| Paper: | PS-2B.48 | ||
| Session: | Poster Session 2B | ||
| Location: | H Fläche 1.OG | ||
| Session Time: | Sunday, September 15, 17:15 - 20:15 | ||
| Presentation Time: | Sunday, September 15, 17:15 - 20:15 | ||
| Presentation: | Poster | ||
| Publication: | 2019 Conference on Cognitive Computational Neuroscience, 13-16 September 2019, Berlin, Germany | ||
| Paper Title: | A mechanistic account of transferring structural knowledge across cognitive maps | ||
| Manuscript: | Click here to view manuscript | ||
| License: |  This work is licensed under a Creative Commons Attribution 3.0 Unported License. |
||
| DOI: | https://doi.org/10.32470/CCN.2019.1324-0 | ||
| Authors: | Shirley Mark, Rani Moran, Thomas Parr, Steve Kennerley, University College London, United Kingdom; Tim Behrens, University of Oxford, United Kingdom | ||
| Abstract: | Animals can transfer knowledge that was learnt previously and infer when this knowledge is relevant. Frequently, the relations between elements in an environment or task follow hidden underlying structure. We suggest that animals represent these underlying structures using abstract basis sets that are generalized over particularities of the current environment, such as its stimuli and size. We show that this type of representation allows inference of important task states, correct behavioural policy and the existence of unobserved routes. We further conducted two experiments in which participants learned three maps during two successive days and asked how the structural knowledge that was acquire during the first day affect participants behaviour during the second day. In line with our model, we show that participants who have a correct structural prior are able to infer the existence of unobserved routes and are able to infer appropriate behavioural policy. Therefore supporting the idea that abstract structural knowledge can be acquired and generalised across different cognitive maps. | ||