Mediofrontal event‐related potentials following observed actions reflect an action prediction error

Abstract

Both execution and observation of erroneous actions have been shown to increase the activity of the anterior cingulate cortex (ACC) as reflected in characteristic event-related potential (ERP) components labelled error-related negativity (ERN) and observer error-related negativity (oERN), respectively. Whereas these labels implicate a modulation of both components by response accuracy, recent findings suggest a more general involvement of the ACC in the detection of unexpected events. In previous studies, a lower frequency of erroneous as compared with correct observed actions resulted in lower expectation of erroneous actions. The present study investigates whether ERPs following observed actions are modulated by response accuracy or violation of expectation. Sixteen human subjects observed a virtual person whose actions in a game were expected or unexpected. Action expectation was independent of accuracy. In both conditions, subjects observed correct and incorrect actions equally often. Whereas ERPs were not modulated by accuracy, we found an enhanced amplitude of a negative frontocentral ERP component in the time window of the oERN for unexpected as compared with expected observed actions, which we suggest reflects an action prediction error. These results propose that the function of the ACC in performance monitoring depends less on accuracy of actions but rather on predictions and their violations. Future research will have to clarify whether the present ERP modulations revealed a feature of the oERN or whether they represent a distinct component.

Probabilistic reward learning in adults with Attention Deficit Hyperactivity Disorder—An electrophysiological study

“…5B shows that the mean amplitude signal is more negative (less positive) for observed correct responses throughout all processing steps. However, between 200 and 400 ms there is a pronounced negative-going component for both observed errors and correct responses, resembling an observer ERN-like signal recently described to reflect an action prediction error (Kobza and Bellebaum, 2013). This signal, when considered relative to the preceding positivity, appears to be more pronounced for observed errors.…”

“…The ERN for actively performed errors and correct responses alike was then defined as the most negative peak amplitude from À 50 ms to 100 ms relative to response onset at electrode site FCz. For the analysis of the observational ERN, we took into account that the signal between 200 and 400 ms resembled the pattern of an FRN, in accordance with one recent study by our group in which we examined action prediction errors (Kobza and Bellebaum, 2013). In contrast to other studies, e.g.…”

“…In contrast to other studies, e.g. by van Schie et al (2004), the observational ERN was thus defined as the most negative peak amplitude between 200 and 400 ms after the observed response, relative to the preceding positive peak (minimum latency 150 ms), also at electrode site FCz (see Kobza and Bellebaum, 2013 for a similar procedure).…”

Feedback delay gradually affects amplitude and valence specificity of the feedback‐related negativity (FRN)

“…As already outlined in the introduction, this signal, and ACC activity in general, has been related to processing of violations of action‐outcome expectations (Alexander & Brown, ; Ferdinand et al, ). Our group described a similar ERP component also for the observation of behavior (Kobza & Bellebaum, ), raising doubts whether mediofrontal negativity is even specific for processing of action outcomes or does more generally indicate the occurrence of something unexpected. Outside the feedback domain, processing of rare or unexpected stimuli (oddballs) has been associated with an early poststimulus negativity (N2) that closely resembles the FRN in time course and topography.…”

Effects of trait empathy and expectation on the processing of observed actions

“…In this game, participants have to track under which of two shells a ball is hidden. Because this task is quite easy, erroneous responses by the observed person were unexpected (Kobza & Bellebaum, 2013). The task, however, also entails a falsebelief condition: in this, observers had exclusive access to task-related knowledge that made correct responses unexpected.…”