Commentary: “Hearing faces and seeing voices”: Amodal coding of person identity in the human brain

In a recent paper, Hasan et al. (2016) report the results of a neuroimaging study on amodal person identity processing. Across multiple testing sessions, five participants were presented with audioonly, video-only and audiovisual stimuli of four familiar people producing the syllable “had.” During the scanning session, participants performed a forced-choice person identification task in response to each stimulus. A univariate fMRI analysis confirmed that visual and auditory cortices where involved in stimulus processing. Using multivoxel pattern analyses (MVPA), the authors found that auditory identities could be successfully classified in temporal areas, while visual identities could be successfully classified in fusiform gyrus and inferior temporal gyrus. In addition, auditory and visual identities could both be classified in overlapping areas of right posterior superior temporal sulcus (pSTS). Successful cross-classification of stimuli was achieved in left inferior frontal gyrus, right supramarginal gyrus and in multiple sites in STS—here classifiers were trained on data from one modality and tested on data from the other, thus probing modalityindependent coding. Hasan et al. (2016) conclude that their results are evidence for increasingly abstracted neural representations of person identity. This study addresses important theoretical questions about the multimodal processing of person identity and attempts to empirically map identity coding from unimodal toward abstracted amodal representations. However, several issues arise from the study design and interpretation.