UC Irvine

We examined the role of top-down influences, including predictive coding, in the action observation network (AON) and extended brain regions. To achieve this, we examined functional connectivity within the network, multivariate pattern classification of regional representations, and diffusion modeling to reveal indicators of predictive decision bias. In the first study, subjects were exposed to three-second action vignettes involving avatars performing one of two different actions.The participant was cued in advance to attend to the action, the outcome of the action (goal) or the identity of the actor. The results indicated that attention could modulate precision with which the actions were encoded in the AON, specifically the right pSTS and IFG. These findings are consistent with the hypothesis that the IFG provides top-down feedback to the pSTS and implicates a right-lateralized system in processing action kinematics. Study 2 seeks to evaluate whether information about anticipated actions is apparent in the neural signal prior to observation, as a potential source action prediction. In this second study, in half of the trials subjects were cued as to which action they would see before observing the action vignettes. The findings revealed that, when given a valid cue, the right IFG had a higher classification accuracy during cued trials than cued trials (uncorrected for multiple comparisons), when measured during the anticipatory interval (prior to encoding). These results suggest that predictive cues can indeed impact the neural decoding processes within the AON. Study 3 further examines this predictive process in terms of the association between response time and brain responses. We used hierarchical diffusion modeling to model trial-wise posterior estimates of decision bias, which showed clear evidence for observer’s using the cue during the anticipatory interval. These studies collectively demonstrate how top-down influences from the IFG enhance the pSTS's reconstruction of action dynamics. These results are also consistent with the proposal that predictive cues can bias both behavioral responses and neural activity, providing a deeper understanding of the neural pathways involved in interpreting socially relevant actions.