Beynin biyolojik hareket algısı ağında tahmine dayalı kodlama

The literature on biological motion processing argues that it occurs in occipitotemporal, parietal and frontal regions of the brain. Nevertheless, the literatureis currently unable to explain how this processing is affected by expectations. Although models exist to explain how biological motion is perceived, they usuallyignore top-down processes. To this end, the current fMRI study presented twopoint-light displays (embedded in noise) on either side of the screen to the participants (N=29). One of the displays was a biological motion (walking or kicking)whereas the other one was its scrambled version. The participants were asked toreport the location of the biological motion. Importantly, before the presentationof motions, the participants were shown a cue about the action type (walking orkicking) which was congruent with the motion 75% of the time. There were alsotwo additional conditions in which the cue was uninformative about the action(neutral condition) or there were no motion stimuli at all. As expected, the action observation network (consisting of pSTS, parietal cortex and IFG) showeda clear and strong activation during the conditions that a motion was present.However, these regions have failed to significantly discriminate between congruent and incongruent conditions. It should be acknowledged that this lack ofsignificant result might be caused by the low number of trials. In order to better investigate the connections within action observation network, a DCM analysiswas conducted. The winning DCM model has successfully shown the presence offeedback connections in the biological motion processing. More specifically, themodel argues that both feedforward and feedback modulatory connections arepresent during congruent, incongruent and neutral conditions. In sum, the studyhighlights the importance of incorporating top-down signals such as expectationsin the computational models of biological motion perception.