Machine learning to detect schedules using spatiotemporal data of behavior: A proof of concept

Aug 5, 2026

Upload by Aurelia

Traditional behavioral analysis often uses the single discrete response paradigm to reveal response patterns such as key pecks or lever presses. Advancing technology enables schedule detection using richer features, motivating this proof-of-concept study that applies spatiotemporal data to compare four machine learning algorithms for identifying the presence and components of time-based schedules in 12 rats. Logistic regression, SVMs, random forests, and neural networks distinguished programmed schedules, differentiating fixed- vs variable-space schedules. However, no approach reliably separated fixed-time from variable-time schedules, and performance was not consistently better across models, supporting spatiotemporal-feature-based schedule detection.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Single discrete response paradigm and its limits

Spatiotemporal features and reinforcement contingencies

Methods and Approach

Supervised machine learning framework

Algorithms compared

Results and Findings

Detection of schedule presence and components

Fixed vs variable space schedules

Fixed vs variable time schedules

Conclusion

Received: 6 October 2023 Accepted: 26 April 2025 DOI: 10.1002/jeab.70029 RESEARCH ARTICLE Machine learning to detect schedules using spatiotemporal data of behavior: A proof of concept Marc J. Lanovaz 1,2 | Varsovia Hernandez 3 | Alejandro Len 3 1cole de psychoéducation, Université de Montréal, Canada 2Centre de recherche de l’Institut universitaire en santé mentale de Montréal, Canada 3Centro de Investigaciones Biomédicas, Universidad Veracruzana, Mexico Correspondence Alejandro Len, Centro de Investigaciones Biomédicas, Universidad Veracruzana, C.P. 91190, Dr. Luiz Castelazo Ayala s/n, Col. Industrial, De las A´nimas, Xalapa, Veracruz, Mexico. Email: [aleleon@uv.mx](aleleon@uv.mx); 􀀁 Marc J. Lanovaz, Ecole de psychoéducation, Université de Montréal, C.P. 6128, succursale Centre-Ville, Montreal, QC, Canada, H3C 3J7 . [Email:](Email: marc.lanovaz@umontreal.ca)[ marc.lanovaz@umontreal.ca](Email: marc.lanovaz@umontreal.ca) Funding information Fonds de Recherche du Québec-Santé, Grant/Award Number: 269462 Editor-in-Chief: Suzanne H. Mitchell Handling Editor: Elizabeth Kyonka Abstract Traditionally, the experimental analysis of behavior has relied on the single discrete response paradigm (e.g., key pecks, lever presses, screen clicks) to identify behavioral patterns. However, the development and availability of new technology allow researchers to move beyond this paradigm and use other features to detect schedules. Thus, our study used spatiotemporal data to compare the accuracy of four machine learning algorithms (i.e., logistic regression, support vector classifiers, random forests, and artificial neural networks) in detecting the presence and the components of timebased schedules in 12 rats involved in a behavioral experiment. Using spatiotemporal data, the algorithms accurately identified the presence or absence of programmed schedules and correctly differentiated between fixed- and variable-space schedules. That said, our analyses failed to identify an algorithm to discriminate fixed-time from variable-time schedules. Furthermore, none of the algorithms performed systematically better than the others. Our findings provide preliminary support for the utility of using spatiotemporal data with machine learning to detect stimulus schedules. KEYW ORDS machine learning, neural network, spatiotemporal data, time-based schedule The experimental analysis of behavior typically involves environments that are manipulated by an experimenter to examine the effects of changes in these environments on the behavior of living organisms (Skinner, 1966) . These experimental manipulations have contributed to the development of a comprehensive, rigorous, and conceptually systematic natural science of behavior (Catania, 2013) . By manipulating environments, researchers have extensively studied the effects of stimulus delivery and behavior-dependent schedules in both human and nonhuman organisms (Kahng & Iwata, 2005; Schlinger et al., 2008) . Traditionally, the identification of these behavioral patterns has relied on the measurement of single discrete responses. This traditional approach, known as the single discrete response paradigm (Henton & Iversen, 1978), relies on behaviors selected for their ease of measurement and applicability such as lever presses for rats, key pecks for pigeons, and screen clicks for humans. However, an organism’s behavior is not confined to discrete responses in noncontrived environments. Organisms exhibit spatiotemporal continuity in their behavior, often displaying a diverse range of ecologically relevant behavioral patterns including foraging, water-seeking, and mating (Hernndez et al., 2023; Len et al., 2020, 2021) . A substantial body of empirical evidence indicates thatspatiotemporal features of behavior are highly sensitive to reinforcement contingencies (Baum & Rachlin, 1969; Hernndez et al., 2023; Len et al., 2021; Silva & Timberlake, 1997; Timberlake & Lucas, 19

FAQs

Why move beyond the single discrete response paradigm in behavioral experiments?

Technology and richer measurement capabilities allow detecting schedule structure using features beyond single discrete responses, improving access to behavioral pattern information.

Which machine learning algorithms were compared for schedule detection?

The study compared logistic regression, support vector classifiers, random forests, and artificial neural networks using spatiotemporal behavioral data.

What schedule-related distinctions did the algorithms achieve successfully?

The algorithms accurately identified whether programmed time-based schedules were present and differentiated fixed-space from variable-space schedules.

What limitation was observed for time-based schedule discrimination?

No algorithm reliably discriminated fixed-time from variable-time schedules, and none performed systematically better than the others.

Aug 5, 2026

Upload by Aurelia