The Action Selection Paradigm


The following seven positions briefly sketch the Action Selection Paradigm of mind. The book, Artificial Minds, argues for these positions.

1) The overiding task of Mind is to produce the next action. This is what minds, natural or artificial, are about. A few cautions seem in order. Don't be misled by the "next". Mind may well operate in parallel, producing more than one action simultaneously. Also, "next" doesn't imply discrete, rather than continuous, action. Finally, don't read purpose into the work "task". Producing actions is just what minds do. A consequence is that minds are properties of autonomous agents.

2) Mind is better viewed as a continuous, as opposed to a boolean notion. That is, it should not be a question of whether or not something has mind or it doesnŐt. It's more useful to think about degrees of mind, and what capabilities of mind a particular organism, system, or agent displays.

3) Mind is aggregate rather than monolithic. All but the most simple minds are comprised of relatively independent competencies or agents with only limited communication bandwidth between them. Communication; must be limited because, with so many different systems, it would be impossible for each to be in contact with every other. Also, there is typically no need for one agent, say the one who contracts your right thumb, to dialog with another, say your .i.respiratory control mechanism;. On the other hand, it's useful to hold your breath when you go under water. So it's useful that the goal agent for swimming under water communicate with respiratory control.

4) Mind is enabled by a multitude of disparate mechanisms. There is a role for all the various modalities and mechanisms we have discussed so far and, no doubt, many, many more. I'm not a fan of unified theories of cognition.

5) Mind operates on sensations to create information for its own use. I don't think of minds as information processing machines in the sense of taking information from the environment and processing it to arrive at its next action. Rather, I think of information as not existing out there in the invironment at all. Information comes into being when minds process sensations. The same scene can provide quite different information to different minds.

6) Mind uses prior information (memories) to produce actions by a reconstrutive process rather than by retreival. Memories are not stored as if in a folder in a filing cabinet, but rather rebuilt when triggered by appropriate associative cues.

7) Mind, to some degree, is implementable on machines. The question is, how much, and more importantly, how do we do it?



Author: Stan Franklin
(to emai clidk herel)
Last Updated: Tuesday, September 19, 1995