Rational causes : the concept of preference in the social sciences
The concept of preference is used in the social sciences to explain and predict behaviour. This thesis investigates the conditions the preference concept has to satisfy in order to operate as explanans. First, it defends the naturalistic position that preferences are causes of behaviour. More specifically, it is argued that preferences are programming properties that are themselves not causally efficacious, but causally relevant in that they realise efficacious properties. Further, the argument that the allegedly intentional nature of preferences poses a problem to such a causal relevance is rejected. Second, methodologies of preference attribution are discussed. The methodology of introspection in its current form is rejected, as well as the Radical Behaviourists' proposal to avoid mental properties altogether. Instead, it is argued that preferences are theoretical concepts. Third, a framework is provided that connects preferences over prospects of different degrees of abstraction. Such a framework allows to attribute specific preferences on the basis of observed actions and derive from these specific preferences more abstract preferences which are employed in the explanation and prediction of behaviour. Fourth, this thesis develops a model of preference change. It is specified under which conditions the inconsistency of an agent's behaviour with the preferences previously assigned to her should be interpreted as a preference change. The model then takes those behavioural observations and predicts how the preferences must have been changed in order to retain consistency. Principles guiding such a change are specified and operationalised, and the ensuing model is compared to existing ones.