Academic journal article Psychonomic Bulletin & Review

Modeling Simple Driving Tasks with a One-Boundary Diffusion Model

Academic journal article Psychonomic Bulletin & Review

Modeling Simple Driving Tasks with a One-Boundary Diffusion Model

Article excerpt

Published online: 3 December 2013

© Psychonomic Society, Inc. 2013

Abstract A one-boundary diffusion model was applied to the data from two experiments in which subjects were performing a simple simulated driving task. In the first experiment, the same subjects were tested on two driving tasks using a PC-based driving simulator and the psychomotor vigilance test. The diffusion model fit the response time distributions for each task and individual subject well. Model parameters were found to correlate across tasks, which suggests that common component processes were being tapped in the three tasks. The model was also fit to a distracted driving experiment of Cooper and Strayer (Human Factors, 50, 893-902, 2008). Results showed that distraction altered performance by affecting the rate of evidence accumulation (driftrate) and/or increasing the boundary settings. This provides an interpretation of cognitive distraction whereby conversing on a cell phone diverts attention from the normal accumulation of information in the driving environment.

Keywords Decision making . Reaction time analysis . Reaction time methods

Diffusion decision models have been successful in dealing with simple two-choice decision-making tasks (Ratcliff, 1978; Ratcliff& McKoon, 2008; Wagenmakers, 2009). There have been applications of these models in a variety of domains, such as psychology, neuroscience (Gold & Shadlen, 2001; Hanes & Schall, 1996; Philiastides, Ratcliff, & Sajda, 2006; Ratcliff, Cherian, & Segraves, 2003; Schall, Purcell, Heitz, Logan, & Palmeri, 2011; Smith & Ratcliff, 2004; Wong & Wang, 2006), neuroeconomics and decision making (Krajbich & Rangel, 2011; Roe, Busemeyer, & Townsend, 2001), and various clinical domains (White, Ratcliff, Vasey, & McKoon, 2010), and with a variety of subject populations, such as children (Ratcliff, Love, Thompson, & Opfer, 2012), older adults (Ratcliff, Thapar, & McKoon, 2010, 2011), aphasics (Ratcliff, Perea, Colangelo, & Buchanan, 2004), and children with ADHD (Mulder et al., 2010) and dyslexia (Zeguers et al., 2011). In these models, evidence toward one or the other of the alternatives is assumed to accumulate over time.

Recently, a diffusion model for one-choice tasks has been developed and fit both to data from the psychomotor vigilance test (PVT), a task used extensively in sleep deprivation research, and to data from a simple (one-choice) brightness detection task (Ratcliff& Van Dongen, 2011). In the PVT, a millisecond timer is displayed on a computer screen, and it starts counting up after delays between 2 and 12 s after the subject's last response. The subject's task is to hit a key as quickly as possible to stop the timer. When the key is pressed, the counter is stopped, and the response time (RT, in milliseconds) is displayed for 1 s.

Ratcliffand Van Dongen (2011) presented fits of the model to data from the PVT. In one analysis, they fit RT distributions (including hazard functions) from experiments with over 2,000 observations per RT distribution per subject. They also fit data in which the PVTwas tested every 2 h for 36 h of sleep deprivation. They found that driftrate was closely related to an independent measure of alertness, and this provided an external validation of the model.

The aim of this article is to examine whether the singlechoice diffusion model can be used to fit driving RT data. In one version, subjects are seated in front of a PC monitor with a gaming steeringwheel and foot pedals (accelerator and brake), and they follow approximately 100 feet behind a lead vehicle at about 65 m.p.h. There are two tasks. One is to brake when the lead vehicle brakes, which is signaled by the lead vehicle slowing and the brake lights turning on. The second is to drive around the lead vehicle into an unoccupied lane when the lead vehicle brakes.

Data from these two driving tasks and the PVT were used to test the one-choicemodel. …

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


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.