BrainInNaturalSound · The anatomy and dynamics of the cortical processing of naturalistic sounds

The anatomy and dynamics of the cortical processing of naturalistic sounds

More than two decades of studies on the psychophysics of hearing document our ability to recognize naturalistic sound sources (e.g., breaking glass), and have detailed the effects on perception of largely diverse types of sound-related information, ranging from the low-level acoustical structure to high-level symbolic knowledge. In contrast, cognitive neuroscience studies of naturalistic sounds have only recently begun to detail the cortical processing of sound sources and focus almost exclusively on the encoding of category-related information (e.g., animal vocalizations vs. tool-action sounds). The overall absence of integration of psychophysics and cognitive neuroscience approaches makes our knowledge of the cortical processing of naturalistic sounds both limited and potentially flawed. For example, it is unknown the extent to which previous observations of category encoding are mediated by category differences in low-level acoustical or high-level semantic structure (e.g., animal vocalizations are more harmonic and less semantically heterogeneous than tool sounds). As such, our current understanding of hearing in the everyday world strongly resembles a “black box” model: we have a knowledge of the perceptually relevant stimulus properties (e.g., velocity of loudness decay) and of the mapping between stimulus properties and perceptual attributes (e.g., metallic objects are recognized in slowly decaying sounds), but have a limited knowledge of the cortical mechanisms that govern this mapping. The neuroimaging experiments carried out during this project significantly advanced our knowledge in the field: 1. We discovered that the spatial pattern of cortical activation contain information about low-level structure and sound category, and assessed the presence of regions in the auditory cortex that represent sound categories in an abstract fashion, i.e., independently of reliable between-category differences in low-level acoustical structure. 2. We discovered that the phase rather than power of time-varying cortical responses to sound: [a] tracks several time-varying acoustical features; [b] contains information for differentiating reliably between natural sounds, and between natural sounds on the one hand, and silence on the other; [c] contains information about the perceptual dissimilarity of natural sounds. Importantly, all of these encoding effects emerged for the phase of low-frequency oscillatory responses of the time-varying cortical response to sound. 3. We discovered that a region in the left prefrontal cortex is involved in the automatic, i.e., task-independent processing of the identity of a wide variety of sound sources: speakers, musical instruments and non-speech non-music sound sources (e.g., vacuum cleaners). We observed that this region tracks the identity of sound sources based on objective, measurable properties of the low-level structure of the sounds. 4. We discovered that the same cortical regions that respond selectively to one category of natural sounds (e.g., human voices) also represent a large deal of information about the same sound category in the spatial pattern of cortical activity: [a] differentiation of activating category from non-activating categories (e.g., human action); [b] differentiation of within-category exemplars (e.g., speech vs. physiological vocalizations such as sneezes); [c] perceptual dissimilarity of within-category exemplars.

Data: CORDIS, © European Union