Time Course of Error Detection and Correction in Humans PDF

The Journal of Neuroscience, November 15, 2002, 22(22):9990–9996 Time Course of Error Detection and Correction in Humans: Neurophysiological Evidence Antoni Rodrı́guez-Fornells, Arthur R. Kurzbuch, and Thomas F. Münte Department of Neuropsychology, Otto von Guericke University, 39112 Magdeburg, Germany Using event-related brain potentials, the time course of error preceded the ERN. Thus, error correction was implemented detection and correction was studied in healthy human sub- before or at least in parallel with the appearance of the ERN jects. A feedforward model of error correction was used to component. Also, the amplitude of the ERN component was predict the timing properties of the error and corrective move- increased for errors, followed by fast corrective movements. ments. Analysis of the multichannel recordings focused on (1) The results are compatible with recent views considering the the error-related negativity (ERN) seen immediately after errors ERN component as the output of an evaluative system engaged in response- and stimulus-locked averages and (2) on the in monitoring motor conflict. lateralized readiness potential (LRP) reflecting motor prepara- Key words: error correction; error detection; error-related tion. Comparison of the onset and time course of the ERN and negativity; ERN; lateralized readiness potential; LRP; response LRP components showed that the signs of corrective activity conflict; event-related brain potentials; ERPs To adapt their behavior to an ever-changing environment, hu- Fencsik, 2001; van Veen et al., 2001). As pointed out by Paus mans need to be able to monitor their performance and to detect (2001), the ERN conflict-detection model emphasizes the “eval- and correct any errors. The present investigation seeks to delin- uative” nature of the anterior cingulate cortex. Other authors, eate the time course of error detection and correction in humans however, have suggested that the ERN or anterior cingulate using event-related brain potentials (ERPs) (Münte et al., 2000). activations are indexing a more general evaluative system that A negative ERP component labeled error-related negativity processes the motivational significance of events, including but (ERN) has been isolated, appearing immediately after commit- not limited to errors and response conflict situations (Bush et al., ting errors (Falkenstein et al., 1990; Gehring et al., 1993, 1995). A 2000). large part of the ERN can be explained by a source in the anterior Less evidence is available about the neural implementation of cingulate cortex (Gemba et al., 1986; Dehaene et al., 1994; Carter error corrections targeted in the current study. Bearing in mind et al., 1998, Luu and Tucker, 2001). Two competing models have that error correction is one of the fastest cognitive processes been proposed to explain the cognitive mechanism underlying the (Rabbitt, 1966a,b; Cooke and Diggles, 1984), it was predicted that ERN. One model associates the ERN to an error-detection mech- the corrective motor command would be triggered as soon as the anism (Gehring et al., 1993; Falkenstein et al., 1995). Because the evaluative system has accumulated enough evidence. Critically, peak of the ERN appears too early (⬃60 msec after the response) the onset of the corrective movement will be indexed with the to depend on sensory and proprioceptive information, it has been onset of the lateralized readiness potential (LRP) (Gratton et al., assumed that the ERN reflects the output of a feedforward 1988; Smid et al., 1992). Thus, the relative timing of the LRP and control mechanism (Bernstein et al., 1995). This mechanism ERN components in the present experiment could be used to compares an internal goal with the predicted consequences of the infer the time course of error correction and detection. Two ongoing movement, made available through the efference copy control experiments (noncorrection instructions and unilateral vs (Shadmehr and Mussa-Ivaldi, 1994; Wolpert et al., 1995; Desmur- bilateral movements) demonstrated the sensitivity of the LRP and get and Grafton, 2000) (see Fig. 1). An internal “error signal” is ERN to error correction. generated if a mismatch is detected by the system. Alternatively, the ERN conflict-detection model holds that the MATERIALS AND METHODS ERN merely reflects the degree of response conflict experienced Error correction. Sixteen right-handed neurologically healthy subjects by subjects (Cohen et al., 2000; Botvinik et al., 2001). This theory (Ss) participated in the experiment (four women; mean ⫾ SD age, 24.6 ⫾ fits well with data showing ERNs or anterior cingulate activity for 4.5 years) after giving informed consent according to the declaration of correct responses (Carter et al., 1998; Luu et al., 2000; Scheffers Helsinki. All of them were paid for their participation. The Eriksen flanker task (Eriksen and Eriksen, 1974) was used in and Coles, 2000; Vidal et al., 2000) and responses showing a which subjects are required to focus on the letter in the center of a visual higher degree of motor conflict (Barch et al., 2000; Gehring and array of five letters, designated as “target,” and to respond with the right or left hand depending on which of two target letters (H or S) had Received Feb. 19, 2002; revised Sept. 3, 2002; accepted Sept. 6, 2002. appeared. Four flanker letters surrounding the target letter either fa- vored the target response (compatible trials, HHHHH or SSSSS) or This work was supported by grants from the Deutsche Forschungsgemeinschaft (T.F.M.). A.R.F. was supported by a postdoctoral fellowship from the Spanish primed the other response (incompatible trials, HHSHH or SSHSS). To Government. We thank Christine Matthes for running the control experiment. optimize the number of errors, 40% of compatible trials and 60% of Correspondence should be addressed to Dr. Thomas F. Münte, Department of incompatible trials were presented. Each stimulus array subtended ⬃2.5° Neuropsychology, Otto von Guericke University, Universitätsplatz 2, Gebäude 24, of visual angle in width, and a fixation line was presented in the middle 39106 Magdeburg, Germany. E-mail: [email protected]. of the computer monitor just below the target letter in the array. In Copyright © 2002 Society for Neuroscience 0270-6474/02/229990-07$15.00/0 addition, in one-half of the trials, the target letter appeared degraded, by Rodrı́guez-Fornells et al. Time Course of Error Correction J. Neurosci., November 15, 2002, 22(22):9990–9996 9991 hands were recruited sequentially. Ss were required to make left–right responses assigned to an H or S presented in the center of the screen (fixed stimulus onset asynchrony of 900 msec; stimuli subtended 0.5° in width; a fixation line was always present just below the target letter). During 400 trials, single-hand responses were required. In the switching hand condition (400 additional trials), Ss had to press first with the assigned response hand and then to switch immediately and respond with the other hand. Therefore, trials required either left–right or right–left responses. Both conditions and letter–hand assignments were counter- balanced. As in the bimanual condition, the motor command of the second (“switch”) response could be prepared in parallel with the exe- cution of the first response, and the comparison of the unilateral and bilateral response LRPs in the control experiment with correct and error-corrective LRPs in the main experiment was thought to be reveal- ing with regard to the point in time at which the corrective response began to be prepared in the main experiment. Eight right-handed subjects, who had not taken part in the main experiment, participated in the control study (age, 23 ⫾ 2.1 years; six women). No analysis of erroneous responses were made in this experi- ment, and only correct trials were included in the averages. ER Ps. The electroencephalogram (EEG) was recorded from 29 scalp locations, including all standard 10/20 system positions against the alge- braic mean of the activity at the mastoid electrodes (bandpass, 0.01–70 Hz; digitization rate, 250 Hz) using tin electrodes mounted in an elastic cap. Vertical eye movements and horizontal eye movements were re- corded by bipolar montages. After artifact rejection based on individu- alized amplitude criteria, the EEG signal was averaged separately for each stimulus–response combination for epochs of 1024 msec (⫺300, 724 msec in response-locked averages; ⫺100, 924 in stimulus-locked averag- Figure 1. Simplified diagram of the different stages of information pro- es). Baselines used for response-locked averages were from ⫺300 to cessing (boxes) and internal representations (ellipses) involved when an ⫺200 msec for LRPs and from ⫺50 to 0 msec for the ERN component. erroneous response ( R) is produced (adapted from Desmurget and A low-pass filter (8 Hz, half-amplitude cutoff) was applied in all of the Grafton, 2000) (Wolpert et al., 1995; Coles et al., 2001). The large dotted computations and averages reported. square delineates the feedforward control mechanism used to inhibit and For each subject, at least 100 artifact-free error trials for each response correct the erroneous response. When a wrong selection of the motor hand were obtained. Thus, at least 200 trials were available for the command is produced, a motor copy of the activated on-line response is computation of the ERN component. The mean number of error re- generated (the efference copy) and compared with the predicted correct sponses included in the averages was 216 ⫾ 80 for the correction- response (Angel, 1976). If a mismatch is produced, an error signal is forbidden session responses and 260 ⫾ 102 for the correction-encouraged elicited, and immediately a stop command is triggered to abort the session (t(15) ⫽ ⫺1.7; p ⫽ 0.098). From this pool of errors, 238 ⫾ 81 were response. If the erroneous response bypasses this command, the system right-hand errors, and 238 ⫾ 79 were left-hand errors (t(15) ⬍ 1). The triggers the corrective motor command as fast as possible. An additional median reaction time of the corrective responses was determined sepa- potential source of information for the corrective system might be pro- rately for every subject using the trials free of artifacts. This median was prioceptive input from the limb (see dotted arrow). It has been suggested used to create response-locked averages for “slow” and “fast” corrective that proprioceptive information can modify EMG responses with a delay responses for every subject. of only 70 msec (Crago et al., 1976). LRPs were assessed by using C3 and C4 electrode locations, in which the amplitude of the readiness potential is maximum (Kutas and Donchin, 1980). The LRP is computed by a double subtraction as shown in the removing ⬃70% of the pixels (the flankers were maintained nonde- following equation: LRP ⫽ left hand(C4 ⫺ C3) ⫺ right hand(C4 ⫺ C3). graded). Duration of the stimuli was 100 msec, and the stimulus onset Left and right hands refer to the expected correct hand, and (C4 ⫺ C3) asynchrony between two successive stimuli was 900 msec. Letter–hand is the difference in electrical potential between these electrodes (Gratton assignments were counterbalanced between subjects. et al., 1988; Coles, 1989; Smid et al., 1992). The resulting LRP compo- Ss participated in two sessions. In the first session, subjects were nent is negative if subjects produce correct responses and positive for initially trained with 400 trials to reach a reaction time (RT) baseline error trials. In some statistical analyses, the polarity of the error-LRP was level that could be used as a starting point to fix the final deadline RT for inverted to perform statistical analyses and to allow visual comparison in each subject. After this baseline period, a series of 40 trials was admin- the figures. istered, and the subjects received feedback about their performance. The For statistical analysis, mean amplitude measures were obtained and goal of this procedure was to aim for a reaction time that would yield entered into ANOVA statistics with the Huynh –Feldt epsilon correction ⬃15% of errors (see below). After six of such blocks, subjects performed applied as necessary. All tests involving electrode ⫻ condition interac- between 20 and 22 experimental blocks of 200 trials each. Between tions were performed on vector normalized data (McC arthy and Wood, blocks, subjects were given at least 2 min to relax and stretch. Correction 1985). Onset latencies of the LRP were determined via a stepwise series of the errors was forbidden. This session will only be used as a control or of one-tailed serial t tests (step size of 4 msec) (Schmitt et al., 2000). For baseline condition (hereafter referred as correction forbidden) to assess each test, data from a time window of 40 msec were averaged (i.e., point the differences between the error-LRPs when correcting and not correct- of measure, ⫾20 msec); the onset latency was defined as the point at ing errors. which five consecutive t tests showed a significant difference from zero In the second session, the main experiment, error correction was ( p ⬍ 0.05). required whenever Ss detected a performance error (Fig. 1). Using the same RT deadline procedure, subjects performed between 24 and 26 experimental blocks. Subjects were encouraged to correct their errone- RESULTS ous responses as fast as possible, before the appearance of the next stimulus. All of the analyses reported in this study are based on this Correcting erroneous responses error-correction data, except for the comparisons with the control ex- The mean overall reaction time was 360 ⫾ 34 msec. The mean periment and the error-correction forbidden condition from the first reaction time needed to correct an error was 212 ⫾ 40 msec. As session. Control e xperiment: responding with and without switching hands. The expected, error responses were faster (⬃36 msec) than correct aim of this control experiment was to mimic the motor activity per- responses: 342 ⫾ 35 msec versus 378 ⫾ 35 (t(16) ⫽ 7.3; p ⬍ 0.001). formed when correcting errors in the previous experiment, in which both Percentage of hits was 82%, and the percentage of errors was 16%. 9992 J. Neurosci., November 15, 2002, 22(22):9990–9996 Rodrı́guez-Fornells et al. Time Course of Error Correction Figure 2. A, Response-locked grand average for correct and error trials. A clear ERN component is elicited immediately after the response. B, The stimulus-locked averages are illustrated for a single frontal electrode. Notice the increased negativity, ⬃300 and 600 msec for the error trials. The ERN component was clearly present in the response- locked averages (Fig. 2 A) and appeared immediately after the response with a frontocentral distribution. It was followed by a positive parietal component (error positivity) (Falkenstein et al., 1995). An ANOVA was performed on the three midline elec- trodes [Fz (midline frontal), Cz (midline central), and Pz (mid- line parietal)] and for type of trial (correct vs error). Errors showed an increased negativity quantified as the mean amplitude in the 0 –100 msec window (F(1,15) ⫽ 51.2; p ⬍ 0.001). The frontocentral distribution of the ERN was reflected in the signif- icant interaction of electrode ⫻ type of trial (F(2,30) ⫽ 14.2; p ⬍ 0.001; at Cz, mean amplitude errors of ⫺2.4 ⫾ 3.3 ␮V and correct responses of 2.75 ⫾ 2.5). The stimulus-locked ERPs are shown in Figure 2 B. Again, an enhanced negativity was observed for the errors with an onset of ⬃250 msec, which corresponds to the ERN Figure 3. A, Response-locked LRP in the correction-encouraged condi- in the response-locked averages. tion for the correct, error, and corrective responses. LRP was obtained by double subtraction from the lateral central electrode sites C3 and C4. B, Time course of error correction The t test plot of the differences tested. Hereafter, the term correct-LRP refers to the LRP for correct responses, error-LRP refers to the LRPs computed for the error responses, and corrective-LRP indicates the LRP to the correc- switching hands was 214 ⫾ 27 msec (second reaction time). tive response after the error. The correct-LRP differed from zero Despite the differences between both tasks (correction vs switch- between ⫺124 and 86 msec (210 msec significant interval for ing), the LRP waveforms look very similar. In the control exper- negative LRP) (Fig. 3A). The error-LRP was significant between iment, both LRPs (one-hand and switching) began to develop at ⫺116 and 12 msec, being more positive than 0 (128 msec inter- the same time, ⬃120 msec before the response, and no differences val). From 56 until 308 msec (252 msec interval), as a sign of the were observed between both conditions in the onset latency corrective response, the error-LRP differed from 0 in the negative (paired t test; t(7) ⬍ ⫺0.55). However, 50 msec later, preparation of direction. the contralateral response starts in the switching condition, and As can be observed in Figure 3A, the onset of the corrective- correct responses from one-hand and switching conditions began to LRP is earlier than that of the correct-LRP. The onset latency of differ at ⫺76 msec (t(7) ⫽ ⫺2.46; p ⬍ 0.05). This difference the corrective-LRP was ⫺156 msec (significant until 84 msec; 240 remained significant until 332 msec (t(7) ⫽ ⫺2.37; p ⬍ 0.05). Notice msec interval) (Fig. 3B). When contrasting the correct-LRP with that this comparison attests to the ability of LRP to differentiate the corrective-LRP, these were significantly different between between the overlapping activity of responses of different hands. ⫺136 and 80 msec. Thus, these data suggest that (electrophysi- In Figure 4 B, data from the correction-forbidden condition of ologically) it takes more time to prepare the corrective response. the main experiment are shown. Notice the large overlap between In Figure 4A (left), the polarity of the error-LRP is inverted to both conditions, correction forbidden and encouraged, in partic- facilitate comparison with the LRP to the correct responses and ular for correct responses. To get an additional estimate for the to determine the onset of the motor activity for the corrective time at which the corrective motor command is triggered, the response. The correct-LRP began to differ from the inverted error-LRPs of the correction-forbidden and correction- error-LRP at ⫺68 msec (t(15) ⫽ ⫺2.29; p ⬍ 0.05). This time point encouraged conditions were compared (Fig. 4 B, right). A statis- can be taken as an estimate for the time at which the corrective tically reliable difference between the error-LRPs was found from command is already initiated. ⫺8 msec (t(15) ⫽ ⫺2.51; p ⬍ 0.05) onward, until 316 msec. In We contrasted this time point with the moment at which contrast, no such early differences between conditions were seen subjects began their second hand movement in the control exper- in the response-locked ERPs to correct or error responses (Fig. iment (Fig. 4 A, right). The mean reaction time for unimanual 4 B). Also, it is noteworthy that the ERN in the correction- movements was faster than for the switching condition (366 ⫾ 35 forbidden condition was smaller than in the correction- vs 405 ⫾ 36 msec; t(7) ⫽ ⫺3.8; p ⬍ 0.01). The time needed for encouraged condition (F(1,15) ⫽ 6.83; p ⬍ 0.02). Rodrı́guez-Fornells et al. Time Course of Error Correction J. Neurosci., November 15, 2002, 22(22):9990–9996 9993 Figure 5. Response-locked LRP and ERN in the correction-encouraged condition computed as a function of whether corrective responses were fast or slow (median split). The error-LRP polarity was inverted for a better comparison with the correct LRP. corrections began to diverge from the LRP for slow corrections at ⫺24 msec (relative to the error response). The two LRPs re- mained different until 160 msec. Figure 5 also shows the ERN component for fast and slow corrections. The amplitude of the ERN is larger in the fast corrections compared with slow correc- tions. We tested this effect at frontocentral locations for the 0 –100 msec time window. Fast-corrected movements showed a signifi- cant increase compared with slow corrections at Fz (F(1,15) ⫽ 9.97; p ⬍ 0.01; fast corrections, ⫺2.6 ⫾ 2.4 ␮V vs slow corrections, ⫺1.9 ⫾ 2.4 ␮V) and at Cz (F(1,15) ⫽ 24.6; p ⬍ 0.001; fast corrections, ⫺3.5 ⫾ 3.5 ␮V vs slow corrections, ⫺1.3 ⫾ 3.4 ␮V). Is speed of error correction depending on the quality of information? Based on the feedforward model (Fig. 1), the correction process Figure 4. A, LRPs for the correction-encouraged condition and the might be triggered by an internal error signal, which depends on control experiment requiring unimanual or successive bimanual re- sponses. The LRP for the error response was inverted for a better the comparison between the actual motor command (efference comparison. Notice the faster onset of the switched response compared copy) and the mental representation of the predicted correct with the corrective response. B, Comparison between LRPs for the response. If the stimulus information is reduced (e.g., when it is error-correction condition and the baseline session, in which error cor- degraded), the “correct mental representation” might be delayed, rection was forbidden. Notice that the corrective command is initiated before the erroneous button press occurs. In contrast, the ERN for and, therefore, the error correction mechanisms should be trig- erroneous responses from both conditions depicted below does not show gered later as well. For testing this prediction, easy trials compris- such early differences. ing incompatible nondegraded stimuli and difficult trials compris- ing incompatible degraded stimuli were compared. Incompatible trials were chosen to yield a sufficient number of errors to obtain Fast corrections versus slow corrections clean LRPs. The corrective responses were divided according to the median of The stimulus-locked LRPs are depicted in Figure 6 for de- the reaction time of the correction in every subject. Thus, ERPs graded and nondegraded stimuli from the correction-encouraged for fast-corrective trials and slow-corrective trials were obtained and correction-forbidden conditions. With regard to the correct (Fig. 5). Note that averages time-locked to the error response stimuli, in both conditions, the LRP onset was earlier in the were used, thus neutralizing the possible RT differences in the nondegraded compared with the degraded trials (forbidden: non- main response (errors). The mean reaction time for the fast- and degraded, 212 msec and degraded, 304 msec; encouraged: non- slow-corrective responses was 160 ⫾ 37 and 264 ⫾ 43 msec, degraded, 216 msec and degraded, 308 msec; all t values ⬎ 2.4; respectively. To rule out the possibility that the latency difference p ⬍ 0.05). Thus, the reaction time and LRP depended on the of fast and slow corrections was not attributable to a difference in quality of the information in the correct trials. This was not the the latencies of the preceding error responses, we computed the case for the error trials. For the correction-encouraged condition, error RTs separately for these two categories. For the errors that the onset of the error-LRP in the nondegraded and degraded were followed by fast corrections, the mean reaction time was trials was virtually identical (nondegraded, 204 msec, degraded, 338 ⫾ 34 msec; the errors followed by slow corrections were 200 msec, all t values ⬎ 2.4; p ⬍ 0.05). As can be derived from slightly but reliably faster, at 308 ⫾ 35 msec (t(15) ⫽ 11.7; p ⬍ Figure 6 (bottom right), the two error-LRPs are clearly different 0.001). Thus, the difference in corrective RTs for fast and slow with regard to the correction phase, with the downswing of the correction (104 msec) cannot be attributed to the difference in the LRP being much earlier in the nondegraded condition, leading to RT of the preceding error (30 msec). a significant difference between the two error-LRPs between 284 Importantly, as shown in Figure 5, the onsets of the LRP and and 400 msec (t ⬎ ⫺2.34; p ⬍ 0.05). Thus, whereas the latency of ERN responses did not follow the same pattern. The LRP for fast the error response did not depend on the quality of information, 9994 J. Neurosci., November 15, 2002, 22(22):9990–9996 Rodrı́guez-Fornells et al. Time Course of Error Correction a dissociation between the elicitation of the correction mecha- nisms and the ERN was found (Fig. 5). Whereas a clear delay was present in the LRP for the slow corrective responses, the ERN latency was not affected. The second comparison pertains to the discriminability of the critical stimuli. We predicted an earlier onset of the correction mechanism in trials that were easier to discriminate (incompatible degraded trials vs the incompatible nondegraded ones), because a mental representation of the pre- dicted correct response should be available faster in these trials. In the correction-encouraged condition, the stimulus-locked error-LRP for the easy trials hardly lateralized above zero (Fig. 6), because the correction mechanism was implemented immedi- ately. In contrast, in the difficult trials, a longer interval of later- alization of the error-LRP was found before the onset of the corrective-LRP. Before accepting this interpretation, a possible alternative ac- Figure 6. Stimulus-locked LRPs for easy (incompatible nondegraded) count might be considered, however. This account states that the and difficult (incompatible degraded) trials. The LRPs are presented for corrective responses obtained in the correction-encouraged con- correct and error trials for both correction-encouraged and correction- dition do not occur in response to, i.e., subsequent to the detec- forbidden conditions. Whereas the onset of the LRP depended on the per- tion of, an error but rather are slow correct responses prepared in ceptual quality of the stimulus for the correct trials, the onset of the error-LRPs was independent of stimulus quality. On the other hand, the la- parallel with the faster errors. A revealing comparison in this tency of the corrective response, indicated by the downswing in the error- regard is shown in Figure 4 B, which depicts the LRPs obtained in LRP, varied as a function of the quality of the eliciting response. the correction-forbidden and correction-encouraged conditions. LRPs to the error trials are virtually identical in both conditions up to 20 msec before the response, rendering it unlikely that the the latency of the corrective response did. In the correction- correct response is prepared in parallel with the error. A second forbidden condition on the other hand, error-LRPs for degraded argument against this alternative account is based on the control and nondegraded stimuli virtually overlapped with no statistical experiment. Here, the LRPs in bimanual and unimanual condi- differences (Fig. 6, bottom left). tions diverged earlier from each other than the correct and DISCUSSION error-LRPs in the correction-encouraged condition of the main To our knowledge, this is the first electrophysiological study experiment (Fig. 4 A). assessing the time course of error detection and correction. Two The nature of the ERN component major points will be outlined: (1) error correction is implemented before or at least in parallel with the appearance of the ERN A second aim of this study was to investigate the influence of the component, and (2) the amplitude of ERN component appears to correction process on the ERN. The amplitude differences of the reflect the degree of motor conflict when correcting errors. ERN to trials followed by fast and slow corrections are most revealing in this respect. The ERN has a higher amplitude for the Time course of error correction trials, which were followed by fast corrective movements com- The feedforward model of error correction (Fig. 1) suggests that pared with the ones that were corrected slower (Fig. 5). This the error-correction process might be triggered immediately after follows directly from the ERN conflict-detection model (see in- the elicitation of an internal error signal, which depends on the troductory remarks) that predicts that coactivation of two con- representation of the predicted “correct response” and efference flicting motor channels (in this case, error response and correc- copy of the on-line motor command. The error-LRP began to tive response) should lead to an increase of the ERN (Botvinik et differ from the correct-LRP as early as 68 msec before the al., 2001). As for fast corrections, the temporal overlap of the response (Fig. 4 A), thus giving an estimate of the time point at motor programs for error and corrective responses is greater, and which the corrective response is triggered. Importantly, this time a larger ERN is expected. In the error-detection model of the point precedes the onset of the ERN component. An alternative ERN, the ERN amplitude is driven by the mismatch between the estimate of the onset of correction was obtained by comparing the representation of the computed correct response and the effer- error-LRPs in both correction conditions (Fig. 6 B), which began ence copy of the actual (error) response (Coles et al., 2001). For to differ 8 msec before the response. In both cases, the conclusion error trials followed by slow corrections, it might be the case that is warranted that the ERN component appears after error cor- not enough stimulus information is available at the time of the rection is already underway or, at the very best, at the same time. error for detection to occur effectively, leading to a smaller ERN This result is incompatible with the assumption that the process and slower corrections. For fast corrections, more information underlying the ERN triggers error correction. Thus, if the ERN might be available, hence a larger ERN and faster corrections. A is taken as an index of (conscious) error monitoring (Scheffers third account of the ERN states that the amplitude of this and Coles, 2000; Coles et al., 2001), a second (nonconscious) component indexes the activity of a general evaluative system internal error signal must precede the ERN and initiate the concerned with the motivational significance of the error rather correction process. This partial independence of error detection than with the response conflict per se (Luu and Tucker, 2001). and correction converges with data on very fast corrections of The present data are compatible with such an account, if one complex movements (Goodale et al., 1986; Castiello et al., 1991) assumes that the motivational level of errors is higher in a and is also supported by two additional comparisons. First, when situation in which correction is allowed and also higher for trials comparing error trials followed by either slow or fast corrections, accompanied by fast corrections. Thus, the current results are Rodrı́guez-Fornells et al. Time Course of Error Correction J. Neurosci., November 15, 2002, 22(22):9990–9996 9995 indifferent with regard to the different accounts of the ERN investigation (Carter et al., 1998) but not in the on-line correc- component. tions in the double-step paradigm (Desmurget et al., 2001). An unresolved issue is the relationship of the ERN component to the conscious detection of errors. Although it has been stated REFERENCES that the ERN is linked to the conscious perception of errors Angel RW (1976) Efference copy in the control of movement. Neurology (Dehaene et al., 1994; Scheffers and Coles, 2000), it might also be 26:1164 –1168. Barch DM, Braver TS, Sabb FW, Noll DC (2000) Anterior cingulate and the case that the ERN is only reflecting an internal unconscious the monitoring of response conflict: evidence from an fMRI study of error signal and that conscious perception is followed later. In the overt verb generation. J Cognit Neurosci 12:298 –309. current data set, the ERN follows the corrective command (in- Bernstein PS, Scheffers MK, Coles MGH (1995) “Where did I go wrong?” A psychophysiological analysis of error detection. J Exp Psy- dexed by the LRP) in time. Because the internal error signal chol Hum Percept Perform 21:1312–1322. needs to precede the corrective command, it cannot underlie the Botvinick MM, Braver TS, Barch DM, Carter CS, Cohen JD (2001) Conflict monitoring and cognitive control. Psychol Rev 108:624 – 652. ERN. This favors the interpretation that the internal error signal Bush G, Luu P, Posner MI (2000) Cognitive and emotional influences in itself is unconscious and that the ERN is probably associated with anterior cingulate cortex. Trends Cogn Sci 4:215–222. the conscious perception of the errors. This view is supported by Carter CS, Braver TS, Barch DM, Botvinick MM, Noll D, Cohen JD (1998) Anterior cingulate cortex, error detection, and the on-line mon- findings from Luu et al. (2000), who found a linear increase in the itoring of performance. Science 280:747–749. amplitude of the ERN with increasingly late responses in a task Castiello U, Paulignan Y, Jeannerod M (1991) Temporal dissociation of defining late responses as errors. motor responses and subjective awareness. Brain 114:2639 –2655. Cohen JD, Botvinick M, Carter CS (2000) Anterior cingulate and pre- The existence of automatic error correction has been demon- frontal cortex: who’s in control. Nat Neurosci 3:421– 423. strated recently (Pisella et al., 2000) (see also Rabbitt, 1968). In Coles MGH (1989) Modern mind-brain reading: psychophysiology, physiology and cognition. Psychophysiology 26:251–269. this study, subjects were making pointing reactions with their Coles MGH, Scheffers MK, Holroyd CB (2001) Why is there an hand. In some trials, the target was displaced unexpectedly to the ERN/Ne on correct trials? Response representations, stimulus-related left or to the right, requiring the inhibition of the pointing components, and the theory of error-processing. Biol Psychol 56:173–189. response. Subjects could not avoid to make automatic hand move- Cooke JD, Diggles VA (1984) Rapid error correction during human arm ment corrections to the displacements, similar to the control movements: evidence for central monitoring. J Motor Behav 16:348 –363. situation, which, in a different group of subjects, required an Crago PE, Houk JC, Hasan Z (1976) Regulatory actions of human adjustment of the movement in the direction of the displacement. stretch reflex. J Neurophysiol 39:925–935. In this situation, a clear conflict emerged between the correction Dehaene S, Posner MI, Tucker DM (1994) Localization of a neural system for error detection and compensation. Psychol Sci 5:303–305. system, apparently acting automatically, and the voluntary motor Desmurget M, Grafton S (2000) Forward modeling allows feedback con- intentions to inhibit the on-line response. trol for fast reaching movements. Trends Cogn Sci 4:423– 431. Desmurget M, Epstein CM, Turner RS, Prablanc C, Alexander GE, With regard to the generality of the current findings, it is Grafton ST (1999) Role of the posterior parietal cortex in updating important to consider that they pertain to a particular kind of reaching movements to a visual target. Nat Neurosci 2:563–567. error correction, i.e., the generation of a new (or corrective) Desmurget M, Grea H, Grethe JS, Prablanc C, Alexander GE, Grafton ST (2001) Functional anatomy of nonvisual feedback loops during motor command in the presence of an erroneous initial motor reaching: a positron emission tomography study. J Neurosci response. This is clearly different from paradigms in the motor 21:2919 –2928. Eriksen BA, Eriksen CW (1974) Effects of noise letters upon the iden- domain, such as so-called double-step experiments involving sub- tification of target letters in a non-search task. Percept Psychophys liminal target displacements (Prablanc and Martin, 1992). In a 16:143–149. recent brain imaging study using this paradigm, a neural network Falkenstein M, Hohnsbein J, Hoormann J, Blanke L (1990) Effects of errors in choice reaction tasks on the ErP under focused and divided including the cerebellum, posterior parietal cortex (PPC), and attention. In: Psychophysiological brain research (Brunia CHM, Gaillard primary motor cortex was demonstrated to be involved in error AWK, Kok A, eds), pp 192–195. Tilburg, The Netherlands: Tilburg UP. Falkenstein M, Hohnsbein J, Hoormann J (1995) Event-related poten- correction (Desmurget et al., 2001). PPC has also been demon- tial correlates of errors in reaction tasks. In: Perspectives of event- strated to be important the correction of hand trajectories (Des- related brain potentials research (Karmos G, Molnár M, Csépe V, murget et al., 1999; Pisella et al., 2000). For example, a patient Czigler I, Desmedt JE, eds), pp 287–296. Amsterdam: Elsevier. Gehring WJ, Fencsik DE (2001) Functions of the medial frontal cortex with bilateral lesions of the PPC failed to show automatic error in the processing of conflict and errors. J Neurosci 21:9430 –9437. corrections of his movements but was able to make slow and Gehring WJ, Gross B, Coles MGH, Meyer DE, Donchin E (1993) A neural system for error detection and compensation. Psychol Sci 4:385– deliberate motor corrections (Pisella et al., 2000). Desmurget and 390. Grafton (2000) consequently have proposed a model of error Gehring WJ, Coles MGH, Meyer DE, Donchin E (1995) A brain poten- correction in which the PPC is controlling the adequacy of the tial manifestation of error-related processing. In: Perspectives of event- related brain potentials research, EEG Suppl 44 (Karmos G, Molnár M, ongoing motor response, predicting the movement end point and Csépe V, Czigler I, Desmedt JE, eds), pp 261–272. Amsterdam: detecting discrepancies between the expected and the predicted Elsevier. Gemba H, Sasaki K, Brooks VB (1986) “Error” potentials in limbic position. The error signal is sent to the cerebellum, which in turn cortex (anterior cingulate area 24) of monkeys during motor learning. corrects ongoing activity in the primary motor cortex. Neurosci Lett 70:223–227. This specific on-line error-correction mechanism differs in Goodale MA, Pelisson D, Prablanc C (1986) Large adjustments in visu- ally guided reaching do not depend on vision of the hand or perception many aspects from the one studied here, both with regard to its of target displacement. Nature 320:748 –750. functional characteristics and to the brain systems involved. Gratton G, Coles MGH, Sirevaag EJ, Eriksen CW, Donchin E (1988) Pre- and poststimulus activation of response channels: a psychophysio- Whereas on-line corrections in the double-step paradigm are logical analysis. J Exp Psychol Hum Percept Perform 14:331–344. performed even if the target displacement is not consciously Kutas M, Donchin E (1980) Preparation to respond as manifested by detected (Prablanc and Martin, 1992), the error-correction mech- movement-related brain potentials. Brain Res 202:95–115. Luu P, Tucker DM (2001) Regulating action: alternating activation of anism studied in the present investigation is related to conscious midline frontal and motor cortical networks. Clin Neurophysiol detection and the intentional release of a corrective motor com- 112:1295–1306. Luu P, Flaisch T, Tucker DM (2000) Medial frontal cortex in action mand using the contralateral limb. The anterior cingulate cortex monitoring. J Neurosci 20:464 – 469. is clearly involved in the type of paradigm used in the present McCarthy G, Wood CC (1985) Scalp distributions of event-related po- 9996 J. Neurosci., November 15, 2002, 22(22):9990–9996 Rodrı́guez-Fornells et al. Time Course of Error Correction tentials: an ambiguity associated with analysis of variance models. Rabbitt PMA (1968) Three kinds of error signalling responses in a serial Electroencephalogr Clin Neurophysiol 62:203–208. choice task. Q J Exp Psychol 21:159 –170. Münte TF, Urbach TP, Düzel E, Kutas M (2000). Event-related brain Scheffers MK, Coles MG (2000) Performance monitoring in a confusing potentials in the study of human cognition and neuropsychology. In: world: error-related brain activity, judgements of response accuracy, Handbook of neuropsychology, Vol 1 (Boller F, Grafmann J, Rizolatti and types of errors. J Exp Psychol Hum Percept Perform 26:141–151. G, eds), pp 139 –235. Amsterdam: Elsevier. Schmitt BM, Münte TF, Kutas M (2000) Electrophysiological estimates Paus T (2001) Primate anterior cingulate cortex: where motor control, of the time course of semantic and phonological encoding during drive and cognition interface. Nat Rev Neurosci 2:417– 424. implicit picture naming. Psychophysiology 37:473– 484. Pisella L, Grea H, Tilikete C, Vighetto A, Desmurget M, Rode G, Boisson Shadmehr R, Mussa-Ivaldi FA (1994) Adaptative representation of dy- namics during learning of a motor task. J Neurosci 14:3208 –3224. D, Rossetti Y (2000) An “automatic pilot” for the hand in human Smid HGOM, Mulder G, Mulder LJM, Brands GJ (1992) A psychophys- posterior parietal cortex: toward reinterpreting optic ataxia. Nat Neu- iological study of the use of partial information in stimulus-response rosci 3:729 –736. translation. J Exp Psychol Hum Percept Perform 18:1101–1119. Prablanc C, Martin O (1992) Automatic control during hand reaching at van Veen V, Cohen JD, Botvinick MM, Stenger VA, Carter CS (2001) undetected two-dimensional target displacements. J Neurophysiol Anterior cingulate cortex, conflict monitoring, and levels of processing. 67:455– 469. NeuroImage 14:1302–1308. Rabbitt PMA (1966a) Error correction time without external error sig- Vidal F, Hasbroucq T, Grapperon J, Bonet M (2000) Is the “error nals. Nature 212:438. negativity” specific to errors? Biol Psychol 52:109 –128. Rabbitt PMA (1966b) Errors and error correction in choice-response Wolpert DM, Ghahramani Z, Jordan MI (1995) An internal model for tasks. J Exp Psychol 71:264 –272. sensorimotor integration. Science 269:1880 –1882.

Time Course of Error Detection and Correction in Humans PDF

Document Details

Tags

Related

Summary

Full Transcript

Upgrade to continue