Computational models of concept formation: Cognitive chunks and neural engrams



Bennett, Dmitry
(2021) Computational models of concept formation: Cognitive chunks and neural engrams. PhD thesis, University of Liverpool.

[img] Text
201013367_Feb2021.pdf - Unspecified
Access to this file is embargoed until 1 August 2024.

Download (7MB)

Abstract

A key issue in cognitive science concerns the fundamental psychological processes that underlie the formation and retrieval of concepts in the short-term and long-term memory (STM and LTM, respectively). This thesis tackled this question from two opposite levels of understanding mental phenomena: the level of cognitive psychology and the level of neuroscience. On the cognitive psychology side, the thesis advanced Chunking Theory and its computational embodiment CHREST to propose a single model that accounts for significant aspects of concept formation in the domains of literature and music. The proposed model inherits CHREST’s architecture with its integrated STM/LTM stores, while also adding a moving attention window and an “LTM chunk activation” mechanism. These additions address the overly destructive nature of primacy effects in discrimination network based architectures and expand Chunking Theory to account for learning, retrieval and categorisation of complex sequential symbolic patterns – namely real-life text and written music scores. The model was trained through exposure to labelled stimuli and learned to categorise classical poets/writers and composers. On the neuroscience side, the thesis replicated the categorisation experiments above with a Deep Learning/Artificial Neural Network (ANN) architecture. The results of both categorisation experiments showed qualitative, quantitative and functional similarities between the cognitive and the neural modelling approaches. Both CHREST and ANN models were then tasked with simulating/predicting human music categorisation performance. Structured interviews with six music conservatory students established their musical history as well as their performance on categorisation of novel musical pieces. Individual models were then made for every participant – these were trained and tested on the same data as their human counterparts. Both models were found to have good overall fit to the human data. These findings offer further support to the mechanisms proposed by Chunking Theory, connect them to the neural network modelling approach, and make further inroads towards integrating concept formation theories into a Unified Theory of Cognition (Newell, 1990).

Item Type: Thesis (PhD)
Divisions: Faculty of Health and Life Sciences
Depositing User: Symplectic Admin
Date Deposited: 06 Sep 2022 10:01
Last Modified: 18 Jan 2023 21:05
DOI: 10.17638/03153266
Supervisors:
URI: https://livrepository.liverpool.ac.uk/id/eprint/3153266