Wednesday, January 17, 2007

Further note on "Cognitive Robotics"

In my previous post, I discussed the preliminaries of a definition of the term "cognitive robotics". There is a few further point I'd like to mention now, which I neglected before.

In the discussion, no mention was made of what sort of implementation the cognitive architecture should use - that is, the definition does specify symbolic rule-based systems or artificial neural networks, or some hybrid. My interpretation of this is that it is the functionality of the system which is important, and not the computational substrate. The current state of affairs seems to indicate that it is the neural network-like architectures which hold the most promise, but that does not necessarily mean (in my opinion) that other approaches are without merit (see this for an alternative point of view, and this for some follow up discussion).

3 comments:

Dave said...

I took an introduction to cognitive science course a year ago and it was mainly concerned with that difference between symbolist and connectionist accounts of intelligence and the histories of each. I definitely agree with you that cognitive science is (or should be) about exploring and understanding the function of intelligence, at all its levels, independent of implementation.

Paul said...

I agree. The point is though that there is debate as to whether a particular implementation (e.g. symbolic, neural, hybrid) is necessary for cognitive functions - or whether it is sufficient to look at only the functionality of the system, and not worry so much about how you get there.

Thanks for your comment!

Deep said...

I have to disagree a littel bit... My point of view is that funtional description is not enougth to understand how a human brain works so it necesary to undestand how these functional behaviour EMERGES from well-known single neuron behavior.

http://www.deepknow.com