From: Ian Wright (wrighti@ACM.ORG)
Date: Sat Nov 25 2006 - 16:25:07 EST
Hi Jerry > The language embodied in robotic software design is purely mathematical, > i.e. it is the language of formal logic. Yet, semantics refers (no pun > intended) to more than formal logic, doesn't it? I make a distinction between the description of a program (e.g. a listing of the code) and the execution of that program. So a running program, embodied in a system with sensors and actuators, is not a purely mathematical or formal description. It's a causal process. For example, consider the simple AIBO robot visually tracking a ball. It has an internal "syntactic" data structure that represents the relative location of the ball. It uses that information to control its head motors to maintain the ball in its visual field. The internal state of the data structure is causally coupled to the external state of the physical ball. The "syntax" gets "semantics" in virtue of this kind of causal coupling. Or let's look at this another way. We find an AIBO-like creature on Mars. We don't know how it was built. It can visually track objects. We open it up and find all kinds of complicated circuitry and memory capacity. Reverse engineering this creature's control program is very difficult. We cannot "see" its programs; it is difficult to extract the formal description of the laws that govern its behaviour. The point is this: we are in the same situation with respect to the human mind. Our cognition is software running on our brains. It's very difficult to deduce the formal logic from simply looking at the material embodiment. -Ian.
This archive was generated by hypermail 2.1.5 : Thu Nov 30 2006 - 00:00:06 EST