The galatean model of psychological classification

The galatean model of human classification is based on a prototype for category representation. Instead of being the central tendency, the prototype is the most perfect example of a category which may realistically exist; it is called a galatea after Pygmalion's ideal woman.

A galatea is a prototype which bases judgments of category membership on the conditional probabilities of the category given cues where the probabilities may be moderated by cue rarity. The conditional probability is a measure of a cue's weight or selective attention and each one competes with the others within a category for a fixed amount of total influence. Selective attention and cue competition are the main phenomena accounting for the galatean model's ability to predict people's ostensibly erratic use of base-rate information during classification tasks.

The model was tested on a series of previously-published experimental results using binary-valued cues and successfully predicted the significant responses of participants (see the abstract of a paper on how the model can be used in clinical decision support systems). However, if it is to have credence as a general model of classification then it needs to be applicable to domains which go beyond binary symptoms. Most importantly, it needs to be able to accommodate dimensional components. It was originally conceived to do this and successfully captured expert classification decisions in the fields of horseracing and assessment of clients' suitability for psychodynamic psychotherapy. The general model represents dimensions by including probabilities for certain salient values, most notably, the perfect ones which maximise the probability of an object being in the class. This results in an approximate probability distribution for the dimension which can generate conditional probabilities for any cue value.

C.D.Buckingham, Computer Science, University of Aston