To PE or not to PE: A P3-like ERP component reflecting ...

Viewer
Transcript

Psychophysiology, 46 (2009), 531–538. Wiley Periodicals, Inc. Printed in the USA. Copyright r 2009 Society for Psychophysiological Research DOI: 10.1111/j.1469-8986.2009.00790.x

To PE or not to PE: A P3-like ERP component reﬂecting the processing of response errors

K. RICHARD RIDDERINKHOF,a JENNIFER R. RAMAUTAR,b and JASPER G. WIJNENa a Amsterdam Center for the Study of Adaptive Control in Brain and Behavior, Department of Psychology, University of Amsterdam, Amsterdam, the Netherlands b Department of Psychiatry, Uniformed Services University, Bethesda, Maryland, USA

Abstract ERP studies have highlighted several electrocortical components that can be observed when people make errors. We propose that the PE reﬂects processes functionally similar to those reﬂected in the P3 and that the PE and P3 should covary. We speculate that these processes refer to the motivational signiﬁcance of rare target stimuli in case of the P3 and of salient performance errors in case of the PE. Here we investigated whether PE amplitude after errors in a Simon task is correlated speciﬁcally to varying target–target intervals in a visual oddball task, a factor known to parametrically affect P3 amplitude. The amplitude of the PE, but not the NE, was observed to covary with the effect of target– target interval on P3 amplitude. The speciﬁcity of this novel ﬁnding supports the notion that the PE and P3 reﬂect similar neurocognitive processes as possibly involved in the conscious processing of motivationally signiﬁcant events. Descriptors: PE, P3, Error processing, Motivational signiﬁcance

vational signiﬁcance, suggesting that the PE reﬂects processes similar to those expressed in another ERP component, the classical P3. Before outlining this perspective and developing an initial empirical approach to verify the predictions derived from such a perspective, we brieﬂy describe the relevant ERP components and their typical interpretation.

Adaptive control of behavior requires not only the ability to regulate performance in accordance with task demands, but also the ability to monitor for signals that indicate the need for adjustment. The cognitive and neural mechanisms underlying performance monitoring have been explored rigorously in recent studies in the cognitive neurosciences. Event-related brain potential (ERP) studies have highlighted two electrocortical components that can be observed when people make errors: the error(-related) negativity (NE or ERN) and the error positivity (PE; Falkenstein, Hoorman, & Hohnsbein, 1999). The sensitivity of the NE to individual differences and experimental manipulations has been described in considerable detail, but the sensitivity of the PE to such factors has not been studied with similar scrutiny (Overbeek, Nieuwenhuis, & Ridderinkhof, 2005). Likewise, the development of theoretical and mechanistic models of the processes reﬂected in the NE (e.g., Holroyd et al., 2002; Yeung, Cohen, & Botvinick, 2004) has not been paralleled by corresponding developments regarding the functional signiﬁcance of the PE. As will be reviewed brieﬂy below, the outlines are beginning to emerge of a theoretical and empirical perspective on the functional signiﬁcance of the PE in terms of error salience or moti-

Performance Monitoring Processes Reﬂected in the NE The NE is a negative deﬂection peaking 60–100 ms following an errant response with a frontocentral scalp distribution. In animals, errors in reward prediction are coded by phasic changes in activity of the mesofrontal/mesolimbic dopamine system: a phasic increase or decrease when ongoing events are suddenly better or worse (respectively) than expected (Schultz, 2002). These phasic dopamine signals are communicated to posterior areas of the medial frontal cortex, clustering in the rostral cingulate zone (RCZ, the border zone between the medial areas BA8, BA6, and BA32 0 with some extension into BA24 0 ). As these signals arrive in the RCZ, they give rise to the NE and are used for improving task performance in accordance with basic reinforcement-learning principles (Holroyd & Coles, 2002). The RCZ may also be involved in the monitoring of response conﬂict (Botvinick, Braver, Barch, Carter, & Cohen, 2001). Thus, the RCZ appears to be engaged when the need for adjustments to achieve action goals becomes evident: Response conﬂict signals a reduced probability of obtaining reward, whereas errors and unexpected negative feedback signal the loss of anticipated reward (Ridderinkhof, Ullsperger, Crone, & Nieuwenhuis, 2004). The electrophysiological correlate of these performance-monitoring functions is the NE.

This work was supported by grants from the Netherlands Organization for Scientiﬁc Research (NWO) to the ﬁrst author. The help of Hilde Huizenga in statistical analysis and Sander Nieuwenhuis in conceptual issues is gratefully acknowledged. Address reprint requests to: K. Richard Ridderinkhof, Amsterdam Center for the Study of Adaptive Control in Brain and Behavior, Department of Psychology, University of Amsterdam, Amsterdam, the Netherlands. E-mail: [email protected] 531

532 Motivational Signiﬁcance Processes Reﬂected in the P3 Probably the most extensively studied ERP component is the P3, and the search for its neural generators has been a topic of considerable interest for many years. The P3 is a broad, positivegoing deﬂection peaking approximately 300–600 ms after the eliciting event. The P3 has a centro-parietal scalp distribution and has been related to cognitive processes that underlie the development of representations in short-term memory such as context updating, context closure, and event categorization (Donchin & Coles, 1988; Kok, 2001; Verleger, 1988). The events that give rise to a P3 can vary widely (from salient, novel, or rare stimuli to the absence of expected stimuli) but appear to have in common that they are motivationally signiﬁcant, that is, they should motivate the individual to initiate or change a course of action in order to keep performance at an optimal level. According to recent views, the P3 comprises the electrocortical expression of the response of the locus coeruleus-norepinephrine (LC-NE) system to the preliminary outcome of internal decisionmaking processes and the consequent effects of the noradrenergic potentiation of information processing (Nieuwenhuis, AstonJones, & Cohen, 2005). Awareness of Errors Reﬂected in the PE The NE is typically followed by the PE, a slow positive wave with a diffuse scalp distribution and maximum amplitude between 200 and 600 ms. The PE appears to consist of an initial component, intimately related to the NE, and a slower component more reminiscent of the P3 (Van Veen & Carter, 2002). A review of the literature revealed that the NE and slow PE are remarkably dissimilar in terms of antecedent conditions (experimental manipulations, individual differences), suggesting that the PE reﬂects aspects of error processing that are largely independent of those manifested in the NE (Overbeek et al., 2005). Rather than reﬂecting the remedial process of behavioral adaptation following errors or the emotional appraisal of the error or its consequences, the PE appears to reﬂect the conscious recognition of the fact that an error was committed (Endrass, Franke, & Kathmann, 2005; Nieuwenhuis, Ridderinkhof, Blom, Band, & Kok, 2001; for a review, see Overbeek et al., 2005). It should be noted, however, that although there is some empirical support for the errorawareness hypothesis, this hypothesis is somewhat descriptive and its predictions have been tested in only a few studies to date. Relation between the PE and the P3 Several authors have noted a resemblance between the PE and the P3 in terms of timing, morphology, scalp topography, and signiﬁcance (e.g., Davies, Segalowitz, Dywan, & Pailing, 2001; Falkenstein et al., 1999; Hajcak, McDonald, & Simons, 2003; O’Connell et al., 2007; Van Veen & Carter, 2002). Thus, the PE might constitute a P3-like response associated with the motivational signiﬁcance of an error (Overbeek et al., 2005). This view is consistent with ﬁndings that a larger PE is observed for more salient errors (Leuthold & Sommer, 1999) and with the ﬁnding that the PE is small or absent when the subject does not explicitly recognize the erroneous response (Endrass et al., 2005; Nieuwenhuis et al., 2001; O’Connell et al., 2007). Furthermore, Davies et al. reported a signiﬁcant correlation between the amplitudes of the stimulus-locked P3 during correct responses in an Eriksen ﬂanker task and the response-locked PE in the same task. In an fMRI study, Hester, Foxe, Molholm, Shpaner, and Garavan (2005) contrasted BOLD signals associated with recognized and unrecognized errors in a go/no-go task. Differential

K.R. Ridderinkhof, J.R. Ramautar, and J.G. Wijnen activation was observed in bilateral prefrontal and inferior parietal cortices that are among the main regions implicated in generating the P3 (Soltani & Knight, 2000; Stevens, Calhoun, & Kiehl, 2005), thus reinforcing the notion that the PE and the P3 may reﬂect similar neural and functional processes. Dipole source modeling of the PE scalp topography has suggested neural generators of the PE in brain areas that differ from those implicated in generating the P3: The PE has been source localized to rostral (van Boxtel, van der Molen, & Jennings, 2005; Van Veen & Carter, 2002) or caudal portions of medial frontal cortex (Herrmann, Ro¨mmler, Ehlis, Heidrich, & Fallgatter, 2004). In general, source modeling results must be interpreted with caution because the dipole source localization problem is underdetermined (the so-called inverse problem). This problem may be speciﬁcally relevant for the PE: Although the broad scalp distribution may suggest the contribution of multiple generators, it is often easy to ﬁt such a distribution artiﬁcially with a limited number of relatively deep dipoles. To examine more rigorously the neurocognitive processes underlying the PE and its similarity to the P3, we follow up on the observation that the amplitudes of the PE and P3 are correlated (Davies et al., 2001). However, rather than examining the amplitude of the P3 per se, we focus on a parametric effect on the P3 for reasons to become clear in the next section. Probing the Processes Underlying the P3 A robust ﬁnding is that P3 amplitude is inversely related to target probability in oddball tasks (requiring the detection of distinct infrequent target stimuli or oddballs that are embedded in a series of frequently presented nontarget or standard stimuli; e.g., Duncan-Johnson & Donchin, 1984). Moreover, it appears that expectations elicited by the recent stimulus history, so called global and local target probability, may inﬂuence P3 amplitude. In other words, P3 to an oddball is more enhanced when the target stimulus is embedded in a train of nontarget stimuli rather than in a train of other targets (Squires, Wickens, Squires, & Donchin, 1976). However, rather than being attributable to target probability per se, these P3 effects are crucially mediated by targetto-target interval (TTI) duration. Croft, Gonsalves, Gabriel, and Barry (2003) investigated whether P3 amplitude varied as a function of global and local target probability while TTI was held constant and vice versa. P3 amplitude turned out to be mainly inﬂuenced by TTI and was relatively unaffected by target probability (see also Gonsalves et al., 1999). Horovitz, Skudlarski, and Gore (2002) replicated the parametric effects of TTI on P3 amplitude and then administered the same oddball task to the same subjects while recording the BOLD response. Areas sensitive to the parametric effects of TTI were found in parietal (viz., supramarginal gyrus) and frontal cortex (viz., right medial frontal cortex and bilateral insula), corresponding roughly to regions implicated in generating the P3 (Soltani & Knight, 2000; Stevens et al., 2005). The Present Study The notion that the PE reﬂects the motivational signiﬁcance of a salient performance error, similar to the P3 reﬂecting the motivational signiﬁcance of a rare target stimulus, would gain considerable support if it could be demonstrated that PE amplitude is correlated speciﬁcally to the parametric effect of TTI on P3 amplitude. To be more speciﬁc, we predict that the amplitude of the PE, but not of the NE, covaries signiﬁcantly with the effect of TTI on P3 amplitude. The P3 was examined in an oddball task, using

To PE or not to PE a TTI manipulation known to parametrically affect speciﬁcally the salience processing as reﬂected in the P3. The amplitude of the P3 as observed in this oddball task was not correlated to the PE directly; instead, PE amplitude was correlated to the parametric effect of TTI on the oddball P3. The speciﬁcity of this predicted ﬁnding provides a more optimal test for the notion that the PE and P3 reﬂect similar neurocognitive processes, as it capitalizes on the processing of motivational salience and is relatively insensitive to the variance in P3 amplitude incurred by factors other than the parametric TTI effect. The present study was designed therefore to examine (in the same subjects) the sensitivity of the P3 to varying TTI in a visual oddball task and the amplitude of the PE in response to errors in a Simon task (a task that typically elicits a considerable number of performance errors that are followed by increased performance control; cf. Ridderinkhof, 2002). Our main hypothesis was that the parametric TTI effect on P3 amplitude is correlated speciﬁcally with PE amplitude (and thus not with NE amplitude). Method Participants Fourteen participants (10 women) were recruited from the ﬁrst-year psychology undergraduate student population of the University of Amsterdam and received course credits for their participation. Participants ranged in age between 18 and 35 years (M 5 20.7, SD 5 4.2), and were all right-handed and had normal (or corrected-to-normal) vision according to self-report. They reported to be free of psychoactive medications and to have no history of serious head injury or other neurological or psychiatric disorders that could be expected to affect cognitive function. Task and Procedure The oddball task consisted of a series of nontarget and target stimuli that were presented on a computer screen in white uppercase letters (Xs and Os) against a gray background. Between stimuli a white ﬁxation cross was visible (0.35 cm 0.35 cm) that stayed on the screen for 1400 ms. The stimuli themselves were presented for 100 ms. For half of the subjects the letter X served as a target and the letter O as a nontarget; for the other half it was the other way around. Four experimental blocks were presented to the subject, each of which contained 300 nontargets and 30 target stimuli. The sequence of target and nontarget trials was varied in such a way that 15 TTI (the number of nontargets between two targets) were created. These TTIs ranged from 3 to 17 nontargets between targets. The sequence of these 15 TTI conditions within blocks was determined randomly by the computer. Subjects were asked to respond as quickly and accurately as possible with their right index ﬁnger to targets only. No reaction was required when a nontarget was presented. Stimuli were 3.8 cm 3.4 cm (subtending a visual angle of 2.421 2.161 when viewed from a distance of 90 cm). The Simon task involved the discrimination between two colors (blue and red). Stimuli were circles presented 1.9 cm above or below a gray ﬁxation square (0.22 cm) against a white background. Stimuli were 1 cm 1 cm (subtending a visual angle of 0.641 0.641 when viewed from a distance of 90 cm). All trials started with the ﬁxation square only, which stayed on the screen for 750 ms, after which a circle stimulus was added either above or below the ﬁxation point. The circle remained on the screen for 2000 ms or until one of the response buttons was pressed. Thus, the trial duration was 2750 ms at most. The response button box

533 with vertically placed buttons was placed 25 cm in front of the subject. Each response button was assigned to one of the two colors (counterbalanced across subjects). Stimulus location was irrelevant (and instructed to be ignored) and corresponded to the location of the correct/incorrect response button on 50%/50% of the trials. Subjects were presented with a practice block of 12 trials, immediately followed by three blocks of 200 trials each, yielding a total of 300 congruent and 300 incongruent trials. Button and color responses were counterbalanced across subjects. Half of the subjects started with the oddball tasks and the other half began with the Simon task. Written informed consent was obtained from all participants before the start of the experiment. The study was approved by the local ethics committee and complied with relevant laws and institutional guidelines.

Psychophysiological Recording and Data Reduction Electroencephalogram (EEG) data were collected from 30 electrodes using an Easycap with sintered Ag-AgCl electrodes referenced to the left earlobe using Scan 4.2. AFz served as the ground electrode. The EEG was recorded continuously with a sample rate of 500 Hz with a low-pass ﬁlter at 30 Hz and a time constant of 1 s. Vertical electrooculogram (EOG) was recorded bipolarly from electrodes placed above and below the left eye, and horizontal EOG was recorded bipolarly from electrodes placed on the outher canthi of the two eyes. Electrode impedance was kept below 10 kO. For the oddball task stimulus-locked epochs starting from 200 ms prestimulus to 1000 ms poststimulus were extracted. For the Simon task response-locked epochs were extracted starting from 400 ms preresponse to 600 ms postresponse, as errorrelated brain potentials are best observed in signals time-locked to the time of the error response. First, the EOG (but not EEG) channels were checked for artifacts using an automatic rejection procedure: Segments were excluded from further analysis when the minimum and maximum amplitude within the segment differed by more than 1000 mV. Furthermore segments were excluded when amplitudes exceeded 11000 mV or 1000 mV and when two subsequent samples differed by more than 50 mV. Eye movements were corrected off-line using the procedure used by Gratton, Coles, and Donchin (1983) as implemented in Brain Vision Analyzer. Subsequently the EEG (but not EOG) channels were checked for artifacts. Segments were excluded when the minimum and maximum amplitude within the segment differed by more than 250 mV, when amplitudes exceeded 1250 mV or 250 mV, and when two consecutive samples differed by more than 50 mV. For the oddball task, averaged waveforms aligned to a 100ms prestimulus baseline were generated for nontarget and target stimuli for each TTI bin separately. The P3 peak was deﬁned as the most positive peak in the 250–550-ms time window after the stimulus, separately for each channel. For the Simon task, averages were generated for correct and for error trials (pooled for congruent and incongruent conditions, which is conventional, as congruent trials generate too few errors to allow reliable average ERPs to be computed; we note that the resulting ERPs for error trials were virtually identical to those obtained when incongruent error trials only were included in the averages) aligned to a 200-ms preresponse baseline. NE was deﬁned as the most negative peak in the 50–150-ms time window after response onset, whereas PE was deﬁned as the most positive peak in the 100–300-ms postresponse time window.

534

K.R. Ridderinkhof, J.R. Ramautar, and J.G. Wijnen

Data Analysis Two generalized linear model repeated-measures analyses of variance (ANOVAs) were used to investigate accuracy and reaction time in the oddball task. The independent variable for this analysis was TTI. Every three consecutive TTIs were collapsed together into one TTI bin, resulting in ﬁve TTI bins (3–5, 6–8, 9–11, 12–14, and 15–17 consecutive nontargets). T tests were used to verify whether the usual effects of compatibility were seen in the Simon task. Mean response times (RT) and error rates were compared for the congruent versus incongruent conditions. For the physiological data, three series of analyses were performed. All analyses centered on midline electrodes, as P3 and error-related components are typically largest there (this was conﬁrmed by visual inspection of the present data). First, a repeated-measures ANOVA with TTI (in the ﬁve TTI bins described above) and Electrode Position (Fz, FCz, Cz, CPz, Pz, and Oz) as within-subjects factors was conducted on P3 amplitude data from the oddball task as the dependent variable. Second, two repeated-measures ANOVAs with Accuracy (correct vs. incorrect) and Electrode Position (Fz, FCz, Cz, CPz, Pz, and Oz) as within-subjects factors were conducted on PE and NE amplitude data from the Simon task as the dependent variables, respectively. Finally, separate ANOVAs were carried out for each of the midline electrodes with TTI as a within-subjects factor and NE amplitude on that electrode as a continuous independent variable in one series of analyses and PE amplitude on that electrode in another series. For these last analyses our main interest was with the covariance between on the one hand the parametric effect of TTI on P3 amplitude (quantiﬁed in the linear contrast analysis of TTI effects obtained in the oddball task) and on the other hand the amplitude of error-related components (NE and PE) as obtained in the Simon task. This analysis of covariance allowed us to examine whether, as predicted, the parametric effect of TTI on P3 amplitude correlates with the amplitude of the PE but not the NE. Whenever necessary, alpha levels were corrected per analysis using the Bonferoni method, and the degrees of freedom were adjusted with the Greenhouse–Geisser correction.

Results Behavioral Results For the oddball task, mean RT for correct target detection responses was 376.32 ms. RT varied slightly as a function of TTI, F(4,52) 5 3.38, po.050 (see Table 1). Post hoc comparisons (Tukey’s least signiﬁcant difference [LSD]) revealed that subjects responded signiﬁcantly more slowly in the two briefest TTI conditions compared to the third condition (see Table 1). Overall Table 1. Mean reaction Times (RT) to Detected Targets for Each of the Five Target–Target Intervals in the Oddball Task No. of targets 3–5 6–8 9–11 12–14 15–17 a,b

RT (SD) 385.48 (43.99)a 383.09 (46.76)b 365.94 (45.56)a,b 373.79 (55.40) 372.99 (48.68)

The mean difference is signiﬁcant at the .05 level (between both a’s and both b’s).

accuracy of target detection was 99.64% and did not vary systematically as a function of TTI, F(4,52) 5 0.19, po.894. For the Simon task, mean RT for correct responses to congruent stimuli were faster than to incongruent stimuli (342.5 vs. 373.2 ms) t(13) 5 9.28, po.001. Likewise, error rates were lower for congruent compared to incongruent trials (4.50 vs. 14.83%) t(13) 5 6.10, po.001. ERP Results The oddball task. Figure 1 depicts the grand average waveforms at all electrode positions for nontarget and target stimuli for each of the ﬁve TTIs. Compared to standard stimuli, target stimuli elicited more pronounced ERP components, most prominently in the P3 time window, as is typical for oddball tasks. P3 amplitudes differed between electrode positions, F(5,65) 5 20.87, po.001, being largest at CPz and Pz, as conﬁrmed by pairwise comparisons (using Tukey’s LSD), indicating that these positions differed signiﬁcantly from each of the other positions (all po.02). Most important, P3 amplitude increased generally as a function of TTI (see Figure 2) F(5,65) 5 3.23 , po.019; linear contrast F(1,13) 5 6.06, po.029, across electrode positions. An interaction effect between electrode position and TTI indicated that this pattern differed signiﬁcantly between electrode positions, F(20,260) 5 2.12, po.004. Subsequent separate ANOVAs for each electrode revealed that the effect of TTI was most robust at centroparietal electrodes, as can be seen in Table 2. These TTI effects were well captured by linear contrasts (reﬂecting the a priori hypothesized parametric effect of TTI; computed only for electrodes at which the TTI effect obtained statistical signiﬁcance; see Table 2), but not by higher-order polynomial contrasts. The Simon task. Figure 3 depicts grand average ERP waveforms, time-locked to response onset, for errors versus correct responses (pooled across congruent and incongruent trials) at midline electrode sites. Compared to correct responses, errors were associated with a pronounced rapid negative deﬂection (identiﬁed as the NE) followed by a pronounced slower positive deﬂection (identiﬁed as the PE). The amplitudes of these errorrelated components depended on electrode position: NE: F(5,65) 5 13.42, po.001; PE: F(5,65) 5 5.01, po.009. In accordance with typical observations, the NE was largest at Fz and FCz whereas the PE was more widely distributed, being largest between frontocentral and parietal sites (as can be seen in Figure 3, and as conﬁrmed by pairwise comparisons for Cz and Pz). Correlation between the TTI effect on the P3 and error-related components. The relation between PE amplitude (in the Simon task) and the parametric effect of TTI on P3 amplitude (in the oddball task) was examined in two ways. First, the TTI effect on P3 amplitude correlated signiﬁcantly with the PE on centroparietal electrode sites (see Table 2). Second, and more important, we entered PE amplitude as a continuous independent variable into the ANOVA on the TTI on P3 amplitude. A signiﬁcant correlation was revealed in the diminution of TTI effects on P3 amplitude in the resulting ANCOVAs (see Table 2). A visual impression of this correlation is shown in Figure 4, which plots PE amplitude as a function of the slope of the linear regression function that best describes the parametric effect of TTI on P3 amplitude for each subject. Individuals whose P3s are more sensitive to TTI effects are associated with greater PEs to response errors.

535

To PE or not to PE

Figure 1. Grand average waveforms from midline leads evoked by nontargets and targets for each of the ﬁve TTI conditions in the oddball task. Negative voltage is plotted upward.

The correlation of the parametric effect of TTI on P3 amplitude with NE amplitude was examined similarly. The TTI effect on P3 amplitude did not correlate signiﬁcantly with the NE on any electrode site (see Table 2). When entering NE amplitude as a covariate into the ANOVA of P3 data, at the electrodes for which TTI effects had been obtained, this effect was abolished only slightly when covariance with NE amplitude was partialled out (see Table 2). Discussion As reviewed by Overbeek et al. (2005), studies that have examined individual differences in and/or the effects of experimental manipulations on both the NE and the PE often report these effects to be dissociated. Several studies indicate that the amplitude of the PE, but not the NE, covaries with the degree of

awareness of the error or the salience of the error-inducing stimulus. Intentional errors, which presumably are not experienced as very salient (affectively or cognitively), elicit a smaller PE than do genuine errors (Stemmer, Witzke, & Scho¨nle, 2001). With the exception of caffeine, all substances (insofar as tested) that directly or indirectly affect dopaminergic activity and produce (enhancing or attenuating) effects on the NE fail to produce such effects on the PE (De Bruijn, Hulstijn, Verkes, Ruigt, & Sabbe, 2004; Ridderinkhof et al., 2002; Tieges, Ridderinkhof, Snel, & Kok, 2004). Various dopamine-deviant populations show altered NE amplitudes (in comparison to controls), but little or no differences in PE (for a review, see Overbeek et al., 2005). Although relying on null ﬁndings carries the usual risks, the patterns appear reasonably consistent across studies and suggest that the PE reﬂects neurocognitive processes that differ from those underlying the NE.

536

K.R. Ridderinkhof, J.R. Ramautar, and J.G. Wijnen

Figure 2. Mean P3 peak amplitudes from midline leads for each of the ﬁve TTI conditions in the oddball task. Error bars represent the standard error of the mean.

Here we examined the hypothesis that the PE reﬂects processes similar to those reﬂected in the P3. For instance, the PE may reﬂect the motivational signiﬁcance of errors (cf. Nieuwenhuis et al., 2005; Ro¨sler, 1983), similar to the P3 reﬂecting the motivational signiﬁcance of eliciting stimulus events (Picton, 1992). The PE has a similar morphology and broad midline scalp distribution as the P3 (cf. Leuthold & Sommer, 1999). The focus of the PE scalp distribution in this study was central, as was the focus of the TTI effect on P3 amplitude. The PE has been found to correlate highly with the skin conductance response associated with an error; the skin conductance response is regarded as an orienting response, as associated with events that typically elicit a P3 (cf. Hajcak et al., 2003). The view that the PE constitutes a P3 associated with the motivational signiﬁcance of the error is consistent with the ﬁnding of a larger PE for more salient errors (Leuthold & Sommer, 1999) and with the ﬁnding that the PE is small or absent when the subject does not explicitly recognize the error (Endrass et al., 2005; Kaiser, Barker, Haenschel, Baldeweg, & Gruzelier, 1997; Nieuwenhuis et al., 2001), although it remains to be clariﬁed whether the PE is the expression of error awareness, or reﬂects the processes that lead to error awareness. We found that the amplitude of the PE, much more than of the NE, covaries signiﬁcantly with the effect of TTI on P3 amplitude. This observed correlation replicates the correlation reported previously by Davies et al. (2001) and extends it in important ways. Rather than scoring the P3 in ERP data from the same task, which was not designed primarily to study the processes underlying the P3, we scored the P3 in a separate oddball task using a TTI manipulation designed expressly to tap into the motivational-salience processing thought to be reﬂected in the P3. Moreover, rather than correlating the PE to the P3 directly, we

correlated it to the parametric effect on TTI, which provides a purer picture, as it excludes much of the variance that is not related directly to the processing of motivational salience. The ﬁnding that the TTI effect on P3 did not covary strongly with the amplitude of the NE attests to the speciﬁcity of the observed effects and strengthens our interpretation of the PE in terms of the motivational signiﬁcance of response errors. The further development and veriﬁcation of our hypothesis regarding the functional signiﬁcance of the PE may be serviced by studying the neural generators of the PE. Any brain region showing differential activation to recognized and unrecognized errors would be an important candidate generator of the PE. In an initial neuroimaging study, Hester et al. (2005) observed regions that displayed such differential patterns in bilateral prefrontal and inferior parietal cortices, corresponding roughly to the main brain regions implicated in generating the P3 (Soltani & Knight, 2000). A more direct comparison of the areas engaged in the parametric effect of TTI and the areas sensitive to aware versus unaware errors might reinforce the notion that the PE and P3 may reﬂect similar neural and functional processes. In a previous study, we observed that post-error slowing occurred only after aware errors, which were characterized by a pronounced PE (Nieuwenhuis et al., 2001). The apparent relationship between the PE and post-error slowing might thus be mediated in part by the awareness of errors, because post-error slowing may reﬂect a deliberate strategy, dependent on the conscious recognition of the error. In other studies, however, posterror slowing has been found to be related to NE amplitude (Debener et al., 2005; Gehring, Goss, Coles, Meyer, & Donchin, 1993), although such ﬁndings are subject to debate (e.g., Gehring & Fencsik, 2001) and wait to be resolved in future work. Posterror slowing has also been observed for errors that were not perceived as such (Rabbitt, 2002). Overbeek et al. (2005) suggested the possibility that post-error adaptations in information processing might be instigated through multiple separate routes, perhaps implemented in parallel systems. On the one hand, a rapid preconscious system involving the basal ganglia and the RCZ computes and signals the likelihood of reward (expressed at the scalp in the NE), thereby guiding adaptive actions (Ridderinkhof et al., 2004; Rushworth, Walton, Kennerley, & Bannerman, 2004). On the other hand, a slower, more deliberate error-signiﬁcance evaluation system (expressed at the scalp in the PE) might come into play when errors are sufﬁciently salient. The possibility of two partially redundant post-error adaptation systems might explain why patients with damage to the RCZ still exhibit adequate post-error slowing in tasks in which errors are

Table 2. Results of ANOVA and Correlation Analysisa

Site

ANOVA F(4,52)

Correlation with PE amplitude (r)

Correlation with NE amplitude (r)

ANOVA linear contrastb F(1,13)

ANCOVA with PE amplitude as covariate F(4,48)

ANCOVA with NE amplitude as covariate F(4,48)

Fz FCz Cz CPz Pz Oz

2.46 3.23nn 3.51nn 4.20nn 3.34nn 0.92

.23 .35 .47 .31 .33 .43

.27 .05 .02 .02 .09 .23

8.24n 6.96nn 5.41 3.19

1.28 0.98 0.30 0.31

1.19 2.13 3.69nn 2.94

a

In case of signiﬁcant ANOVA outcomes: linear contrast of TTI effects and of ANCOVA with PE amplitude and NE amplitude as covariates. None of the higher-order polynomial contrasts reached statistical signiﬁcance. n Signiﬁcant AN(C)OVA result at a 5 .05 (after Bonferroni correction). nn Trendwise AN(C)OVA signiﬁcant at a 5 .10 (after Bonferroni correction). b

537

To PE or not to PE

Figure 3. Grand average waveforms from the midline leads time-locked to the response onset in the Simon task. Negative voltage is plotted upward.

particularly salient, such as Stroop and go/no-go tasks (Fellows & Farah, 2005). In sum, we propose that the PE reﬂects the motivational signiﬁcance of a salient performance error, similar to the P3

reﬂecting the motivational signiﬁcance of a rare target stimulus. Although the present study does not allow for a direct test of an interpretation in terms of motivational signiﬁcance, results were obtained in support of the hypothesis that the PE and P3 reﬂect

Figure 4. PE amplitude for each of the midline leads plotted as a function of the slope of the linear regression function for the parametric effect of TTI on P3 amplitude for each subject.

538

K.R. Ridderinkhof, J.R. Ramautar, and J.G. Wijnen

similar processes. PE amplitude (but not NE amplitude) is correlated speciﬁcally to factors known to parametrically affect P3 amplitude. The speciﬁcity of this novel ﬁnding supports the no-

tion that the PE and P3 reﬂect similar neurocognitive processes, possibly as involved in the conscious processing of motivationally signiﬁcant events.

REFERENCES Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conﬂict monitoring and cognitive control. Psychological Review, 108, 624–652. Croft, R. J., Gonsalves, C. J., Gabriel, C., & Barry, R. J. (2003). Targetto-target interval versus probability effects on P300 in one- and twotone tasks. Psychophysiology, 40, 322–328. Davies, P. L., Segalowitz, S. J., Dywan, J., & Pailing, P. E. (2001). Errornegativity and positivity as they relate to other ERP indices of attentional control and stimulus processing. Biological Psychology, 56, 191–206. Debener, S., Ullsperger, M., Siegel, M., Fiehler, K., von Cramon, D. Y., & Engel, A. K. (2005). Trial-by-trial coupling of concurrent electroencephalogram and functional magnetic resonance imaging identiﬁes the dynamics of performance monitoring. Journal of Neuroscience, 24, 11730–11737. de Bruijn, E. R. A., Hulstijn, W., Verkes, R. J., Ruigt, G. S. F., & Sabbe, B. G. C. (2004). Drug-induced stimulation and suppression of action monitoring in healthy volunteers. Psychopharmacology, 177, 151–160. Donchin, E., & Coles, M. G. H. (1988). Is the P300 component a manifestion of context-updating? Behavioural and Brain Sciences, 11, 357–374. Duncan-Johnson, C. C., & Donchin, E. (1984). The P300 event-related potential as an index information processing. Biological Psychology, 14, 1–52. Endrass, T., Franke, C., & Kathmann, N. (2005). Error awareness in a saccade countermanding task. Journal of Psychophysiology, 19, 219–229. Falkenstein, M., Hoorman, J., & Hohnsbein, J. (1999). ERP components in go/no-go tasks and their relation to inhibition. Acta Psychologica, 101, 267–291. Fellows, L. K., & Farah, M. J. (2005). Is anterior cingulate cortex necessary for cognitive control? Brain, 128, 788–796. Gehring, W. J., & Fencsik, D. E. (2001). Functions of the medial frontal cortex in the processing of conﬂict and errors. Journal of Neuroscience, 21, 9430–9437. Gehring, W. J., Goss, B., Coles, M. G. H., Meyer, D. E., & Donchin, E. (1993). A neural system for error detection and compensation. Psychological Science, 4, 385–390. Gonsalves, C. J., Gordon, E., Grayson, S., Barry, R. J., Lazzaro, L., & Bahramali, H. (1999). Is the target-to-target interval a crtical determinant of P3 amplitude? Psychophysiology, 36, 643–654. Gratton, G., Coles, M. G. H., & Donchin, E. (1983). A new method for off-line removal of ocular artifact. Electroencephalography and Clinical Neurophysiology, 55, 468–484. Hajcak, G., McDonald, N., & Simons, R. F. (2003). To err is autonomic: Error-related brain potentials, ANS activity, and post-error compensatory behavior. Psychophysiology, 40, 895–903. Herrmann, M. J., Ro¨mmler, J., Ehlis, A.-C., Heidrich, A., & Fallgatter, A. J. (2004). Source localization (LORETA) of the error-relatednegativity (ERN/Ne) and positivity (Pe). Cognitive Brain Research, 20, 294–299. Hester, R., Foxe, J. J., Molholm, S., Shpaner, M., & Garavan, H. (2005). Neural mechanisms involved in error processing: A comparison of errors made with and without awareness. NeuroImage, 27, 602–608. Holroyd, C. B., & Coles, M. G. H. (2002). The neural basis of human error processing: Reinforcement learning, dopamine, and the errorrelated negativity. Psychological Review, 109, 679–709. Horovitz, S. G., Skudlarski, P., & Gore, J. C. (2002). Correlations and dissociations between BOLD signal and P300 amplitude in an auditory oddball task: A parametric approach to combining fMRI and ERP. Magnetic Resonance Imaging, 20, 319–325. Kaiser, J., Barker, R., Haenschel, C., Baldeweg, T., & Gruzelier, J. H. (1997). Hypnosis and event-related potential correlates of error processing in a Stroop-type paradigm: A test of the frontal hypothesis. International Journal of Psychophysiology, 27, 215–223. Kok, A. (2001). On the utility of P3 amplitude as a measure of processing capacity. Psychophysiology, 38, 557–577. Leuthold, H., & Sommer, W. (1999). ERP correlates of error porcessing in spatial S-R compatibility tasks. Clinical Neurophysiology, 110, 342–357.

Nieuwenhuis, S., Aston-Jones, G., & Cohen, J. D. (2005). Decision making, the p3, and the locus coeruleus-norepinephrine system. Psychophysiolog Bulletin, 131, 510–532. Nieuwenhuis, S., Ridderinkhof, K. R., Blom, J., Band, G. P. H., & Kok, A. (2001). Error-related brain potentials are differentially related to awareness of response errors: Evidence from an antisaccade task. Psychophysiology, 38, 752–760. O’Connell, R. G., Dockree, P. M., Bellgrove, M. A., Kelly, S. P., Hester, R., Garavan, H., et al. (2007). The role of cingulate cortex in the detection of errors with and without awareness: A high-density electrical mapping study. European Journal of Neuroscience, 25, 2571– 2579. Overbeek, T. J. M., Nieuwenhuis, S., & Ridderinkhof, K. R. (2005). Dissociable components of error processing: On the functional signiﬁcance of the Pe vis-a`-vis the ERN/Ne. Journal of Psychophysiology, 19, 319–329. Picton, T. W. (1992). The P300 wave of the human event-related potential. Journal of Clinical Neurophysiology, 9, 456–479. Rabbitt, P. (2002). Consciousness is slower than you think. The Quarterly Journal of Experimental Psychology: Section A, 55, 1081–1092. Ridderinkhof, K. R. (2002). Micro- and macro-adjustments of task set: Activation and suppression in conﬂict tasks. Psychological Research, 66, 312–323. Ridderinkhof, K. R., de Vlugt, Y., Bramlage, A., Spaan, M., Elton, M., Snel, J., et al. (2002). Alcohol consumption impairs the detection of performance errors by mediofrontal cortex. Science, 298, 2209–2211. Ridderinkhof, K. R., Ullsperger, M., Crone, E. A., & Nieuwenhuis, S. (2004). The role of medial frontal cortex in cognitive control. Science, 306, 443–447. Ro¨sler, F. (1983). Endogenous ER ‘Ps’ and cognition: Probes, prospects, and pitfalls in matching pieces of the mind-body puzzle. In A. W. K. Gaillard & W. Ritter (Eds.), Tutorials in event-related potential research: Endogenous components (pp. 9–35). Amsterdam: Elsevier. Rushworth, M. F., Walton, M. E., Kennerley, S. W., & Bannerman, D. M. (2004). Action sets and decisions in the medial frontal cortex. Trends in Cognitive Sciences, 8, 410–417. Schultz, W. (2002). Getting formal with dopamine and reward. Neuron, 36, 241–263. Soltani, M., & Knight, R. T. (2000). Neural origins of the P300. Critical Review of Neurobiology, 14, 199–224. Squires, K. C., Wickens, C., Squires, N. K., & Donchin, E. (1976). The effects of stimulus sequence on the waveform of the cortical eventrelated potential. Science, 193, 1142–1146. Stemmer, B., Witzke, W., & Scho¨nle, P. W. (2001). Losing the errorrelated negativity in the EEG of human subjects: An indicator for willed action. Neuroscience Letters, 308, 60–62. Stevens, M. C., Calhoun, V. D., & Kiehl, K. A. (2005). fMRI in an oddball task: Effects of target-to-target interval. Psychophysiology, 42, 636–642. Tieges, Z., Ridderinkhof, K. R., Snel, J., & Kok, A. (2004). Caffeine strengthens action monitoring: Evidence from the error-related negativity. Cognitive Brain Research, 21, 87–93. van Boxtel, G. J. M., van der Molen, M. W., & Jennings, J. R. (2005). Differential involvement of the anterior cingulate cortex in performance monitoring during a stop-signal task. Journal of Psychophysiology, 19, 1–10. Van Veen, V., & Carter, C. S. (2002). The timing of action-monitoring processes in the anterior cingulate cortex. Journal of Cognitive Neuroscience, 14, 593–602. Verleger, R. (1988). Event-related potentials and cognition: A critique of the context updating hypothesis and an alternative interpretation of P3. Behavioural and Brain Science, 11, 343–357. Yeung, N., Cohen, J. D., & Botvinick, M. M. (2004). The neural basis of error detection: Conﬂict monitoring and the error-related negativity. Psychological Review, 111, 931–959. (Received July 24, 2007; Accepted July 1, 2008)

download basis, it'd come to less than a pe - Libsyn