# Top-down modulation of visual action perception: distinct task effects in the action observation network

## TL;DR

This study shows how attention influences brain regions involved in observing others' actions, revealing distinct effects on different areas of the Action Observation Network.

## Contribution

The study demonstrates that top-down attention modulates the Action Observation Network in a feature- and region-specific manner.

## Key findings

- The task model correlated significantly with neural similarity in all AON regions, showing task alignment.
- Different tasks uniquely affected each AON node, indicating specific feature-region interactions.
- Top-down attention alters neural representations in the AON, highlighting attention's role in action perception.

## Abstract

Perceiving others’ actions is essential for survival and social interaction. Cognitive neuroscience research has identified a network of brain regions crucial to visual action perception, known as the Action Observation Network (AON), comprising the posterior superior temporal cortex (pSTS), posterior parietal cortex, and premotor cortex. Recent research highlights the importance of integrating top-down processes, such as attention, to gain a deeper understanding of action perception. This study investigates how attention modulates the AON during human action perception. We conducted a two-session fMRI experiment with 27 participants. They viewed eight videos of pushing actions, varying in actor (female vs. male), effector (hand vs. foot), and target (human vs. object). In the first session, participants focused on specific features of the videos (actor, effector, or target). In the second, they passively viewed the videos. From the passive viewing session data, we defined regions of interest (ROIs) in the pSTS, parietal, and premotor cortices for each hemisphere. We then performed model-based representational similarity analysis (RSA) and decoding analysis. RSA results showed that only the task model, among all tested models, exhibited a significant correlation with neural representational similarity matrices (RDMs) across all ROIs, indicating a specific alignment between AON nodes and the ongoing task. Decoding analysis further showed that different task types uniquely affected each AON node, indicating feature- and region-specific interactions. These findings underscore that top-down attentional processes significantly alter neural representations within the AON, highlighting the dynamic interplay between attention and action perception in the brain.

The online version contains supplementary material available at 10.1007/s00429-025-03042-z.