Sensitivity to spectral and temporal properties of sound in human non-primary auditory cortex

Relative to a silent baseline, sound-evoked activation engages much of the auditory cortex irrespective of the acoustic properties of the signal. However, particular response preferences can be localized within discrete subregions by contrasting sound conditions that possess different values of the acoustical property of interest. Using these simple subtraction and linear regression approaches, numerous functional magnetic resonance (fMRI) studies have measured the relative distribution across the human auditory cortex of sensitivities to a range of fundamental sound properties including frequency, bandwidth, sound level, temporal structure, spatial location, and auditory motion (see Hall et al., 2003, for a general review). In perhaps every case except that of sound intensity encoding (Hart et al., 2002, 2003a), these features evoke differential activation that clearly extends across the non-primary, as well as the primary, auditory cortex. Spectral and temporal sound features convey information about the identity and source of that sound. Although many computational transformations have been completed before the acoustic signal reaches the auditory cortex, the cortex significantly contributes to auditory analysis. The neural computations performed by nonprimary auditory cortex are uncertain but they may involve stages in identifying auditory objects, including the integration of spectral and temporal sound features (Zatorre et al., 2002) and the matching of these components with learned spectrotemporal representations (Griffiths & Warren, 2002).