Academic journal article Psychonomic Bulletin & Review

The Urgency-Gatingmodel Can Explain the Effects of Early Evidence

Academic journal article Psychonomic Bulletin & Review

The Urgency-Gatingmodel Can Explain the Effects of Early Evidence

Article excerpt

Published online: 24 September 2015

© Psychonomic Society, Inc. 2015

Abstract In a recent report, Winkel, Keuken, van Maanen, Wagenmakers & Forstmann (Psychonomics Bulletin and Review 21(3): 777-784, 2014) show that during a random-dot motion discrimination task, early differences in motion evidence can influence reaction times (RTs) and error rates in human subjects. They use this as an argument in favor of the drift-diffusion model and against the urgency-gating model. However, their implementation of the urgency-gating model is incomplete, as it lacks the low-pass filter that is necessary to deal with noisy input such as the motion signal used in their experimental task. Furthermore, by focusing analyses solely on comparison of mean RTs they overestimate how long early information influences individual trials. Here, we show that if the urgency-gating model is correctly implemented, including a low-pass filter with a 250 ms time constant, it can successfully reproduce the results of the Winkel et al. experiment.

Keywords Decision making . Response time models . Computational modeling . Drift-diffusion model . Perceptual discrimination

(ProQuest: ... denotes formulae omitted.)

Introduction

Many current models of decision-making (Bogacz et al., 2006; Busemeyer & Townsend, 1993; Gold & Shadlen, 2007; Ratcliff, 1978; Usher & McClelland, 2001) suggest that during deliberation, the brain integrates evidence in favor of each choice until the total integrated evidence reaches a threshold, the setting of which determines an accuracy criterion. The most influential of these is called the Bdrift-diffusion model" (DDM) (Ratcliff, 1978), and has been used to explain reaction times (RTs) and error rates as well as neural activity in a wide variety of decision-making tasks (Churchland, Kiani, & Shadlen, 2008; Domenech & Dreher, 2010; Gold & Shadlen, 2007; Heekeren, Marrett, & Ungerleider, 2008; Palmer, Huk, & Shadlen, 2005; Ratcliffet al., 2007; Roitman & Shadlen, 2002). One feature of the DDM is that, like any integrator, it is robust in the presence of noise.

However, in most natural situations the environment can change rapidly, and animals must be able to respond to such changes quickly. Integrators are not ideal in this regard because in order to change an ongoing decision they must first reverse the previously integrated evidence for the old choice. In contrast, a low-pass filter with a short time constant can respond more quickly to changes in sensory evidence while remaining robust to noise. For this reason, we and others have suggested that sensory evidence is not integrated continuously but rather low-pass filtered and combined with an independent signal related to the urgency for making a choice - and that together these bring neural activity to a decision threshold (Cisek, Puskas, & El-Murr, 2009; Ditterich, 2006; Thura et al., 2012). We have called this the Burgency-gating model" (UGM).

Importantly, nearly all of the experiments typically cited in support of the drift-diffusion model have used tasks in which subjects view stimuli in which the informational content pertinent to the decision is held constant throughout each trial. For random-dot motion tasks this means that the underlying direction and degree of coherent motion are constant and always present during any given trial. However, if evidence is held constant in this way, no conclusive distinction can be made between the DDM and the UGM because under such conditions both of these models make very similar predictions that are difficult to distinguish experimentally. Thus, the best way to distinguish these models is to design tasks in which the evidence changes within trials - a scenario for which the models make qualitatively different predictions.

Specifically, because it keeps a running sum of all sensory evidence, the DDM predicts that adding or removing sensory evidence at any point prior to the decision will affect the time at which the integrated sum of evidence crosses the threshold, thereby affecting the response time. …

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.