Perfect error processing: Perfectionism-related variations in action monitoring and error processing mechanisms

International Journal of Psychophysiology 97 (2015) 153–162

Contents lists available at ScienceDirect

International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho

Perfect error processing: Perfectionism-related variations in action monitoring and error processing mechanisms Jutta Stahl a,⁎, Manuela Acharki a, Miriam Kresimon a, Frederike Völler a, Henning Gibbons b a b

University of Cologne, Germany University of Bonn, Germany

a r t i c l e

i n f o

Article history: Received 3 December 2014 Received in revised form 3 June 2015 Accepted 4 June 2015 Available online 10 June 2015 Keywords: Event-related potentials Error processing Action monitoring Individual differences Perfectionism

a b s t r a c t Showing excellent performance and avoiding poor performance are the main characteristics of perfectionists. Perfectionism-related variations (N = 94) in neural correlates of performance monitoring were investigated in a flanker task by assessing two perfectionism-related trait dimensions: Personal standard perfectionism (PSP), reflecting intrinsic motivation to show error-free performance, and evaluative concern perfectionism (ECP), representing the worry of being poorly evaluated based on bad performance. A moderating effect of ECP and PSP on error processing – an important performance monitoring system – was investigated by examining the error (-related) negativity (Ne/ERN) and the error positivity (Pe). The smallest Ne/ERN difference (error–correct) was obtained for pure-ECP participants (high-ECP–low-PSP), whereas the highest difference was shown for those with high-ECP–high-PSP (i.e., mixed perfectionists). Pe was positively correlated with PSP only. Our results encouraged the cognitive-bias hypothesis suggesting that pure-ECP participants reduce response-related attention to avoid intense error processing by minimising the subjective threat of negative evaluations. The PSPrelated variations in late error processing are consistent with the participants' high in PSP goal-oriented tendency to optimise their behaviour. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The trait perfectionism reflects the stable disposition to show error free performance. Different perspectives are employed in perfectionism research to conceptualize this trait, in particular, group-based models classifying types of perfectionists (e.g., Hamachek, 1978: normal perfectionism vs. neurotic perfectionism) and dimensional trait models of perfectionism (e.g., Frost et al., 1990; Hewitt and Flett, 1991). Frost et al. (1990) suggested six dimensions of perfectionism which are, at least partially, independent (see Frost et al., 1990). Concern over mistakes describes the tendency to equate errors with personal failure and represents the expectation of negative consequences such as negative evaluation by others. Personal standards reflects the demand of very high criteria to evaluate one's own performance. Doubts about action represents the characteristic of not being satisfied with the quality of one's own performance irrespective of the actual, objective outcome. The two dimensions, parental expectations and parental criticism, refer to the individual's mind-set regarding the impressions of extremely high parental demands and excessive criticism in case of imperfect

⁎ Corresponding author at: Department of Psychology, University of Cologne, Pohligstr. 1, D-50969 Cologne, Germany. Tel.: +49 221 470 6289 (Office); fax: +49 221 470 5105. E-mail address: [email protected] (J. Stahl).

http://dx.doi.org/10.1016/j.ijpsycho.2015.06.002 0167-8760/© 2015 Elsevier B.V. All rights reserved.

behaviour. Finally, organisation describes the preference for order and precision. Two perfectionism dimensions are crucial to reliably predict variations in behavioural tendencies related to performance and performance evaluation. By combining these key dimensions, evaluative concern perfectionism (ECP) and personal standards perfectionism (PSP), in their 2 × 2-model of perfectionism, Gaudreau and Thompson (2010, 2011, see also Gaudreau, 2013), their model amalgamates the advantages of the dimensional and the group-based approaches in ways that four perfectionism subtypes eventuate: pure-PSP (low concerns and high standards), mixed perfectionism (high concerns and high standards), non-perfectionism (low concerns and low standards), and pure-ECP (high concerns and low standards). Considering the dimensions separately, people with high ECP are often more anxious, show higher neuroticism scores, as well as avoidant coping styles, and they tend to develop depression and obsessive symptoms more frequently. Conversely, people with higher PSP show higher positive affect, higher conscientiousness, endurance, less external locus of control and active coping styles (for review see, Stoeber and Otto, 2006). In this context, the interactionist view of Gaudreau and Thompson's model (2010, 2011) made a valuable initial contribution by showing that high ECP participants benefit from high PSP as a kind of protecting factor compared to pure-ECP participants. Brand and Altstötter-Gleich (2008) further supported the benefit of an interactionist view by their finding

154

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

that the ECP-by-PSP interaction was a better predictor for decisionmaking performance than the single traits. As perfectionism comprises cognitive, emotional and motivational characteristics linked to performance and to the evaluation of its quality, we aimed at investigating the neurophysiological correlates of perfectionism-related variations in error processing using an interactionist approach. As individual behaviour cannot be free of errors constantly, it is an important question whether the different subtypes of perfectionists diverge in action monitoring and error processing. 2. Error processing Error processing can be investigated by means of two dissociable components of the event-related potential (ERP), the error (related) negativity (Ne/ERN, Falkenstein et al., 1991; Gehring et al., 1993), and the error positivity (Pe, Falkenstein et al., 1991). Usually, the peaks of the two components can be observed at approximately 100 ms after an erroneous response with a negative medial–frontal scalp distribution (Ne/ERN), and at around 300 ms with a positive centro-parietal scalp distribution (Pe). Several studies showed that the two components are associated with partially distinct neural sources (e.g. Ne/ERN: anterior cingulate cortex; see Ridderinkhof et al., 2004; Pe: anterior cingulate cortex, posterior cingulate cortex, parietal, and insular cortices; Herrmann et al., 2004; Orr and Hester, 2012; Van Veen and Carter, 2002; Vocat et al., 2008). Falkenstein et al. (1991) were the first to suggest that the Ne/ERN reflects the activity of an internal error-detection mechanism that responds to mismatches between representations of the actually given response and the required response (for further development, see also Gibbons et al., 2011). The underlying activity is mediated by the midfrontal dopaminergic system by reinforcement learning processes (Holroyd and Coles, 2002). The Ne/ERN is also affected by more general action monitoring processes (e.g., Vidal et al., 2000), such as ongoing response conflict monitoring (Armbrecht et al., 2010; Stahl and Gibbons, 2007; Yeung et al., 2004), or force production monitoring (Armbrecht et al., 2012, 2013; De Bruijn et al., 2003). A component similar to the Ne/ERN with respect to topography and time course can be observed after correct responses (CRN, correct response negativity, Vidal et al., 2000). It is still debated as to whether the Ne/ERN and CRN reflect the same action monitoring process (Hoffmann and Falkenstein, 2010) or whether they constitute different mechanisms (Endrass et al., 2012). As the Pe amplitude is higher when participants are aware of an error, as compared with when they are not, the Pe is interpreted as the neural correlate of conscious error processing (e.g., Nieuwenhuis et al., 2001; O'Connell et al., 2007). Moreover, the Pe amplitude correlates with indicators of post-error behaviour, which indicates a relationship to strategic response adaptation processes (e.g., Overbeek et al., 2005; see also Danielmeier and Ullsperger, 2011; Schroder and Infantolino, 2013). An accumulator model (Steinhauser and Yeung, 2010, 2012) describes the Pe amplitude as reflecting the accumulated evidence of error commission, which is required in a post-response decision concerning the correctness of a response.

anxious characteristics, especially anxious apprehension (i.e., worrying and rumination) in common. According to Moser et al.'s (2013) analyses, apprehension and the non-task related cognitive workload are the crucial characteristics explaining the relationship between anxiousness and error processing. Interestingly, to date, no relationship between the concern-related perfectionism (ECP) and the Ne/ERN has been revealed (Pieters et al., 2007; Schrijvers et al., 2010; Tops et al., 2013). This is surprising since the concern over mistakes scale assesses the tendency of worrying about being poorly evaluated by others based on imperfect (erroneous) behaviour. Thus, according to Moser et al.'s (2013) theory a relationship would be expected. In a reply to Proudfit et al. (2013) comments on the study by Moser et al. (2013), however, the authors (Moser et al., 2014) stated that the concern over mistakes scale assesses several different aspects of anxiety-related characteristics, that is, not only worry-related anxiety. The three above-mentioned studies (Pieters et al., 2007; Schrijvers et al., 2010; Tops et al., 2013), however, demonstrated an effect of some perfectionism sub-traits on error-related ERP components. Pieters et al. (2007) used perfectionism as a moderator variable in a design contrasting anorexia nervosa patients with a healthy control group. In the healthy control group (N = 19), the authors observed a more negative Ne/ERN amplitude in healthy individuals with a higher overall perfectionism score (i.e., the sum of the Multidimensional Perfectionism Scale (FMPS) developed by Frost et al., 1990), but there was no such variation in the anorexia nervosa group (N = 17). The second study (Schrijvers et al., 2010) showed higher Ne/ERN amplitudes in individuals with higher doubts-about-action scores, and higher Pe amplitudes in participants with higher ECP scores (medicated participants with major depressive disorder; N = 39). In contrast to these studies, Tops et al. (2013) investigated the specific relationship between ECP and the three components [Ne/ERN, early Pe (150–350 ms), and late Pe (400–500 ms)] in 16 healthy participants and reported a positive relationship between ECP and late frontal Pe. However, the authors could not preclude that this effect resulted from an overlap of the stimuluspreceding negativity (i.e., feedback preceding), an ERP component which is a response to the anticipation of upcoming stimuli, with the Pe. The feedback stimuli (happy or disgusted faces for positive or negative feedbacks, respectively) were presented, on average, 565 ms after response onset. As these feedbacks may have had an ECP-related emotional arousing effect, the stimulus-preceding negativity could have been decreased in the crucial time period of the late Pe and, thus, could have resulted in a relative increase of late Pe for participants with high ECP. Remarkably, the Ne/ERN and the early Pe did not correlate with ECP in their study. None of the above-mentioned studies, however, accounted for an interaction of perfectionism sub-traits or other traits, although at least five decades of personality research show that related personality traits, such as anxiety and impulsivity interact (e.g., Reinforcement-Sensitivity theory; Gray, 1970; revised Gray and McNaughton, 2000). Therefore, we set out to investigate whether an interaction of ECP and PSP predicts variations in error processing better than the single trait dimensions.

3. Perfectionism and error processing

4. The objective of the present study

Considerable research on individual differences in error processing (including personality traits and clinical groups) has provided evidence of highly anxious individuals with higher Ne/ERN amplitudes in response conflict tasks (e.g., flanker task, and Stroop task). In a metaanalysis, Moser et al. (2013) reported a small-to-medium effect of anxiety on Ne/ERN. The authors considered studies that investigated error processing and general anxiety disorder (e.g., Weinberg et al., 2010, 2012), obsessive–compulsive disorder (Gehring et al., 2000; Zambrano-Vazquez and Allen, 2014), trait anxiety (Aarts, and Pourtois, 2010), and the behavioural inhibition system (Boksem et al., 2006). Although these groups differ in several aspects, they have

From Moser and colleagues' (2013) study, we know that one general factor of anxiety (i.e., worrying) is related to the Ne/ERN. The authors, however, did not consider moderating the effects of other traits in their meta-analysis. This is not surprising as most of the contributing studies did not use an interactionist approach. Our study aimed to bridge this gap by including the ECP-by-PSP interaction (Gaudreau and Thompson, 2010) in our design and to account for some further restrictions of the previous studies (Pieters et al., 2007; Schrijvers et al., 2010), among them (medicated) clinical samples or small sample sizes that restrict the ability to draw general conclusions on personality traits and to reliably examine dimensional trait variations.

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

Based on the findings of Schrijvers et al. (2010), we expected a negative relationship (note that Ne/ERN has a negative sign) between Ne/ERN amplitude and PSP, which can be explained by variations in motivation. High motivation to perform faultlessly has increasing effects on Ne/ERN amplitudes (e.g., Gehring et al., 1993). As mentioned above, previous studies reported Pe to be a correlate of post-error adaptation (Overbeek et al., 2005) as well as a correlate of conscious error processing (e.g., Nieuwenhuis et al., 2001; O'Connell et al., 2007), both of which are important processes for improving the performance outcome. Participants with high PSP who are highly motivated to show excellent performance should use the conscious error processing system more effectively compared to participants with low PSP; wherefore we supposed a positive relationship between PSP and Pe. Although no ECP-related variations with the Ne/ERN have been reported thus far, we aimed to investigate whether ECP, that is, the worry about being poorly evaluated by others (here, the experimenter) influenced the postulated relationship between PSP and Ne/ERN. As Schrijvers et al. (2010) found a positive correlation between ECP and Pe in a clinical sample (anxiety and major depression), we also expected a positive correlation between ECP and Pe. 5. Method 5.1. Participants Ninety-four participants (51 females; 8 left handed; all students from different faculties) were recruited through e-mail announcements via mailing lists provided by different student councils of the University of Cologne. All were recruited by a computer-aided registration tool for experiments (CORTEX, Elson and Bente, 2009). Participants were paid €8.00 per hour or received course credits for their participation. Their ages ranged from 19 to 42 years (mean ± standard deviation (SD): 25.1 ± 5.8 years). None of the recruited participant had to be excluded for any reason: all participants reported normal or corrected-to-normal vision; none of the participant reported a neurological or psychiatric disorder, nor did we have to exclude participants because of personality-based criteria. Informed written consent was given prior to the investigation and the study was approved by the ethics committee of the German Psychological Association. 5.2. Psychometric assessment The German version of the FMPS (Altstötter-Gleich and Bergemann, 2006) was used to assess PSP by the scores of the personal standard scale (seven items, Cronbach's alpha reliability of the present data: rα = .773) and ECP by the scores of the concern over mistakes scale (nine items: rα = .881). To compare the present results with the previous findings of Schrijvers et al. (2010) and Pieters et al. (2007), we further assessed the four FMPS subscales: parental expectation scale (five items: rα = .895), parental criticism (four items: rα = .825), doubt about action (four items: rα = .725), organisation (six items: rα = .910) and the overall score (including all items except the items of organisation, i.e., 29 items: rα = .904). The item scales ranged from 0 (never) to 5 (always) and the individual scores were computed by averaging the item scores. The responses were checked for missing values. All participants responded to all items on the scales. 5.3. Procedure and experimental task An Eriksen flanker task (Eriksen and Eriksen, 1974) was employed in the present study. The stimuli (five white letters on a black background) were presented in the centre of a 17″ VGA monitor (HHHHH, HHSHH, SSHSS, SSSSS). The participants were instructed to respond to the central letters with the left or right index finger as fast and as accurately as possible, and to ignore the flanking letters. The stimulus–response assignment was balanced across participants. The experiment consisted of

155

eight blocks (48 trials per block with stimuli randomised within the blocks). The stimulus duration was 84 ms. The response interval was set from 0 to 1000 ms after stimulus onset. A reaction time (RT) error was defined as a response longer than 1000 ms. A hand error was defined as the use of the incorrect hand. A feedback was presented for 400 ms. To prevent an overlap of ERP components like the SPN such as in the study of Tops et al. (2013; see the Introduction section) and to minimise a reducing impact on the Ne/ERN component of a trial-by trial feedback (Olvet and Hajcak, 2009), the feedback onset was 1000 ms after the response interval (i.e., 2000 ms after stimulus onset). We aimed to investigate the impact of different feedback types (but see below), thus, we used symbols and words as feedback to indicate the accuracy of the response. In Blocks 1 to 4, symbols were used, and in Blocks 5 to 8 German words were used for correct responses (‘+’ or ‘perfekt’ — perfect), hand errors (‘#’ or ‘falsch’ — wrong) and RT errors (‘!’ or ‘schneller’ — faster). The next trial began after 200 ms. 5.3.1. Behavioural data Behavioural data were recorded by means of force-sensitive keys. Each key was composed of a plastic cuboid (110 × 19 × 2 mm) attached to a spring steel plate and held by an adjustable metal clamp at one end. During the experiment, the fingertip of the index finger rested on the cuboid at the open end. The response force of the index finger was measured by strain gauges at the fixed end. The analogous signal was digitised at a sampling rate of 500 Hz. Each apparatus was fixed to a board to the left and to the right of the computer screen. The forearms rested on these boards and the key position was adjusted to the length of the forearms. In order to maintain the participant's posture during force production, an adjustable chin rest was provided (fixed 50 cm from the screen). The RT was defined as the temporal interval from stimulus onset to the first point in time exceeding a response force of 50 cN. The posterror behaviour was assessed by two indicators of post-error adaptation (e.g., Schroder and Infantolino, 2013). Post-error slowing was defined as the difference between the mean RT in trials following an error and the mean RT in trials preceding an error (i.e., MRTE+1 minus MRTE−1). This estimate was suggested by Dutilh et al. (2012) as a more robust measure than the traditional ones because it accounts for confounding effects based on the differences in the distributions of errors across time in an experimental session. To test for difference between post-error RT and post-correct RT, we additionally assessed the measures of post-error slowing separately (post-correct RT: mean RT in trials following a correct response, and post-error RT: mean RT in trials following an erroneous response). Post-error accuracy was defined as the percentage of correct performances in trials following an error (i.e., 0% implies no correct response after error commission; 100% implies that all responses were correct after error commission). Analogously, postcorrect accuracy was defined as the percentage of correct performances in trials following a correct response (i.e., 0% implies no correct response after correct performance; 100% implies that all responses were correct after correct performance). 5.3.2. Electrophysiological data The EEG was recorded from 61 scalp electrode sites (Fp1, Fp2, FPz, F7, F3, Fz, F4, F8, FC5, FC1, FC2, FCz, FC6, T7, C3, C3′, Cz, C4, C4′, T8, CP5, CP1, CP2, CP6, AF7, AF3, AF4, AF8, F5, F1, F2, F6, FT7, FC3, FC4, FT8, C5, IZ, C6, TP7, CP3, CPz, CP4, TP8, P5, P1, P2, P6, P7, P3, Pz, P4, P8, PO7, PO3, POz, PO4, PO8, O1, Oz, O2). The active Ag/AgCl electrodes (actiCAP, Brain Products) were referenced against the left mastoid. Vertical and horizontal electrooculograms (EOGs) were recorded from electrode positions infra-orbital to the right eye and 2 cm lateral from the outer canthi. Electrodes were re-referenced off-line against algebraically linked mastoid. The EEG was continuously recorded at a sampling rate of 500 Hz using a BrainAmp DC (Brain Products) amplifier. A band-pass filter (DC — 70 Hz) was employed for all channels.

156

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

The electrophysiological data was averaged by using the response onset. The EEG was analysed off-line with epochs ranging from −100 ms before, until 400 ms after response onset. A baseline period of 100 ms preceding the response onset was used before averaging. Influences of eye blinks were eliminated by applying an ocular correction algorithm (Gratton et al., 1983). All data were screened for artefacts and contaminated trials exceeding maximum/minimum amplitudes of ± 150 μV were rejected. There was no difference in the number of artefacts between conditions and the artefacts did not correlate with one of the assessed perfectionism scores. A current source density (CSD) analysis of the ERP waveforms was performed. The CSD signals were computed for each electrode site by taking the second derivative of the scalp voltage distribution. The CSD analysis accounts for the curvature of the head using a spline algorithm (Pernier et al., 1988; Perrin et al., 1987, 1989). The advantage of CSD analyses is that the activity overlap of adjacent sources can be reduced. CSD analyses showed similar results to traditional ERP analyses but we used CSD analyses because the activity can be more precisely assigned to the neural source. For the sake of brevity, we report only the CSD-based results. Unfortunately, the different feedback types could not be analysed separately as the number of error trials was not sufficient to compute reliable ERPs (see below) for the two experimental parts. However, preliminary analyses showed no difference between the feedback conditions (words vs. symbols) in RT, error rates, the amplitude of the ERP components (for the sake of brevity the statistics are not reported in detail). Therefore, we averaged all data across the eight blocks. Due to the different perceptual and symbolical characteristics of feedback types we did not average the ERPs time-locked to the feedback, thus, we could not examine the feedback negativity. Consequently, for the two Response Types (correct, hand errors) we determined CRN and Ne/ERN amplitudes as well as Pc and Pe amplitudes as dependent electrophysiological measures from the individual mean CSD–ERP waveforms. Each individual averaged ERP waveform contained 10 or more trials per condition, as more than six trials are required to regard the Ne/ERN and the Pe as reliable measures (see Pontifex et al., 2010; see also Baldwin et al., 2015). Due to the low number of slow responses (2.04 ± .32%, RT errors), we investigated only ERP results for hand errors not RT errors (Stahl, 2010). The Ne/ERN was determined at electrode site FCz where it was most prominent (see the Results section). Based on visual inspection of the waveforms, the CRN amplitude and the Ne/ERN amplitude were defined as the mean amplitude of the ±10 data points around the local peak for the CSD–ERP waveforms within a time window ranging from 0 to 150 ms after the onset of correct responses and error responses, respectively. The search interval for the PC amplitude and Pe amplitude at electrode site Cz, where it was most prominent, ranged from 150 to 300 ms after the onset of correct responses and error responses, respectively. 5.4. Statistical analyses Several general linear model analyses (GLMs) with Response Type (correct vs. hand errors) as a within-subject factor were performed using centralised personal standard scores (PCP) and concern over mistake scores (ECP) as continuous predictors. Further GLMs were used as post-hoc tests for analyses of interactions with the continuous variables, and Scheffé tests for within-subject effects. In cases of significant interaction of the continuous predictors, we performed simple slope analyses (Bauer and Curran, 2005; Cohen et al., 2003) and presented the confidence intervals for the estimated slopes (Preacher et al., 2006). All reported correlation coefficients are Pearson's correlations. 6. Results 6.1. Psychometric data The sample's means ± SD of the FMPS subscales were: concern over mistake scores (ECP) 2.30 ± 0.80 (range: 1.0 to 4.3), personal standard

scores (PSP) 3.44 ± 0.71 (ranging from 1.7 to 4.9), organisation 3.67 ± 1.66 (ranging from 1.16 to 5.0), doubt about action 2.55 ± .88 (ranging from 1.0 to 4.5), parental control 1.90 ± 0.95 (ranging from 1.0 to 4.3), and parental expectation 2.33 ± 1.07 (ranging from 1.0 to 4.8). 6.2. Behavioural data The percentage of errors (mean ± standard error of means) was 8.04 ± .62%. A GLM including PSP and ECP as continuous predictors did not show an effect of the perfectionism scores on the error rate (Fs b 1.0; ps N .10). A further GLM revealed that Response Type had a significant effect on RT, F(1, 90) = 26.31, p b .001, η2 = .23. Hand errors were faster (416 ± 6.7 ms) than correct responses (443 ± 5.1 ms). No perfectionism related effect on RT was shown (all ps N .10). The first GLM for post-response slowing (post-error RT, post-correct RT) revealed a significant effect of Response Type, F(1, 90) = 65.18, p b .001, η2 = .42. On average, responses were slower after errors (473 ± 8.13 ms) compared to responses after correct responses (426 ± 5.80 ms). The two continuous predictors did not show an effect of perfectionism scores (Fs b 1.0; ps N .10), which was further supported by the more robust difference measure of post-error slowing (see the Method section; Dutilh et al., 2012) in a second GLM (Fs b 1.0; ps N .10). The GLM for post-response accuracy (post-error accuracy, postcorrect accuracy) showed that there were less correct responses following errors (80.1 ± 1.6%) than correct responses following correct performance (91.5 ± 0.6%), F(1, 90) = 50.7, p b .001, η2 = .36. Although, no relationship was found for post-correct accuracy (all ps N .10), there was a significant effect of PSP on post-error accuracy, F(1, 90) = 17.06, p b .001, η2 = .16, as well as a significant effect of ECP, F(1, 90) = 10.89, p b .002, η2 = .11. The positive beta coefficient for PSP (β = .51; p = .001) indicates a higher post-error accuracy for higher PSP scores, whereas the negative beta coefficient for ECP (β = − .42; p b .002) implies a higher post-error accuracy for lower ECP scores. The interaction was not significant, F(1, 90) b 0.01. 6.3. Event-related potentials The GLM for mean CRN and Ne/ERN amplitudes revealed a significant effect of Response Type, F(1, 90) = 73.31, p b .001, η2 = .45. Mean Ne/ERN amplitudes were higher (−0.483 ± 0.028 μV/cm2) compared to CRN (−0.215 ± 0.016 μV/cm2). Furthermore, there was a significant three-way interaction between Response Type, PSP, and ECP, F(1, 90) = 4.24, p b .05, η2 = .05. Three post-hoc regression analyses were performed to explain the interaction separately for CRN amplitude, for Ne/ERN amplitudes and for Ne/ERND amplitude; the difference between error and correct trials. The analyses revealed neither differences in the slopes for correct trials (R2 = .051, F(3, 90) = 1.63, p = .19; t(93) = 0.99, p N .10), or for error trials (R2 = .076, F(3, 90) = 2.47, p = .07; t(93) = −1.82, p = .07). However, the analysis of the Ne/ERND obtained a significant difference in the slopes between participants with high ECP and low ECP depending on PSP (R2 = .081, F(3, 90) = 2.74, p b .05, t(93) = −6.60, p b .001). We illustrate this moderating effect of ECP on PSP in terms of simple slope analyses (Fig. 1A–E). The lines in the left panels (A–B) represent the estimated slopes for participants with low ECP (i.e., one standard deviation below mean), medium ECP (i.e., mean scores) and high ECP (i.e., one standard deviation above mean). Fig. 1A depicts the estimated slopes for correct trials (thin lines) and error trials (bold lines). The corresponding confidence intervals for the simple slopes demonstrate (Fig. 1C–D) that the slopes did not significantly vary for any ECP score. Like the post-hoc regression analyses, these findings indicate that there was no significant relationship between PSP, ECP, and the CRN and Ne/ERN amplitudes, considering correct trials and error trials separately. However, Fig. 1B and E shows clearly that the slopes of the regression lines vary with the degree of ECP for the Ne/ERND. For low

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

157

Fig. 1. Simple slope analyses. Predicted CRN and Ne/ERN amplitude based on the moderated regression analysis for the three-way interaction of the within-subject factor Response Type and the between-subject factors PSP and ECP (the predicted valued based for the mean, and plus/minus one standard deviation of PSP and ECP). (A) Correct trials and error trials, (B) difference between errors and correct responses, (D) the simple slopes for ECP as a function of PSP and confidence intervals for correct trials, (E) the simple slopes for ECP as a function of PSP and confidence intervals for error trials, and (F) the simple slopes for ECP as a function of PSP and confidence intervals for correct trials. The region of significance is defined by a confidence range not including zero. The range of the observed centralised ECP scores and centralised PSP scores in our sample is indicated by the arrow heads.

ECP participants, Ne/ERND did not vary as a function of PSP, whereas for high ECP participants (and slightly for mean ECP) the Ne/ERND is higher with higher PSP scores. The confidence interval shows that ECP significantly moderates the relationship between PSP and Ne/ERND with scores above the sample's mean. Fig. 2 shows difference waveforms (error minus correct) separately for the four perfectionism groups (here just for visualisation, divided by median-split). Unfortunately, the presentation of ERP waveforms requires group-based averaging, thus, some of the above mentioned effects cannot be seen as clearly as in the more appropriate presentation of the simple slope analyses. No further significant effects were obtained (all ps N .1). For Pe amplitude, the GLM analysis showed a significant effect of Response Type, F(1, 90) = 12.01, p b .001, η2 = .12, with higher Pe amplitudes after incorrect responses (0.136 ± 0.018 μV/cm2) compared to PC after correct trials (0.068 ± 0.014 μV/cm2). There was also a significant interaction between Response Type and PSP, F(1, 90) = 6.48, p b .05, η2 = .07. The post-hoc regression analyses yielded neither a

significant relationship between PSP and Pe (β = .22; p = .08) nor a significant relationship between PSP and PC (β = −.02; p N .10), but a positive relationship between the PeD amplitude difference (error minus correct) and PSP (β = .24; p b .05), indicating higher error-specific positivity with higher PSP scores (see Fig. 2). No further significant effects were obtained (all ps N .1). 6.4. Correlational analyses Table 1 shows the correlation coefficients between the perfectionism scales (six FMPS subscales, overall scale), the experimental data (RT, error rate, post-error slowing, post-error accuracy, CRN, Ne/ERN and Pc, and Pe) and age. 6.4.1. Experimental data Our analysis revealed significant correlations between the behavioural measures: Error RT correlated positively with correct response RT (r = .772) and negatively with error rate (r = − .247), post-error

158

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

Fig. 2. (A) The CSD–ERP difference waveforms time-locked to the response (0 ms) as a function of PSP and ECP at electrode site FCz for Ne/ERN. (B) The CSD–ERP-difference waveforms time-locked to the response (0 ms) as a function of PSP at electrode site Cz for error positivity (Pe). Note that the group-assignment based on median-split which explains the slight discrepancies with Fig. 1.

accuracy correlated negatively with error rate (r = −.243). There were also significant correlations between the behavioural measures and the ERP components: CRN and RT in correct trials correlated negatively (r = − .212), Ne/ERN in error trials correlated positively with error rate (r = .304) and negatively with post-error accuracy (r = −.268). Finally, the CRN and Ne/ERN, as well as PC and Pe, correlated positively (r = .563 and r = .336, respectively).

6.4.2. Perfectionism scales The correlation coefficients between the different perfectionism scores ranged between − .116 and .726 (see Table 1). The correlation between the scales of interest, PSP and ECP, was r = .513. Age correlated significantly with PSP (r = .206) and CRN (r = .216). There were positive relationships between PSP and post-error accuracy (r = .251) as well as between parental criticism and post-error slowing (r = .241); no further significant correlations between perfectionism scores and behavioural data were revealed (all ps N .1). We found a significant correlation between Ne/ERN amplitude in error trials and personal standards (r = −.252; no significant change after controlling for age or post-error accuracy: r = −.234 and r = −.198, respectively), that is, a more negative Ne/ERN with higher PSP scores. Personal standards also correlated significantly with Pe amplitude in error trials (r = .203; no significant

change after controlling for age or post-error accuracy: r = .223 and r = .100, respectively). 7. Discussion Although people usually try to avoid errors, we observe individual differences in dealing with errors in everyday life. These differences depend on the individual's capabilities as well as a person's attitude towards flawless performance. In the current study, we investigated variations in error processing depending on two core facets of perfectionism (PSP and ECP, Gaudreau and Thompson, 2010) using a flanker task. Interestingly, although perfectionism reflects a general tendency towards error avoidance (e.g., Brand and Altstötter-Gleich, 2008; Shafran et al., 2002) we did not find a perfectionism-related difference in the general error rate, response speed or post-error slowing. Participants with higher PSP scores, however, performed a better after error commission compared to low PSP participants. Furthermore, ECP was correlated with post-error accuracy, yet negatively (the higher the ECP score, the worse post-error accuracy). Thus, although the two sub-traits correlated positively, they had a reverse impact on posterror behaviour. Interestingly, we also found several perfectionismrelated variations in error-processing activity. These variations cannot be explained by a simple bias on the basis of different number of error

Pe

.009 .336⁎⁎ .104 −.049 −.121 .119 .199 −.185

.563⁎⁎⁎ .180 .082 .216⁎

.011 .200 .115

7.1. Perfectionism and early error processing

.003 −.114 −.150

−.035 −.095 .005 .118

−.247⁎ .157 .045 −.212⁎

a

−.031 .203⁎ −.206⁎

.034 −.042 .026 −.015 .241⁎ −.089 .052 −.091 .052 .118

Percentage of correct performances in trials following an error. b Difference between the mean RT in trials following an error and the mean RT in trials preceding an error (Dutilh et al., 2012). ⁎ p b .05. ⁎⁎ p b .01. ⁎⁎⁎ p b .001.

.772⁎⁎⁎ −.161 −.031 −.036 −.108 .058 −.071 −.030 .158 −.001 −.065 .037 .004 .151 −.161 −.111 −.096 .056 −.075 .634⁎⁎⁎ .125 .029 .169 −.125 .096 −.057 .024 −.094 −.007 −.142 .470⁎⁎⁎ .768⁎⁎⁎ −.081 −.119 −.105 −.098 −.019 −.167 −.201 −.005 −.053 −.110 .613⁎⁎⁎ .208⁎ .548⁎⁎⁎ −.039 −.011 −.036 .251⁎ −.064 −.162 −.252⁎ .099 .402⁎⁎⁎ .350⁎⁎ .762⁎⁎⁎

−.116 .342⁎⁎ .146 .100 .072 .082 .101 −.168 .090 −.015 −.199 −.086 −.139 .072 −.139 −.106 .638⁎⁎⁎ .190 .302⁎⁎ .194 .726⁎⁎⁎

Parental expectation (PExp) Organisation (Org) Parental criticism (PC) Personal standards (PS) Concern over mistakes (CM) Doubt about action (DAA) Overall perfectionism (OAP) Response Timecorrect (RTc) Response Timeerror (RTe) Error rate (ER%) Post-error accuracy (PEA)a Post-error slowing (PES)b Correct response negativity (CRN) Error negativity (Ne/ERN) Correct response positivity (CRP) Error positivity (Pe) Age

159

trials contributing to the ERPs but rather, by variations in error processing and post-error behaviour. This topic will be outlined below.

.167 −.177 −.268⁎ −.047 .075 .068 −.243⁎ .022 .106 .304⁎⁎

PEA ER% RTc OAP DAA CM PS PC Org

Table 1 The correlation coefficients of the assessed perfectionism subtraits, behavioural data, electrophysiological variables and age.

−.050 −.075 .045 .035 .193 −.103 −.067 −.093 .080 .017

ERP components

CRN PExp

PES Behavioural data

RTe

Perfectionism scales

Ne/ERN

CRP

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

Concerning early error processing, we expected larger Ne/ERN amplitudes for high PSP participants compared to low PSP participants (see Schrijvers et al., 2010); the negative correlation between Ne/ERN in error trials and PSP confirmed our assumption. However, the GLM revealed an interesting moderating effect of ECP. With low ECP, PSP was not related to the error-specific activity which is defined as Ne/ERN amplitude difference (i.e., error minus correct). A negative relationship between PSP and the error-specific activity was revealed only for participants with ECP scores above the mean. As a result, pure-ECP participants (high ECP, low PSP) showed the smallest error-specific activity, whereas mixed-perfectionists (high ECP, high PSP) showed the highest error-specific activity. This finding has several implications for different theoretical perspectives. First, for perfectionism research, the finding that pure-ECP participants (high ECP, low PSP) and nonperfectionists (low ECP, low PSP) clearly differ with regard to basic error-related processes validates the central idea of the 2 × 2 model by Gaudreau and Thompson (2010). The authors considered pure-ECP participants and non-perfectionists separately, though some perfectionism models (i.e., 2 + 1 models) amalgamate these two subtypes into one: the non-perfectionists (e.g., Stoeber and Otto, 2006). Previous studies investigating Ne/ERN with regard to perfectionism (Pieters et al., 2007; Schrijvers et al., 2010; Tops et al., 2013) found no relationship between Ne/ERN and ECP. Moser et al. (2013) demonstrated that worry is the only aspect of anxiety related to Ne/ERN, and argue that perfectionistic concerns reflect several aspects of anxiety but not the specific worry-related ones. This argument seems quite astonishing as ECP reflects worries about being poorly evaluated by others because of erroneous performance (Frost et al., 1990). Our results might provide some explanations for some discrepancies. Although, the present study also did not show an ECP main effect, we found an interaction of ECP and PSP. In combination with high intrinsic motivation to perform error-free (i.e., high PSP), the high ECP participants showed high error-specific activity, whereas with low intrinsic motivation (i.e., low PSP) this activity is the smallest. As previous studies (Pieters et al., 2007; Schrijvers et al., 2010; Tops et al., 2013) did not investigate the interaction of these sub-traits, this could explain the diverging findings. In line with many studies (for a review, see Hajcak, 2012; Moser et al., 2013), it is not surprising that the highest error-specific activity was shown for mixed-perfectionists as these participants are characterized by worry-moderated motivation to show excellent performance. The missing variation with PSP for participants with low ECP, thus low worries, also links in well with the above mentioned assumption of Moser et al. (2013). The least negative error-specific activity for pure-ECP participants, however, is a challenging finding from several perspectives. Based on the trait characteristics of ECP, motivation- and attention-related effects on error processing may account for these results. In pure-ECP participants, the motivation to perform faultlessly is not to fulfil their own standards but to avoid negative evaluations from others. This motivational tendency could result in a reduction of error salience and a lack of error-specific attention and might explain the smaller error-specific activity (Maier et al., 2011, 2012). The two studies of Maier and colleagues demonstrated that, when participants increased their posterror attention, the performance was improved in trials following error commissions. This post-error attention shift was related to the Ne/ERN amplitude, such as the higher Ne/ERN amplitude preceded post-error trials in which participants showed increased selective attention. Transferring these general results to our findings, pure-ECP participants might have avoided intensive error processing because errors and the anticipated threat of negative evaluations are aversive for these individuals. Thus, they may generally reduce selective attention

160

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

to response processing and post-error adaptation and this could serve as a kind of avoidance strategy. The idea is in line with observations that high ECP participants tended towards experiential avoidance and avoidance coping strategies in performance situations (e.g., in sports; Santanello, and Gardner, 2007; Weiner and Carton, 2012). Such a coping by avoidance corresponds to Moser et al.'s (2013) postulate that worrying requires task unrelated cognitive resources and that an attention shift to the task-specific goal obligations increases the Ne/ERN. However, pure-ECP participants describe themselves as having low performance standards and low goal orientation (e.g., Frost et al., 1990) and therefore reallocation of attention might not be applied by them. Hence, despite major worries, they show lower error-specific activity. The findings of worsening post-error accuracy with higher ECP and low PSP scores further supported the notion that pure-ECP perfectionist did increase their attention to the task requirements after error commission. Rather than the attention hypothesis, low performance expectations could explain the small Ne/ERN amplitudes in pure-ECP participants. Particularly, as violations of expectations can increase the neural activity of the action monitoring system (Holroyd and Coles, 2002). A reduced self-efficacy expectation was observed for high ECP participants (e.g., Beauregard, 2012; Hadjistavropoulos et al., 2007), which might serve as some kind of a coping strategy to minimise the concerns of not fulfilling a task perfectly. This means that the individual's low expectation was not violated in cases of error commission because of selfrelated cognitions such as, 'once a loser, always a loser', and, therefore, no additional activity was produced. An important aspect regarding the discussion above is that we observed that the ECP-by-PSP interaction only for error-specific activity which was reflected by the difference between Ne/ERN in correct and in error trials but not for the Ne/ERN in error trials. Amplitude differences, however, are always more difficult to interpret. Although the different slopes for the single response type conditions did not reach the level of significance, Fig. 1A indicates that the main contribution of high error-specific activity in mixed perfectionists resulted from the higher Ne/ERN in error trials; the low error-specific activity in pureECP perfectionists resulted from the low Ne/ERN in error trials. Fig. 1A nicely illustrates that with lower ECP scores the difference between error trials and correct trials for low PSP increased with increased Ne/ERN and decreased CRN. 7.2. Perfectionism, late error processing, evidence accumulation, and behavioural adaptation In line with our hypothesis, we found a positive relationship between PSP scores and Pe amplitude in error trials. Furthermore, the PSP-by-Response Type interaction demonstrated a positive relationship between the Pe difference (error minus correct: error specific positivity) and PSP scores. Previous work suggested the Pe as the neural correlate of error evidence accumulation (Steinhauser and Yeung, 2010, 2012) and behavioural adaptation (e.g., Overbeek et al., 2005). According to Steinhauser and Yeung's (2012) model, the more error evidence is accumulated in a short period of time, the higher the Pe amplitude is. Following this notion, in order to optimise their behaviour more efficiently in post-error trials (e.g., by being more attentive), one can assume that participants with high PSP accumulate more error evidence than participants with low PSP. This hypothesis fits very well with our findings of a higher post-error accuracy for high PSP participants and the general behavioural tendency of participants with high PSP who describe themselves as focused and eager to reach their goals (Stoeber, 2011). Our ECP-related results challenge the findings of Schrijvers et al. (2010) who showed positive relationships between ECP and Pe. To start with, one obvious explanation is that there are some general differences between the design of our study and that of Schrijvers and colleagues. As mentioned above, the authors recruited medicated participants diagnosed with major depression, which may have affected the neurophysiological processes differently. The authors reported the

personality scores to be relatively high compared to healthy samples. Unfortunately, as Schrijvers and colleagues did not report the mean ECP value and SD of their sample, we were unable to compare our data with that from theirs. Presumably, as our healthy sample was more than twice as large as their sample, our data should be less sensitive to outliers. Tops et al. (2013) also reported a relationship between ECP and late Pe for a healthy sample but, as they investigated only 16 participants and the may have had overlapping processing activity of the trial-by-trial feedback (for details, see Introduction), their findings have to be interpreted with caution. Moreover, from a more theoretical perspective, the above postulated lack of attention in terms of avoiding intensive early error processing in pure-ECP participants, might also be responsible for the missing relationship between ECP and Pe. Reallocated attention in the stimulus – response interval as well as in the post-response interval because of worrying (Moser et al., 2013) in pure-ECP participants should have detrimental effects on error-evidence accumulation reflected by Pe (Steinhauser and Yeung, 2010). Thus, if early error processing did not provide sufficient error evidence, the late error processing system might not obtain enough evidence to efficiently affect the Pe and challenge the chance of post-error adaptation. This view is supported by the finding that high ECP participants demonstrated worse post-error performance than low ECP participants. From this line of reasoning, however, it still remains unclear as to why we could not find smaller Pe amplitudes for pure-ECP participants. Experimentally varying attention shift will help to learn more about the underlying mechanisms in future research.

8. Conclusions and implications As outlined above, we found different effects of two perfectionism sub-traits on two ERP components reflecting unaware and aware error processing (Nieuwenhuis et al., 2001). These findings are also helpful for the debate as to which of the constellations of perfectionism is the most maladaptive in terms of unhealthy behaviour. In early research on perfectionism, many authors postulated that mixedperfectionism is the unhealthiest type (for a review see Stoeber and Otto, 2006), whereas Gaudreau and Thompson (2010) argued that high PSP might serve a protective function. This suggestion was supported by several studies (Douilliez and Lefèvre, 2011; Gaudreau, 2012; Stoeber and Yang, 2010). Based on the present results, high PSP might be protective by preventing unhealthy avoidance strategies. Instead, participants with high PSP aimed at optimising behaviour that offers the opportunity to get more positive outcome in the future. Based on our findings, further questions have to be considered: Are participants with high PSP actually more successful in error detection than participants with low PSP? If so, does that explain their improved post-error performance? And can attention-related training for pureECP participants affect their error processing activity? To end with, the investigation of trait interactions, on an emotional as well as a motivational basis, can help to increase the understanding of variations of the errorrelated ERP components. Likewise, perfectionism research can benefit from the ERP approach as a means of learning more about variations in executive functions which might help to develop individual training methods to reduce unhealthy and maladaptive behavioural strategies.

Acknowledgment None of the authors have potential conflicts of interest to be disclosed. We are grateful to Bernd Kuderer for help with data acquisition and to Katharina Banscherus for helpful comments and suggestions to the manuscript. This research was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) STA 1035/3-1.

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

References Aarts, K., Pourtois, G., 2010. Anxiety not only increases, but also alters early errormonitoring function. Cogn. Affect. Behav. Neurosci. 10, 479–492. http://dx.doi.org/ 10.3758/CABN.10.4.479. Altstötter-Gleich, C., Bergemann, N., 2006. Testgüte einer deutschsprachigen Version der Mehrdimensionalen Perfektionismus Skala von Frost, Marten, Lahart und Rosenblate (MPS-F). Diagnostica 52 (3), 105–118. http://dx.doi.org/10.1026/0012-1924.52.3. 105. Armbrecht, A.S., Wöhrmann, A., Gibbons, H., Stahl, J., 2010. Temporal dynamics of conflict monitoring and the effects of one or two conflict sources on error-(related) negativity. Biol. Psychol. 85 (1), 112–123. http://dx.doi.org/10.1016/j.biopsycho.2010.06.002. Armbrecht, A.S., Gibbons, H., Stahl, J., 2012. Monitoring force errors: medial–frontal negativity in a unimanual force‐production task. Psychophysiology 49 (1), 56–72. http:// dx.doi.org/10.1111/j.1469-8986.2011.01282.x. Armbrecht, A.S., Gibbons, H., Stahl, J., 2013. Effects of response force parameters on medial–frontal negativity. PLoS ONE 8 (1), e54681. http://dx.doi.org/10.1371/ journal.pone.0054681. Baldwin, S.A., Larson, M.J., Clayson, P.E., 2015. The dependability of electrophysiological measurements of performance monitoring in a clinical sample: a generalizability and decision analysis of the ERN and Pe. Psychophysiology 52 (6), 790–800. http:// dx.doi.org/10.1111/psyp.12401. Bauer, D.J., Curran, P.J., 2005. Probing interactions in fixed and multilevel regression: inferential and graphical techniques. Multivar. Behav. Res. 40, 373–400. http://dx. doi.org/10.1207/s15327906mbr4003_5. Beauregard, T.A., 2012. Perfectionism, self-efficacy and OCB: the moderating role of gender. Pers. Rev. 41 (5), 590–608. http://dx.doi.org/10.1108/00483481211249120. Boksem, M.A.S., Tops, M., Wester, A.E., Meijman, T.F., Lorist, M.M., 2006. Error-related ERP components and individual differences in punishment and reward sensitivity. Brain Res. 1101 (1), 92–101. http://dx.doi.org/10.1016/j.brainres.2006.05.004. Brand, M., Altstötter-Gleich, C., 2008. Personality and decision-making in laboratory gambling tasks: evidence for a relationship between deciding advantageously under risk conditions and perfectionism. Personal. Individ. Differ. 45 (3), 226–231. http://dx.doi.org/10.1016/j.paid.2008.04.003. Cohen, J., Cohen, P., West, S.G., Aiken, L.S., 2003. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. 3rd ed. Erlbaum, Hillsdale. Danielmeier, C., Ullsperger, M., 2011. Post-error adjustments. Front. Psychol. 2, 233. http://dx.doi.org/10.3389/fpsyg.2011.00233. De Bruijn, E.R., Hulstijn, W., Meulenbroek, R.G., Van Galen, G.P., 2003. Action monitoring in motor control: ERPs following selection and execution errors in a force production task. Psychophysiology 40 (5), 786–795. http://dx.doi.org/10.1111/ 1469-8986.00079. Douilliez, C., Lefèvre, F., 2011. Interactive effect of perfectionism dimensions on depressive symptoms: a reply to. Personal. Individ. Differ. 50 (7), 1147–1151. Dutilh, G., van Ravenzwaaij, D., Nieuwenhuis, S., van der Maas, H.J., Forstmann, B.U., Wagenmakers, E., 2012. How to measure post-error slowing: a confound and a simple solution. J. Math. Psychol. 56 (3), 208–216. http://dx.doi.org/10.1016/j. jmp.2012.04.001. Elson, M., Bente, G., 2009. CORTEX — COmputer-aided Registration Tool for EXperiments. University of Cologne, Cologne, Germany (Retrieved from: http:// cortex.uni-koeln.de/). Endrass, T., Klawohn, J., Gruetzmann, R., Ischebeck, M., Kathmann, N., 2012. Response‐ related negativities following correct and incorrect responses: evidence from a temporospatial principal component analysis. Psychophysiology 49 (6), 733–743. http://dx.doi.org/10.1111/j.1469-8986.2012.01365.x. Eriksen, B.A., Eriksen, C.W., 1974. Effects of noise letters upon the identification of a target letter in a nonsearch task. Percept. Psychophys. 16 (1), 143–149. http://dx.doi.org/10. 3758/BF03203267. Falkenstein, M., Hohnsbein, J., Hoormann, J., Blanke, L., 1991. Effects of crossmodal divided attention on late ERP components. II. Error processing in choice reaction tasks. Electroencephalogr. Clin. Neurophysiol. 78 (6), 447–455. http://dx.doi.org/10.1016/ 0013-4694(91)90062-9. Frost, R.O., Marten, P., Lahart, C., Rosenblate, R., 1990. The dimensions of perfectionism. Cogn. Ther. Res. 14 (5), 449–468. http://dx.doi.org/10.1007/BF01172967. Gaudreau, P., 2012. A methodological note on the interactive and main effects of dualistic personality dimensions: an example using the 2 × 2 model of perfectionism. Personal. Individ. Differ. 52 (1), 26–31. http://dx.doi.org/10.1016/j.paid.2011.08.022. Gaudreau, P., 2013. The 2 × 2 model of perfectionism: commenting the critical comments and suggestions of Stoeber (2012). Personal. Individ. Differ. 55 (4), 351–355. http:// dx.doi.org/10.1016/j.paid.2013.03.021. Gaudreau, P., Thompson, A., 2010. Testing a 2 × 2 model of dispositional perfectionism. Personal. Individ. Differ. 48 (5), 532–537. http://dx.doi.org/10.1016/j.paid.2009.11.031. Gaudreau, P., Thompson, A., 2011. Erratum to ‘testing a 2 × 2 model of dispositional perfectionism’ [personality and individual differences 48 (2010) 532–537]. Personal. Individ. Differ. 50 (1), 126. http://dx.doi.org/10.1016/j.paid.2010.09.001. Gehring, W.J., Goss, B., Coles, M.G., Meyer, D.E., Donchin, E., 1993. A neural system for error detection and compensation. Psychol. Sci. 4 (6), 385–390. http://dx.doi.org/ 10.1111/j.1467-9280.1993.tb00586.x. Gehring, W.J., Himle, J., Nisenson, L.G., 2000. Action-monitoring dysfunction in obsessive– compulsive disorder. Psychol. Sci. 11 (1), 1–6. http://dx.doi.org/10.1111/1467-9280. 00206. Gibbons, H., Fritzsche, A.S., Bienert, S., Armbrecht, A.S., Stahl, J., 2011. Percept-based and object-based error processing: an experimental dissociation of error-related negativity and error positivity. Clin. Neurophysiol. 122 (2). http://dx.doi.org/10.1016/j. clinph.2010.06.031 (299–31).

161

Gratton, G., Coles, M.G., Donchin, E., 1983. A new method for off-line removal of ocular artifact. Electroencephalogr. Clin. Neurophysiol. 55 (4), 468–484. http://dx.doi.org/ 10.1016/0013-4694(83)90135-9. Gray, J.A., 1970. The psychophysiological basis of introversion–extraversion. Behav. Res. Ther. 8 (3), 249–266. Gray, J.A., McNaughton, N., 2000. The Neuropsychology of Anxiety: An Enquiry Into the Functions of the Septo-hippocampal System. Oxford University Press, Oxford. Hadjistavropoulos, H., Dash, H., Hadjistavropoulos, T., Sullivan, T., 2007. Recurrent pain among university students: contributions of self-efficacy and perfectionism to the pain experience. Personal. Individ. Differ. 42 (6), 1081–1091. http://dx.doi.org/10. 1016/j.paid.2006.09.013. Hajcak, G., 2012. What we've learned from mistakes: Insights from error-related brain activity. Curr. Dir. Psychol. Sci. 21 (2), 101–106. http://dx.doi.org/10.1177/ 0963721412436809. Hamachek, D.E., 1978. Psychodynamics of normal and neurotic perfectionism. Psychology 15 (1), 27–33. Herrmann, M.J., Römmler, J., Ehlis, A.C., Heidrich, A., Fallgatter, A.J., 2004. Source localization (LORETA) of the error-related-negativity (ERN/Ne) and positivity (Pe). Cogn. Brain Res. 20 (2), 294–299. http://dx.doi.org/10.1016/j.cogbrainres.2004.02.013. Hewitt, P.L., Flett, G.L., 1991. Perfectionism in the self and social contexts: conceptualization, assessment, and association with psychopathology. J. Pers. Soc. Psychol. 60 (3), 456. http://dx.doi.org/10.1037/0022-3514.60.3.456. Hoffmann, S., Falkenstein, M., 2010. Independent component analysis of erroneous and correct responses suggests online response control. Hum. Brain Mapp. 31 (9), 1305–1315. http://dx.doi.org/10.1002/hbm.20937. Holroyd, C.B., Coles, M.G., 2002. The neural basis of human error processing: reinforcement learning, dopamine, and the error-related negativity. Psychol. Rev. 109 (4), 679–709. http://dx.doi.org/10.1037/0033-295X.109.4.679. Maier, M.E., Yeung, N., Steinhauser, M., 2011. Error-related brain activity and adjustments of selective attention following errors. NeuroImage 56 (4), 2339–2347. http://dx.doi. org/10.1016/j.neuroimage.2011.03.083. Maier, M.E., Di Pellegrino, G., Steinhauser, M., 2012. Enhanced error‐related negativity on flanker errors: error expectancy or error significance? Psychophysiology 49 (7), 899–908. http://dx.doi.org/10.1111/j.1469-8986.2012.01373.x. Moser, J.S., Moran, T.P., Schroder, H.S., Donnellan, M.B., Yeung, N., 2013. On the relationship between anxiety and error monitoring: a meta-analysis and conceptual framework. Front. Hum. Neurosci. 7. http://dx.doi.org/10.3389/fnhum.2013.0046 (ArtID: 466). Moser, J.S., Moran, T.P., Schroder, H.S., Donnellan, M.B., Yeung, N., 2014. The case for compensatory processes in the relationship between anxiety and error monitoring: a reply to Proudfit, Inzlicht, and Mennin. Front. Hum. Neurosci. 8. http://dx.doi.org/ 10.3389/fnhum.2014.00064 (ArtID: 6). Nieuwenhuis, S., Ridderinkhof, K.R., Blom, J., Band, G.P., Kok, A., 2001. Error-related brain potentials are differentially related to awareness of response errors: evidence from an antisaccade task. Psychophysiology 38 (5), 752–760. http://dx.doi.org/10.1111/ 1469-8986.3850752. O'Connell, R.G., Dockree, P.M., Bellgrove, M.A., Kelly, S.P., Hester, R., Garavan, H., Foxe, J.J., 2007. The role of cingulate cortex in the detection of errors with and without awareness: a high-density electrical mapping study. Eur. J. Neurosci. 25 (8), 2571–2579. http://dx.doi.org/10.1111/j.1460-9568.2007.05477.x. Olvet, D.M., Hajcak, G., 2009. The effect of trial-to-trial feedback on the error-related negativity and its relationship with anxiety. Cogn. Affect. Behav. Neurosci. 9 (4), 427–433. http://dx.doi.org/10.3758/CABN.9.4.427. Orr, C., Hester, R., 2012. Error-related anterior cingulate cortex activity and the prediction of conscious error awareness. Front. Hum. Neurosci. 6. http://dx.doi.org/10.3389/ fnhum.2012.00177 (ArtID: 177). Overbeek, T.J., Nieuwenhuis, S., Ridderinkhof, K.R., 2005. Dissociable components of error processing: on the functional significance of the Pe vis-à-vis the ERN/Ne. J. Psychophysiol. 19 (4), 319–329. http://dx.doi.org/10.1027/0269-8803.19.4.319. Pernier, J., Perrin, F., Bertrand, O., 1988. Scalp current density fields: concept and properties. Electroencephalogr. Clin. Neurophysiol. 69 (4), 385–389. http://dx.doi.org/10. 1016/0013-4694(88)90009-0. Perrin, F., Bertrand, O., Pernier, J., 1987. Scalp current density mapping: value and estimation from potential data. Biomed. Eng. IEEE Trans. 4, 283–288. http://dx.doi.org/10. 1109/TBME.1987.326089. Perrin, F., Pernier, J., Bertrand, O., Echallier, J.F., 1989. Spherical splines for scalp potential and current density mapping. Electroencephalogr. Clin. Neurophysiol. 72 (2), 184–187. http://dx.doi.org/10.1016/0013-4694(89)90180-6. Pieters, G.L., de Bruijn, E.R., Maas, Y., Hulstijn, W., Vandereycken, W., Peuskens, J., Sabbe, B.G., 2007. Action monitoring and perfectionism in anorexia nervosa. Brain Cogn. 63 (1), 42–45. http://dx.doi.org/10.1016/j.bandc.2006.07.009. Pontifex, M.B., Scudder, M.R., Brown, M.L., O'Leary, K.C., Wu, C.-T., Themanson, J.R., Hillman, C.H., 2010. On the number of trials necessary for stabilization of errorrelated brain activity across the life span. Psychophysiology 47 (4), 767–773. http:// dx.doi.org/10.1111/j.1469-8986.2010.00974.x. Preacher, K.J., Curran, P.J., Bauer, D.J., 2006. Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. J. Educ. Behav. Stat. 31, 437–448. Proudfit, G.H., Inzlicht, M., Mennin, D.S., 2013. Anxiety and error monitoring: the importance of motivation and emotion. Front. Hum. Neurosci. 7. http://dx.doi.org/10.3389/ fnhum.2013.00636. Ridderinkhof, K.R., Ullsperger, M., Crone, E.A., Nieuwenhuis, S., 2004. The role of the medial frontal cortex in cognitive control. Science 306 (5695), 443–447. http://dx.doi. org/10.1126/science.1100301. Santanello, A.W., Gardner, F.L., 2007. The role of experiential avoidance in the relationship between maladaptive perfectionism and worry. Cogn. Ther. Res. 31 (3), 319–332. http://dx.doi.org/10.1007/s10608-006-9000-6.

162

J. Stahl et al. / International Journal of Psychophysiology 97 (2015) 153–162

Schrijvers, D.L., De Bruijn, E.R., Destoop, M., Hulstijn, W., Sabbe, B.G., 2010. The impact of perfectionism and anxiety traits on action monitoring in major depressive disorder. J. Neural Transm. 117 (7), 869–888. http://dx.doi.org/10.1007/s00702-010-0419-2. Schroder, H.S., Infantolino, Z.P., 2013. Distinguishing between types of errors and adjustments. J. Neurosci. 33 (47), 18356–18357. http://dx.doi.org/10.1523/JNEUROSCI. 4078-13.2013. Shafran, R., Cooper, Z., Fairburn, C.G., 2002. Clinical perfectionism: a cognitive-behavioural analysis. Behav. Res. Ther. 40 (7), 773–791. http://dx.doi.org/10.1016/S00057967(01)00059-6. Stahl, J., 2010. Error detection and the use of internal and external error indicators: an investigation of the first-indicator hypothesis. Int. J. Psychophysiol. 77 (1), 43–52. http://dx.doi.org/10.1016/j.ijpsycho.2010.04.005. Stahl, J., Gibbons, H., 2007. Dynamics of response-conflict monitoring and individual differences in response control and behavioral control: an electrophysiological investigation using a stop-signal task. Clin. Neurophysiol. 118 (3), 581–596. http://dx.doi. org/10.1016/j.clinph.2006.10.023. Steinhauser, M., Yeung, N., 2010. Decision processes in human performance monitoring. J. Neurosci. 30 (46), 15643–15653. http://dx.doi.org/10.1523/JNEUROSCI.1899-10. 2010. Steinhauser, M., Yeung, N., 2012. Error awareness as evidence accumulation: effects of speed-accuracy trade-off on error signaling. Front. Hum. Neurosci. 6, 240. http://dx. doi.org/10.3389/fnhum.2012.00240. Stoeber, J., 2011. The dual nature of perfectionism in sports: relationships with emotion, motivation, and performance. Int. Rev. Sport Exerc. Psychol. 4 (2), 128–145. http://dx. doi.org/10.1080/1750984X.2011.604789. Stoeber, J., Otto, K., 2006. Positive conceptions of perfectionism: approaches, evidence, challenges. Personal. Soc. Psychol. Rev. 10 (4), 295–319. http://dx.doi.org/10.1207/ s15327957pspr1004_2. Stoeber, J., Yang, H., 2010. Perfectionism and emotional reactions to perfect and flawed achievements: satisfaction and pride only when perfect. Personal. Individ. Differ. 49 (3), 246–251. http://dx.doi.org/10.1016/j.paid.2010.03.044.

Tops, M., Koole, S.L., Wijers, A.A., 2013. The Pe of perfectionism: concern over mistakes predicts the amplitude of a late frontal error positivity. J. Psychophysiol. 27 (2), 84–94. http://dx.doi.org/10.1027/0269-8803/a000090. Van Veen, V., Carter, C.S., 2002. The anterior cingulate as a conflict monitor: fMRI and ERP studies. Physiol. Behav. 77 (4), 477–482. http://dx.doi.org/10.1016/S00319384(02)00930-7. Vidal, F., Hasbroucq, T., Grapperon, J., Bonnet, M., 2000. Is the ‘error negativity’ specific to errors? Biol. Psychol. 51 (2), 109–128. http://dx.doi.org/10.1016/S03010511(99)00032-0. Vocat, R., Pourtois, G., Vuilleumier, P., 2008. Unavoidable errors: a spatio-temporal analysis of time-course and neural sources of evoked potentials associated with error processing in a speeded task. Neuropsychologia 46 (10), 2545–2555. http://dx.doi.org/ 10.1016/j.neuropsychologia.2008.04.006. Weinberg, A., Olvet, D.M., Hajcak, G., 2010. Increased error-related brain activity in generalized anxiety disorder. Biol. Psychol. 85, 472–480. http://dx.doi.org/10.1016/j. biopsycho.2010.09.011. Weinberg, A., Klein, D.N., Hajcak, G., 2012. Increased error-related brain activity distinguishes generalized anxiety disorder with and without comorbid major depressive disorder. J. Abnorm. Psychol. 121 (4), 885–896. http://dx.doi.org/10.1037/a0028270. Weiner, B.A., Carton, J.S., 2012. Avoidant coping: A mediator of maladaptive perfectionism and test anxiety. Personal. Individ. Differ. 52 (5), 632–636. http://dx.doi.org/10.1016/ j.paid.2011.12.009. Yeung, N., Botvinick, M.M., Cohen, J.D., 2004. The neural basis of error detection: conflict monitoring and the error-related negativity. Psychol. Rev. 111 (4), 931. http://dx.doi. org/10.1037/0033-295X.111.4.931. Zambrano-Vazquez, L., Allen, J.J.B., 2014. Differential contributions of worry, anxiety, and obsessive compulsive symptoms to ERN amplitudes in response monitoring and reinforcement learning tasks. Neuropsychologia 61, 197–209. http://dx.doi.org/10.1016/j. neuropsychologia.2014.06.023.