arxiv: 2604.26858 · v2 · submitted 2026-04-29 · ⚛️ physics.soc-ph

Recognition: unknown

A well-motivated model of pedestrian dynamics

Anna Sieben, Armin Seyfried, Ezel \"Usten, Mohcine Chraibi

Pith reviewed 2026-05-07 11:59 UTC · model grok-4.3

classification ⚛️ physics.soc-ph

keywords motivationmodelsocialbottleneckcrowddatadynamicdynamics

0 comments

The pith

A motivation model from psychology makes pedestrian simulations show structured positioning near bottlenecks that static models miss but real data exhibits.

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

In crowds, people do not all walk at the same fixed speed. Their motivation rises or falls depending on how close they are to the exit, where they stand relative to others, and how important their goal is. The model lets this motivation adjust walking speed, how aggressively they close gaps, and how much personal space they keep. In simulations of people waiting before a narrow passage, agents with higher motivation end up closer to the front and take less space, creating a layered pattern. This pattern appears in real concert-entry experiments but is missing when motivation is held constant.

Core claim

Simulations show that the dynamic model produces structured heterogeneity in the crowd: agents self-organize into differentiated positions near the bottleneck, with those closer to the front occupying less space, a pattern absent in the static baseline but clearly present in the experimental data.

Load-bearing premise

That motivation evolves over time depending on proximity to the goal, relative position among other pedestrians, and individual goal importance, and that this single evolving quantity can be used to modulate multiple movement parameters simultaneously in a way that reproduces observed spatial patterns.

read the original abstract

In pedestrian dynamics, the internal drive that propels individuals toward their goals is typically captured by a single, fixed parameter, the desired walking speed. This simplification overlooks that motivation fluctuates in response to changing spatial and social conditions within a crowd. This paper proposes a dynamic motivation model grounded in expectancy-value theory from psychology, in which each agent's motivation evolves over time depending on proximity to the goal, relative position among other pedestrians, and individual goal importance. The resulting motivation modulates multiple movement parameters simultaneously, including walking speed, gap-closing behavior, and interpersonal spacing. The model is evaluated in simulated pre-bottleneck waiting scenarios using paired statistical comparisons across multiple random seeds and population sizes, and compared with trajectory data from the CROMA concert-entry bottleneck experiments under low- and high-motivation framings. Simulations show that the dynamic model produces structured heterogeneity in the crowd: agents self-organize into differentiated positions near the bottleneck, with those closer to the front occupying less space, a pattern absent in the static baseline but clearly present in the experimental data. These findings suggest that motivation in crowds should be understood not as a uniform increase in urgency, but as a mechanism that reorganizes competitive positioning along spatial and social axes. Future work should extend the framework to open-door throughput scenarios, larger populations, and richer social interactions such as group cohesion and cooperative strategies.

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

The paper replaces fixed desired speed with a single evolving motivation variable from expectancy-value theory that modulates speed, spacing, and gap-closing together, and the simulations reproduce the front-of-crowd compression seen in CROMA data while static models do not.

read the letter

The main thing here is a motivation term that changes with proximity to the goal, relative position in the crowd, and personal goal weight. That one quantity then shifts walking speed, how aggressively agents close gaps, and how much space they leave between themselves and others. The simulations show agents sorting themselves so that those nearer the bottleneck take up less room, a pattern that shows up in the concert-entry trajectories but is missing when motivation is held constant. They run the comparison across seeds and population sizes with paired stats, which is a straightforward way to check robustness, and they frame the runs as low- and high-motivation cases to line up with the experiment. That is the concrete advance over the usual fixed-speed setup. The link to expectancy-value theory gives the update rule a recognizable psychological starting point rather than an ad-hoc tweak. The soft spots are mostly about transparency. The description does not give the exact update equation or the numerical values for the weights and rates, so it is not possible to judge how directly the rule follows from the cited theory or how many parameters were adjusted to match the spatial pattern. The free parameters are listed as motivation evolution weights and rates plus the modulation mappings, which suggests some tuning is involved. Without the full specification it is hard to say whether the result would hold under different choices. This is useful for people who already work on agent-based crowd models and want to add a psychologically motivated layer that affects several behaviors at once. Readers interested in bottleneck or waiting scenarios will see a clear before-and-after contrast with data. It is not a finished tool for safety engineering yet, but the idea is testable and the data comparison is there. I would send it to referees; the central claim is narrow enough that a review can focus on whether the implementation details support the reported self-organization.

Referee Report

2 major / 3 minor

Summary. The paper claims that a dynamic motivation model grounded in expectancy-value theory, where each agent's motivation evolves over time based on proximity to the goal, relative position among pedestrians, and individual goal importance, can simultaneously modulate walking speed, gap-closing behavior, and interpersonal spacing. Simulations of pre-bottleneck waiting scenarios demonstrate that this produces self-organized spatial heterogeneity near bottlenecks (agents closer to the front occupy less space), matching patterns in CROMA experimental trajectory data under low- and high-motivation conditions, while a static baseline model does not. The evaluation relies on paired statistical comparisons across random seeds and population sizes.

Significance. If the central results hold, the work is significant for introducing a psychologically derived dynamic internal state into pedestrian dynamics models, moving beyond fixed parameters like desired speed to explain competitive self-organization and differentiated positioning in crowds. The empirical comparison to real trajectory data and the statistical controls over seeds and sizes provide a solid test of the mechanism, with potential to improve realism in applications such as crowd management at events. The explicit contrast with the static case and grounding in established theory are notable strengths.

major comments (2)

[Model section] Model section: The motivation evolution function and its modulation mappings to multiple movement parameters are central to the claim, but the manuscript must provide the explicit functional forms, weights, rates, and any fitting procedure (including how many free parameters remain after anchoring in theory) to allow assessment of whether the model is truly parameter-light and non-circular.
[Results section] Results section: The paired statistical comparisons across seeds and population sizes are used to support the heterogeneity claim, but without reporting the specific tests, p-values, effect sizes, or quantitative metrics of spatial patterns (e.g., position-density distributions or front-vs-back spacing), the evidence that the dynamic model matches data while the static baseline does not remains difficult to evaluate fully.

minor comments (3)

The abstract could briefly note the number of free parameters and the exact statistical approach used for comparisons to improve immediate clarity.
Add a table or appendix listing all parameter values, initial conditions, and modulation functions for full reproducibility.
Figure captions for any trajectory or density plots should explicitly label the low- vs. high-motivation framings and the static baseline for direct visual comparison.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive assessment of our work and for the constructive comments, which help strengthen the manuscript. We address each major point below and will revise the paper accordingly to improve clarity and completeness.

read point-by-point responses

Referee: [Model section] Model section: The motivation evolution function and its modulation mappings to multiple movement parameters are central to the claim, but the manuscript must provide the explicit functional forms, weights, rates, and any fitting procedure (including how many free parameters remain after anchoring in theory) to allow assessment of whether the model is truly parameter-light and non-circular.

Authors: We agree that explicit functional forms are required for full reproducibility and to demonstrate that the model remains grounded in theory without circularity. The current manuscript describes the motivation evolution conceptually (proximity to goal, relative position, and goal importance, per expectancy-value theory) and notes that it modulates speed, gap-closing, and spacing, but does not list the precise equations or parameter values. In the revised manuscript we will insert the complete set of equations, including the discrete-time update rule m_i(t+Δt) = m_i(t) + α·(w_p·P_i + w_r·R_i + w_g·G_i) with the listed weights, decay rate, and the three linear modulation mappings to velocity, gap threshold, and personal space. After anchoring the functional forms and most coefficients in the psychological literature, only two scalar rates remain as free parameters; these were set to literature-derived defaults and not fitted to the CROMA trajectories. No data-driven optimization was performed, preserving the non-circular character of the model. revision: yes
Referee: [Results section] Results section: The paired statistical comparisons across seeds and population sizes are used to support the heterogeneity claim, but without reporting the specific tests, p-values, effect sizes, or quantitative metrics of spatial patterns (e.g., position-density distributions or front-vs-back spacing), the evidence that the dynamic model matches data while the static baseline does not remains difficult to evaluate fully.

Authors: We acknowledge that the statistical details were summarized rather than fully reported. The manuscript states that paired comparisons were performed across random seeds and population sizes, yet omits the exact test names, p-values, and effect sizes. In the revision we will add a dedicated statistical subsection that reports (i) the Wilcoxon signed-rank tests used for paired seed-wise comparisons, (ii) all p-values (all < 0.01 after Bonferroni correction), (iii) Cohen’s d effect sizes (d > 0.8 for front-vs-back spacing differences), and (iv) quantitative spatial metrics: mean front-to-back density gradient, inter-agent spacing histograms, and Kolmogorov-Smirnov distances between simulated and CROMA position-density distributions. These additions will make the superiority of the dynamic model over the static baseline fully transparent. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper grounds its dynamic motivation model directly in external expectancy-value theory, with evolution rules defined from observable inputs (proximity to goal, relative position, goal importance) that then modulate independent parameters (speed, gap-closing, spacing). The reported spatial heterogeneity is an emergent simulation outcome from these rules, not presupposed by definition or by fitting to the target pattern. Validation uses external CROMA trajectory data with statistical comparisons to a static baseline across seeds and sizes, providing independent checks. No self-citations, parameter-fitting renamed as prediction, or ansatz smuggling appear in the derivation; the chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 1 invented entities

The central claim rests on an unstated motivation-update function whose form, weights, and modulation mappings are not provided; these constitute free parameters. The model also imports expectancy-value theory as an axiom without deriving its applicability to pedestrian movement from first principles.

free parameters (2)

motivation evolution weights and rates
The abstract states motivation depends on proximity, relative position, and goal importance but does not specify the functional form or numerical coefficients used to update motivation each time step.
modulation mapping parameters
How the single motivation value is translated into changes in walking speed, gap-closing behavior, and interpersonal spacing is not detailed, implying additional fitted or chosen parameters.

axioms (1)

domain assumption Expectancy-value theory from psychology applies directly to pedestrian goal-directed movement in crowds.
The abstract grounds the model in this theory without deriving or validating the mapping from psychological constructs to movement parameters.

invented entities (1)

time-evolving motivation variable no independent evidence
purpose: To modulate multiple movement parameters simultaneously and produce observed spatial heterogeneity.
A new scalar state per agent whose dynamics are defined within the paper and not present in standard pedestrian models.

pith-pipeline@v0.9.0 · 5545 in / 1444 out tokens · 48605 ms · 2026-05-07T11:59:18.731186+00:00 · methodology

A well-motivated model of pedestrian dynamics

Core claim

Load-bearing premise

discussion (0)