Use this URL to cite or link to this record in EThOS:
Title: From transitive inference to exhaustive search : towards self-regulating models of developmental processes
Author: St Johnston, Benedict
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 1994
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
This thesis is sympathetic to Piaget's view of development as a dynamic interactive process between an autonomous agent and its environment allowing cognitive growth through self-regulation and resulting in an extended behavioural repertoire. This thesis explores developmental issues through computer modelling. Piaget thought transitive inferences relied on logic, which he considered the basis of rationality, and as such marked the end-point of cognitive development. More recently, empirical studies have shown that the ability to make transitive inferences does not rely on logic, and is close to the lower ontological bounds of the system, requiring us to reassess our theories for rationality, ontology, and ontogeny. Using Piagetian principles, it is argued that transitive inference is likely to be primitive, and should be applied as a default assumption for pragmatic reasons but that this assumption must be defeasible. Basic decision-making models are produced, based on Sutton and Barto's self-supervised learning models. They show how easy it is to generate a simple order over a set of binary choices leading to a strong transitive bias. Elaborating the representation in the model with units that learn stimulus-stimulus relationships, the model can also learn a circular relationship and generalise appropriately. Thus, transitivity is captured as a default yet defeasible assumption giving the agent the ability to exploit its environment as much as possible whilst still remaining adaptable. The model also shows similar scope to subjects in that its performance on triadic choices is worse than on binary ones. Unlike some subjects, the model cannot repair its own performance. Seriation is a more transparent ordering skill which develops later. Empirical work determining the basis of seriation skills showed that the important development is a data-reducing strategy, based on picking up redundant information in the task, that reduces seriation to a simple search task.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available