Analysis of qualitatively different learning forms by testing multi-modality and other catastrofe flags
Verena Schmittmann, AiO
Developmental Psychology,
Department of Psychology
University of Amsterdam (UvA)
Roetersstraat 15, Room 1002
1018 WB Amsterdam
Phone: +31 20 525 6082 / 6830 (secr.)
E-mail: Verena Schmittmann
Supervisors: Prof. dr H.L.J. Van der Maas, dr M.E.J. Raijmakers
Project running from: 1 October 2003 – 1 October 2008
Summary:
Recently, nonlinear dynamic models of sudden, qualitative change have successfully been applied in the area of developmental psychology. One particular subclass of models, catastrophe models, predicts specific features, such as multimodality, of a system in the vicinity of a qualitative change. Several statistical techniques have been developed to test these hypotheses. In psychology, these models have mainly been applied to investigate Piagetian, cognitive development. The aim of this project is to adapt these techniques for studies of discrimination learning. The question whether two qualitatively different learning forms exist, rather than a continuum of learning forms is a topical subject of discussion in the field of explicit and implicit learning. Research on this subject has mainly concerned the nature of the learning forms, but has not produced definite results. In the context of development, usually mean scores per age groups have been inspected, or other ad-hoc analyses have been applied to describe individual differences. Statistical techniques related to catastrophe theory, particularly finite mixture analysis for multimodality analysis, are appropriate for investigating the existence of qualitatively different forms of learning, and to model individual differences. In this project these techniques will be integrated with existing, mostly mathematical psychological, learning models and research paradigms for discrimination learning.
Date of defence: 27 October 2009
Title of thesis: A model-based analysis of the development of learning