Your exam at 10:30, Thurs (room to be announced) will be a subset of these questions.

R. Port

1. Explain the notion of intentionality.

2. What is digitalness, according to Haugeland?

3. What are deductive vs. inductive inference?

4. What are `classical categories' (or `definitional categories')? In what ways do they fail to have the right properties as models of natural categories created by humans?

5. Describe several fundamental issues on which the embedded or embodied (or situated) theorists differ from the `classical'' theorists about AI. (cf. Haugeland 25-27 and Smith reading)

6. Dreyfus says the problem with AI is that it must treat ``the world as an object and our know-how as knowledge'' (p. 181). Clarify his points here. Why does make this charge and why does he think these are deep problems?

7. Differentiate icon, index and symbol as described by Deacon. (See the webpage on Deacon in the Syllabus.) Why are symbols all but impossible for nonhumans?

8. Explain the concept of `optic array' and showing why it is important to Gibson's theory of vision.

9. Contrast Gibson's five `perceptual systems' with the traditional five senses (light, sound, touch, taste, smell). How similar and how different?

10. Describe several ecological cues for visual perception, including the notion of tau (time to contact).

11. Describe lateral inhibition and what practical functions it can serve on a topological map (eg, a frequency map in audition or retinotopic map in vision).

12. Describe the characteristics in common among clear-cut cases of emergent phenomena. Use some specific example to illustrate.

13. Describe the functioning of a genetic algorithm, explaining how crossover and mutation work and what their relative strengths and weaknesses are as operators.

14. Randomness plays an important role in many learning techniques that we have seen. In what specific way is randomness used? What are some strengths and weaknesses in using randomness?

15. Describe the Hebbian learning rule impressionistically and the way it can be implemented mathematically.

16. Explain why a letter can be recognised faster when embedded in a word than when displayed all by itself, according the Rumelhart-McClelland interactive activation network model.

17. What are some reasons for supposing that dynamical systems may be the most appropriate tool for modelling human cognition?

18. What are the main reasons that the issue of time plays such an important role in discussions of the difference between classical cognitive science and dynamical cognitive science?

19. What is Thelen's argument for the embodied nature of general cognition - of thinking about things and events?

20. Explain the distinction Chalmers makes between the `hard' and `easy' problems of consciousness. What makes the easy easy and the hard hard?