Representing more about an agent's capability

 

Here are some examples of introspective free will and some considerations. They need to be represented in logic so that a robot could use them to learn from its past and plan its future.

1. Did I make the wrong decision just now? Can I reverse it?
2. ``Yesterday I could have made my reservation and got a cheap fare.''
3. ``Next year I can apply to any university in the country. I don't need to make up my mind now.''.
4. ``If I haven't studied calculus, I will be unable to take differential equations.''
5. ``If I learn to program computers, I will have more choice of occupation.''
6. ``It is better to have an increased set of choices.''
7. ``I am not allowed to harm human beings.'' Asimov imagined his three laws of robotics, of which this is one, as built into his imaginary positronic brains. In his numerous science fiction stories, the robots treated them as though engraved on tablets and requiring interpretation. This is necessary, because the robots did have to imagine their choices and their consequences.
8. Some of a person's behavior is controlled by reflexes and other automatic mechanisms. We rightly regard reflexive actions as not being deliberate and are always trying to get better control of them.
   • The coach helps the baseball player analyze how he swings at the ball and helps him improve the reflexive actions involved.
   • I'm a sucker for knight forks and for redheads and need to think more in such chess and social situations.

9. In the introduction I wrote

    

   What kind of an entity is x? In situation calculus, the simplest operand of can is a situation as treated above, but also we can consider an action itself or a propositional fluent. A propositional fluent p is a predicate taking a situation argument, and an agent can reason that it can (or cannot) bring about a future situation in which p holds.

[McCarthy 1996] includes an extensive discussion of what consciousness, including consciousness of self, will be required for robots.