Neural and behavioral correlates of similarity-based categorization: An event-related potential approach

Categorization is an important mechanism in human information processing and a central issue in psychology. By most accounts, almost every aspect of cognition can be described by the basic processes of categorization. This dissertation is an assessment of neural and behavioral correlates of similarity-based categorization in perceptual, conceptual, and semantic contexts. The utility of cognitive electrophysiology in stimulus classification tasks is well known. This methodology is a non-invasive procedure for recording neural responses from healthy participants. Brain responses that are time-locked to a stimulus event are referred to as event-related potentials (ERPs). The P300 component is a late-positive component of the ERP that is elicited in response to the detection of improbable stimuli. This component has been widely investigated in the so-called oddball paradigm and has provided much of the human neurophysiological data on brain mechanisms in categorization. In series of separate studies, participants were instructed to classify visually presented stimuli into target and non-target categories. Participants classified stimuli by counting every occurrence of the target stimuli. Targets were intermixed with stimuli that varied according to their target-like characteristics. While probability and detection were expected to affect targets, non-target stimuli were independent of these effects and were presented under identical task conditions, This approach allowed a systematic evaluation of differences between similar and dissimilar non-target stimuli. As predicated, irrespective of context, target stimuli produced the usual P300 components maximally distributed at central-parietal electrode sites. Non-target stimuli that were contextually similar-to-target elicited intermediate P300s that were topographically target-like, but smaller in amplitude than those produced by targets. Overall, similar patterns of results were recorded for behavioral responses. Taken together, these studies support that the P300 amplitude can be utilized as a good neural marker in assessing the spontaneous categorization schema of the human brain.