|Computational Analysis of Behavior
|Monday, May 29, 2017
|8:00 AM–8:20 AM
|Hyatt Regency, Centennial Ballroom A
|Instruction Level: Advanced
|Chair: Don (Yuhan) Li (The University of Auckland)
|The Computational Analysis of Behaviour: Multivariate Models of Behaviour
|DON (YUHAN) LI (The University of Auckland)
|Abstract: In Skinner’s (1950) view, a study of behaviour investigates the relationship between experimentally controlled variables and the likelihood of response. Skinner proposed that the canonical datum for indexing the likelihood of response ought to be response rate. However, one may argue that response rates do not dominate all other dependent variables for indexing the likelihood of response.
Although contemporary behaviour analysis places less emphasis on rates, models of behaviour typically only account for one type of dependent variable. This constrains those theories to particular dimensions of behaviour. An alternative approach is to construct a multivariate model that links environmental variables to a constellation of dependent variables to produce a more general account of “behaviour”. Computational models of behaviour form a class of models that have this multivariate property. These models output punctuate responses and as a result, almost any arbitrary dependent variable may be calculated. Hence, a computational model allows one to make statements about behaviour in general.
The present paper outlines the Computational Analysis of Behaviour and provides an illustration of multi-objective optimisation with Catania’s Operant Reserve (Catania, 2005).