Disentangling Error-and Frequency-driven Slowing Effects:the Evidences Of Neural Dynamics And Neural Oscillation

Error process is an important element in cognitive control and behavioral monitoring. The behavioral performance should be monitored continuately for completing the task successfully. Since only the behavioral performance is under the continuate monitoring, brain can detect the errors and adjust the response strategy. Previous studies have affirmed that there is a monitor system in the brain. After committing errors, the error monitor system will send the error signals to the cognitive control system to enhance the cognitive control for avoiding a similar error, manifesting as slower response in trial following the error trial. This effect terms as post-error slowing (PES).Execution errors indicate the rapid and impulse errors due to the incomplete processing of task-related stimuli. The behavioral adjustment depends on the enhanced cognitive control induced by error signal. People tend to adopt a much cautious response strategy (slower response and more accuracy). However, Notebaert and his colleagues put forward an alternative account in recent years. They assert that PES is just a special example of slowing effect. Slowing occurs after infrequent events, no matter the correctness of events. Infrequent events draw away the attentionalresource, and then participants need to take more time to reorient the subsequent task. Therefore, both error signal and infrequent event may lead to slowing effect. To elucidate their specific contributions in the generation of slowing effect, we investigated the modulations of error signal and event frequency.To address this problem, the error rate has to be effectively manipulated, since errors usually occur at a low rate in the dairy life, even in the lab environment. Thus, we adopted a look-to-do design to investigate this issue. Previous studies had affirmed that observation errors and execution errors activated a similar neural system, and both resulted in post-error slowing, no matter the observation errors were activated by computers or by the real partners. Moreover, Dandan Tang et al had demonstrated that previous "look" trials can effectively modulate the subsequent "do" trials.In the experiment1, we adopted two error rate conditions50%and80%. Since in the50%error rate condition, the proportion of correct trials and error trials is equal, the influence of infrequency information can be controlled. While in the80%error rate condition,80%is the error trials and20%is the correct trials. The infrequent correct trials can effectively distinguish the role of error signals and error frequency, as the most challenge of orienting account is the novel finding in trials following the infrequent correct trials, that is, a post-correct slowing effect is reduced by the infrequent correct trials. As a result, the post-error slowing was found under both50%and80%error rate condition, suggesting error signals played an important role in post-error behavioral adjustment, which is in favor of the prediction of cognitive accounts.In the experiment2, to verify the reliability of the results from the experiment1, and examine post-error slowing after infrequent errors in the novel design, we recruited an additional independent cohort to perform the same experimental procedure with an additional infrequent error condition (20%). As a result, we found the post-error slowing occurred under both three error rate condition (20%,50%and80%). The effect of PES in the80%and20%were both larger than that in the50%error rate condition, however, the effect of PES was comparable between20%and80%error rate condition. Therefore, current data replicated the results of the experiment1and previous studies, extended the findings of previous studies, and affirmed error signals played an important role in post-error behavioral adjustment.In the experiment3, we further investigated the modulations of error signals and event frequency on the EEG perspective. The experimental procedure was the same with the experiment1, including50%and80%error rate condition. The behavioral results were similar to the experiment1, post-error slowing occurred under both50%and80%error rate condition. However, the EEG results showed distinct patterns (ERP and time-frequency analysis). ERP results showed that only in the50%error rate condition, the neural activationof observed-error trials were significantly larger than those of observed-correct trials for both ERN and Pe. Whilst the time-frequency results showed that only in the80%error rate condition, the neural oscillations of observed-error trials were significantly larger than those of observed-correct trials for both theta and alpha bands. These findings suggested error signals and event frequency both contributed to the post-error behavioral adjustment. However, the generation mechanism was distinct, the role of error signals mainly presented in the ERP components, reflecting the event-related modulation; the role of event frequency mainly presented in the time-frequency bands, reflecting the modulation of overall state of the processing network.In the present study, although the attentional process was involved in the post-error behavioral adjustment, the result was inconsistent with the finding of Notebeart et al. The possible reason is that attentional process induced by internal causes is different in nature from that induced by external causes. The former depends on the top-down modulation and the latter depends on the bottom-up modulation. We adopted the look-to-do design based on the Flanker task, thus the task process might depend on the cognitive modulation instead of the perception input, resulting in a faster response after the infrequent correct trials (the PES of80%error rate condition benefited from this effect).