Cognitive Science 170
Natural and Artificial Symbol-Using Systems
SHORT PAPER #1:
Questions: (3-6 pages). (i) Choose question 1 or 2! (ii)
indicate at beginning of paper which question you have chosen!
(iii) hand in a paper printout in class, and (iv) staple the pages
together (don't send an email attachment).
In both cases, you can argue either side of the issue. I won't penalize
or reward you for taking either side (or for arguing for something
in between). The most important thing is to cite at least one
example of a connectionist model and one example of a more traditional
artificial intelligence model (like the ones presented in class but not
limited to them) to support your arguments. Examples will make your
task easier. Note that Fodor and Pylyshyn cite only trivial examples,
so I am asking you to go beyond what they have done.
- 1. Fodor and Pylyshyn criticize many different features of neural
network models. Pick out *one* of their main arguments, describe it,
and apply it to at least one example of a 'real' neural network model
and a 'real' AI model.
- 2. Turing originally conceived of computation as automated
conscious calculation of the sort that a human can do. Several decades
later, practical computer architectures (von Neumann architecture)
and high-level programming languages (e.g., LISP, production system
languages) were developed. People such as Allen Newell suggested that
the upper level cognitive architecture implemented in these
high-level languages (as opposed to low-level machine architecture or
language) was crucial to understanding human cognition. Support or
argue against this idea using example of at least one 'real' neural
network model and one 'real' AI model.
Finally, it is almost always a good idea to start each paragraph with
a topic sentence. A good test of the organization of your paper is to
read aloud just the topic sentences, one after the other, and see if it
makes reasonable sense.
due beginning of class Mon, 10/23/06