Distinct Functional Contributions of Primary Sensory and Association Areas to Audiovisual Integration in Object Categorization

Multisensory interactions have been demonstrated in a distributed neural system encompassing primary sensory and higher-order association areas. However, their distinct functional roles in multisensory integration remain unclear. This fMRI study dissociated the functional contributions of three cortical levels to multisensory integration in object categorization. Subjects actively categorized or passively perceived noisy auditory and visual signals emanating from everyday actions with objects. The experiment included two 2x2 factorial designs that manipulated either (i) the presence/absence or (ii) the informativeness of the sensory inputs. These experimental manipulations revealed three patterns of audiovisual interactions. (1) In primary auditory cortices (PAC), a concurrent visual input increased the stimulus salience by amplifying the auditory response irrespective of task-context. Effective connectivity analyses demonstrated that this automatic response amplification is mediated via both, direct and indi rect (via STS) connectivity to visual cortices. (2) In superior temporal (STS) and intraparietal (IPS) sulci, audiovisual interactions sustained the integration of higher-order object features and predicted subjects’ audiovisual benefits in object categorization. (3) In the left ventrolateral prefrontal cortex (vlPFC), explicit semantic categorization resulted in suppressive audiovisual interactions as an index for multisensory facilitation of semantic retrieval and response selection. In conclusion, multisensory integration emerges at multiple processing stages within the cortical hierarchy. The distinct profiles of audiovisual interactions dissociate audiovisual salience effects in PAC, formation of object representations in STS/IPS and audiovisual facilitation of semantic categorization in vlPFC. Furthermore, in STS/IPS, the profiles of audiovisual interactions were behaviorally relevant and predicted subjects’ multisensory benefits in performance accuracy