An event-related potential investigation of the neural representations that support familiarity-based picture recognition

Abstract

<p>Most models of recognition memory assume that familiarity results from the matching of stimuli to the contents of memory. This matching process accumulates "evidence" that the stimulus was seen before, and when the evidence exceeds a criterion, a feeling of familiarity is experienced. Such models do not specify what constitutes "evidence," and therefore offer limited insight into the specific attributes that make stimuli feel familiar. In two experiments, this dissertation examined the type of pictorial attributes that serve as "evidence" for familiarity-based picture recognition. Participants encoded briefly presented, masked pictures while event-related brain potentials (ERPs) were recorded. Of primary interest were ERPs for pictures that participants could not identify by name, as previous behavioral research suggests that such items are recognized on the basis of familiarity. Analysis of these encoding ERPs revealed that the global shape of subsequently recognized yet unidentified pictures was fully extracted during the picture's brief presentation, but that their global object shapes were not successfully matched to object representations in memory (Exp 1 & 2). This result indicated that the memory trace for unidentified pictures contained limited conceptual information, and perceptual details that were abstract rather than detailed/episodic. ERPs recorded during retrieval revealed that the neural correlate of familiarity-based retrieval, the FN400, was present for unidentified pictures (Exp 1 & 2), and that the FN400 was more pronounced when participants were oriented toward processing perceptual, as opposed to conceptual, attributes of pictures during encoding (Exp 2). The behavioral measure of familiarity was consistent was consistent with this finding, which together implied that the largely perceptual representations in the pictorial memory trace were sufficient for later recognition of the unidentified picture, and that the pictures were more familiar when perceptual processing was greatest at encoding. The data presented in this dissertation indicate that familiarity-based picture recognition can be based on evidence that is largely perceptual and abstractly represented. The results are discussed within the context of perirhinal cortex models of familiarity, which suggest that picture familiarity is based on conjunctive features represented by the perirhinal cortex within the medial temporal lobe.</p>