A dynamic approach to recognition memory.

Abstract

ii Acknowledgements First, I extend my thanks to my committee members for their time, consideration, and guidance. I would especially like to thank my advisors, Rich Shiffrin and Mike Jones, for first convincing me to come to Indiana, making me feel welcome, and guiding me into the world I now inhabit. I must also thank Isaiah Harbison, Michael Dougherty, and Sharona Atkins for introducing me to the very idea of mathematical psychology and for teaching me and trusting me. show how the dynamic model can be extended to account for paradigms like associative recognition and list discrimination, leading to another novel test of the presence of recall-like processes. Associative recognition, list discrimination, recognition of similar foils, and source exclusion are all better explained by the formation of a compound cue rather than recall, although source memory is found to be better modeled by a recall process.

Information and Processes Underlying Semantic and Episodic Memory Across Tasks, Items, and Individuals

“…This similarity could arise either because items and associations are encoded using similar information or because the processes used to retrieve item and associative information are related, or both. We are not in a position to adjudicate this question here, although we note that other work from our laboratory supports the idea that item and associative information are both stored and retrieved using similar processes (Cox & Criss, 2017), and that memory for associative information involves elaborating or interrelating information about items (Cox & Shiffrin, 2017;see also McGee, 1980;Dosher, 1984;Dosher & Rosedale, 1989, 1991, 1997. Consistent with the idea that item and associative information are encoded in a similar manner, we found that the same item properties, like concreteness and semantic specificity, that make a single word easy to recognize (and to reject when unstudied) also make it easy to recognize a pair in which that word appears.…”

“…This factor identifies a trade-off between two sets of parameters in associative recognition: boundary separation and drift, on the one hand, versus residual time and bias on the other. While we again do not wish to over-interpret this factor, we note that in the model proposed by Cox and Shiffrin (2017), item information tends to be retrieved earlier than associative information, such that the information needed to distinguish between intact and rearranged pairs in AR is only available later in retrieval. As a result, accumulating evidence earlier-which would correspond to a shorter residual time prior to the onset of evidence accumulation-would yield more positive recognition evidence-higher drift rate-because the items in both intact and rearranged pairs had been studied.…”

“…A simple reason why residual time may be higher in AR relative to SR is that participants must process two words in AR rather than just one, although response-signal studies suggest that the cost of reading two vs. one word is relatively minor compared to the cost of requiring associative information rather than only item information (Gronlund & Ratcliff, 1989). Other than the difference in the number of stimuli in AR, response demands (pressing a mouse button) are equivalent across the binary-choice tasks, suggesting that many of the differences in residual time can be attributed to additional processes engaged while accessing the relevant evidence in memory (see Cox & Shiffrin, 2017, for additional discussion on differences in residual times and additional processing in AR relative to SR).…”

“…Strength of semantic features must also play a role at retrieval, in that it is associated with discriminability in single-item recognition (i.e., not just recognition of studied words, but the ability to reject an unstudied but semantically distinctive word). Feature distinctiveness and the resulting differentiation between event memories is a core component of likelihoodbased models of episodic memory (McClelland & Chappell, 1998;Shiffrin & Steyvers, 1997) and has been found to be critical for explaining a variety of memory phenomena across a variety of stimuli (not just words; Cox & Shiffrin, 2017;Kılıç, Criss, Malmberg, & Shiffrin, 2017;Nosofsky & Zaki, 2003;Steyvers & Malmberg, 2003).…”

Sequential Sampling Models Without Random Between-Trial Variability: The Racing Diffusion Model of Speeded Decision

“…Although, the mapping of early stimulus components onto within-trial variability and late decision components onto between-trial drift rate variability may not be so clear. For example, see some recently developed dynamic models of memory (Cox & Shiffrin, 2017). There is a principled reason to assume that within-trial drift rate variability is…”