Datensatz

Zusammenfassung

In the framework of visual cognition and predictive coding models, we tested the hypotheses that visual prediction influences detection of stimuli violating expectation, and that early and late (conscious) behavioral and brain responses to deviant stimuli are related to processing of different aspects of prediction error. To this aim combined recordings of saccadic eye movements and fMRI data were performed on twelve participants performing a visual detection task. Participants were required to detect moving stimuli that were suddenly displaced with respect to their current trajectory (deviant stimuli). Displacement varied in amplitude and orientation. Psychophysical reverse correlation analysis evidenced different perceptual levels of prediction error processing. Analysis of conscious responses revealed reduced detection of visual inputs for stimuli with small deviation from expected behavior with respect to large deviant stimuli as indicated by increased eccentricity of the psychophysical kernel. fMRI data analysis showed that higher-level late conscious processing, mainly associated with cortical activity in fronto-parietal areas as well as subcortical regions such as caudate nucleus and thalamus, seems to be required to detect prediction error and to assess the degree of violation of expectations. Lower-level early processing, associated with dorsal activity in the right angular gyrus, also enables detection of violation of prediction but it does not permit discrimination among large and small deviating stimuli as indicated by almost null eccentricity of psychophysical kernel. These findings highlight at least two primary brain mechanisms, an early and a late stage, subserving detection of visual inputs deviating from perceptual expectations.