Abstract
In real life, newly arriving information is continuously integrated with prior information gathered over different timescales. Using Inter-subject correlation (ISC), a measure of reliability of shared responses across subjects to a temporally complex stimulus, revealed hierarchical information accumulation in visual and auditory cortical circuits. Regions with short temporal receptive windows (TRW), processing information across short timescales, are found in early sensory areas, whereas higher-order areas present long TRW (Hasson et al., 2015). Hierarchical neural organization was previously hypothesized for motor circuits as well. Accordingly, basic motor primitives are represented at lower levels, a combination of primitives, building motor sequences, at subordinate levels, and an overarching action goal at superordinate levels (Hamilton 2009). However, there is currently no consensus regarding the neural substrates of such hierarchical action organization. Using fMRI, we measured ISC to examine the neural topography of hierarchical action representation while observers (n=24) viewed sequential goal-directed hand movements presented in their intact chronological order or piecewise-scrambled at three timescales pertaining to action evolution (~2, ~4, and ~10 s). The results show high ISC in early visual and somatomotor areas in all three temporal disruption levels while ISC gradually decreased for short timescales on a caudal-rostral and lateral-medial axis. Thus, compatible with previous findings, early visual and somatomotor areas present short TRWs and are less sensitive to disruption in temporal structure while higher visual and motor regions, show a gradual increase in their TRW such that sensitivity to the temporal disruption of observed actions was low in the SPL and high in the precuneus. Furthermore, the ventral premotor cortex, associated with the processing of basic motor attributes, showed long TRW thus, indicating its role in information integration during observation of high-level goal-directed actions. Together, the current results reveal the cortical topography of increasing complexity levels of the action observation network.