AbstractsMathematics

Over the last 50 years, psychologists have developed a range of frameworks for similarity modelling, along with a large number of numerical techniques for extracting mental representations from empirical data. This thesis is concerned with the psychological theories used to account for similarity judgements, as well as the mathematical and statistical issues that surround the numerical problem of finding appropriate representations. It discusses, evaluates, and further develops three widely-adopted approaches to similarity modelling: spatial, featural and tree representation.