Identity Conditions for Indicator State Types within Dretskes Theory of :: Psychology Dretske Papers

Identity Conditions for Indicator State Types within Dretske's Theory of Psychological Content Naturalization ABSTRACT: Within the context of Dretske’s theory of psychological content naturalization, as laid out in Explaining Behavior, the concept of an indicator state type plays a pivotal role. Providing a general (and non-circular) description of the identity conditions for being a token of an indicator state type is a prerequisite for the ultimate success of Dretske’s theory. However, Dretske fails to address this topic. Thus, his theory is incomplete. Several different approaches for specifying these identity conditions are possible; however, each is inadequate. Of the various theories for psychological content naturalization put forward within the past two decades, I believe that a Dretske-style approach that explains the content of a mental state in terms of the causal history of past tokens of that state holds out the most promise of giving us a workable theory describing the role that content plays in learned behavior. While I favor this general approach, the particular theory laid out by Dretske in Explaining Behavior has a shortcoming that must be addressed before his theory can be applied to real systems: Dretske fails to provide an analysis of identity conditions for being a token of an indicator state type. The shortcoming is serious because of the critical role that past tokens of an indicator type play in fixing the content of a current token of the indicator type — without identity conditions, there is no way to specify which previously tokened states among the many that have been instantiated during the learning period of the organism are of that indicator type. I begin with a very brief review of Dretske's theory from Explaining Behavior. Some organisms possess indicator states (i.e., internal states that indicate whether some external conditions hold). For example, organism O may token an instance of I (the internal indicator state type) whenever external conditions F obtain. Prior to learning, I indicates F does not mean F. Let's suppose that external conditions F are relevant in some manner to O's continued functioning, perhaps because environments in which F obtains are environments that are relatively inhospitable for O. Let's also suppose that O is capable of learning using reinforcement information (via operant conditioning), such that future tokenings of I come to cause movements that are appropriate to conditions F. (My use the evaluative term "appropriate" here rests on two assumptions: (1)