Colloquium: Novelty rejection in episodic memory – Department of Cognitive Science (carleton.ca)
Date: Wednesday, 16th Mar 2022 – 3:00pm-4:30pm
Title: Novelty rejection in episodic memory
Location: Online
Speaker: : Dr. Adam Osth
Abstract:
Episodic memory theories have postulated that in recognition memory, a given probe is accepted or rejected on the basis of its global similarity to studied items. Mewhort and Johns (2000) directly tested global similarity predictions by manipulating the feature compositions of probes — novelty rejection was facilitated when probes contained novel features even when other features strongly matched, an advantage dubbed the extralist feature effect, which greatly challenged global matching models. In this work, we conducted similar experiments using continuously-valued separable- and integral-dimension stimuli. Analogs of extralist lures were constructed where one stimulus dimension contained a value that was more novel than the other dimensions while overall similarity was equated to another class of lures. Facilitated novelty rejection for lures with extralist features was only found for separable-dimension stimuli. While integral-dimension stimuli were well described by a global matching model, the model failed to account for extralist feature effects with separable-dimension stimuli. We constructed variants of the exemplar-based linear ballistic accumulator (EB-LBA) that employed different means of novelty rejection afforded by separable-dimension stimuli, including decisions based on the global similarity of the individual dimensions, and selective attention being directed toward novel probe values (a diagnostic attention model). While these variants produced enhanced novelty rejection for lures containing extralist features, only the diagnostic attention model succeeded in providing a sufficient account of all of the data. The model was also able to account for extralist feature effects in an experiment with discrete features similar to those from Mewhort and Johns (2000).
Bio:
Adam Osth is a Senior Lecturer in the School of Psychology at the University of Melbourne specializing in the computational modeling of human memory and decision-making. He received his Ph D from The Ohio State University in 2014, followed by a postdoctoral fellowship at The University of Newcastle in Australia. In 2016, he began a Lecturer position at The University of Melbourne and in 2017 was the recipient of the Discovery Early Career Researcher Award fellowship from the Australian Research Council. He has two cats that regularly make uninvited appearances in his presentations and Zoom meetings.