Knowing When to Move: Evidence Accumulation Models of Human Behavior in Traffic

Arkady Zgonnikov; Bhargav Nallapu; Felix van Waveren; Floor Bontje; Gustav Markkula; Leendert van Maanen

arxiv: 2606.00727 · v2 · pith:XF3V7Z5Ynew · submitted 2026-05-30 · 💻 cs.HC

Knowing When to Move: Evidence Accumulation Models of Human Behavior in Traffic

Floor Bontje , Felix van Waveren , Leendert van Maanen , Bhargav Nallapu , Gustav Markkula , Arkady Zgonnikov This is my paper

Pith reviewed 2026-06-28 18:08 UTC · model grok-4.3

classification 💻 cs.HC

keywords evidence accumulation modelstraffic behaviordecision makinghuman factorsperception-action couplingreview

0 comments

The pith

Evidence accumulation models applied to traffic behavior differ systematically by modeling level and architecture.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This review synthesizes 28 studies applying evidence accumulation models to traffic-related decisions from 2014 onward. It groups the work along two axes: discrete decision-making versus continuous action control, and stand-alone models versus those embedded in larger perception-action systems. These groupings align with consistent differences in how the models are built, parameterized, fed data, and checked against observations. The synthesis aims to show how task demands shape model choices and to flag open methodological issues for traffic modeling and broader decision research.

Core claim

Organizing the literature by modeling level (discrete versus continuous) and model architecture (stand-alone versus embedded) produces systematic differences in architecture, parameterization, data usage, and validation strategies that reflect the specific demands of traffic tasks.

What carries the argument

Two organizing dimensions: modelling level (discrete decision-making versus continuous action control) and model architecture (stand-alone decision model versus embedded component within perception-action frameworks).

If this is right

Discrete decision models and continuous control models will require different data streams and fitting procedures.
Embedded models will draw on different validation checks than stand-alone models.
Task demands in traffic will continue to dictate whether evidence accumulation is treated as the full explanation or only one module.
Laboratory extensions to sustained and time-varying tasks can test whether the same distinctions appear outside driving.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The patterns may guide model selection when new traffic scenarios are studied, such as multi-driver interactions.
Neurophysiological recordings could be used to test which perceptual signals supply the evidence in embedded models.
Extending the two dimensions to other embodied domains might reveal whether traffic is special or follows a general rule.

Load-bearing premise

The 28 selected studies represent the main ways evidence accumulation models have been used in traffic and the two chosen dimensions capture the primary variations across them.

What would settle it

A new study that applies an evidence accumulation model to traffic behavior but shows no alignment between its modeling level or architecture and its choices of parameterization, data, or validation.

read the original abstract

Evidence accumulation models provide a formal framework for studying decision making as a dynamic process unfolding over time. While these models have been extensively developed and reviewed in laboratory paradigms, their structured application in complex, ecologically valid domains has received comparatively little attention. Road traffic is a particularly relevant context for studying sustained, embodied perception action behavior, where decisions unfold under time pressure and involve continuous control and ongoing perception-action coupling. Examining how EAMs have been applied in this domain may therefore offer insights beyond discrete laboratory tasks toward decision making in real-world behavior. This semi-systematic review synthesizes 28 studies (2014-2026) applying EAMs to traffic-related behavior. We organize the literature along two dimensions: 1) modelling level, distinguishing models at the level of discrete decision-making and models at the level of continuous action control, and 2) model architecture, distinguishing evidence accumulation as either a stand-alone decision model or an embedded component within broader perception-action or interaction frameworks. These distinctions are associated with systematic differences in model architecture, parameterization, data usage, and validation strategies, reflecting task specific demands. By providing a structured overview of these patterns, this review clarifies how EAMs are currently instantiated in traffic contexts and highlights methodological challenges and future directions both in traffic modelling and in modelling of decision-making more broadly. Promising directions include laboratory work on evidence accumulation in sustained and time-varying tasks, interactive multi-individual decision-making, and the use of neurophysiological measures to identify the perceptual evidence underlying complex perception-action behavior.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

A useful 2D organizing frame for EAMs in traffic that highlights task differences, but the synthesis depends on an unverified sample of 28 studies.

read the letter

The paper's core move is to sort existing EAM work in traffic along two axes—discrete decision versus continuous control, and stand-alone versus embedded in larger architectures—and to note that these choices track differences in parameterization, data, and validation. That framing is the main contribution.

It does a reasonable job pulling the literature together and pointing out that traffic tasks often require sustained, coupled perception-action loops rather than isolated choices. The suggested future directions, such as neurophysiological grounding or multi-agent extensions, follow logically from the review.

The main limitation is the lack of detail on how the 28 studies were chosen. The abstract labels it semi-systematic but supplies no search strategy, databases, inclusion rules, or bias checks. If the sample over-represents certain model types or misses key papers that fall outside the two dimensions, the claimed associations could be artifacts. The stress-test concern about representativeness therefore lands; without the methods section it is hard to judge whether the two axes really capture the primary variation.

This is aimed at researchers in human factors, transportation modeling, or anyone adapting EAMs to continuous real-world tasks. It is not a new empirical result or derivation, so it will not reshape the core theory, but the organization can help people pick models for applied work.

It deserves peer review. The topic is relevant and the framing is clear enough that referees can check the selection process and tighten the claims.

Referee Report

1 major / 1 minor

Summary. The manuscript is a semi-systematic review of 28 studies (2014-2026) applying evidence accumulation models (EAMs) to traffic-related human behavior. It organizes the literature along two dimensions—modelling level (discrete decision-making versus continuous action control) and model architecture (stand-alone EAM versus embedded component within broader perception-action frameworks)—and reports that these distinctions correlate with systematic differences in model architecture, parameterization, data usage, and validation strategies that reflect task-specific demands. The review discusses methodological challenges and outlines future directions including laboratory studies on sustained tasks, multi-individual decision-making, and neurophysiological measures.

Significance. If the synthesis and reported associations hold, the review supplies a structured framework for extending EAMs beyond discrete laboratory paradigms into sustained, embodied, real-world domains such as traffic behavior. This could usefully inform both applied transportation modeling and theoretical work on perception-action coupling, while the suggested directions (neurophysiological grounding, interactive settings) offer concrete avenues for integration with empirical data.

major comments (1)

[Methods (review procedure section)] The manuscript provides no description of search strategy, databases, inclusion/exclusion criteria, or bias assessment for the semi-systematic review that yielded the 28 studies. Without these details it is impossible to determine whether the sample is representative or whether the patterns attributed to the two chosen dimensions are robust versus selection artifacts. This directly affects the central claim that the distinctions are associated with systematic differences reflecting task-specific demands.

minor comments (1)

[Abstract] The date range '2014-2026' in the abstract includes future years; clarify whether this reflects publication dates, preprint dates, or an error and ensure consistency throughout the text.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback. We address the major comment below and will revise the manuscript to incorporate the suggested improvements.

read point-by-point responses

Referee: [Methods (review procedure section)] The manuscript provides no description of search strategy, databases, inclusion/exclusion criteria, or bias assessment for the semi-systematic review that yielded the 28 studies. Without these details it is impossible to determine whether the sample is representative or whether the patterns attributed to the two chosen dimensions are robust versus selection artifacts. This directly affects the central claim that the distinctions are associated with systematic differences reflecting task-specific demands.

Authors: We agree that a transparent description of the review procedure is required. The manuscript currently omits these methodological details, which limits evaluation of the sample's representativeness and the robustness of the observed associations. In the revised manuscript we will add a dedicated 'Review Procedure' subsection that specifies the search strategy (keywords and Boolean terms), databases searched, inclusion/exclusion criteria applied to identify the 28 studies, and any steps taken to mitigate selection bias. This addition will directly strengthen the evidential basis for the central claims. revision: yes

Circularity Check

0 steps flagged

No circularity: literature review with no derivations or fitted quantities

full rationale

This is a semi-systematic review synthesizing 28 external studies (2014-2026) on evidence accumulation models in traffic. It organizes literature along two descriptive dimensions (modelling level and architecture) but contains no equations, parameter fittings, predictions, or derivation chains. All claims rest on the cited external studies rather than internal definitions, self-citations, or renamings. No load-bearing step reduces to the paper's own inputs by construction. The representativeness concern raised by the skeptic is a question of review methodology, not circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The review rests on the domain assumption that EAMs are applicable to sustained traffic behavior and on the authors' choice of two organizing dimensions as the primary lens for the literature.

axioms (1)

domain assumption Evidence accumulation models can be meaningfully applied to traffic-related perception-action behavior
The entire synthesis presupposes that EAMs, originally developed for discrete lab tasks, extend usefully to continuous, embodied traffic decisions.

pith-pipeline@v0.9.1-grok · 5829 in / 1173 out tokens · 32515 ms · 2026-06-28T18:08:42.480551+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages · 1 internal anchor

[1]

Bachmann, D., & van Maanen, L. (2024). Towards the application of evidence accumulation models in the design of (semi-)autonomous driving systems – an attempt to overcome the sample size roadblock. International Journal of Human-Computer Studies, 185, 103220. https://doi.org/10.1016/j.ijhcs.2024.103220 Boag, R. J., Strickland, L., Heathcote, A., Neal, A.,...

work page doi:10.1016/j.ijhcs.2024.103220 2024
[2]

Active inference as a unified model of collision avoidance behavior in human drivers

https://doi.org/10.1038/nn.3248 Pekkanen, J., Giles, O. T., Lee, Y . M., Madigan, R., Daimon, T., Merat, N., & Markkula, G. (2022). Variable-Drift Diffusion Models of Pedestrian Road-Crossing Decisions. Computational Brain & Behavior, 5(1), 60–80. https://doi.org/10.1007/s42113-021-00116-z Ratcliff, R. (1978). A theory of memory retrieval. Psychological R...

work page internal anchor Pith review Pith/arXiv arXiv doi:10.1038/nn.3248 2022

[1] [1]

Bachmann, D., & van Maanen, L. (2024). Towards the application of evidence accumulation models in the design of (semi-)autonomous driving systems – an attempt to overcome the sample size roadblock. International Journal of Human-Computer Studies, 185, 103220. https://doi.org/10.1016/j.ijhcs.2024.103220 Boag, R. J., Strickland, L., Heathcote, A., Neal, A.,...

work page doi:10.1016/j.ijhcs.2024.103220 2024

[2] [2]

Active inference as a unified model of collision avoidance behavior in human drivers

https://doi.org/10.1038/nn.3248 Pekkanen, J., Giles, O. T., Lee, Y . M., Madigan, R., Daimon, T., Merat, N., & Markkula, G. (2022). Variable-Drift Diffusion Models of Pedestrian Road-Crossing Decisions. Computational Brain & Behavior, 5(1), 60–80. https://doi.org/10.1007/s42113-021-00116-z Ratcliff, R. (1978). A theory of memory retrieval. Psychological R...

work page internal anchor Pith review Pith/arXiv arXiv doi:10.1038/nn.3248 2022