Academic journal article Psychonomic Bulletin & Review

Moving beyond Qualitative Evaluations of Bayesian Models of Cognition

Academic journal article Psychonomic Bulletin & Review

Moving beyond Qualitative Evaluations of Bayesian Models of Cognition

Article excerpt

Published online: 19 September 2014

© Psychonomic Society, Inc. 2014

Abstract Bayesian models of cognition provide a powerful way to understand the behavior and goals of individuals from a computational point of view. Much of the focus in the Bayesian cognitive modeling approach has been on qualitative model evaluations, where predictions from the models are compared to data that is often averaged over individuals. In many cognitive tasks, however, there are pervasive individual differences. We introduce an approach to directly infer individual differences related to subjective mental representations within the framework of Bayesian models of cognition. In this approach, Bayesian data analysis methods are used to estimate cognitive parameters and motivate the inference process within a Bayesian cognitive model. We illustrate this integrative Bayesian approach on a model of memory. We apply the model to behavioral data from a memory experiment involving the recall of heights of people. A cross-validation analysis shows that the Bayesian memory model with inferred subjective priors predicts withheld data better than a Bayesian model where the priors are based on environmental statistics. In addition, the model with inferred priors at the individual subject level led to the best overall generalization performance, suggesting that individual differences are important to consider in Bayesian models of cognition.

Keywords Bayesian models of cognition · Bayesian data analysis · Episodic memory · Individual differences

(ProQuest: ... denotes formulae omitted.)

Introduction

Bayesian models of cognition (BMCs) have experienced a recent upsurge in popularity in the cognitive sciences. These models have made significant theoretical contributions to cognitive science in their accounts of why people behave as they do. The strength of BMCs (sometimes also referred to as rational models) is that they can be used to characterize the computational problems people face when trying to make sense of the world given the sparse and noisy input from our senses. Assuming that the mind solves inference problems in a Bayesian way, BMCs give a principled account of how we update our beliefs about the world given observed data, and how our prior knowledge about the world influences our judgment. These models have been applied to a broad range of areas in human cognition (Anderson, 1990) and specific areas such as reasoning (Oaksford & Chater, 1994), generalization (Tenenbaum & Griffiths, 2001), number concepts in children (Lee & Sarnecka, 2010), categorization (Huttenlocher, Hedges & Vevea, 2000), episodic memory (Shiffrin & Steyvers, 1997; Steyvers & Griffiths, 2008), and semantic memory (Hemmer & Steyvers, 2009; Steyvers, Griffiths, & Dennis, 2006). For an overview of BMCs of cognition see Oaksford & Chater (1998), but also see Mozer, Pashler, & Homaei (2008), Jones & Love (2011), Marcus & Davis (2013), and Bowers & Davis (2012a,b) for critiques of BMCs.

Traditionally, the focus in the Bayesian cognitive modeling approach has been on qualitative model evaluations, where predictions from the models are compared to data that is averaged over participants. At this qualitative level, BMCs can provide useful insights into the cognitive goals and computational mechanisms of average individuals. In many cognitive tasks, however, there are pervasive individual differences, e.g., in working memory (e.g., Unsworth, 2007), judgment and decision making (e.g., Vickers et al., 2006), and reinforcement learning (Steyvers, Lee, & Wagenmakers, 2009). While averaging across participants provides a powerful tool for analysis, it might also lead to mischaracterizations of the behavior of individuals, and of the underlying cognitive goals and processes. Estes (1956) cautioned against the uncritical use of averaged curves to determine effects of experimental treatments in the study of learning. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.