lang, A. Buchner, G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences, Behav. Res. Methods 39 (2007) 175–191. [68] S.J. Luck, N. Gaspelin, How to Get Statistically Significant Effects in Any ERP Experiment (and Why You Shouldn't), Psychophysiology, 2016 (in press). [69] S.J. Luck, An Introduction to the Event-related Technique, MIT Press, Cambridge, 2005. [70] R.R. Wilcox, Modern Statistics for the Social and Behavioral Sciences: A Practical Introduction, 2012. [71] D.M. Pfabigan, E.-M. Seidel, K. Paul, A. Grahl, U. Sailer, R. Lanzenberger, ... C. Lamm, Context sensitivity of the feedback-related negativity for zero-value feedback outcomes, 2015. Biol. Psychol. 104 184–192, http://dx.doi.org/10.1016/j.biopsycho. 2014.12.007.

Suggest Documents