We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Signal neutrality, scalar property, and collapsing boundaries as consequences of a learned multi-timescale strategy.
- Authors
Manneschi, Luca; Gigante, Guido; Vasilaki, Eleni; Del Giudice, Paolo
- Abstract
We postulate that three fundamental elements underlie a decision making process: perception of time passing, information processing in multiple timescales and reward maximisation. We build a simple reinforcement learning agent upon these principles that we train on a random dot-like task. Our results, similar to the experimental data, demonstrate three emerging signatures. (1) signal neutrality: insensitivity to the signal coherence in the interval preceding the decision. (2) Scalar property: the mean of the response times varies widely for different signal coherences, yet the shape of the distributions stays almost unchanged. (3) Collapsing boundaries: the "effective" decision-making boundary changes over time in a manner reminiscent of the theoretical optimal. Removing the perception of time or the multiple timescales from the model does not preserve the distinguishing signatures. Our results suggest an alternative explanation for signal neutrality. We propose that it is not part of motor planning. It is part of the decision-making process and emerges from information processing on multiple timescales. Author summary: Humans and animals integrate sensory information before making a decision. The integration rate varies depending on the task. While driving could require quick reactions, evaluating the authenticity of a painting typically requires long observations. Consequently, the concept of representations created over multiple timescales appears necessary. Nevertheless, there is a lack of theoretical research that exploits multiple timescales, despite experimental evidence for the variety of integration rates. We, therefore, developed a decision-making model based on simple integrators with multiple characteristic times. We analysed its behaviour on a highly volatile, biologically relevant task. Through reward maximisation based on trial and error, the model discovers an effective strategy that is surprisingly different and more robust than the "classical" single timescale approach. This learned strategy exhibits a remarkable agreement with experimental findings, suggesting a fundamental role of multiple timescales for decision-making. Our abstract model achieves a degree of biological realism while performing robustly in different environments.
- Subjects
TIME perception; BUILDING reinforcement; NEUTRALITY; DECISION making; INFORMATION processing; REINFORCEMENT learning
- Publication
PLoS Computational Biology, 2022, Vol 18, Issue 8, p1
- ISSN
1553-734X
- Publication type
Article
- DOI
10.1371/journal.pcbi.1009393