A simple and flexible Bayesian method for inferring step changes in cognition

Abstract

Human behavioral data often shows patterns of sudden change over time. Sometimes the causes of these step changes are internal, such as learning curves changing abruptly when a learner implements a new rule. Sometimes the cause is external, such a people's opinions about a topic changing in response to a new relevant event. Detecting change points in sequences of binary data is a basic statistical problem, with many existing solutions, but they seem rarely to be used in psychological modeling. We develop a simple and flexible Bayesian approach to modeling step changes in cognition, implemented as a graphical model in JAGS. The model is able to infer how many change points are justified by the data, as well as the location of the change points. The basic model is also easily extended to include latent-mixture and hierarchical structures, allowing it to be tailored to specific cognitive modeling problems. We demonstrate the adequacy of the basic model by applying it to the classic Lindisfarne Scribes problem, and the flexibility of the modeling approach through two new applications. The first involves a latent-mixture model to determine if individuals learn categories incrementally or in discrete stages. The second involves a hierarchical model of crowd-sourced predictions about the winner of the U.S. National Football League's Most Valuable Player for the 2016-2017 season.

Applying Bayesian cognitive models to decisions to drive after drinking

“…Our modeling approach builds upon the Bayesian change‐point method of Lee [13]. Given a sequential (ordered) set of repeated choices, this methodology is designed to estimate periods in which the probability of selecting one alternative over another changes.…”

“…Given a sequential (ordered) set of repeated choices, this methodology is designed to estimate periods in which the probability of selecting one alternative over another changes. We now summarize our application of the model and refer the reader to Lee [13] for additional, technical details. First, we describe the model under a single cost condition.…”

“…The model parameters across all cost conditions, for each participant, are estimated jointly in a fully Bayesian fashion. We apply a ‘spike and slab’ [13] prior over the six possible values of τ to simultaneously determine whether or not a change point exists (if the change point is estimated to be τ = 1, then one set of ride and drive probabilities are sufficient) and what values of τ produce the change point. The values of the probabilities and change points are selected according to the posterior mean and mode, respectively.…”

“…The Bayesian modeling framework we employ [13] is designed to distinguish between compensatory and non‐compensatory decision‐making strategies [14]. Compensatory decision strategies involve consideration of all the attributes involved and necessitate trade‐offs between these attributes when making choices [15].…”

Decision Making and Alcohol: Health Policy Implications

“…The costs (in either money or time) of the safe alternatives were systematically varied. Analysis tested whether individuals’ choice patterns could be classified into one of two general strategy types (using a hierarchical Bayesian model for inferring two-step changes; Lee, 2018): (a) a compensatory (“rational”) strategy, where the choice between AID and the safe alternative integrated all choice factors (e.g., number of drinks, cost of rides, time difference between riding and driving), or (b) a noncompensatory (“heuristic”) strategy, where the probability of choosing is based on a simple rule, ignoring some aspects of the decision for some choices. Multiple samples found that the majority (> 80%) of participants’ choice behavior could be clearly classified into one of these two overall strategies.…”