- Poster presentation
- Open Access
Cooperation/supervision of a habit by a cognitive strategy in a goal-directed navigational paradigm
© Hanoune et al. 2015
- Published: 4 December 2015
- Cognitive Strategy
- Unitary Code
- Sensory Motor
- Versus Figure
- Learning Stage
The Stimulus-Response (S-R) theory and Tolman's Cognitive Theory of behavior control both issued from behaviorism in the early 20th century still provide a relevant general framework to account for animal reward-based adaptive behavior. In this paper, we propose a new paradigm for representing and implementing both the cognitive strategy and the S-R habit strategy within a unitary coding frame. Based on a parallel learning of both strategies, the model explains how the fast learning cognitive strategy can supervise and accelerate the slow learning S-R habit strategy; and also how. In late learning stages, the habit strategy can overcome the cognitive. This parallel representation is inspired by the cortico-basal functional loops  and the cooperation between the cognitive associative loop, including the dorso-medial striatum and the mPF; and the sensory-motor loop, associated to the sensory motor cortex in relation with the dorso-lateral striatum.
The implementation of S-R habit strategy is based on a neural modified version of the classical Q-learning and is based on the model of , emulating the functioning of the sensory-motor loop. The states of the model are represented by hippocampal transitions, representing associations between two consecutive place-cells during the exploration of the environment, learned in the CA1-CA3 regions of the hippocampus. The cognitive strategy is based on a map representation of the environment namely the cognitive map . Based on the association between learned transitions, the cognitive map allows the back-propagation of a reward within a tree, allowing the selection of the shortest path to the goal. While the cognitive map is quickly learned, the Q-values associated with the Q-learning are slower to acquire. On the other hand, the Q-learning tends to be more accurate than the cognitive map when fully learned.
This work was supported by the ANR-NEUROBOT project (ANR-BLAN-SIMI2-LS-100617-13-01).
- DeLong MR, Strick PL: Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annual Rev of Neurosci. 1986, 9 (1): 357-381.View ArticleGoogle Scholar
- Hirel J, Gaussier P, Quoy M, Banquet J-P, Save E, Poucet B: The Hippocampo-cortical Loop: Spatio-Temporal Learning and Goal-oriented Planning in Navigation. Neural Networks. 2013, 43: 8-21.PubMedView ArticleGoogle Scholar
- O'Keefe J, Nadel L: The hippocampus as a cognitive map. Oxford University. 1978Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.