Bernard Baars, In the Theater of Consciousness. Oxford U. P. 1997. Chapters 5-9. [Comments in square brackets are my own thoughts.] Ch.5: "C O N T E X T S" (p.115-129) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ By "contexts" Baars means the unconscious assumptions, habits and expectations that come into play in all the situations we encounter and the activities we engage in, and which powerfully shape our conscious experience and understanding of those situations and activities. [From an AI perspective, these are *knowledge* and *know-how*, perhaps organized as contextual "frames" relevant to particular situations.] He illustrates the role of prior experience through the examples of - sea legs and land legs when taking a sailing trip; - later interpretation of an image, once it has been "understood" - verbal priming (PAGES: book; ARREST: book) - garden path sentences [He gets the analysis of "The ship sailed past the harbor sank" a bit wrong; he thinks this is syntactically valid as a version of "The ship (which) sailed past the harbor sank"; he should have said it is valid as a version of "The ship (which was) sailed past the harbor sank", i.e., a passive relative clause. Compare with "The rope stretched past its limit snapped"] Certainly language understanding is known to be strongly affected by "distributional properties" (frequency of various words, phrases and patterns), semantic habituation, world knowledge, conversational expectations, etc. p.123-5 he talks about expectations in language, particularly "tip of the tongue phenomena" and our intentions to express particular ideas and concepts; such an intentional state is contentful in that we easily judge whether a candidate that rises to consciousness is suitable or not; it also occupies space in working memory, potentially displacing other conscious items; but its content [i.e., what the "query" consists of] is not experienced. [Well, we sometimes have a clear sense of images, letters, associated entities, etc., we're trying to use for retrieval, don't we? E.g., "the name of that actress that played Leonardo DiCaprio's beloved in "Titanic" -- "I can *see* her and Leonardo facing into the wind at the bow of the ship, and floating on that plank in the icy water near the end ..." etc.] He suggest (p.126) that the stream of consciousness consists of a sequence of "contexts", each primed by its predecessor. [But wait -- weren't contexts "behind the scenes"?!] He also talks about postmodernism as an attack on ingrained contexts (habits of thought), and about 1-sided attentional neglect caused by 1-sided parietal lobe damage, as a kind of context loss. p.128 he also extends the notion of context to scientific paradigms, and the resistance of established paradigms to change (such as the Newtonian view of space, time and matter). Finally (p.129) he rejects relativism [correctly, I say!] arguing that some perspectives are more nearly in accord with reality than others; but he says we need to be humble about our views, never forgetting that they too are powerfully shaped by our personal "contexts" [indeed: a good scientist remains ever skeptical, even about his or her own views; though many ideas can and should be "played with", only ideas backed by compelling mathematical/logical arguments and by empirical evidence should be accepted in the long run; still, one needs to follow one's own instincts in what ideas to pursue, lest one be cast adrift in a vast sea of unresolved possibilities. As Einstein put it, "Every individual ... has to retain his way of thinking if he does not want to get lost in a maze of possibilities. However, nobody is sure of having taken the right road -- myself least of all."] Ch.6: "V O L I T I O N" (Conscious Control of Action, p.130-141) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ The general themes in this chapter are - the difference between actions carried out voluntarily, and similar actions carried out spontaneously - the compatibility between William James' "ideomotor" theory of voluntary action, and his theater metaphor. The first theme serves Baars' "contrastive phenomenology" strategy. He certainly argues convincingly that voluntary and spontaneous actions are quite different, both in the way they are subjectively perceived, and in the brain processes that go with them. E.g., - intentional vs unintentional stuttering; - forced smile vs. spontaneous smile. The voluntary actions require motor cortex activity, involuntary ones are subcortical. Sadly but interestingly, either voluntary or involuntary facial control can be lost through tumors or stroke, so a patient may be able to smile on request but not spontaneously, or vice versa. James' ideomotor ("from images to action") theory posits that a goal image in consciousness, in the absence of competing ones, automatically triggers action. Getting out of bed on a chilly morning happens, according to James and Baars, once "the spotlight" is occupied by the goal image of being out of bed, together with representations of the day's intended activities (which require being out of bed and thus bring this goal back into the spotlight). Together, these items are presumed to have displaced any thought of staying in bed, and thus the getting-out-of-bed action is promptly turned into motor commands. [Remember, for James there is no unconscious thinking, so "action" is entirely thoughtless action. Though Baars finds James congenial, I would think that the goal-achievement process from Baars' perspective can involve a good deal of unconscious thought, not just action; although it's not entirely clear to me just what Baars means by "automatism" in referring to these processes. Certainly in his discussion of the Langer and Imber letter-coding experiments (p. 135-6) he seems to say that unconscious processes are often just well- practiced conscious ones -- they no longer need conscious monitoring, or less of it.] [Now on the face of it, isn't this at odds with Libet's findings? If the appearance of a goal in the spotlight *triggers* the action, how can the action's readiness potential start up before actual consciousness of the goal? It seems to me we have to conclude it is certain members of the "audience" itself (unconscious processes) that determine what goes into working memory -- and they *notify* us -- or rather, all the other processes -- that a certain decision to act HAS BEEN MADE, where preparation for that action is already in progress. In other words, the *real* decision-making is subconscious, but is broadcast into consciousness and thus to the entire working brain. So it is NOT the appearance of the "goal image" in the spotlight that triggers the action -- it was already triggered and the spotlight merely REPORTS the decision. To put it a bit differently, it is the audience itself that is running the show! After all, who else but the unconscious processes -- the audience -- could be doing the stage direction and spotlight control?! Instead, Baars writes as if the latter processes are somehow separate from and prior to the initiation of action; that's an instance, I think, where the theater metaphor tends to mislead. In the next chapter, "self" is identified as the director running the show, but after all, what I MYSELF do is again just what the system as a whole -- in particular the vast unconscious part -- does. I think even Libet's "free won't" can be explained in this modified picture. Suppose, as I do above, that decisions to act are made subconsciously and at the same time as the preparation for an action begins, the process of abstracting a representation of the desision to act, for placement in the spotlight (or at least on stage), also begins. This representation arrives in working memory a short time before actual motor commands are sent out (so it seems from Libet's experiments). Now assume that at the same time other processes have been working subconsciously to try to inhibit actual follow-through on the foregoing decision (because there are negative consequences associated with it). Suppose they "win", and choke off the about-to happen action, and they also place a representation of their inhibitory action in working memory. Then we may well become conscious of a decision to act, immediately followed by a decision not to act after all. In other words, both the "free will" and "free won't" decisions were actually made unconsciously! That's not to say that representations of decisions that become conscious can't have later effects. Particularly for decisions about actions further in the future, such "broadcasting" of decisions can still have profound effects on further thinking and acting.] Chevreul Pendulum, p. 137-8 as illustration of ideomotor theory (thinking of the pendulum swinging, even without wanting it to swing, causes it to swing). What I've said parenthetically above seems consistent with Baars' remarks at the end of the chapter: he says that the brain seems to function as if governed according to a federal system -- certain major decisions are globally broadcast and thus globally influential, but much of the more specific work is carried out "locally", i.e., by unconscious processes specialized for this work. However, Baars' characterization of the role of the federal government as setting general policy may again conflict with the implications of Libet's lagging will. What we subjectively perceive as conscious policy- setting may again just be a high-level abstraction from policy already set by unconscious processes! Ch.7: "T H E D I R E C T O R " ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ (Self as the Unifying Context of Consciousness) This chapter tries to make a case for identifying "the self" with "deep context" -- i.e., the most fundamental, ingrained knowledge, expectations and habits that guide our interpretation of and responses to the situations we encounter. [I think there's something right about this, and something wrong about it. What's wrong, I think, is the consistent failure to even *consider* that we should perhaps be talking about a SELF MODEL (or in some cases, multiple ones) rather than a SELF, that is somehow something other than the whole person. The self-model MODELS the person and the person's relation to the world, and to its own parts and processes; it ISN'T the person. This seems so very obvious from a knowledge representation perspective, and Marvin Minsky sketched the essential idea in 1965 -- "Matter, mind and models", reprinted in Minsky (ed.), Semantic Information Processing. What's right about it is the idea that events that run totally against our expectations and normal experience may "shake up" the self model, at least temporarily -- confidence in its presumptions and ability to guide action are called into question and perhaps suspended. Note that your model of OTHERS may also be shaken up by unexpected behavior. For instance, if an astronaut, mother of 3, suddenly is found to have carefully prepared to murder a romantic rival, wouldn't that shake up your model of astronauts -- those exceptionally smart, well-adjusted, highly educated, brave, level-headed, self-controlled individuals who fly dangerous billion-dollar rockets into space, after ever- so-careful screening and training?] Baars (p. 143) claims "we disavow responsibility for slips and unintentional errors". [But we still don't deny that it was OURSELVES that slipped up or erred accidentally!] Gilbert Ryle's homunculus (p.143); [to me, this is an illustration of what happens when one confuses self with model of self. Minsky, too, had nested entities -- but they were nested models, and unlike nested selves, these are perfectly sensible (I have knowledge about myself, and also know something about that self-knowledge, etc.).] p.145, "The anatomy of the brain looks like a physical refutation of Ryle's position." [Yes, if we acknowledge that the "sensorimotor homunculus" is a sort of analogue MODEL of ourselves, not THE self!] p. 146, losing someone dear. [I think the notion of losing a part of oneself is metaphoric: one loses a part of one's accustomed *life* -- unless we literally thought of the other person as part of the same "dual individual"!] p.147-8, the narrative self. [My conclusion e.g., from the split brain experiments, is that our self-narrative is an attempt at explanation of what we observe about ourselves and our experiences -- we are "explanation seekers", always. It has been proposed by Hobbs et al. that language understanding is based entirely on explanation-seeking (abduction).] p. 148 -- paralyzed arms. [I think we have an inborn disposition to count as part of ourselves that which we can *feel* -- while discounting the rest.] p. 149-51 -- depersonalization, fugue, multiple selves. [I think initiation of new self-models perfectly explains all this, including not being able to recall alter-egos' experiences.] Ch.8: "C O N C L U D I N G T H O U G H T S" (p. 157-164) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Here Baars enumerates all the crucial functions performed by consciousness (using the "snorting bull behind your back" thought experiment). He mentions - prioritizing (e.g., in coping with danger) - problem-solving (providing access to unconscious resources) - decision-making & executive control - recruitment & control of action - error detection - role in concept of self (dissociated to protect self-concept?) - learning and adaptation - creating contexts for understanding - creating access between otherwise separate knowledge domains [But, one may ask, especially when thinking along AI lines, why should ANY of these functions be accompanied by conscious experience??] E P I L O G U E: A Bit of Philosophy (p. 165-7) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ This could be called "a bit of chastisement of philosophers" for raising airy-fairy questions like those raised by Chalmers, instead of getting some real scientific work done. He says asking what's left after taking away the physical processes is like asking what's left of the Matterhorn's north face after taking away all the rocky material of the Matterhorn. [But do the philosophers really ask, "what's left after you take away all the physically observable processes?", I can't think of many who actually *deny* that taking away the material substrate of thought, or the physical processes, would leave consciousness intact; those who do tend to be mystics or supernaturalists, not people like Chalmers or Blackmore. What would Blackmore say about Ch. 8 and the Epilogue? By now, I think you know.]