Australasian Mathematical Psychology Conference 2019

Revisiting collapsing boundaries

Roger Ratcliff
Ohio University
Philip L. Smith
Melbourne School of Psychological Sciences, University of Melbourne

Recently O’Connell, Shadlen, Wong-Lin, and Kelly (2018) have argued strongly for evidence accumulation models of decision-making with collapsing bounds or urgency signals. They present some arguments based on human behavioral data, but then argue for nonstationarity boundaries or drift rates based on neurophysiological data. Conditional accuracy functions are a way of examining whether accuracy falls over the time course of processes as it would with urgency of collapsing bounds. We present evidence from both human and monkey data that show little evidence for such nonstationarity. We also present results from single and dual random walk processes (surrogates for diffusion processes) showing that paths prior to hitting a decision boundary show similar patterns to neural firing rates in monkeys. We finally review the literature on neurophysiological data and find that although the overall patterns support evidence accumulation models, there are significant differences in the patterns from individual studies.