Neural Bases Of Unconscious Error Detection In A Chinese Anagram Solution Task

In our everyday life, error detection and correction plays an important role in improving our ongoing performance in a changing Environment. Therefore, monitoring and correcting errors is one of the key functions of the cognitive control system. If task performance deflects from what the task required or expected, the cognitive control system would activated. Cognitive neuroscience techniques such as event-related potentials (ERPs), functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) have made it possible for us to precisely record the spatiotemporal cortical activation patterns associated with error detection.Detecting an error is not always conscious when we make a complex decision. Many previous ERP studies focused on the brain electrical events and activities after erroneous responses (response-locked), but there is a lack of research on the temporal patterns of brain activation related to unconscious error detection before the erroneous responses (stimulus-locked). Therefore, in the present study, fMRI and event-related brain potentials (ERPs) were used to explore the electrophysiological correlates of unconscious error detection when subjects performed a complex Chinese anagram solution task.The present research focused on the unconscious error (UE) response type, the no error (NE) response type and the nolexicial-charactered correct(NC) response type. In the NE response type, the anagram was a lexical anagram, and subjects correctly responded "yes" to the anagram. The UE response type was defined as an explicit "yes" response to a nonlexical anagram.The NC response type was defined as an "no" response to a nonlexical anagram and not change their response. Thus, for the UE response type, on the explicit level, subjects actually missed the error radical in the nonlexical anagram, but at the neurological or implicit level, they responded differently than in the NE and NC response type, suggesting that they detected the error radical or identified the nonlexical anagram unconsciously.In the Experiment 1,16 subjects participated the character anagram solution task which was similar to the English angram task. In the formal experiment,there were 200 trials, subjects were presented with either a lexical or a nonlexical anagram(100 lexical anagrams,100 nonlexical anagrams) and required to determine as fast as they could whether they could reconstruct the anagram into a lexical character.In the Experiment 2,20 subjects participated the character anagram solution task. The procedure in the ERP experiment was similar to that of the fMRI experiment, except that the duration of the stimulus was different and a phase that the subject could change their former response to the Chinese anagram was contained. The formal ERP test consisted of 264 trials (132 lexical anagrams,132 nonlexical anagrams).In the fMRI experiment, The mean number of trials for NE and UE response types were 76±8 and 34±12, respectively. The mean RTs for NE and UE response types were 1132±215 ms and 1199±236 ms, respectively. Repeated-measures Analyses of variance (ANOVA) for the mean RTs showed that the main effect of response type was significant.The analysis of UE of the fMRI imaging data mainly resulted in the activation of the left/right posterior lobe, the left/right inferior/middle occipital gyrus (BA 18/19), the right inferior/superior parietal lobule (BA 40/7), the left precuneus (BA 7), and the left/right inferior/middle/superior frontal gyrus (BA 9/46/6). The analysis of the NE fMRI imaging data showed that this condition mainly resulted in the activation of the left/right insula (BA 13), the left inferior parietal lobul (BA 40), the left medial frontal gyrus (BA 6), the right inferior frontal gyrus (BA 9/47), the left inferior occipital gyrus (BA 18), and the left/right lentiform nucleus. The main contrast between UE and NE using a random effects analysis resulted in the activation of the left medial superior frontal gyrus (BA 6/8). A reverse analysis (NE-UE) revealed mainly significant effects in the right cuneus (BA 18), the right/left superior/middle temporal gyrus (BA 42), the left anterior cingulate (BA 32), the right cingulate gyrus (BA 24), and the left precentral gyrus.The ERP data showed that the mean number of trials for the NE, UE and NC response types were 81±5,38±6 and 40±6 respectively. Mean RTs for the NE, UE and NC were 1410±66 ms,1433±75 ms and 1498±67 ms. Repeated-measures analyses of variance for mean RTs showed that the main effects of response type were significant. The mean RTs of the UE response type was significantly shorter than that of the NC response type and the mean RTs of the NE response type was also significantly shorter than that of the NC response type. Subjects would reevaluation and affirmance that the stimuli could not be converted into a correct Chinese character in working memory under the NC response type, so the RTs of the NC was significantly longer than the UE and the NC response type. However, there was no significant difference between the mean RTs of the UE and the NC response type. Two-way repeated- measures ANOVAs results revealed that there was a significant main effect of response type for the main amplitude in the time window of 300-400 ms. The UE elicited a more negative ERP component than did the NC and the NE. In addition, the results of the ANOVAs showed that the main effect of response type for the main amplitude in the time window of 900-1200 ms was also significant. The NC elicited a significant greater late positive component (LPC) than did the UE and the NE.Our fMRI results showed that, compared with the NE Condition, the UE condition revealed significantly stronger brain activation in the left medial superior frontal gyrus (mSFG:BA 6/8), suggesting an unconscious error detection in the ongoing performance. The ERP data showed that the UE response type elicited a more ERP negative component (N300-400) than did the NE and the NC response types in the time window of the 300-400 ms. Taken together with the result of dipole source analysis, the N300-400 (the anterior cingulate cortex) might reflect unconscious error detection (a nonlexical anagram could not be converted into a correct Chinese character) at the early stage of word generation when participants insisted on that the nonlexical anagram can be converted into a correct Chinese character. Then the NC elicited a greater late positive component (LPC:the medial frontal gyrus) than did the UE and the NE between 900 and 1200 ms after the onset of the anagram stimuli, which might reflect reevaluation and affirmance that the stimuli could not be converted into a correct Chinese character in working memory.