| Our daily lives are pervaded by sounds, predominantly speech, music, and environmental sounds. We readily recognize and categorize such sounds. Our understanding of how the brain so effortlessly recognizes and categorizes sounds is still rudimentary. The central focus of this thesis is elucidation of the neural mechanisms underlying auditory categorization. To this end, the thesis mainly employed multivoxel pattern-based analysis techniques (MVPA) applied to functional magnetic resonance imaging (fMRI) data. The first study revealed differential neural patterns for the representation of different auditory object categories (e.g., animate vs. inanimate at superordinate level, human vs. dog at basic level). Importantly, the categorical neural patterns were not just confined within the classical auditory cortex. Rather, we were able to find the categorical responses throughout the brain beyond the early sensory area. A second study revealed both auditory and visual responses to distinguish between animate and inanimate categories within the same anatomical regions far downstream from the early sensory cortex, suggesting that those areas may be involved in object processing independent of modality. A third study identified melodic contour processing areas (e.g., rSTS, lIPL, and ACC) in the music domain. Neural patterns in these areas differ between ascending and descending melodies. A fourth study revealed several left-lateralized cortical loci where different phonetic categories were distinguished with differentiable neural patterns. Further, the findings demonstrated that there was difference between low-level vs. high-level speech processing regions in their role of simple acoustic feature detection vs. complex categorical processing. Taken together, the findings presented in this thesis provide evidence that the brain uses a unifying strategy - categorical neural response - for auditory categorization in all three sub-domains. Further, throughout the studies, not only modality-specific but also modality-independent high-level processing regions were often found for auditory processing. These findings may help us move toward an improved understanding of how received signals progress from low-level processing (e.g., frequency extraction) to high-level processing (e.g., understanding the concept). |
