The role of cingulate cortex in the detection of errors with and without awareness: a high-density electrical mapping study.

Abstract

Error-processing research has demonstrated that the brain uses a specialized neural network to detect errors during task performance but the brain regions necessary for conscious awareness of an error are poorly understood. In the present study we show that two well-known error-related event-related potential (ERP) components, the error-related negativity (ERN) and error positivity (Pe) have a differential relationship with awareness during performance of a manual response inhibition task optimized to examine error awareness. While the ERN was unaffected by the participants' conscious experience of errors, the Pe was only seen when participants were aware of committing an error. Source localization of these components indicated that the ERN was generated by a caudal region of the anterior cingulate cortex (ACC) while the Pe was associated with contributions from a more anterior ACC region and the posterior cingulate-precuneus. Tonic EEG measures of cortical arousal were correlated with individual rates of error awareness and showed a specific relationship with the amplitude of the Pe. The latter finding is consistent with evidence that the Pe represents a P3-like facilitation of information processing modulated by subcortical arousal systems. Our data suggest that the ACC might participate in both preconscious and conscious error detection and that cortical arousal provides a necessary setting condition for error awareness. These findings may be particularly important in the context of clinical studies in which a proper understanding of self-monitoring deficits requires an explicit measurement of error awareness.

Full text

Page 1

The role of cingulate cortex in the detection of errors with
and without awareness: a high-density electrical mapping
study
Redmond G. O’Connell,1,2Paul M. Dockree,1,2Mark A. Bellgrove,3Simon P. Kelly,2,4Robert Hester,3Hugh Garavan,1
Ian H. Robertson1and John J. Foxe1,2,4,5
1Department of Psychology and Trinity College Institute of Neuroscience, Trinity College Dublin, Dublin 2, Ireland
2The Cognitive Neurophysiology Laboratory, St Vincents Hospital, Fairview, Dublin 3, Ireland
3School of Behavioural Sciences, University of Melbourne, Victoria, Australia
4The Cognitive Neurophysiology Laboratory, Nathan S. Kline Institute for Psychiatric Research, Orangeburg, New York, NY 10962,
USA
5Program in Cognitive Neuroscience, Department of Psychology, The City College of the City University of New York, 138th St. &
Convent Avenue, New York, NY 10031, USA
Keywords: cortical arousal, ERP, error positivity, error processing, error-related negativity
Abstract
Error-processing research has demonstrated that the brain uses a specialized neural network to detect errors during task
performance but the brain regions necessary for conscious awareness of an error are poorly understood. In the present study we
show that two well-known error-related event-related potential (ERP) components, the error-related negativity (ERN) and error
positivity (Pe) have a differential relationship with awareness during performance of a manual response inhibition task optimized to
examine error awareness. While the ERN was unaffected by the participants’ conscious experience of errors, the Pe was only seen
when participants were aware of committing an error. Source localization of these components indicated that the ERN was generated
by a caudal region of the anterior cingulate cortex (ACC) while the Pe was associated with contributions from a more anterior ACC
region and the posterior cingulate–precuneus. Tonic EEG measures of cortical arousal were correlated with individual rates of error
awareness and showed a specific relationship with the amplitude of the Pe. The latter finding is consistent with evidence that the Pe
represents a P3-like facilitation of information processing modulated by subcortical arousal systems. Our data suggest that the ACC
might participate in both preconscious and conscious error detection and that cortical arousal provides a necessary setting condition
for error awareness. These findings may be particularly important in the context of clinical studies in which a proper understanding of
self-monitoring deficits requires an explicit measurement of error awareness.
Introduction
Since the initial behavioural work of Rabbitt (1966), understanding
how the brain detects and processes performance failures has been a
major challenge for cognitive neuroscience. A clear message emerging
from this literature is that error processing represents a discrete
component of a broader action-monitoring system and is reliant on a
specialized brain network. Functional imaging and event-related
potential (ERP) studies have consistently pointed to the anterior
cingulate cortex (ACC) as a critical brain region responsible for the
initial detection of unfavourable performance outcomes (Ridderinkhof
et al., 2004; Kennerley et al., 2006). Importantly, however, investi-
gators have not typically made the distinction between error detection
and conscious error awareness. As a result, it is often not clear how the
error-related brain activations that have been demonstrated actually
relate to the conscious experience of making an error. The present
study aims to identify some of the key electrophysiological markers
for error awareness.
The first of only two ERP studies to explicitly measure error
awareness was conducted by Nieuwenhuis and colleagues (Nieu-
wenhuis et al., 2001) who focused on two components commonly
seen in response-locked ERP waves that have been time-locked to
erroneous responses; the error-related negativity (ERN) and the error
positivity (Pe). The early onset of the ERN, before the erroneous
response has been completed, is suggestive of a rapid internal
detection mechanism that is not dependent on conscious processing of
the error (Falkenstein et al., 2000). Rather than directly detecting
errors, the ERN may reflect monitoring processes that are sensitive to
response conflict (Van Veen & Carter, 2002) or changes in reward
probability (Holroyd et al., 2004). Several studies have shown that
ERN amplitude predicts short-term posterror compensatory adjust-
ments (Gehring et al., 1993; Rodriguez-Fornells et al., 2002; Debener
et al., 2006). In contrast, the functional significance of the Pe, maximal
over parietocentral scalp sites, is poorly understood but this compo-
nent peaks sufficiently late (300–500 ms posterror) for sensory or
proprioceptive information to be available and is therefore more likely
to index conscious aspects of error processing. Indeed, Nieuwenhuis
et al. (2001), who asked participants to perform an antisaccade task,
found that consciously perceived errors elicited far larger Pe
Correspondence: Dr R. G. O’Connell, Trinity College Institute of Neuroscience, as above.
E-mail: reoconne@tcd.ie
Received 4 September 2006, revised 7 February 2007, accepted 12 February 2007
European Journal of Neuroscience, Vol. 25, pp. 2571–2579, 2007
doi:10.1111/j.1460-9568.2007.05477.x
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd

Page 2

amplitudes than unperceived errors while the ERN remained unaffec-
ted. This finding, which was replicated more recently by Endrass et al.
(2005), provided the first evidence that the Pe, and not the ERN,
specifically reflects conscious aspects of error processing. However,
the generalisability of this relationship beyond the occulomotor
modality has yet to be verified.
Source localization studies of the ERN and Pe have confirmed the
critical role of the ACC in error detection by identifying generators in
this region for both the ERN and the Pe. The ERN has been
consistently localized to caudal regions of the ACC thought to govern
performance monitoring processes such as conflict detection and
interference control (Dehaene et al., 1994; Van Veen & Carter, 2002;
Hermann et al., 2004; Van Boxtel et al., 2005). Attempts to identify
the precise ACC generator for the Pe have been less consistent. While
Van Veen & Carter (2002) and Van Boxtel et al. (2005) located the
source of the Pe in a rostral portion of the ACC associated with
motivation and affective processing (Bush et al., 2000; Taylor et al.,
2006), the source identified by Hermann et al. (2004) was located in a
far more caudal region. Interestingly, in the only fMRI study to
explicitly distinguish between consciously perceived and unperceived
errors (Hester et al., 2005) there was no additional ACC activity when
participants were aware of their errors. This finding appears to be at
odds with data suggesting an ACC source for the Pe. Thus the neural
processes necessary for conscious error awareness have not yet been
clearly identified.
The present study was designed to shed further light on the neural
substrates of error awareness by addressing three key questions. First,
given that error-related brain activity has primarily been investigated
while participants performed manual response tasks the present study
aimed to verify the generalisability the findings of Nieuwenhuis et al.
(2001) beyond the oculomotor modality using the same manual
response task developed by Hester et al. (2005). A recent study has
indicated that consciously detected errors may also be distinguished
by the presence of a strong autonomic reaction (O’Keeffe et al., 2004).
Here, measures of autonomic system activity are acquired to further
validate the sensitivity of this task to error awareness. Second, in order
to further explore the role of the ACC in error awareness the present
study sought to localize the neural generators of the ERN and Pe while
distinguishing between errors made with and without conscious
awareness. Finally, a limitation of purely event-related approaches to
error processing is that we can learn little about potentially critical
brain states that could provide the setting conditions for conscious
awareness. Research has previously demonstrated a clear association
between alertness deficits and failures in conscious error detection in
brain-injured populations (McAvinue et al., 2005). In addition, several
authors have suggested that the Pe may be part of the same evaluative
process as the stimulus-locked P3 (Nieuwenhuis et al., 2001;
Overbeek et al., 2005), a component which is reliably modulated by
arousal (Nieuwenhuis et al., 2005). A hypothesis that emerges from
this work is that one’s basal level of cortical arousal, as measured by
the ratio of slow (alpha and theta) to fast (beta) wave oscillations
across the task duration, will be associated with the extent of
conscious error processing. Therefore the third aim of this study was
to investigate the relationship between tonic electroencephalogram
(EEG) measures of cortical arousal and error awareness.
Materials and methods
Subjects
Nineteen (one female, one left-handed) neurologically normal volun-
teers participated. Participants received a gratuity of €32 to cover
expenses incurred on the day of testing. Subjects were aged between
18 and 30 years (mean ± SD age 22.07 ± 2.85 years). All participants
gave written informed consent and all procedures were approved by
the ethical review boards of St Vincents Hospital, Fairview, and
Trinity College Dublin. All subjects reported normal or corrected-to-
normal vision. Ethical guidelines were in accordance with the
Declaration of Helsinki.
Error awareness task (EAT) and procedure
We used an error awareness paradigm developed by Hester et al.
(2005). The EAT (see Fig. 1) is a motor Go⁄No-go response inhibition
task in which participants are presented with a serial stream of single
colour words with congruency between the word and its font colour
manipulated. Subjects were trained to respond to each of the words
with a single ‘Go trial’ button press and to withhold this response
when either of two different circumstances arose. The first circum-
stance was if the same word was presented on two consecutive trials
(Repeat No-go), and the second was if the word and its font colour did
not match (Incongruent No-go). In the event of a commission error
Incongruent No-Go
Withhold
response
Repeat No-Go
Withhold
response
Go Trials
Button press
Fig. 1. The error awareness task required subjects to respond with a button press to a stream of colour words and withhold their response when either a word was
repeated on consecutive trials or the font and word were incongruous. Subjects were trained to press a different button following any commission errors.
2572 R. G. O’Connell et al.
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579

Page 3

(failure to withhold to either of these No-go scenarios) subjects were
trained to press a second ‘awareness button’ on the next Go trial after
the error and were not required to make the standard Go response.
Participants were instructed to time their button presses to the offset
(i.e. end) of each stimulus. This kind of ‘response-locking’ has been
shown to reduce interindividual variability and eliminate speed–
accuracy tradeoffs (Stuss et al., 2003). In the present experiment
response-locking allowed us to rule out the possibility that certain
undetected errors could be attributed to an overemphasis on speed
over accuracy.
To maximize the number of errors for ERP averaging participants
completed an average of 11.2 blocks of the EAT (range 8–14). Each
EAT block consisted of 225 stimuli of which 200 were Go stimuli and
25 were No-go stimuli (of which 13 were Repeat No-gos and 12 were
Incongruent No-gos or vice versa). All stimuli were presented for
600 ms followed by an interstimulus interval of 900 ms and appeared
0.25? above a white fixation cross on a grey background at a distance
of ?150 cm. Participants were instructed to focus on the fixation
cross during the task in order to minimize eye movements.
Electrodermal activity (EDA) acquisition and analysis
EDA is recorded as changes in electrical conductance due to sweat
gland activity and is controlled by sympathetic innervation of the
autonomic nervous system. EDA has been commonly used as an index
of the psychological processing of stimulus properties such as
significance, novelty, emotional relevance and effortful processing.
Previous work has demonstrated that EDA responses are absent when
participants are not aware that they have made an error (O’Keeffe
et al., 2004). In the present study we acquire EDA as a further means
of verifying that errors after which there was no ‘awareness’ response
were not associated with significant amounts of conscious processing.
EDA measurements were taken from all participants during EAT
testing with a five-channel BIOPAC MP30B unit, calibrated to skin
conductance responses (SCRs) in microsiemens (lS). Two Ag⁄AgCl
BIOPAC electrodes, with contact areas of ?6 mm, were filled with
SIGNA electrode gel and secured with a velcro strap to the volar
surface of the distal phalanges of the index and middle fingers of the
participant’s nondominant hand. After a 5-min rest period to ensure
skin hydration by the gel, the BIOPAC software was calibrated to the
participants’ own electrodermal parameters before EAT testing
began.
EDA data were analysed using Matlab 6.1 according to previously
established criteria (Dawson et al., 2000). A rise in skin conductance
level (SCL) was considered to be a response (SCR) if its onset was
between 0.5 and 4.5 s after a particular event (presentation of No-go
stimulus or alert). SCRs were measured by subtracting the SCL at
stimulus onset from the peak SCL within the latency period. The
criterion for the smallest acceptable SCR was set at 0.02 lS. Any trials
on which the SCL exceeded a response threshold of 0.02 lS in the
first 0.5 s after the event were thus rejected. The SCR for a given
accepted trial was measured by taking the maximum within the
interval 0.5–4.5 s and subtracting the nearest preceding local
minimum within that interval. For the plotting of EDA waveforms,
the SCL time courses were simply averaged across all trials for each
condition (see Fig. 2).
EEG data acquisition and statistical analysis
Continuous EEG was acquired through the ActiveTwo Biosemi?
electrode system from 72 scalp electrodes, digitized at 512 Hz.
Vertical eye movements were recorded with two vertical (V)
electrooculogram (EOG) electrodes placed below the left and right
eye, while HEOG electrodes at the outer canthus of each eye recorded
horizontal movements.
Data were analysed using BESAVersion 5.1 (Brain Electric Source
Analysis) software (http://www.besa.de). For analysis and display
purposes, data were average-referenced and filtered with a low-pass
0-phase shift 48 dB 30 Hz filter after acquisition. Response-locked
data were segmented into epochs of 400 ms before to 500 ms after
button press and were baseline-corrected relative to the interval )400
to )200 ms. All electrode channels were subjected to an artifact
criterion of ±100 lV from )400 to +500 ms to reject trials with
excessive electromyogram or other noise transients. The single-trial
EEG signals were also corrected for horizontal and vertical EOG
artefacts by means of an eye movement correction procedure
developed by Berg & Scherg (1994). Accepted trials were then
averaged separately for correct Go presses and commission errors
(Repeat and Incongruent No-gos) after which the participants
indicated awareness (Aware Error) and commission errors after which
the participants did not indicate awareness (Unaware Error). Unaware
Errors were rejected if the participant failed to make any response on
the next Go trial or if they pressed the ‘awareness’ button within three
trials of the No-go error.
Inspection of the grand-average waveforms revealed a clearly
defined ERN following both commission error types peaking at
?80 ms after the button press. A smaller ERN-like component was
also evident at the same latency after correct Go presses. For all
three conditions the maximal peak amplitude of the ERN was seen
at FCz with little topographical variation across participants. The
ERN was therefore defined as the most negative peak at FCz
occurring in a window from 50 to 120 ms postresponse. In all three
response conditions (correct Go, aware error and unaware error) the
ERN was immediately followed by a positive deflection with
maximalamplitudealso atFCz
postresponse. On aware errors this early positive peak was followed
by the classic Pe component in the form of a large positive wave
over posterior scalp regions and maximal amplitude at CPz. The
early positive component was calculated for all three response
conditions as the most positive peak at FCz between 140 and
240 ms postresponse. Because the Pe is a more sustained low-
frequency component the mean amplitude at CPz between 300 and
500 ms postresponse was used.
Finally, the average power spectrum over the first eight blocks of
EAT testing was calculated for each participant using the discrete
Fourier transform (eight blocks was the minimum number of EAT
blocks that were completed by all participants). Each participant’s
tonic theta, alpha and beta powers (lV2) were calculated as the power
in the 4–7, 8–12 and 13–29 Hz ranges, respectively. Theta⁄beta and
alpha⁄beta ratios were subsequently calculated.
The behavioural and EEG measures reported were calculated
over the first eight blocks for all participants. Due to the low
number of unaware error trials, ERP averages were calculated for
the full number of blocks completed by each participant (between 8
and 14) to maximize the number of single trials included in each
average. Only those participants who made at least 20 aware and
20 unaware errors were included in the ERP analysis. This led to a
reduced sample of 12 participants who made an average of 76
aware errors and 35 unaware errors. Because of the relatively low
data yield for unaware errors it was important to equate the
number of single trials that contributed to the averages for these
comparisons. Consequently half the aware errors were removed by
selecting only even matches to ensure that they were not
anda peakat
?190 ms
Error awareness2573
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579

Page 4

overrepresented either quantitatively or temporally in these com-
parisons.
Mean values are quoted ±SD.
Results
Behavioural performance
Participants successfully withheld their response on 69.3% of No-go
targets. A significantly greater proportion of total commission errors
occurred on Incongruent No-gos (61%) than on Repeat No-gos
(39%; t18¼ 5.69, P < 0.001). Participants were aware of 75.8% of
all commission errors (aware errors⁄total errors). Although more
errors were made on Incongruent No-go targets, participants were
significantly less likely to be consciously aware of errors on Repeat
No-gos (Repeat No-gos, 67.7%;
t18¼ 3.301, P < 0.01). Mean reaction time (RT) to Go stimuli
was 625 ± 96.87 ms, indicating that participants were successfully
timing their responses to stimulus-end. Participants had significantly
longer RTs for unaware errors (mean RT unaware, 676.3 ±
112.8 ms) than for aware errors (mean RT aware, 593.4 ±
119.3 ms; t18¼ )4.0, P < 0.01).
Incongruent No-Gos,82%;
Modulation of EDA by awareness
As Fig. 2 illustrates, strong SCRs were elicited by aware errors
(0.243 ± 0.19 lS) and correct withholds (0.203 ± 0.16 lS) on No-go
Trials. In contrast, this response was absent following unaware errors
(0.080 ± 0.083 lS). An anova on the magnitude of SCRs for the
three No-go conditions (aware error, unaware error and correct
withhold) revealed a significant main effect of No-go condition
(F1,11¼ 7.96, P < 0.01). Post hoc Bonferonni tests indicated that
Correct Withhold
Aware Error
Unaware Error
Time, secs
µS
Fig. 2. Skin conductance responses (SCR) averaged separately for aware
errors, unaware errors and correct withholds on No-go trials and time-locked to
stimulus end (time-point 0). Here we see that, as expected, unaware errors did
not elicit the autonomic response typically seen following significant events
such as errors, confirming that conscious processes were not active.
ms
V
µ
-2
0
2
4
6
Cz
ms
V
µ
-2
0
2
4
6
Pe
CPz
ms
-2000
200400
-200
0
200
400
-200
0
200400
V
µ
-2
0
2
4
6
ERN
FCz
Early Positivity
Aware Error
Unaware Error
Correct go
(b)
(a)
(c)
Fig. 3. Grand-average ERP waveforms at (a) FCz, (b) Cz and (c) CPz time-locked to button press (time-point 0). Panel (a) shows that the ERN and early
positivity were not modulated by participants’ awareness of errors. In panel (c) we note the large Pe wave following aware errors that is absent on unaware errors and
correct-Go presses.
2574 R. G. O’Connell et al.
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579

Page 5

participants had similar SCRs following aware errors and correct
withholds (P ¼ 0.17) but SCRs following unaware errors were
significantly smaller than those elicited by aware errors (P < 0.01)
or correct withholds (P < 0.05). These data indicate that the
cognitive–emotional processes indexed by EDA were absent on
unaware errors.
Modulation of error-related ERPs by awareness
An anova on the ERN amplitudes for the three Response conditions
(see Fig. 3a) indicated a significant main effect of Response (mean
amplitudes:aware error,
)2.7 ± 1.5 lV;
)2.9 ± 1.4 lV; correct go, )1.9 ± 1.2 lV; F2,22¼ 4.6, P < 0.05).
Tests of within-subjects contrasts revealed that the ERN was
significantly larger following both error types than following correct
Go presses (Go vs. Aware F1,11¼ 6.37, P < 0.05; Go vs. Unaware
F1,11¼ 6.407, P < 0.05) but importantly there was no difference
between ERN amplitude for aware and unaware errors (F1,11¼ 0.015,
P ¼ 0.872).
An anova on the early positive peak immediately following
the ERNrevealed no reliable
(meanamplitudes:aware error,
3.78 ± 2 lV; correct go, 3.0 ± 1.2 lV; F2,22¼ 2.88, P ¼ 0.083;
see Fig. 3a).
Finally, an anova on the Pe amplitudes revealed a significant
maineffect ofResponse(mean
4.9 ± 1.2 lV; unaware error, 0.5 ± 0.13 lV; correct Go, 0.14 ±
0.7 lV; F2,22¼ 68.73, P < 0.001). Tests of within-subjects contrasts
confirmed that this late positive wave was significantly smaller on
correct Go presses and unaware errors than on aware errors (correct
Go vs. aware F1,11¼ 112.87, P < 0.001; unaware vs. aware
F1,11¼ 72.94, P < 0.001). Unaware error and correct Go ERPs
did not differ in terms of this late positivity (F1,11¼ 0.025,
P ¼ 0.877; see Fig. 3c).
unawareerror,
differences
4.4 ± 2 lV;
across
unaware
conditions
error,
amplitudes: awareerror,
Source localization of error-related ERPs
Strong ERN and early positive waves were evident in all three
response conditions including correct Go presses, and their scalp
topographies did not vary (see Fig. 4a). Due to these similarities,
source localization of the ERN and early positivity was conducted on
aware error ERPs instead of a difference waveform. Source localiza-
tion was implemented by BESA 5.1 using a four-shell spherical head
model approximation. First the ERN was selected by highlighting a
20-ms interval around its negative peak. A single dipole model located
in the ACC accounted for most of the variance in the ERN
(x ¼ )14.7, y ¼ 0.1, z ¼ 45 mm; Residual Variance (R.V.) ¼ 6.9%,
Best (minimum variance within the fit interval) ¼ 5.6%). When
selecting a 20-ms interval around the peak of the early positivity the
same ACC dipole accounted for most of the variance in this
component (R.V. ¼ 9.2%, Best ¼ 7.9%; see Fig. 4b). Keeping the
location of the ERN source constant, free-fit source localization was
performed on the 20-ms interval around the peak of the error
positivity. The resulting additional dipole did not explain a significant
amount of variance in the early positivity (change in R.V. < 1%).
Finally, the ERN source was removed and a broader interval extending
from 20 ms prior to the peak of the ERN to 20 ms after the peak of the
early positivity was selected. Again, the free-fit algorithm indicated a
generator in a very similar caudal region of the ACC (x ¼ )16.3,
y ¼ )13.6, z ¼ 50.6 mm). Although the ERN and early positivity
share similar frontocentral scalp distributions, the early positivity does
appear to be distributed more centrally (see Fig. 4a). It is possible
therefore that these two components are generated by areas of the
ACC that are very close together and that the spatial resolution of
source localization was insufficient to prise them apart.
Keeping the location and orientation of the ERN and early positivity
dipole constant, source localization was then performed on the Pe
between 300 and 500 ms after response. A two-source model was
indicated for the Pe (R.V. 3.1%, Best 2.1%) with one dipole located in
the ACC but more anterior to the ERN and early positivity dipole
(x ¼ 2.9, y ¼ 20.5, z ¼ 42.5) and the other located in the vicinity of
the posterior cingulate cortex and precuneus (x ¼ )4.5, y ¼ )37.6,
z ¼ 39.9). Given the proximity of the two ACC dipoles a further test
was conducted to verify the accuracy of this source model. The grand
average three-dipole model and a simpler, two-dipole, model which
did not include the anterior ACC dipole, were applied to individual
subject ERPs. Removal of the anterior ACC dipole caused an increase
of 8.9% in the average R.V. (three dipoles: mean R.V., 16.2%; range,
8.6–30.5; two dipoles: mean R.V., 25.1%, range, 10.4–57.5). An
additional source analysis of the Pe was also conducted on a difference
waveform subtracting the unaware-error waveform from the aware-
error waveform, thus isolating activity specifically associated with
conscious error perception. This analysis revealed a very similar two-
source Pe model with one dipole located in the same ACC region
(x ¼ 6.9, y ¼ 13.1, z ¼ 46.5) and another located around the
posterior cingulate cortex and precuneus, though in a slightly more
inferior location than that indicated by the first solution (x ¼ )5,
y ¼ )59.1, z ¼ 20.7). The dipole models and source waveforms for
the aware error and difference ERPs are compared in Fig. 4b and c.
Tonic cortical arousal and error awareness
Finally, we investigated whether there was a relationship between
tonic slow⁄fast wave ratios in the EEG power spectrum and errors of
commission, error awareness and the amplitudes of the three error-
related components (ERN, early positivity and Pe) on aware errors.
Fifteen of our 19 participants made sufficient aware errors (at least 20)
to be included in the correlation between the ERPs and slow⁄fast
wave ratios but the remaining correlations included all 19 participants.
A significant relationship was observed between tonic theta⁄beta
ratios and the percentage of aware errors (aware errors⁄total errors:
r ¼ –0.478, P < 0.05) and the amplitude of the Pe (r ¼ –0.661,
P < 0.01). Alpha⁄beta ratios were also correlated with Pe amplitude
(r ¼ –0.546, P < 0.05) and there was a close-to-significant relation-
ship with percentage of aware errors (r ¼ –0.44, P ¼ 0.059). Thus a
low ratio of slow⁄fast wave activity in the EEG spectrum (indicating
increased cortical arousal) was associated with better awareness of
one’s errors and a larger Pe amplitude. No significant correlations were
observed between theta⁄beta or alpha⁄beta ratios and errors of
commission or the amplitudes of the ERN or early positivity.
Discussion
In the present study, participants performed a Go⁄No-go response task
and were asked to indicate any commission errors by pressing a
second ‘awareness’ button. Although this requirement imposes a dual-
task element which could conceivably contaminate ‘unaware’ errors
with dual-task failures we argue that error awareness cannot be
explicitly verified without requiring a secondary response. Moreover,
the EDA data (Fig. 2) indicates a marked absence of the cognitive–
emotional response that is seen following conscious recognition of
Error awareness 2575
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579

Page 6

significant events such as action errors (O’Keeffe et al., 2004),
suggesting that the EAT does provide a good measure of error
awareness. It has been suggested that autonomic changes provide
‘somatic markers’ that guide our behaviour and may also contribute to
the emergence of conscious awareness (Damassio et al., 1991). An
interesting question for future work would be to investigate the
temporal relationship between the onset of autonomic responses and
behavioural indices of error awareness.
The behavioural data indicated that, while participants made more
commission errors on Incongruent trials, a significantly greater
proportion of unaware errors were made on Repeat trials. This
difference may be explained by the fact that when participants make
an error on Incongruent trials, the current stimulus can be immediately
identified as a No-go target. In contrast, when an error is made on a
Repeat trial the participant must have a memory of the preceding trial
in order to identify the current stimulus as a No-go target. Given that
participants appear to have found it easier to withhold on Repeat No-
go trials the increased rate of unaware errors suggests that failed
inhibitions on Repeat trials may have been accompanied by a more
dramatic lapse of attention than errors made on Incongruent trials. In
addition, it was found that response times were significantly slower on
unaware errors than on aware errors. This finding is similar to that
reported by Endrass et al. (2005) and also appears consistent with the
view that unaware errors are precipitated by lapses of attention.
ms
-40
-20
0
20
40
60
Anterior ACC
Posterior Cingulate
(c) Difference Waveform Sources
ERN
Early
Positivity
Pe
(a) Scalp topographies
ms
-200
0200
400
-200
0
200
400
-40
-20
0
20
40
60
Caudal ACC
Anterior ACC
Posterior Cingulate
nAmp nAmp
(b) Aware Error Sources
Fig. 4. ERP scalp topographies (a), source time courses and dipole locations for the aware error (b) and difference waveform models (c). The caudal ACC dipole
(red) explains the ERN and the early positivity; the anterior ACC dipole (green) and the posterior cingulate–precuneus dipole [brown in (b), blue in (c)] accounts for
the Pe.
2576R. G. O’Connell et al.
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579

Page 7

When participants made an error of commission a response-locked
waveform was generated which exhibited three clear components. The
ERN was the earliest component and its amplitude was not modified
by awareness of errors. Although the ERN was most pronounced for
errors, an attenuated ERN-like component was also evident following
correct Go responses. This phenomenon, known as the ‘correct
response negativity’, is thought to reflect levels of continuous
performance monitoring during the task (Vidal et al., 2000; Rid-
derinkhof et al., 2003). Thus, while the processes reflected in the ERN
are enhanced when one makes an error, they are also engaged in an
ongoing manner throughout task performance. The latter probably
facilitates the detection of subthreshold levels of conflict or uncer-
tainty that do not necessarily result in errors but signal the need for
fine-grained performance adjustments. The second component noted
in the present study was a strong early positive deflection immediately
following the ERN and maximal over frontocentral scalp sites. A
similar positive deflection has been noted in previous studies and
previous analyses using prestimulus baselines and analyses within the
frequency domain have failed to distinguish it from the ERN (Luu
et al., 2004; Debener et al., 2006). In addition, a previous study has
reported that the ERN and early positivity share the same cortical
generator (caudal ACC), leading to the suggestion that they are part of
a single performance-monitoring component (Van Veen & Carter,
2002). Interestingly, in the present study the amplitude of this early
positive wave did not distinguish between any of the response types
(correct Go, aware error, unaware error). The strong ERN and early
positive waveforms on correct trials may be explained by the
introduction of two competing No-go conditions in the EAT task
(Repeat and Incongruent No-gos), resulting in a higher degree of
uncertainty on Go trials and therefore increased engagement of
monitoring processes.
In contrast, the classic Pe, which followed the early positive wave
and had a more posterior distribution, was only present when
participants were aware of their errors. The marked absence of a Pe
when participants were not aware of their errors provides clear
confirmation of the findings of Nieuwenhuis et al. (2001) and
therefore we have shown that modulation of the Pe by error awareness
is not limited to the oculomotor modality but generalizes to manual
responses. The presence of the ERN and early positivity whether or
not participants were aware that they had made an error indicates that
the brain possesses preconscious detection mechanisms while later
processing stages indexed by the Pe only occur when an error has been
consciously detected.
Although there is now a wealth of evidence implicating the ACC in
error processing, and in the generation of the ERN and Pe, there has
been little investigation of the role that this region plays in error
awareness. In the present study we conducted the first source analysis
of electrophysiological activity that was specific to errors made with
and without awareness. Our results appear to confirm a role for the
caudal ACC in early aspects of error processing reflected in both the
ERN and early positivity and in this respect confirm previous reports
(Dehaene et al., 1994; Van Veen & Carter, 2002; Hermann et al.,
2004; Van Boxtel et al., 2005; Debener et al., 2006). In common with
two previous studies, we found that the ERN and early positivity were
generated by the same ACC region (Van Veen & Carter, 2002;
Debener et al., 2006), thus supporting the contention that they are part
of the same monitoring component. However, the early positivity did
have a more central scalp distribution than the ERN, suggesting that it
may be generated by a distinct region of the ACC. Further work will
be required to specify the functional significance of this component.
Source localization of the Pe also indicated an ACC generator but in a
distinctly more anterior region. Our Pe model also indicated a second
source around the posterior cingulate–precuneus, a region which has
been previously attributed a role in posterror processing (Badgaiyan &
Posner, 1998; Menon et al., 2001) and more broadly in self-awareness
and consciousness (Cavanna & Trimble, 2006). The same regions
were indicated when the unaware error waveform was subtracted from
the aware error waveform leaving only activity relating to error
awareness.
There have been four previous attempts to localize the generator(s)
of the ERN and Pe simultaneously. Source analyses performed by Van
Veen & Carter (2002) and by Van Boxtel et al. (2005) also identified
an ACC generator for the Pe although in a more rostral region than in
the present study. Hermann et al. (2004) also found separate ACC
sources for the ERN and Pe but the source of the Pe was found in a
more caudal ACC region. Imaging and lesion studies indicate that the
ACC can be broadly divided into two functionally distinct subregions,
a caudal region associated with basic cognitive processes such as
conflict and uncertainty detection and a more rostral region thought to
process the subjective or emotional significance of events and stimuli
(Bush et al., 2000). A recent study by Taylor et al. (2006) has noted
clear individual differences in the ACC subregions activated during
error processing. Individual variations in affective and motivational
responsiveness to making an error may explain in part why there has
been a certain degree of inconsistency in localizing the source of the
Pe. It is also important to acknowledge that variation across studies
may also arise from the limited spatial resolution of source localization
associated with the inverse problem. Nevertheless the findings of the
present study are consistent with those of Van Veen & Carter (2002)
and of Van Boxtel et al., 2005) such that the ERN and Pe do appear to
be generated by distinct regions within the ACC. Understanding the
neural basis of conscious error detection may provide an important
avenue to a better understanding of the failures of self-awareness seen
in clinical groups in which reduced awareness of one’s deficits can
represent a significant barrier to rehabilitative interventions and
independent living (Hart et al., 1998).
The source model obtained in the present study does appear to be at
odds with the fMRI findings of Hester et al. (2005), who found no
additional ACC or posterior cingulate–precuneus activations when
participants were aware of their errors. Such differences are not
necessarily surprising given the ability of ERPs to separate minute
portions of trial activity. The comparatively limited temporal resolu-
tion of fMRI may cause subtle effects, particular to finite portions of
processing within a trial, to go undetected. For example, the subtle
increases in regional ACC activity that are evident in source analysis
might not be detected by fMRI if averaging across a longer epoch
causes rostral and caudal activations to become convolved. Never-
theless, caution should be exercised when interpreting the present
source findings. These data highlight the potential pitfalls in making
direct comparisons between different measures across studies. More
direct evidence that the ERN is generated by the ACC has been
provided through trial-by-trial couplings of EEG and fMRI signals
(Debener et al., 2006) and through intracerebral recording studies
(Brazdil et al., 2005). Further work of this kind will be required in
order to elucidate the discrepancy between fMRI correlates of
conscious error processing and repeated attempts to localize the
generators of the Pe through source analysis.
It is important to note that the observed modulatory role of
awareness does not necessarily imply causality in the generation of the
Pe. Our results are also consistent with alternative accounts of the Pe
as a reflection of compensatory adjustments (Hajcak et al., 2003) or
subjective and⁄or emotional appraisal (Van Veen & Carter, 2002) but
suggest that these processes may be reliant on conscious awareness.
The finding that the Pe was correlated with tonic EEG measures of
Error awareness 2577
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579

Page 8

arousal may be particularly illuminating in the context of recent
models linking the Pe to the same underlying process as the P3, a
component which can be elicited by any motivationally significant
stimulus and which is thought to reflect the evaluation of that stimulus
(Nieuwenhuis et al., 2005; Overbeek et al., 2005). The P3 and Pe
share several obvious characteristics including centroparietal scalp
topography, positive polarity and relative peak latency (300–500 ms
relative to stimulus and response onset, respectively). These similar-
ities have led to speculation that the two components may be part of
the same process (Falkenstein et al., 2000; Nieuwenhuis et al., 2005)
whereby the Pe would reflect an additional P3-like evaluation of the
incorrect response. This possibility raises a number of interesting and
testable questions. In the present paper, we examined the influence of
arousal on the Pe.
Recently, Nieuwenhuis et al. (2005) reviewed the accumulating
evidence that the P3 indexes a phasic arousal response, originating in
the locus coeruleus system, which is designed to increase the speed of
information processing by enhancing neural responsivity in task-
relevant cortical regions. Thus when a motivationally significant
stimulus has been detected, an arousal response is triggered which
facilitates further processing of that stimulus. Locus coeruleus (LC)
neuronal activity has been shown to precede changes in behavioural
states and appears to play a modulatory role in maintaining the alert
state, most probably via the action of the neurotransmitter noradren-
aline. Experimental alterations in LC activity indicate a causal
relationship between LC activity and cortical arousal reflected in
EEG power spectra (Swick et al., 1994). In addition, manipulations of
tonic LC activity have been shown to affect the amplitude of the P3.
Suppression of LC activity, associated with drowsiness and hypo-
arousal, precedes increases in slow wave (theta, alpha) EEG activity,
decreases in fast wave (beta) activity and attenuation of the P3 (Swick
et al., 1994). When LC activity is enhanced these effects are reversed.
Our results show a significant relationship between tonic levels of
cortical arousal (as measured by the ratio of slow to fast wave activity
during task performance) and the amplitude of the Pe. Hence the
present study offers new evidence that the Pe and P3 may share a key
functional characteristic: modulation by cortical arousal. The absence
of any such relationship with the ERN in the current study further
distinguishes these two error-related ERP components. Recent work
has demonstrated that the ERN, and not the Pe, is sensitive to changes
in dopaminergic neurotransmission relating to reinforcement learning
(reviewed in Overbeek et al., 2005). Therefore, evidence now exists
that the ERN and Pe represent functionally distinct elements of error
processing that are dependent on different brain regions and different
neurotransmitter systems. An obvious question which remains is how
Pe-related processes actually influence task performance.
With some exceptions, the majority of studies have failed to show
any relationship between the Pe and posterror performance adjust-
ments (Falkenstein et al., 2000; Fiehler et al., 2005; Ullsperger & von
Cramon, 2006). Most of these studies, however, have relied on a short-
term measure of corrective behaviour, such as immediate RT slowing,
that may not necessarily reflect a definite change in performance
strategy. For example, Gehring & Knight (2000) showed that patients
with prefrontal lesions exhibited normal posterror slowing but were
less likely to correct their errors on the next target. In addition there is
evidence that posterror slowing can occur even when errors have not
been consciously perceived (Rabbitt, 2002; Hester et al., 2005). These
findings tell us that there may be dissociable forms of posterror
control. One hypothesis is that while the ERN reflects short-term
increases in cognitive control that are not reliant on awareness and
result in remedial action on the current trial, the conscious error
processes indexed by the Pe may engender broader adaptations of
performance strategy that are likely to result in longer term changes in
behaviour. Experiments that use a wider variety of posterror correction
measures (e.g. short-term measures such as response force on error
trials and posterror RT vs. longer term measures such as performance
on the next target or changes in RT variability) will be necessary to
answer this question.
Our data also show a relationship between EEG measures of cortical
arousal and individual rates of error awareness. Again, this finding is
consistent with evidence of LC functioning which indicates that low
levels of tonic activity can lead to the absence of the normal phasic
response to motivationally significant events (Aston-Jones & Cohen,
2005; Nieuwenhuis et al., 2005). Reduced tonic activity in the LC,
reflected in our tonic EEG measures, would lead to a dampening of the
phasic response to error feedback which, as a result, may be too weak
for the error to reach conscious awareness. Thus we speculate that
when tonic levels of arousal fall below a certain threshold, the
erroneous response will not trigger the phasic facilitation of cortical
error processors, reflected in the Pe, that are necessary for awareness.
Further investigation using trial-by-trial couplings of error awareness,
Pe amplitude and both tonic and phasic LC activity would be desirable
to confirm this relationship.
The present study clearly demonstrates that a proper understanding
of the error processing system requires differentiating error-related
brain activations in terms of their relationship with conscious
awareness. Verifying levels of error awareness may be particularly
important in the context of studies that have used these ERP
components to investigate self-monitoring deficits in clinical popula-
tions such as attention-deficit hyperactivity disorder and schizophrenia
(e.g. Mathalon et al., 2002; Wiersema et al., 2005).
Acknowledgements
The work reported herein was supported by the Government of Ireland
Research Scholarship in the Humanities and Social Sciences to R.G.O’C. and is
administered by the Irish Research Council for the Humanities and Social
Sciences. This work was also supported in part by a grant from the US National
Institute of Mental Health (MH65350 to J.J.F.). The authors would like to thank
Dr Sherlyn Yeap for her excellent technical support and all those who
participated in this study.
Abbreviations
ACC, anterior cingulate cortex; Best, minimum variance within the fit interval;
EAT, error awareness task; EDA, electrodermal activity; EEG, electroencepha-
logram; EOG, electrooculogram; ERN, error-related negativity; ERP, event-
related potential; LC, locus coeruleus; Pe, error positivity; R.V., residual
variance; RT, reaction time; SCR, skin conductance response.
References
Aston-Jones, G. & Cohen, J.D. (2005) Adaptive gain and the role of the locus
coeruleus–norepinephrine system in optimal performance. J. Comp. Neurol.,
493, 99–110.
Badgaiyan, R.D. & Posner, M. (1998) Mapping the cingulate cortex in response
selection and monitoring. Neuroimage, 3, 255–260.
Berg, P. & Scherg, M. (1994) A multiple source approach to the correction of
eye artifacts. Electroencephalogr. Clin. Neurophysiol., 90, 229–241.
Brazdil, M., Roman, R., Daniel, P. & Rektor, I. (2005) Intracerebral error-
related negativity in a simple Go ⁄ NoGo task. J. Psychophysiol., 19, 244–
255.
Bush, G., Luu, P. & Posner, M. (2000) Cognitive and emotional influences in
anterior cingulate cortex. Trends Cognitive Sci., 4, 215–222.
Cavanna, A.E. & Trimble, M.R. (2006) The precuneus: a review of its
functional anatomy and behavioural correlates. Brain, 129, 564–583.
Damassio, A.R., Tranel, D. & Damassio, H. (1991) Somatic markers and the
guidance of behaviour: Theory and preliminary testing. In Levin, H.S.,
2578 R. G. O’Connell et al.
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579

Page 9

Eisenberg, H.M. & Benton, A.L. (eds), Frontal Lobe Function and
Dysfunction. Oxford University Press, New York.
Dawson, M.E., Schell, A.M. & Filion, D.L. (2000) The Electrodermal System.
In J.T., Cacioppo (Ed.), Handbook of Psychophysiology. USA: Cambride
University Press.
Debener, S., Ullsperger, M., Siegel, M., Fiehler, K., von Cramon, D. & Engel,
A. (2006) Trial-by-trial coupling of concurrent electroencephalogram and
functional magnetic resonance imaging identifies the dynamics of perfor-
mance monitoring. J. Neurosci., 25, 11730–11737.
Dehaene, S., Posner, M.I. & Tucker, D.M. (1994) Localisation of a neural
system for error detection and compensation. Psychol. Sci., 5, 303–305.
Endrass, T., Franke, C. & Kathmann, N. (2005) Error awareness in a saccade
countermanding task. J. Psychophsiol., 19, 275–280.
Falkenstein, M., Hoormann, J., Christ, S. & Hohnsbein, J. (2000) ERP
components on reaction errors and their functional significance: a tutorial.
Biol. Psychol., 51, 87–107.
Fiehler, K., Ullsperger, M. & Von Cramon, D.Y. (2005) Electrophysiological
correlates of error correction. Psychophysiology, 42, 72–82.
Gehring, W.J. & Knight, R.T. (2000) Prefrontal-cingulate interactions in action
monitoring. Nature Neuroscience, 3, 516–520.
Gehring, W.J., Gross, B., Coles, M.G., Meyer, D.E. & Donchin, E. (1993) A
neuralsystemforerrordetectionandcompensation.Psychol.Sci.,4,385–390.
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.
Hart, T.,Giovannetti,M., Montgomery, M.& Schwartz,M.A.(1998) Awareness
of errors in naturalistic action after traumatic brain injury. J. Head Trauma,
13, 16–28.
Hermann, M., Rommler, J., Ehlis, A., Heidrich, A. & Fallgatter, A. (2004)
Source localisation (LORETA) of the error-related negativity (ERN) and
positivity (Pe). Cogn. Brain Res., 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., Niuwenhuis, S., Mars, R.B. & Coles, M.G.H. (2004) Anterior
cingulate cortex, selection for action, and error processing. In Posner, M.I.
(ed.), Cognitive Neuroscience of Attention. Guilford, New York.
Kennerley, S.W., Walton, M.E., Behrens, T.E.J., Buckley, M.J. & Rushworth,
M.F. (2006) Optimal decision making and the anterior cingulate cortex. Nat.
Neurosci., 9, 940–947.
Luu, P., Tucker, D.M. & Makeig, S. (2004) Frontal midline theta and the error-
related negativity: neuropsychological mechanisms of action regulation.
Clin. Neurophysiol., 115, 1821–1835.
Mathalon, D.H., Fedor, M., Faustman, W.O., Gray, M., Askari, N. & Ford, J.M.
(2002) Response-monitoring dysfunction in schizophrenia: an event-related
brain potential study. J. Abnorm. Psychol., 111, 22–41.
McAvinue, L., O’Keeffe, F.M., McMackin, D. & Robertson, I.H. (2005)
Impaired sustained attention and error awareness in traumatic brain injury:
implications for insight. Neuropsychol. Rehabilitation, 15, 569–587.
Menon, V., Adleman, N.E., White, C.D., Glover, G.H. & Reiss, A.L. (2001)
Error-related brain activations during a Go ⁄ No-go response inhibition task.
Hum. Brain Mapp., 12, 131–143.
Nieuwenhuis, S., Aston-Jones, G. & Cohen, J.D. (2005) Decision making, the
P3, and the locus coeruleus–norepinephrine system. Psychol. Bull., 131,
510–532.
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,
752–760.
O’Keeffe, F.M., Dockree, P.M. & Robertson, I.H. (2004) Poor insight in
traumatic brain injury mediated by impaired error processing?: Evidence
from electrodermal activity. Cognitive Brain Research, 22, 101–112.
Overbeek, T.J.M., Nieuwenhuis, S. & Ridderinkhof, K.R. (2005) Dissociable
components of error processing. J. Psychophsiol., 19, 319–329.
Rabbitt, P.M.A. (1966) Errors and error correction in choice–response tasks.
J. Exp. Psychol., 71, 264–272.
Rabbitt, P.M.A. (2002) Consciousness is slower than you think. Q. J. Exp.
Psychol.: Section A, 55, 1081–1092.
Ridderinkhof, K.R., Nieuwenhuis, S. & Bashore, T.R. (2003) Errors are
foreshadowed in brain potentials associated with action monitoring in
cingulate cortex in humans. Neurosci. Lett., 348, 1–4.
Ridderinkhof, K.R., Ullsperger, M., Crone, E.A. & Nieuwenhuis, S. (2004) The
role of the medial frontal cortex in cognitive control. Science, 306, 443–447.
Rodriguez-Fornells, A., Kurzbuch, A. & Munte, T. (2002) Time course of error
detectionandcorrectionin humans:
J. Neurosci., 22, 9990–9996.
Stuss, D.T., Murphy, K.J., Binns, M.A. & Alexander, M.P. (2003) Staying on
the job: the frontal lobes control individual performance variability. Brain,
126, 2363–2380.
Swick, D., Pineda, J.A. & Foote, S.L. (1994) Effects of systemic clonidine on
auditory event-related potentials in squirrel monkeys. Brain Res. Bull., 33,
79–86.
Taylor, S.F., Martis, B., Fitzgerald, K.D., Welsh, R.C., Abelson, J.L., Liberzon,
I., Himle, J.A. & Gehring, W.J. (2006) Medial frontal cortex activity and
loss-related responses to errors. J. Neuroscience, 26, 4063–4070.
Ullsperger, M. & von Cramon, D.Y. (2006) How does error correction differ
from error signaling? An event-related potential study. Brain Res., 1105,
102–109.
Van Boxtel, G., Van der Molen, M. & Jennings, J. (2005) Differential
involvement of the anterior cingulate cortex in performance monitoring
during a stop-signal task. J. Psychophysiol., 19, 1–10.
Van Veen, V. & Carter, C.S. (2002) The timing of action-monitoring processes
in the anterior cingulate cortex. J. Cogn. Neurosci., 14, 593–602.
Vidal, F., Hasbrouc, J. & Grapperonc & Bonnet, M. (2000) Is the ‘error
negativity’ specific to errors? Biol. Psychol., 51, 109–128.
Wiersema, J.R., Van der Meere, J. & Roeyers, H. (2005) ERP correlates of
impaired error monitoring in children with ADHD. J Neural Transm., 112,
1417–1430.
neurophysiologicalevidence.
Error awareness2579
ª The Authors (2007). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 25, 2571–2579