Bayesian modeling of the Mnemonic Similarity Task using multinomial processing trees

Abstract

The Mnemonic Similarity Task (MST: Stark et al., 2019) is a modified recognition memory task designed to place strong demand on pattern separation. The sensitivity and reliability of the MST make it an extremely valuable tool in clinical settings. We develop new cognitive models, based on the multinomial processing tree framework, for two versions of the MST. The models are implemented as generative probabilistic models and applied to behavioral data using Bayesian graphical modeling methods. We demonstrate how the combination of cognitive modeling and Bayesian methods allows for flexible and powerful inferences about performance on the MST. These demonstrations include latent-mixture extensions for identifying individual differences in decision strategies, and hierarchical extensions that measure fine-grained differences in the ability to detect lures. One key finding is that the availability of a "similar" response in the MST reduces individual differences in decision strategies and allows for more direct measurement of recognition memory.

Cognitive modeling of the Mnemonic Similarity Task as a digital biomarker for Alzheimer’s Disease

“…Previously, we demonstrated that the traditional REC correlated with π, while LDI correlated with λ (Lee and Stark, 2023, previously denoted as τ). Our first goal was to assess the relationship among the traditional and modeled metrics in the two datasets (Stark et al, 2013; Trelle et al, 2021).…”

“…Cognitive modeling provides a useful tool for inferring latent psychological variables beyond traditional measurements. Previously, we used cognitive modeling to model subject-level performance on the MST in young adults (Lee and Stark, 2023) using the multinomial processing tree (MPT) framework, a common approach for cognitive modeling of recognition memory tasks. The MPT framework assumes that cognitive processes can be divided into discrete categories or decision points (Fig 1B).…”

“…Cognitive modeling of individual’s memory can give a richer understanding of mechanisms (Norman et al, 2001) and how these are altered by aging or cognitive impairments (Lee et al, 2020; Chwiesko et al, 2023; Mulhauser et al, 2023). Recently we developed a cognitive model to analyze performance on the MST using Bayesian methods that both fit individual participant performance and identified individual differences in memory and response strategies (Lee and Stark, 2023).…”

A response time model of the three-choice Mnemonic Similarity Task provides stable, mechanistically interpretable individual-difference measures.

“…Our work fits in with other recent process models of performance in this task. One such approach used Multinomial Processing Tress (MPT) to distinguish remembering and discrimination from each other and from guessing [18]. A key advantage of this model is that it leverages previous psychometric calibrations of lure item similarity [9] to support the distinction of discrimination-based processing, as discrimination should get progressively more difficult with higher-similarity items.…”

A New Measure of Mnemonic Discrimination Applicable to Recognition Memory Tests with Continuous Variation in Novel Stimulus Interference

“…The MST LDI is generally quantified as the probability that a lure image is correctly classified as "similar," minus the probability that a foil image is mistakenly classified as "similar". Other analytical methods have been proposed for the MST, but they all require clear distinctions between which images in the test set are lures and foils (21,22). However, in clinically applicable memory tests, the differences between testing images may not be so significant that one group can be classified as "foils" while the others are "lures."…”