A methodological framework to incorporate psychophysiological indicators into transportation modeling

Abstract

Reporting and hypothetical biases are inherent to canonical methods of transportation data collection and had implied that analysis in this field has often neglected aspects that are strong behavioral drivers, such as uncertainty, physical effort or stress. Granular information on these aspects would allow measuring their valuation and/or addressing a pervasive source of endogeneity. Recent advances in miniaturization and data processing, as well as evidence that indicators from biosensors correlate with psychophysiological states and emotions, suggest that there is an opportunity to close this gap by collecting a new type of data from transportation users. This research works on leveraging this opportunity by putting forward, illustrating and testing a methodological framework to incorporate psychophysiological indicators gathered from biosensors into transportation choice behavioral modeling. The proposed framework adapts the integrated choice and latent variable approach by incorporating the psychophysiological responses as additional indicators of a latent psychophysiological state that enters the utility. For the practical implementation of the proposed framework we also consider a specific form of aggregation of the indicators across time to avoid the curse of dimensionality arising from the unmanageably large number of folds for integration. The proposed framework is illustrated and validated using Monte Carlo simulations. Besides, a prototype field experiment was designed and performed to confirm the validity of three crucial components of the proposed framework: i) the relation between transportation markers and emotions; ii) the possibility of measuring those emotions trough biosensors installed on travelers, iii) and the validity of the proposed aggregation needed for practicality. In the experiment, a public transport user travelled wearing a bracelet with a Printed Circuit Board that integrated tiny biosensors to capture electrodermal activity, heart rate variation, temperature and acceleration. Results confirm the hypotheses, enabling future massive data collection efforts to take full advantage of the proposed framework.