Perceptual categorization and decision making in the human brain has, in recent years, become a popular topic among the neuroimaging community. The majority of studies that have embarked upon this problem use functional magnetic resonance imaging (fMRI) to identify the cortical regions involved in decision making. However due to the low temporal resolution of fMRI, little can be said about the relative timing of the cortical processes underlying decision making in humans.;In this dissertation, we choose a different approach to decipher the spatiotemporal characteristics of the neural correlates of perceptual categorization. Specifically, we use single-trial analysis of the electroencephalogram (EEG) to map out the temporal characteristics of this process first. Using a machine learning approach and signal detection theory we identify temporally-specific components that are predictive of decision accuracy and task difficulty during a face categorization task. As a result we are able to report the first non-invasive neural measurements of perceptual decision making that lead to neurometric functions predictive of psychophysical performance. We subsequently use these results to develop a timing diagram for perceptual categorization and relate the component activities to parameters of a diffusion model for decision making.;Secondly, we utilize the relative strengths of these components across different experimental conditions in design an EEG-informed fMRI study in order to characterize their spatial extent as well. In doing so, we demonstrate that a cascade of events associated with perceptual categorization takes place in a highly distributed neural network. Furthermore, taking into consideration evidence from anatomical and functional connectivity experiments we are able to argue for the interconnectivity between the participating regions, where both bottom-up and top-down influences exist. We use these results to develop a comprehensive spatiotemporal diagram for perceptual categorization and decision making.;In this thesis we also consider the application of parametric spectral analysis techniques to multichannel EEG data collected from our face discrimination task. We are able to identify causal influences between a distributed set of electrode locations the timing of which correspond to previously reported EEG face-selective components. More importantly we present evidence that there are both feedforward and feedback influences, a finding that is in direct contrast to current computational models of perceptual discrimination and decision making which tend to favor a purely feedforward processing scheme.;Finally, we design a novel gender discrimination task to identify the cortical regions sensitive to different dimensions of decision difficulty during face categorization. Categorization of faces is often thought to involve matching against internal representations, however where these comparisons are made and how these representations are retrieved remains unknown. We manipulated the difficulty of a gender categorization task along two distinct stimulus dimensions by either morphing or adding noise to male and female images. We then used fMRI to identify cortical regions selective to differences in these dimensions. We found activations suggesting that a distributed set of areas, some of which are found outside the general face processing system, are activated during the retrieval and comparison processes relevant to face categorization. In accordance with the main theme of this thesis we are also able to use this paradigm to identify cortical regions that are directly related to categorical decision making.
展开▼
