Evidence for a Fixed-Point Property in Categorization Response Time Data

A/Prof Daniel Little was the First speaker for the 2019 Complex Human Data Hub Seminar Series. In this seminar, he presented one of his own papers titled “Evidence for a Fixed-Point Property in Categorization Response Time Data” (2018) which examined the properties of mixture models in the strong inference methodology, Systems Factorial Technology (SFT).

This paper was chosen because many different attentional and decision making strategies can be conceptualized as mixture models. For instance, shifts between automatic and controlled processes could be considered mixtures of parallel and serial processing. This explanation has been favored in recent investigations of categorization using separable dimensions. However, one caveat is that mixtures of parallel and serial processing can often be mimicked by parallel models with inhibitory interactions.

The seminar explored the systematic predictions of mixtures of processing architectures and compared them to systems of interactive parallel models. It also examined mimicry using a novel method designed to tease apart these explanations. Mixture models have an important 'fixed-point' property such that there is a single point of cross-over in the probability density functions generated from different mixture proportions. This property can be used to distinguish mixture models from other explanations. Two experiments that used separable dimensions were presented as evidence for this fixed point property. The discussion also focused on how more complex models, currently beyond the scope of Systems Factorial Technology (SFT) might be identified using combinations of methods.

The presentation provided valuable insight into the nature of categorization response time data and stimulated a thorough discussion on the topic. We thank A/Prof Little for his presentation and for kicking off the 2019 CHDH Seminar Series.

Reference: Little, D. R., Eidels, A., Houpt, J. W., Garrett, P. M., & Griffiths, D. W. (2018). Systems Factorial Technology analysis of mixtures of processing architectures. Journal of Mathematical Psychology.