Identifying and modelling cognitive biases in mobility choices

TL;DR

The paper investigates whether daily mobility decisions are fully rational and how cognitive biases shape them. It combines a survey-based calibration with an agent-based simulator implemented in the GAMA platform to model four mobility modes across six criteria using the weighted-score approach $score_i(m) = \sum_{c \in crits} val(m,c) \cdot prio_i(c)$. By analyzing responses from 625 participants, the authors identify biases such as confirmation and overestimation that shift perceived mode values and priorities. Two simulation experiments show that including biases yields distributions closer to observed real-world shares, validating the approach and highlighting the need for bias-aware designs in serious games about mobility transitions.

Abstract

This report presents results from an M1 internship dedicated to agent-based modelling and simulation of daily mobility choices. This simulation is intended to be realistic enough to serve as a basis for a serious game about the mobility transition. In order to ensure this level of realism, we conducted a survey to measure if real mobility choices are made rationally, or how biased they are. Results analysed here show that various biases could play a role in decisions. We then propose an implementation in a GAMA agent-based simulation.

Figures (9)

• Figure 1: Distribution of habitual mobility mode over 625 respondents
• Figure 2: Priority profiles of the 4 subpopulations wrt usual mobility mode
• Figure 3: Comparing the evaluation of modes by users / non-users
• Figure 4: Rational choice of mobility for users of the 4 modes, based on personal DECLARED values
• Figure 5: Rational choice of mobility for users of the 4 modes, based on average CROWDSOURCED values