ÃÛÌÒTV

We have used a variety of methods and tools to investigate configural processing:

1. Responses to detect differences between two faces are faster when two features (e.g. eyes and mouth) are changed relative to when a single feature is changed. However, these speeded responses are not due to super capacity in processing both features, but rather result from a horse-race between the two feature changes.
2. General recognition theory (GRT) is a multi-dimensional version of signal detection theory, which allows the disambiguation of the perceptual and decisional components involved in visual cognition. By using GRT, configurality can be determined through the interdependency between dimensions, and the nature of the interdependency (perceptual versus decisional) can be identified. In this project, the use of GRT has shown that, contrary to previous research, configurality can arise via decisional processes rather than perceptual processes alone.
3. A well-acknowledged challenge for GRT analyses is the problem of model identifiability: essentially the problem of a one-to-many mapping from empirical data to inferred model. Such situations can potentially lead to inferential errors. A multi-measure approach can potentially reduce the potential for inferential errors. One approach in this project is to use multivariate multinomial probit models to identify the theoretical structure underlying the empirical data in terms of GRT.
4. Another aim of this project is to investigate the developmental differences in the ability to process a face that is presented at unusual orientations. The results of experiments with children have shown that the range of orientations that can be processed configurally increases with development.

Funding: .