Variable-Drift Diffusion Models of Pedestrian Road-Crossing Decisions

Abstract

Human behavior and interaction in road traffic is highly complex, with many open scientific questions of high applied importance, not least in relation to recent development efforts toward automated vehicles. In parallel, recent decades have seen major advances in cognitive neuroscience models of human decision-making, but these models have mainly been applied to simplified laboratory tasks. Here, we demonstrate how variable-drift extensions of drift diffusion (or evidence accumulation) models of decision-making can be adapted to the mundane yet non-trivial scenario of a pedestrian deciding if and when to cross a road with oncoming vehicle traffic. Our variable-drift diffusion models provide a mechanistic account of pedestrian road-crossing decisions, and how these are impacted by a variety of sensory cues: time and distance gaps in oncoming vehicle traffic, vehicle deceleration implicitly signaling intent to yield, as well as explicit communication of such yielding intentions. We conclude that variable-drift diffusion models not only hold great promise as mechanistic models of complex real-world decisions, but that they can also serve as applied tools for improving road traffic safety and efficiency.

Subtle motion cues by automated vehicles can nudge human drivers’ decisions: Empirical evidence and computational cognitive model

“…Taken together, these findings strengthen the existing notion that noisy accumulation of evidence coming from relevant perceptual variables (distance, TTA, and acceleration) is a key mechanism underlying human tactical decisions in traffic (Pekkanen et al, 2022;Markkula et al, 2022). Besides added explanatory value, we believe our seemingly straightforward addition of the acceleration term to the previously proposed model makes an important step towards applications of the model for understanding and managing human-AV interactions in traffic.…”

“…Recently, drift-diffusion models have been applied to human decisions in traffic, e.g. gap acceptance in pedestrian crossing (Markkula et al, 2022;Pekkanen et al, 2022) and unprotected left turns .…”

“…The key assumption of these versions of the model is that the drift rate tracks the dynamically changing size of the gap, relative to a certain critical value. The decision boundary is usually assumed to be fixed (Ratcliff et al, 2016;Pekkanen et al, 2022) but can also be collapsing over time to reflect choice urgency Drugowitsch et al, 2012).…”

Race Lévy flights: A mathematically tractable framework for studying heavy-tailed accumulation noise

“…The DDM has also been useful in answering questions about different contexts within each of these areas of decision-making, such as the ability of humans to acheieve performance optimality (Bogacz et al, 2006;Evans and Brown, 2017;Drugowitsch et al, 2012;Tajima et al, 2016), and the impacts of aging on performance in a range of different cognitive tasks (Theisen et al, 2021;Ratcliff et al, 2003Ratcliff et al, , 2006Servant and Evans, 2020;Thapar et al, 2003;Ratcliff et al, 2004b,c;Starns and Ratcliff, 2010). Moreover, the DDM has also been applied to data from more real life situations, such human economical decisions (Clithero, 2018) and driver behavior (Pekkanen et al, 2022;Tillman et al, 2017b). For more applications, we refer the interested readers to (Ratcliff et al, 2016;Forstmann et al, 2016;Evans and Wagenmakers, 2020;Evans, 2021).…”

Numerical Approximation of the First-Passage Time Distribution of Time-Varying Diffusion Decision Models: A Mesh-Free Approach

“…Importantly, these psychologically meaningful parameters have allowed the DDM to answer questions of interest in the psychology, cognitive science, and neuroscience literature [8,11,9]. For example, the DDM has been used to better understand human economical decisions (e.g., gambling behavior [28], value-based decision making [29], and preference-based decision making [30,31]), applied decisions such as driving behavior [32,33,34], clinical questions (e.g., aging [35,36], autism spectrum disorder [37,38,39], attention deficit hyperactivity disorder [40,41], and dyslexia [42]), and theoretical questions such as performance optimality [43,44,45,46], just to name a few. While the standard DDM has been successful in capturing and explaining a wide range of behavioral patterns, several new variations of the model have been proposed based on ideas of neural plausibility, such as models with evidence leakage (e.g., the Ornstein-Uhlenbeck model [47]; see also [16,19]), boundaries that collapse over time (i.e., collapsing boundaries models [48,49]), and/or an urgency signal that amplifies the accumulated evidence with increasing time (i.e., urgency signal/gating models [19,17]).…”