HOMEWORK 2, CSC 290A, 2007 ========================== DUE: Fri Feb 22/08, 5:00pm NOTE: Questions should be answered independently by each student. (That's not to say you should talk about the ideas with others -- by all means do!) Any text taken from elsewhere should be quoted and attributed (to avoid those dreaded charges of plagiarism). Keep your answers as crisp and tightly reasoned as you can. It's better to say a few things clearly and persuasively than to keep writing just to fill the page. ALL WORK SHOULD BE IN ELECTRONIC FORM, BUT PRINTED OUT FOR HANDING IN - THANKS! -=-=-=-=-=-=-=-=-=-=- 1. Baars' theater metaphor ~~~~~~~~~~~~~~~~~~~~~~~ a. Imagine that a stone-age hunter in Africa (say, 40,000 years ago) suddenly encounters a rhinoceros and decides to scramble up a tree. Try to explain the entire brain process leading to the hunter's action in terms of the theater metaphor. Be as specific as you can, for example concerning the types of unconscious processes that come into play, what's "on stage", what's in "the spotlight", etc. [Answer in a couple of paragraphs.] b. Do you consider this kind of account, based on Baars' theater metaphor, reconcilable with Daniel Dennett's insistence (as per Blackmore's book) that there is no "Cartesian theater" -- no "homunculus", not even a distributed one, viewing a "movie" in the brain? Refer to footnote 1, p.viii-ix in Baars' book, and also to Baars' discussion of Stephen Kosslyn's findings concerning mental imagery. [Answer in a paragraph or two.] 2. Perception and action in robots, animals, and humans ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The following is a fairly plausible outline of how a "cognitive robot" (one capable of thinking and planning) might react to what it sees at a given moment: - light falls via lenses onto the artificial "retinas", and this is registered as a color- and brightness-coded pixel array; - various local features (edges, points, textures, colors, etc, and changes in these) are computed, and merged in several stages into larger curves, regions, and surfaces; - based on the results of the previous step, various objects and their relationships are hypothesized, expressed symbolically, e.g., an apple on a table, a waste basket next to the table, and so on; - in addition to performing such perceptual processing, the robot also pursues goals; for example, it might have the goal (represented symbolically in its goal-directed planning system) of bringing a snack to the "master", and its perception of the apple would then cause it to plan to pick up the apple, and bring it to the master; this planning may involve reasoning, for instance, the inference that an apple may be regarded as a "snack"; - the plan execution system would then begin to translate this plan into action, by calling up existing routines for picking up and object, moving from one point to another, and so on. Make some comparisons between this outline and Baars' discussion of human and animal cognition. In particular, relate the outline to Baars' theater metaphor, to his discussion of vision (the regions of the visual cortex, and the dorsal and ventral pathways from there to other brain regions), and to his discussion of the similarities between human and animal cognition (here you might focus on the role of symbolic processing). [Write about a page.] DO ONE OF QUESTIONS 3, 4, 5, writing an essay of a page or two: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3. FREE WILL: Baars seems to find William James' "ideomotor" theory of how conscious thought is turned into action quite compatible with the theater metaphor (see ch. 6): a "goal image" (such as being out of bed and having breakfast) appears in the "spotlight", and this is "observed" by the unconscious processes that can achieve that goal (such as getting-out-of-bed routines), causing them to take control and automatically achieve the goal. But isn't this incompatible with Libet's delay (not mentioned by Baars)? I.e., the idea that a conscious goal or image *causes* the action seems incompatible with Libet's observation that action preparation *precedes* our consciousness of deciding to act on a particular goal (suggesting that the decision to act was itself made unconsciously, and our consciousness merely reflects -- with a time delay-- that unconscious decision). To put it bluntly, it seems that Baars' view is consistent with the idea of "free will" (conscious decisions causing actions), while Libet's findings strongly suggest otherwise. Discuss this apparent conflict, either resolving it (in a way that you suppose Baars might endorse), or sharpening it, by showing that it's not really feasible to resolve it within Baars' explanatory framework. 4. SELF: Baars views "the self" as a "deep context", while Blackmore viewed it as an illusion constructed by the brain. The course instructor has emphasized the key role played by a self-*concept*, i.e., a model that a person has of him- or herself. Such a self-concept (in symbolic reasoners like ourselves) could be implemented through a self-symbol ("ME"), serving as a "hook" for symbolic knowledge about ourselves -- just as we have symbols, and associated knowledge, for myriads of other entities in the world around us. Try to articulate your own position on the issue of "self", taking account of (and arguing for or against) any or all of the above views (and there is no expectation that you'll agree with the instructor's views, only that you'll argue as clearly and convincingly as you can!) 5. EXPLANATORY GAP: In Ch. 8 Baars summarizes what he regards as the numerous and crucial roles played by consciousness in human cognition; in the Epilogue he then chastises philosophers for still worrying that consciousness has no apparent causal role to play in cognition (all we see, according to some philosophers, is neural activations, leaving as a mystery which of these processes are conscious and which are unconscious, and just how and where the transition to consciousness happens). Try to sort out this apparent clash of views, keeping in mind not only Baars' but also Blackmore's discussion of the (presumed) "explanatory gap".