School of Computing

Using e-greedy reinforcement learning methods to further understand ventromedial prefrontal patients' deficits on the Iowa gambling task

Kiran Kalidindi, Howard Bowman, and Brad Wyble

Technical Report 12-05, University of Kent, Computing Laboratory, Canterbury, Kent, CT2 7NF, UK, October 2005.

Abstract

An important component of decision making is the process of choosing from a set of options. This choice is often based on an evaluation of the potential costs or benefits of selecting an option, its estimated reward value. The foundation for these estimations of costs and benefits is frequently past experience. The way past experience is used to predict future rewards and punishments can have profound effects on choice. Current literature suggests an important role for the orbitofrontal cortex (OFC), in both humans and non-human primates, in representing the reward value of stimuli. The role played by the ventromedial prefrontal cortex (a region which includes the medial OFC) in decision making has previously been tested with the Iowa Gambling Task (IGT), by comparing the performance of patients with bilateral ventromedial prefrontal lesions (VMF) and normal healthy controls (NHCs). A number of theories in the literature offer potential explanations for the underlying cause of the deficit(s) found in VMF patients on the IGT. An \emph{error-driven} $\epsilon$-greedy reinforcement learning method has been used to model both normative and VMF patient groups from a number of studies, offering support and, more importantly, a deeper understanding of the causes for the `myopia' for future consequences explanation of VMF patients deficits on the IGT.

Download publication 1411 kbytes (PDF)

Bibtex Record

@techreport{2288,
author = {Kiran Kalidindi and Howard Bowman and Brad Wyble},
title = {Using e-greedy reinforcement learning methods to further understand ventromedial prefrontal patients' deficits on the {I}owa Gambling Task},
month = {October},
year = {2005},
pages = {182-196},
keywords = {determinacy analysis, Craig interpolants},
note = {},
doi = {},
url = {http://www.cs.kent.ac.uk/pubs/2005/2288},
    publication_type = {techreport},
    submission_id = {29564_1131460917},
    number = {12-05},
    address = {Canterbury, Kent, CT2 7NF, UK},
    institution = {University of Kent, Computing Laboratory},
}

School of Computing, University of Kent, Canterbury, Kent, CT2 7NF

Enquiries: +44 (0)1227 824180 or contact us.

Last Updated: 21/03/2014