Post-error adaptations are reflected in modulations of theta and alpha band power as revealed by the auditory condensation task
Commission of error causes the adjustments in a number of brain systems related to goal-directed behavior. Errors may be caused by failures of motor inhibition or by general attentional lapses, which lead to the different neural adjustments with their specific electrophysiological and behavioral correlates (van Driel et al., 2012; Danielmeier and Ullsperger, 2011). Thus, post-error adjustments may lead both to non-specific increase of motor threshold or to specific improvement of stimulus processing and decision making, with different brain systems involved in these processes (King et al., 2010). In the present study, we aimed at the investigation of error-related theta and alpha band power modulations and of corresponding behavioral adjustments.
An auditory two-choice version of the condensation task was used in the experiment (Posner, 1964; Chernyshev et al., 2015). Subjects were presented with random sequence of tones; each tone was either 500 Hz (‘low’) or 2000 Hz (‘high’), either a pure tone (‘pure’) or the same tone intermixed with broadband noise. The participants were instructed to respond to stimuli with pressing left or right button on a gamepad, according to the memorized rule (see Table 1). Correct responses after stimulus onset were immediately followed by a positive feedback (a schematic smiling face) presented for 500 ms after the response. This task is highly demanding for sustained attention, but implies no to-be-inhibited “automatic” responses. We analyzed modulations of non-phase-locked theta (4 – 7 Hz) and alpha (8 – 12 Hz) EEG power that occurs on erroneous and post-error correct trials. We used data-driven approach with threshold-free cluster enhancement (TFCE)-based permutational correction for multiple spatial-time-frequency bins in order to avoid a priori ROI selection. Also, we analyzed correlations between post-error spectral modulations and behavioral variables (percentage of errors and post-error slowing), using Spearman’s correlation coefficient.
Response times (RT) on erroneous trials was significantly larger than on correct trials (t = 9.48, p < 0.001). No significant post-error slowing (PES) was found (t=-0.53, p=0.60). Errors (compared to correct trials) lead to significantly (p < 0.05) increased frontal midline theta (FMT) power (0 – 400 ms), followed by the enhanced alpha band suppression in the parietal (400 – 700 ms) and the left central regions (500 – 1000 ms) (Fig. 1A, top row). Based on these results, we selected three regions of interest: R1 – frontal midline theta, R2 – posterior alpha, R3 – left central alpha (Fig. 1A, top row). Stronger parietal alpha suppression was associated with better task performance, stronger left central alpha suppression was associated with more pronounced PES, and FMT increase positively correlated with both behavioral variables (Fig. 1B). On post-error correct trials (compared to post-correct ones), the following significant (p < 0.05) effects were found (Fig. 1A, bottom row): stronger pre-response left-central alpha suppression (-1000 – -250 ms); stronger generalized alpha suppression around the response (-150 – 500 ms), weaker post-response FMT power (0 – 600 ms).
We believe that our results suggest the occurrence of the conflict / error detection signal, followed by the signals of attentional reconfiguration and motor threshold adjustment. These adjustments resulted in optimized performance on the subsequent trials, accompanied by the reduced uncertainty of the response and decreased conflict. Our findings presumably indicate post-error adaptations in several brain systems, and extend the literature on sustained attention lapses and cognitive control.