English
 
User Manual Privacy Policy Disclaimer Contact us
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT
  Individual animals use distinct strategies in a changing spatial memory task

Kastner, D., Gillespie, A., Dayan, P., & Frank, L. (2018). Individual animals use distinct strategies in a changing spatial memory task. In Computational and Systems Neuroscience Meeting (COSYNE 2018) (pp. 50-51).

Item is

Basic

show hide
Item Permalink: http://hdl.handle.net/21.11116/0000-0003-E859-4 Version Permalink: http://hdl.handle.net/21.11116/0000-0003-E85A-3
Genre: Meeting Abstract

Files

show Files

Locators

show

Creators

show
hide
 Creators:
Kastner, D, Author
Gillespie, A, Author
Dayan, P1, Author              
Frank, L, Author
Affiliations:
1External Organizations, ou_persistent22              

Content

show
hide
Free keywords: -
 Abstract: Behavior is typically highly variable across individuals. To deal with this complexity, average data from many an- imals, grouped by genotype or experience, is analyzed for differences. This approach, while powerful, obscures the individual differences within groups. To link the workings of individual brains to their behavioral outputs, we have to identify and quantify inter-individual behavioral differences. Therefore, we designed an automated six- arm behavioral apparatus to minimize and control extraneous variables. Initially, wild type (WT) and fragile-X syndrome (FXS) model Long-Evans rats received reward on all arms, allowing us to examine their innate biases. Subsequently, reward was restricted to just three arms, and was only delivered if the rats visited the arms in a specific sequence. Four different sets of rewarded arms were employed in sequential blocks. We then developed a generative model for the entire behavior, using a variant of a reinforcement learning (RL) model with short-term memory. Conventional RL implementations learned markedly slower than the animals. The addition of two types of arm biases that were apparent during the exploratory behavior produced a model with three free parameters– learning, forgetting, and temporal discounting rates–that could be fit to describe the behavior of each animal. A 2D subspace of the parameters identified systematic variation of parameters between individuals, such as where a group of animals had similar parameters for their learning rates, but varied in their temporal discounting rates. These analyses revealed that individual differences far outweighed genotypic differences across the full cohort of FXS and WT animals, but that when subsets of animals with similar sets of model parameters were identified, genotype-related differences emerged. These findings suggest that relatively simple behavioral models can ex- plain individual differences and thereby reveal otherwise hidden patterns of variability across different groups of animals.

Details

show
hide
Language(s):
 Dates: 2018-03
 Publication Status: Published online
 Pages: -
 Publishing info: -
 Table of Contents: -
 Rev. Method: -
 Identifiers: -
 Degree: -

Event

show
hide
Title: Computational and Systems Neuroscience Meeting (COSYNE 2018)
Place of Event: Denver, CO, USA
Start-/End Date: 2018-03-01 - 2018-03-04

Legal Case

show

Project information

show

Source 1

show
hide
Title: Computational and Systems Neuroscience Meeting (COSYNE 2018)
Source Genre: Proceedings
 Creator(s):
Affiliations:
Publ. Info: -
Pages: - Volume / Issue: - Sequence Number: T-42 Start / End Page: 50 - 51 Identifier: -