eCite Digital Repository

A Flexible and Efficient Hierarchical Bayesian Approach to the Exploration of Individual Differences in Cognitive-model-based Neuroscience


Ly, A and Boehm, U and Heathcote, A and Turner, BM and Forstman, B and Marsman, M and Matzke, D, A Flexible and Efficient Hierarchical Bayesian Approach to the Exploration of Individual Differences in Cognitive-model-based Neuroscience, Computational Models of Brain and Behavior, Wiley-Blackwell Publishing Ltd., AA Moustafa (ed), United States, pp. 467-480. ISBN 978-1-119-15906-3 (2018) [Research Book Chapter]

Copyright Statement

Copyright 2018 John Wiley & Sons

DOI: doi:10.1002/9781119159193.ch34


Cognitive‐model‐based neuroscience provides powerful methods of finding the brain areas supporting latent psychological processes. One of these methods is to identify areas whose activation is associated with individual differences in the parameters of cognitive models. We describe how to apply this approach based on Bayesian hierarchical models that are estimated without reference to neural covariates. This enables efficient exploration of the rich sets of covariates without the computational difficulties associated with refitting the cognitive model. Our approach, based on methodology originating from educational surveys (e.g., Mislevy, 1991; Mislevy, Beaton, Kaplan, & Sheehan, 1992), avoids overconfidence in inferences that is associated with performing frequentist tests on posterior point estimates (Boehm, Marsman, Matzke, & Wagenmakers, submitted). We show how to extend this approach to take account of uncertainty in generalizing from a sample of participants to the population (Ly, Marsman, & Wagenmakers, 2015), providing an assessment of whether findings will generalize to new samples. We illustrate the application of our methods to Forstmann et al.'s (2008) fMRI study of the relationship between activation in pre‐SMA and the Basal Ganglia and threshold setting in the LBA model (Brown & Heathcote, 2008), comparing their individual participant maximum‐likelihood estimates to both individual and hierarchical Bayesian estimates obtained using Differential‐Evolution Markov Chain Monte Carlo sampling (Turner, Sederberg, Brown, & Steyvers, 2013).

Item Details

Item Type:Research Book Chapter
Keywords:evidence accumulation model, correlations, plausible values, Bayesian estimation, hierarchical model, speed-accuracy trade-off, basal ganglia
Research Division:Psychology
Research Group:Cognitive and computational psychology
Research Field:Memory and attention
Objective Division:Expanding Knowledge
Objective Group:Expanding knowledge
Objective Field:Expanding knowledge in psychology
UTAS Author:Heathcote, A (Professor Andrew Heathcote)
ID Code:123607
Year Published:2018
Deposited By:Psychology
Deposited On:2018-01-15
Last Modified:2019-04-15

Repository Staff Only: item control page