Neurocomputational models of corticostriatal interactions in action selection

Caso, Andrea (2019) Neurocomputational models of corticostriatal interactions in action selection. Doctoral thesis, Birkbeck, University of London.

Abstract

Schema theory is a framework based on the idea that behaviour in many areas depends on abstractions over instances called schemas, which work in a cooperative or sequential fashion, but also compete with each other for activation. Cooper & Shallice (2000) provide an implementation of schema-theory with their model that simulates how routine actions works in healthy and neurologically-impaired populations. While schema theory is helpful in representing functional interactions in the action-perception cycle, it has no commitment to a specific neural implementation. Redgrave et al.’s (2001) model of the basal ganglia is, in principle, compatible with a device that regulates the competition among schemas, carrying out action selection. This thesis is mainly concerned with improving the neurobiological plausibility of the schema theoretic account of action selection without sacrificing its theoretical underpinning. We therefore start by combining an implementation of schema-theory with a reparametrised version of the original basal ganglia model, building the model from the ground up. The model simulates two widely used neuropsychological tasks, the Wisconsin Card Sorting Test (WCST), and the Brixton Task (BRX). In order to validate the model, we then present a study with 25 younger and 25 over-60 individuals performing the WCST and BRX, and we simulate their performance using the schema-theoretic basal ganglia model. Experimental results indicate a dissociation between loss of representation (present in older adults) and perseveration of response (absent in older adults) in the WCST, and the model fits adequately simulate these findings while grounding the interpretation of parameters to the neurobiology of aging. We subsequently present a further study with 50 participants, 14 of whom have an ADHD diagnosis, performing the WCST under an untimed and a timed condition, and we then use our model to fit response time. Results indicate that impulsivity traits, but not inattention ones, predict a slower tail of responses in the untimed task and an increased number of missed responses and variability across subtasks. Using the model, we show that these results can be produced by variation of a combination of two parameters representing basal ganglia activity and top-down excitation. We conclude with recommendations on how to improve and extend the model.

Metadata

Statistics

DownloadsShow export options

Export as XMLJSONCSV

Activity Overview

6 month trend

209Downloads

6 month trend

189Hits

Additional statistics are available via IRStats2.

Archive Staff Only (login required)

Edit/View Item