From chalmers@paradox.soc-sci.arizona.edu Thu Sep 23 18:13:31 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id SAA10855 for ; Thu, 23 Sep 1999 18:13:31 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGBHXDFZFKB8VN0F@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Thu, 23 Sep 1999 18:13:35 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGBHXAZU9SB8V4JG@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Thu, 23 Sep 1999 18:13:31 -0700 (MST) Received: (from chalmers@localhost) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) id SAA10848 for scicon; Thu, 23 Sep 1999 18:13:22 -0700 Date: Thu, 23 Sep 1999 18:13:22 -0700 From: David Chalmers Subject: What does a neural correlate of consciousness explain To: scicon@paradox.soc-sci.arizona.edu Message-id: <199909240113.SAA10848@paradox.soc-sci.arizona.edu> MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=ISO-8859-1 Content-transfer-encoding: 8BIT Status: R >From Kniels1@aol.com Thu Sep 23 15:57:22 1999 From: Kniels1@aol.com Date: Thu, 23 Sep 1999 18:57:14 -0400 (EDT) Subject: What does a neural correlate of consciousness explain? To: Chalmers@Arizona.EDU Cc: lnielsen@u.arizona.edu X-MIME-Autoconverted: from quoted-printable to 8bit by paradox.soc-sci.arizona.edu id PAA10743 Hi Dave, I'm out of town and don't have access to my own e-mail accounts, so I can't post to the list. Would you please submit it for me? Thanks, Lis --------------------------------- In his talk of September 10, Dave accorded a prominent place to the search for the NCC in an overall science of consciousness. He viewed it as playing an important role in leading to the discovery of bridging principles linking the physical and phenomenal facts. My comments here address why I think this expectation is premature, and why approaches like those of Logothetis and Crick & Koch, while clever and interesting, are unlikely to give us much new information about consciousness. (I make no claims regarding the originality of these ideas, as they are greatly influenced by comments made by others in class. Writing them out has merely helped to focus my thinking on these issues.) Whenever we have found specific brain correlates for particular psychological functions, we seem to feel that we have better understood those functions. In fact, often all that we have understood is why damage to those brain areas leads to deficits of this or that sort. In the current climate where high-tech brain mapping is so fervently pursued, this feeling of making explanatory progress is reinforced by numerous social factors that drive research. I remain skeptical that any substantial explanatory progress is actually made in this field, beyond the continued confirmation and refinement of our existing hypotheses about brain localization. The following (recently encountered) quote on the relevance of neural correlates to the shaping of developmental theory expresses part of my worry as regards how this kind of evidence can inform a theory of consciousness: "Suppose we have a psychological theory that says that function A leads ontogenetically into function B. Being able to suggest brain correlates for A and B has great merits, but it does not by itself advance developmental theory. This can only be done if there is information about those brain correlates that can be used to inform the developmental theory, such as the relationship between physiological substrates of A and B. ... [P]hysiology on its own has never been a prerequisite to developmental theory, since sometimes the psychological and neurological facts match on levels and sometimes they do not." (Segalowitz, 1994): What is said here in terms of developmental theory seems to apply just as well to discussions of a theory of consciousness. Suppose we have an area IT that we know codes for a certain type of visual information X, because cells in IT fire when subjects are viewing X. We then discover that firing in IT correlates with phenomenal experience of X under conditions of binocular rivalry. We can assume that this is either because some top-down process has focused attention on X (as opposed to the rivalrous stimulus Y) or because some bottom-up process has caused the processing of X to override the processing of Y. What does this tell us about consciousness that we didn't know before? Arguably little, unless "there is information about those brain correlates that can be used to inform the … theory [of consciousness]." For this to be the case, we need to learn something new about the role of IT in relation to these other aspects of visual processing. For example, something new about its position in the visual processing chain or in informational feedback loops, or something new about its connectivity to other brain areas that sheds light on how these top-down or bottom-up processes operate. Presumably, this is the kind of thing we would like to learn from our neural investigations, namely how the information that (we already know that) the brain encodes is made accessible to consciousness. We want to know WHAT makes them accessible, and what this consciousness IS to which they are accessible. Only insofar as a discovery about the NCC of conscious content can shed light on these issues does it inform our theory in any new sense. Crick and Koch place certain connectivity demands on any potential NCC (that it have connections to the prefrontal cortex), somehow putting the cart before the horse in this research program. The problem with their approach is that once we assume that we understand the functional connectivity of any NCC, then we are unable to learn anything new about the nature of consciousness from finding its neural correlates. Their research program leaves us with two options. Either we discover that all areas connected with PFC are NCCs for different contents of consciousness. Or we discover that only a subset of these areas are NCCs and have to explain what makes them different from the remainder. We are still left without a clue as to how PFC connections generate consciousness. Until we have an answer to that question, we are simply engaged in more brain mapping. The approaches of both Logothetis and Crick and Koch make further assumptions about the NCC that are not necessarily justified. First, they assume that phenomenal consciousness and neural firings will correlate on temporal dimensions. But, as Amanda pointed out, they have established no guidelines for what degree of correlation is acceptable. One might be tempted to conclude that binocular rivalry studies confirm the temporal correlation hypothesis, but in fact all they show is that there is a place in the processing stream where there is a (rough) temporal correlation between phenomenal consciousness and visual information processing. It could be that the neural correlate of consciousness resides in the circuits independent of content-related processing that are, perhaps, firing all the time, i.e., those circuits that show no deviation from their baseline when we are conscious of a stimulus. That's the activity that usually gets subtracted out of any brain imaging or mapping study - the activity that is constant across all trials. Content-related firings (such as those in IT) might just feed into whatever background circuits are maintaining this overall conscious state, accounting for their differential localization as a function of stimulation or attention. But this is not surprising, once one already knows that the brain compartmentalizes its content in various ways. We had a good idea about this already, from numerous lesion and experimental studies that continue to be confirmed and refined by brain mapping techniques. What neural activity or circuitry determines whether and how a particular content gets into consciousness seems to me a far more interesting question that what neural areas are active when that particular content is represented. The more piecemeal approaches of the binocular rivalry type are likely to leave us with a huge and unwieldy catalogue of NCCs in multiple modalities and at multiple levels of processing. My guess is that they will tell us little NEW about consciousness. In my opinion, our search for the neural correlates of consciousness should focus on questions of access. They should be guided by our current best understanding of the function of consciousness, and of why certain contents are made available to consciousness at any given time. Finally, until we have a more fine-grained mapping of the phenomenal realm, our questions about what phenomenal content states map onto what neural states would seem premature. From switanek@U.Arizona.EDU Thu Sep 23 20:32:37 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id UAA10960 for ; Thu, 23 Sep 1999 20:32:37 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGBMSTUZHCB8VLW8@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Thu, 23 Sep 1999 20:32:41 MST Received: from orion.U.Arizona.EDU by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGBMSRKIWGB8VGV0@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Thu, 23 Sep 1999 20:32:38 -0700 (MST) Received: from localhost (switanek@localhost) by orion.U.Arizona.EDU (8.8.6 (PHNE_17190)/8.8.6) with ESMTP id UAA04202; Thu, 23 Sep 1999 20:32:36 -0700 (MST) Date: Thu, 23 Sep 1999 20:32:36 -0700 (MST) From: Nicholas J Switanek In-reply-to: <199909240113.SAA10848@paradox.soc-sci.arizona.edu> To: David Chalmers Cc: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO I'd like to know whether I'm thinking of NCCs in the right way. It appears that an NCC is a sequence of neural firings. Then, characterizing any NCC requires the identification of a set of neurons and the order in which they fire. If this is on the right track, then a horde of issues descends upon me. 1. First, a small point: Dave wrote in his last email that the argument for the sufficiency of an NCC N might be damaged if the brain is. There could be, he suggests, activity in N without the conscious state that N correlates ever arising. He might mean that there is activity in the set of neurons that partially make up an NCC. But there's no reason to expect that firing alone should give rise to the conscious experience N gives rise to, because N only becomes sufficient for the conscious experience when N is fully instantiated, the neurons in the relevant set fire in the precisely ordered way. The strengths of the argument for sufficiency are not harmed in the case Dave suggests. 2. Second, let there be a lesion. Now, I might imagine that the set of neurons relevant to a particular NCC N might be unscathed by this nasty lesion and the consciousness-bestowing, magical, exact sequence of firings in this set, call the set {N}, takes place, but the conscious experience the mind was wont to have does not arise. Intuitively, we might say that lesions leave us woozy; it's difficult having the same experience after undergoing them. What this situation suggests to me is that there might be important relational properties adhering the neurons of {N}. How the neurons are related to one another and to other parts of the brain--how off(or off-off) Cartesian broadway they are--might give clues to why they and not another set of neurons are used to instantiate N. Presumably we couldn't string together the elements of {N}(in a string that might cross itself), watch them firecrack in sequence, and expect the conscious experience to arise. The geometry of the set is relevant. Indeed, this idea might suggest an explanation how the damaged brain finds another set of neurons, {M}, in which to allow a certain conscious experience to take place, one that N used to correlate. The relational properties of the neurons in {M} might be required to be isomorphic to those of the elements of {N}, for M and N to be able to give rise to the same conscious experience. 3. Last, I think Lis is right, if I understand her correctly. Apparently, talk of neural correlates just means the brain mapping project is getting closer to the neural level of description and not necessarily that we are any closer to an explanation of consciousness. Maybe sets of neurons that might serve as NCC-platforms might be isolated and identified. Maybe isomorphic functions between neural sets might be described, to explain the flexibility of NCCs. Perhaps rules will be found that govern the interaction of neurons participating in an NCC, rules that might be strongly correlated to certain types of conscious experience. Won't the hard question remain? A last, interesting paragraph of Lis's is on the extension of experience and its neural correlate through time. It calls to my mind arguments about the nature of time. We think ourselves to be instantaneously conscious, but clearly we wouldn't say that the neurons firing in an NCC at a particular instant give rise to that instant's experience. Nick ps Sorry for all the wanton speculation. I hope the anologues weren't too loose. I am grateful to any who would explain me my folly. From hvorecky@U.Arizona.EDU Fri Sep 24 11:51:43 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id LAA12164 for ; Fri, 24 Sep 1999 11:51:43 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGCIWB49EOA62REM@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Fri, 24 Sep 1999 11:51:45 MST Received: from f1n2.u.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGCIW82KBKB8VIVJ@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Fri, 24 Sep 1999 11:51:41 -0700 (MST) Received: from localhost (hvorecky@localhost) by f1n2.u.arizona.edu (8.8.8/8.8.8) with ESMTP id LAA20926; Fri, 24 Sep 1999 11:51:40 -0700 Date: Fri, 24 Sep 1999 11:51:40 -0700 (MST) From: Juraj Hvorecky Subject: Neural correlate of consciousness In-reply-to: <199909240113.SAA10848@paradox.soc-sci.arizona.edu> To: David Chalmers Cc: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO As I was thinking about the problem David rises concerning the issues of lesion and their relation to discrimination of the NCC, it now looks to me that if his reasoning is correct (and I believe it is), the distinction between "good" and "bad" lesions is nmore than necessary. Why? Well, because it not only applies to the search for NCC, but to any neurological correlate of any disorder. Take any dissociation studies, such as prosopagnosia. In order to study it, neuroscientists have determind a possible "prosopagnosia correlate", that is a part of bring, which, if dmagaed, patients are unable to recognize familiar faces. But what if the reasoning is mistaken and the thing they have identified is only somehow part of a causal chain, which leads to face recognition. That is, the "real" correlate is more downstream, but this particular place is so important, that it in fact bears all the important sings of the real correlate. Does it make sense? Well, if it does, then we have to face consequences. One of them would be, that the proposals Dave's talk about as correct ways to determine correlates (unusual percepts and direct brain stimulation) might not also be so helpful. How do you, for example, want to test "correlate" of faces recognition with UNUSUAL stimulus? What is an unusual face. If the correlate functions only in usual conditions, then such cases might not work. We would still be safe as far as the brain stimulation works, but concerning the overall technical difficulties, the sceptic might conclude that the time we determine the real correlate is infinitely far away... Great weekend to everyone! From logant@U.Arizona.EDU Sat Sep 25 13:48:13 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id NAA13855 for ; Sat, 25 Sep 1999 13:48:13 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGE1974Q7KB8VWPU@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Sat, 25 Sep 1999 13:48:20 MST Received: from orion.U.Arizona.EDU by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGE195EN80B8VT5T@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Sat, 25 Sep 1999 13:48:18 -0700 (MST) Received: from localhost (logant@localhost) by orion.U.Arizona.EDU (8.8.6 (PHNE_17190)/8.8.6) with ESMTP id NAA14264; Sat, 25 Sep 1999 13:48:16 -0700 (MST) Date: Sat, 25 Sep 1999 13:48:16 -0700 (MST) From: Logan T Trujillo Subject: Re: What does a neural correlate of consciousness explain In-reply-to: <199909240113.SAA10848@paradox.soc-sci.arizona.edu> To: David Chalmers Cc: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO Hello All: I have a couple of comments to make concerning the discussion of NCCs. First, Lis Nielsen recently made a post that I believe is relevant to the arguments Brad and I made on Tuesday concerning the difficulty in localizing a NCC. Here is an excerpt from her posting: > The approaches of both Logothetis and Crick and Koch make further assumptions > about the NCC that are not necessarily justified. First, they assume that > phenomenal consciousness and neural firings will correlate on temporal > dimensions. But, as Amanda pointed out, they have established no guidelines > for what degree of correlation is acceptable. One might be tempted to > conclude that binocular rivalry studies confirm the temporal correlation > hypothesis, but in fact all they show is that there is a place in the > processing stream where there is a (rough) temporal correlation between > phenomenal consciousness and visual information processing. > > It could be that the neural correlate of consciousness resides in the > circuits independent of content-related processing that are, perhaps, firing > all the time, i.e., those circuits that show no deviation from their baseline > when we are conscious of a stimulus. That's the activity that usually gets > subtracted out of any brain imaging or mapping study - the activity that is > constant across all trials. Content-related firings (such as those in IT) > might just feed into whatever background circuits are maintaining this > overall conscious state, accounting for their differential localization as a > function of stimulation or attention. But this is not surprising, once one > already knows that the brain compartmentalizes its content in various ways. > We had a good idea about this already, from numerous lesion and experimental > studies that continue to be confirmed and refined by brain mapping > techniques. Lis' comment seems to accord with the metaphor I put forward about IT being the sculptor that acts upon the clay (lower visual cortical areas, i.e. V1-5 et al) to produce a conscious visual percept. What reason do we have to think that visual qualia are restricted to only one area of the visual pathway and not another? The example of blindsight may seem to suggest that certain portions of the visual stream have qualia producing aspects in that damage to these areas leads to a loss of phenomenology on the part of the blindsight patient. However there are two things to consider: (1) if we remove processing at lower levels (as in blindsight) there is no lower level visual information for higher level areas to act upon to create a visual percept (i.e. no clay). Thus both lower and higher visual areas are needed for the creation of a conscious percept. Given this, does it make sense to say that one area is a NCC and the other is not?; (2) as we will see in the case of blindsight, damage to lower visual areas does not preclude the possibility that visual information can still reach conscious awareness. It is just that the phenomenological manifestation of this information within the blindsighted individual is extremely impoverished. Depending upon the extent of the damage, the phenomenology may range from absolutely nothing to vague (almost indescribable) sensations. In the case of the latter, could it be that the vague sensations are produced from higher order visual areas that receive the information via roundabout loops that bypass the lower visual areas? Since this information is not coming through the visual stream proper, activation of the higher areas may lead to the production of qualia that reflects the higher level activity. Since the lower areas are damaged (and not producing qualia) this "higher level" qualia would not be bound to the "lower level" qualia, and thus the resulting sensations would be "vague" or "nebulous". In light of the fact that we still do not know how neurons produce qualia at all, views that ascribe qualia production to only limited areas of the brain are extremely premature. My final comment is the following: I think the difficulty with the definition of an NCC that we have been discussing (and it seems to be the core of Lis' argument above) is in the notion of "correlation". Weak correlations and strong correlations are both correlations; what reason do we have to believe a priori that only strong correlations qualify as the proper criteria for a NCC? Logan T. From franzen@U.Arizona.EDU Sat Sep 25 17:45:41 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id RAA14211 for ; Sat, 25 Sep 1999 17:45:41 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGE9JMG140B8W6S1@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Sat, 25 Sep 1999 17:45:48 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGE9JKXN0WB8VOSV@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Sat, 25 Sep 1999 17:45:46 -0700 (MST) Received: from f1n7.u.arizona.edu (IDENT:franzen@f1n7.U.Arizona.EDU [128.196.137.107]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id RAA14205 for ; Sat, 25 Sep 1999 17:44:09 -0700 Received: from localhost (franzen@localhost) by f1n7.u.arizona.edu (8.8.8/8.8.8) with ESMTP id RAA31970 for ; Sat, 25 Sep 1999 17:44:18 -0700 Date: Sat, 25 Sep 1999 17:44:18 -0700 (MST) From: Peter L Franzen To: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO It is concluded from the empirical work of Logothetis on vision in monkeys that "V1 is unlikely to be or involve an NCC, for example, due to the failure of V1 cells to correlate with the contents of consciousness." I don't follow the reasoning here. What is being lost is the issue of networks in the brain, and that these networks, through interaction, may bring about the counsciousness phenomenon. And, how exactly can the inferior temporal (IT) cortex be the NCC, and not V1, when the IT receives projections, either directly or indirectly, from V1? With all the rush toward localization, do we risk losing ground in the search for NCC by missing the bigger picture? A person can lose a part of their brain that is thought to be important in some function through lesions, but not lose that function completely. There are often multiple pathways, and losing part of the circuit does not guarantee that there are not other ways the circuit can be completed. Of course, it is always possible that there is more than one NCC, and that a lesion in one does not disrupt the other. Or perhaps there is remaping of a function elsewhere in the brain. How would that change a NCC? As for whether the NCC state is required to be necessity/sufficient for the conscious state... If an NCC state is only required to be sufficient for the corresponding state, but not necessary, are we any further in understanding how the NCC leads to consciousness or an experience of consciouss state? If it is only sufficient, and not necessary, than how do we know its truly improtant for the conscious state? From press@U.Arizona.EDU Sun Sep 26 16:08:57 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id QAA16864 for ; Sun, 26 Sep 1999 16:08:57 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGFKH2GMDSB8W1WI@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Sun, 26 Sep 1999 16:09:06 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGFKH1DY34B8VWRV@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Sun, 26 Sep 1999 16:09:05 -0700 (MST) Received: from f1n2.u.arizona.edu (IDENT:press@f1n2.U.Arizona.EDU [128.196.137.102]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id QAA16859 for ; Sun, 26 Sep 1999 16:08:51 -0700 Received: from localhost (press@localhost) by f1n2.u.arizona.edu (8.8.8/8.8.8) with ESMTP id QAA20664 for ; Sun, 26 Sep 1999 16:09:01 -0700 Date: Sun, 26 Sep 1999 16:09:01 -0700 (MST) From: Joel K Press Subject: Minimal sufficiency and redundancy To: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO All - When Dave endorses minimal sufficiency as the proper criterion for identifying the NCC, he notes that on this criterion it will be possible that there will be more than one NCC. He seems untroubled by this, and in one sense I share that attitude. Not only should we not be surprised if there is no "Cartesian Theater" where all the representations involved in phenomenal consciousness occur together, but we should not be surprised if there are redundant NCCs. Given the prevalence of redundant systems throughout the body (kidneys, lungs, etc.) it seems to make good evolutionary sense, other things being equal, to have multiple systems for consciousness. In fact, I think there is strong evidence that this is the case which can be found in studies of split brain patients. Im sure that the psychologists among us know much more about these experiments than I do, and perhaps some of the philosophers too, but just in case some dont, heres a brief (and somewhat idealized) synopsis: When the bundle of nerve fibers (corpus callosum) that connect the right and left cerebral hemispheres is severed, the patients suffer a dissociation phenomenon. Because each hemisphere receives input from and has control over (primarily) the opposite side of the body, it is possible to create situations in which one hemisphere knows something that the other doesnt. For example, if the experimenter were to show the word "handbag" to one of these patients in such a way that the "hand" part would only be perceived by the left hemisphere and the "bag" part only by the right hemisphere, and if the patient were then asked to point at the object named (from a more or less random collection of objects provided) , the right hand (controlled by the left hemisphere) would point to, say, a mannequin hand, and the left hand would point to, say, a paper bag. The patient would not point out the purse. Many would say that this is evidence that in the split brain patient there are two independent streams of conscious states. Though the split brain cases are fascinating in themselves, what they suggest about normal cases is more relevant to my point here. Since cutting the corpus callosum has no effect on the structure of the rest of the brain, it seems reasonable to conclude that the structures responsible for consciousness must be duplicated in each hemisphere even in normal patients. Since each of these structures can function on its own in the split brain cases, each would seem to be an NCC. In normal patients, the corpus callosum seems to have an information sharing function, so that the two NCCs end up with the same content. (For the sake of simplicity, assume that there is a single Cartesian-Theater-style NCC in each hemisphere). Dave doesnt seem to be worried about this, but I think that perhaps he should. The worry I have is that the existence of multiple NCCs may raise a set of questions to which he cannot provide an answer. For example, if every normal person has a right NCC and a left NCC, how could we decide which of the following views is true. 1) A normal persons phenomenal consciousness is produced by both NCCs running in parallel. 2) A normal persons phenomenal consciousness is produced by his left NCC. The right NCC performs all the same operations as the left NCC but does so without producing any phenomenal properties. 3) 2) but with right and left reversed. 4) A normal person has two independent sets of phenomenal consciousness (two minds), one produced by each NCC. Since the two run in parallel, the phenomenal contents of each mind is identical, but they are nevertheless separate minds. The problem would only get worse if we found that there were other redundancies within the hemispheres, or if the Cartesian Theater model turned out to be inaccurate in such a way that redundant versions of all sorts of representations of conscious contents were scattered about in the brain. The reason that I think Dave will have a hard time defending one of the options above is that he has no causal explanation of phenomenal properties, merely a correlation. If we had a theory of consciousness that told us that the NCC causes consciousness by emitting "phenomenal particles" which we could detect with a "phenomometer," we might be able to verify that, say, 2) was the correct option because normal subjects only emit "phenomenal particles" from their left hemispheres. But, of course, this is precisely what Dave doesnt think we will ever be able to do. I suppose he could appeal to criteria of simplicity, appeal to the best explanation, and so on, but it is not clear to me that any one of these possible claims is any simpler, or a better explanation, etc. Of course, maybe a theory of consciousness that narrows things down to one of the above alternatives is the best we can do given our epistemological situation, but if so, I think a critic could justifiably claim that our "explanation" of consciousness is much too weak. After all, if we cannot even tell how many minds (streams of consciousness, conscious states) a person has, how much do we really know about the phenomena we claim to have explained? Joel From serobert@U.Arizona.EDU Sun Sep 26 18:33:00 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id SAA17248 for ; Sun, 26 Sep 1999 18:32:59 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGFPHNTK00B8VRQR@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Sun, 26 Sep 1999 18:33:08 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGFPHM3E3KB8VZ5E@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Sun, 26 Sep 1999 18:33:06 -0700 (MST) Received: from f1n3.u.arizona.edu (IDENT:serobert@f1n3.U.Arizona.EDU [128.196.137.103]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id SAA17243 for ; Sun, 26 Sep 1999 18:32:38 -0700 Received: from localhost (serobert@localhost) by f1n3.u.arizona.edu (8.8.8/8.8.8) with ESMTP id SAA32280 for ; Sun, 26 Sep 1999 18:32:49 -0700 Date: Sun, 26 Sep 1999 18:32:48 -0700 (MST) From: Simon E Roberts-Thomson Subject: Correlation and Lesion studies In-reply-to: To: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO Dear all, Many of the recent mails have tended to be rather scathing of either the use of correlation in the search for NCC, or the use (or interpretation) of lesion studies. I thought that it might be interesting to try and defend both of these - perhaps it will help us to focus more clearly on their problems. (1) Correlation The criticisms of correlation seem to be centred on the idea that we are failing to advance our knowledge of *consciousness* when we correlate neuronal firing with specific conscious experience. Nicholas seems to suggest that all we are doing is improving our 'maps', and not answering the real questions. In a similar vein, Lis is worried that the *real* work of consciousness might be taking place in an area where firing is unrelated to content, and hence would fail to sow up in these mapping studies. I will try to answer each charge in turn. With respect to Nicholas' concern, it may well be that all we are doing when we study neuronal firing is improving our 'maps' of the brain. However, it does not seem to me that this is a problem. Presumably any theory of consciousness will need to account for the correlations between specific content and firing patterns - a part of what such studies are doing is defining the things that a brain-based theory of consciousness needs to address. Thus whilst these studies do not in themselves give us a theory, they are essential in that they define part of the problem. Turning now to Lis' point, it is certainly possible that the real is such that its firing is not specifically related to content. Instead, areas of the brain whose firing is content-related feed into this (continually?-)firing area, which then produces consciousness. Whilst this view is possible, however, it seems to me to be implausible. If the real NCC does not change its firing rate or pattern when confronted with differing stimuli, then how do we have differing content in consciousness? It seems reasonable to assume that if there is a central NCC, then it must use something like differential firing rates to engender different conscious experiences. If this is so, then presumably such changes should appear of the scans (if not, then this must be due to the limits of the equipment, and not the theory behind their use). In other words, there will be some sort of correlation between the firing and the experience. (2) Lesion Studies. One of the problemswith the use of lesion studies that was discussed both here and in class, was the concern that whilst a particular lesion might result in the loss of a certain type of consciousness, there is no reason to assume that the area in which the lesion is situated is the NCC for that particular type of consciousness. It may be that the region is merely a part of a chain which leads to the particular NCC. Likewise, there might be multiple pathways to an NCC, such that the loss of one does not result in the loss of that particular consciousness. In response to these concerns, I would like to say that this is not a reason not to use lesion studies, but merely a reason to be careful of how such studies are interpreted. Thus if we had one lesion study which suggested that region A was involved in visual consciousness, then we should not necessarily assume that visual cionsciousness takes place in A, but rather the much weaker claim that A is involved in visual consciousness. If there were to be multiple studies of related lesion studies, all of which seemed to result in a loss of visual consciousness, then perhaps we could 'map out' the visual pathway (of course, if the brain turns out to be very plastic, then this would be more difficult). In a similar way, it might be possible to identify redundant pathways. Thus these studies will help us to identify what parts of the brain are necessary and/or sufficient for consciousness. I hope that I have managed to put together a passable defence of the above concepts. I think that they are likely to be very important to the study of consciousness, and should not be dismissed lightly. Simon. From anhabib@U.Arizona.EDU Sun Sep 26 19:17:36 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id TAA17273 for ; Sun, 26 Sep 1999 19:17:36 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGFR1Z3UW0B8VQA5@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Sun, 26 Sep 1999 19:17:46 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGFR1X4GM8B8WMAC@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Sun, 26 Sep 1999 19:17:43 -0700 (MST) Received: from f1n7.u.arizona.edu (IDENT:anhabib@f1n7.U.Arizona.EDU [128.196.137.107]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id TAA17268 for ; Sun, 26 Sep 1999 19:17:26 -0700 Received: from localhost (anhabib@localhost) by f1n7.u.arizona.edu (8.8.8/8.8.8) with ESMTP id TAA30020 for ; Sun, 26 Sep 1999 19:17:37 -0700 Date: Sun, 26 Sep 1999 19:17:37 -0700 (MST) From: Allen N Habib Subject: Some thoughts on the NCC To: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO Hey all, I have three thoughts on the NCC that have been bothering me since our class talk. The first is this, why are we limiting our discussion solely to considerations of necessity and sufficiency? If we are willing to admit that there may be more than one NCC for even a very specific conscious state (what Dave would call a content state), and if we are willing (pace Dave) to admit the possiblity that an NCC could be instantiated, even concomitently with the requisite background state C, without the conscious state arising? (I should note here that Dave argues for this not from his thesis that any physical state can be seperated from any conscious state because consciousness is not reducible to physicality, but rather from the hypothetical situation of 'downstream lesions', lesions that cut off an NCC from other, later processing that might result in a loss of consciousness, even though the NCC was intact and in state N, Sarah has something to say against this possibility, but even if her argument is correct, Dave could always rely on his earlier claim of seperability.) Anyway, the point I want to make is that if we take these two possibilities seriously, from a practical standpoint we might never arrive at the 100% correlation that necessity, sufficiency or even minimal sufficieny demands. Mightn't we be better off (epistemically) if we were to take a reliabilist stance, and judge putative NCCs by their correlation percentages, positive and negative, with conscious states? So, for example, NCC1 might have a 85% positive correlation with Conscoius State 1 (CS1), meaning that 85% of the time that NCC1 is in N, CS1 occurs, and a 90% negative correlation, meaning that 90% of the time that NCC1 is not in state N, CS1 does not occur. Of course, these ratios can come apart because of the possibilities of multiple realisability and full physical instantiation without phenomenal (conscious) instantiation. Then we could compare different candidate NCCs by seeing how their numbers stack up, and we could figure out what sort of shortcomings an NCC might have given its different performances on the two scales, e.g. an NCC with a high positive but low negative correlation might be one of a number of individually sufficient NCCs for that CS. Also, we could use these scores to settle questions about hwere to draw the boundaries of an NCC, by referring to the changes in the numbers that occur when we include (or exclude) certain substructures of an NCC. For these reasons, I think it might behoove us to focus on this probabilistic correlation as the defining one of NCCs, and not the all-or-nothing one we have been using so far. My second worry is over Dave's suggestion that artificial brain stimulation might be used to exclude non-core elements of an NCC, in the following way: If we can artificially stimulate a neural network 'downstream' from a putative NCC, and we get the same CS as we did when the NCC was supplying the inputs, then we have a case for eliminating the earlier NCC, since it no longer seems necessary for the CS. What worries me here is that it seem that what we are doing when we 'artificially stimulate' the downstream system is taking on the functional role of the earlier system. But if this is the case, why is it impossible to take on the functional role of part of an NCC? And if it is possible, how will we differentiate between 'merely' background or enabling states and actual (parts of) NCC states? It seems to me that unless there is some a priori reason that a machine cannot assume the functional role of part of an NCC, then we have to come up with a way to tell them apart, otherwise we risk arbitrarily excluding neuronal groups that are possibly part of an NCC. Which brings me to my third point. One might argue that one way to tell a background or non-core neural state from an NCC one is via temporal indexing, using the onset of consciousness as a benchmark. So, we might say, anything that occurs prior to the onset of consciousness is a possible background system, but not part of an NCC, and anything that occurs afterward is at least possibly part of one. But this is insufficient in two ways: Firstly, even if it were a viable method of parsing, it would only answer questions about upstream neural candidates, and second, it doesn't even seem a viable method, to me, because it tacitly assumes a symmetry between conscious temporality and physical temporality. If we are going to use the onset of consciousness as our temporal benchmark, and exclude neural candidates for NCChood on the basis of their preceeding this benchmark, then we have to assume that the onset of consciousness is co-temporaneous with the NCC processing that is responsible for it, and this is not necessarily the case. It might be that the onset of consciousness lags behind NCC processing, and as a result the subject enters the CS after the crucial NCC processing has occured. We can imagine further that some other processing, downstream of the NCC, occurs co-temporaneously with the onset of consciousness. In such a scenario (which seems completely possible to me) if we were to employ the benchmarking method outlined above, we might well associate the CS with the later, non-NCC processing, because the actual NCC processing occurs before the onset of consciousness, and is therefore 'upstream'. As to why consciousness might lag behind the NCC processing that underlies it, who knows, but the relationship between the physical and the phenomenal is certainly mysterious enough for this to be possible. (I apologise to Dave for the lateness and to everybody for the long-windedness and pedantry) Al. From sawright@U.Arizona.EDU Sun Sep 26 19:31:38 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id TAA17287 for ; Sun, 26 Sep 1999 19:31:37 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGFRJCXLTSB8VIC9@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Sun, 26 Sep 1999 19:31:47 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGFRJBBX28B8VZMG@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Sun, 26 Sep 1999 19:31:44 -0700 (MST) Received: from pavo.U.Arizona.EDU (pavo-2.U.Arizona.EDU [128.196.137.196]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id TAA17282 for ; Sun, 26 Sep 1999 19:31:18 -0700 Received: from localhost (sawright@localhost) by pavo.U.Arizona.EDU (8.8.6 (PHNE_17190)/8.8.6) with ESMTP id TAA20849 for ; Sun, 26 Sep 1999 19:31:29 -0700 (MST) Date: Sun, 26 Sep 1999 19:31:29 -0700 (MST) From: Sarah A Wright Subject: Downstream Lesions To: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO I wanted to dwell a bit on the worries about lesions in "What is the Neural Correlate of Consciousness?" and from last class. It seems that lesion studies should be scrutinized with care both because they can lead us to exclude some of the NCC from our NCC, and because they can lead us to include some non-NCC in our NCC. (They seem to make the NCC both too wide and too narrow.) In the first case, we have the worries about plasticity of the brain. If, after a lesion, the brain is capable of rerouting its processing in such a way that the NCC actually changes location, then the information from the study of such lesions might give us a sufficient NCC, but not the NORMAL sufficient NCC. Relying only on lesion studies might cause us to reject as the NCC an area of he brain that we normally use as the NCC, but which in marginal cases we can work around. In the second case, we have the putative NCC in a petri dish, or other less extreme versions of this problem. In these cases it seems that the putative NCC would not be conscious. However we donUt want to hold this against the NCC we have isolated, since there are certain background conditions we want to hold fixed in our search for the NCC. These background conditions might include getting appropriate (not processed) sensory data, or, if there is an Ron switchS for consciousness, having that switch RonS. Requiring the NCC to operate despite all lesions (down to amputation) seems to ignore the fact that we are looking for the MINIMAL sufficient NCC. Now the impetus for each of the worries above seem to be directed at lesions that occur upstream of (or in) the NCC. Sensory input should occur far upstream of the NCC, and, whatever an "on switch" might turn out to be it should be activated either before or at the time of consciousness. Likewise, it seems that lesions before or in the NCC are the ones we could use plasticity to get around. These concerns have brought us to the the question of where to draw the upstream line on the NCC, or how far back we need to go. There is another related question of how far forward we should draw the downstream line of the NCC. In particular I was bothered by this claim in "What is the Neural Correlate of Consciousness?" >Less radically, one can imagine placing lesions immediately downstream >from a candidate NCC N, so that NUs effects on the rest of the brain are >significantly reduced. In such a case, it is probable that N can be >active without the usual behavioral effects associated with >consciousness,and *quite plausibly without consciousness itself*. I am willing to admit that the downstream limit of the NCC need not be as far down as behavior; we certainly donUt want to say that someone who is paralyzed cannot have a full NCC simply because they cannot report it. But it seems to me that a lesion downstream of the NCC should not interfere with consciousness itself. In the petri dish case we discount certain background conditions because while we can't have consciousness without them, they also don't seem relevant to conscious experience. Sensory input and brain "on"ness are plausible examples. But what could the background conditions here be? These would be conditions that occur after consciousness, but also enable it. I cannot think of any plausible examples of this sort of background condition. If an event downstream from our NCC is needed for consciousness, rather than discount this event, we should update our NCC to include the substrate of that later event. (Note that the background conditions needed here aren't those (epistemological ones) that make the consciousness of others available to us, but rather those (ontological ones) that allow conscious experience.) If you all can think of any examples of a downstream background condition that should be discounted, please let me know. From lan@U.Arizona.EDU Sun Sep 26 22:20:56 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id WAA17539 for ; Sun, 26 Sep 1999 22:20:55 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGFXG9R51CB8UKZ0@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Sun, 26 Sep 1999 22:21:05 MST Received: from orion.U.Arizona.EDU by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGFXG8P4JKB8VZUA@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Sun, 26 Sep 1999 22:21:04 -0700 (MST) Received: from localhost (lan@localhost) by orion.U.Arizona.EDU (8.8.6 (PHNE_17190)/8.8.6) with ESMTP id WAA26118; Sun, 26 Sep 1999 22:21:03 -0700 (MST) Date: Sun, 26 Sep 1999 22:21:03 -0700 (MST) From: Lonnie A Nelson Subject: Considering conditions C In-reply-to: <199909230807.BAA09334@paradox.soc-sci.arizona.edu> To: David Chalmers Cc: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO Given that the necessity and sufficiency issue has lead the discussion to speculate about possible versions of "conditions C" which would be the "normal conditions under whcih we would expect a NCC to operate". The question has become "what are normal conditions?" As Dave mentioned in class one could start as far away from the brain as the type of stimulus that one is exposed to and only allow "environmentally valid" stimuli. This has coherence under the "natural functioning" band of thought, but it won't really allow us to play with the conscious/subliminal distinction that would need to be experimentally allowed in order to have any data whatsoever to procede with. While I believe that we should leave the brain and perceptual systems in as pristine a state as possible so that we can observe the system's "normal functioning", I would question what is referred to as "environmentally valid" regarding the allowed for stimuli set. WHile binocular rivalry may be a supremely unlikely situation to find oneself in, we have in no way said that the stimuli needed to be "evolutionarily valid" as the only monkeys finding themselves needing to discern binocular rivalrous stimuli are in a lab with electrodes in their brains, hence not a "naturally occurring" condition. However, if we take Environmentally valid, and make it into "natural reception system". thus cutting the world off, for the time being at the level of the organism, we have a reasonably wide array of possible manipulations that we are able to subject the system to and that we may monitor its activity in response to. For these "natural modes of signal reception" we have (as should be obvious) our five senses. The most easily controlled of the phenomenal consciousness set. The case being that psychophisics has given us a pretty good idea of general thresholds for the various senses with regard to their respective stimuli; we should be able to monitor the difference between JND (Just noticeable difference) below and above the threshold for conscious expperience. The difference in activity of the system between these two inputs should give us a reasonable approximation of the brain subsystems involved in "phenomenal experience" of a selected modality. Given that the actual energy difference in the light input for the two stimulus strengths can be known, a subtraction from the output values (in units expressed by activity) could be used to determine the generative potential of conscious experience, and the localization of this activity can be carried out via PET imaging, or fMRI (as long as you aren't using audition as your mode of exploration). And while these imaging techniques certainly have their drawbacks in terms of fine versus rough grained localization, we are given the added benefit of not having to disturb the system or "sacrifice" it post experiment to make sure that we were measuring the right (fine grained) place. While this methodology will surely only give us a rough picture of a NCC it will also be an accurate one in the "natural state of the system", which may be of some extra value, given that we would not have to deal directly with the plasticity problem. Just some thoughts, --Lonnie ___________________ It is a common fate of all knowledge to begin as heresy and end as orthodoxy. -Thomas Huxley Lonnie A Nelson Department of Psychology Human Energy Systems Laboratory University of Arizona lan@u.arizona.edu From lachter@u.arizona.edu Mon Sep 27 10:45:27 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id KAA18548 for ; Mon, 27 Sep 1999 10:45:27 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGGNGD03Y8B8R43M@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Mon, 27 Sep 1999 10:45:37 MST Received: from trifid.u.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGGNGBR23KB8VV3N@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Mon, 27 Sep 1999 10:45:36 -0700 (MST) Received: from [128.196.99.98] ([128.196.99.98]) by trifid.u.arizona.edu (8.8.8/8.8.8) with ESMTP id KAA26716; Mon, 27 Sep 1999 10:45:34 -0700 Date: Mon, 27 Sep 1999 10:45:45 -0700 From: Joel Lachter Subject: Re: Considering conditions C In-reply-to: X-Sender: lachter@pop.u.arizona.edu To: Lonnie A Nelson Cc: David Chalmers , scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: text/plain; charset=us-ascii References: <199909230807.BAA09334@paradox.soc-sci.arizona.edu> Status: RO Lonnie proposes that we look for NCCs by doing imaging studies where we subtract visual distinctions which are just below one just noticeable distance (JND) from those that are just above it. While this seems plausible in principle there are a couple things to keep in mind. First, notions like "JND" and "threshold" are really statistical. They are defined as the point where an observer can get some percentage of trials correct (usually 75%). Thus, just below a JND, the subject will usually be getting 70% correct and just above they will be getting 80%. The whole system is graded, and the parts necessary for consciousness are not likely to be much different. Second, you are likely to get bigger differences as you move downstream independent of the location of any NCC. This is because all that downstream stuff that is hanging out thinking "Is that something? Where is it? Oh, I think that was it? Maybe not...," is going to act differently when it actually is getting information from when it is just hanging out. Third, it is not clear that objective and subjective thresholds are the same or should be measured the same way (you are likely to get a lot of flack on this no matter how you do it). In short, while this is an appealing idea, I don't think it could actually be made to work. Joel From lan@U.Arizona.EDU Mon Sep 27 14:00:14 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id OAA18930 for ; Mon, 27 Sep 1999 14:00:13 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGGU9S73LCB8VXUE@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Mon, 27 Sep 1999 14:00:23 MST Received: from orion.U.Arizona.EDU by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGGU9QXJNKB8LOKA@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Mon, 27 Sep 1999 14:00:20 -0700 (MST) Received: from localhost (lan@localhost) by orion.U.Arizona.EDU (8.8.6 (PHNE_17190)/8.8.6) with ESMTP id OAA27715; Mon, 27 Sep 1999 14:00:18 -0700 (MST) Date: Mon, 27 Sep 1999 14:00:17 -0700 (MST) From: Lonnie A Nelson Subject: Re: Considering conditions C In-reply-to: To: Joel Lachter Cc: David Chalmers , scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO On Mon, 27 Sep 1999, Joel Lachter wrote: I was aware that the threshold was a statistical construct, however, it should not take all that long to test for each subjects individual threshold (perhaps 30 trials, to assume a random distribution). and you would still have a difference even if there were only 30% that the subject was "not seeing" you could compare imaging on that 30% to the trials in which they "saw" the stimuli > Second, you are likely to get bigger differences as you move downstream > independent of the location of any NCC. This is because all that downstream > stuff that is hanging out thinking "Is that something? Where is it? Oh, I > think that was it? Maybe not...," is going to act differently when it > actually is getting information from when it is just hanging out. Wonderful, these are the differences that we will be looking for. What we want to subtract is the "just hanging out" from the "getting information", the larger the differences, the happier we should be. One would expect that the largest difference should be able to be found in the actual NCC itself, as it "got new information" and is no longer "just hanging out". > Third, it is not clear that objective and subjective thresholds are the > same or should be measured the same way (you are likely to get a lot of > flack on this no matter how you do it). This can be done in some way similar to the first reply I made (see above) on individual thresholds. And when you are trying to do something in a novel way, you will always catch flack, the object should be to get the most valid form of information that you can for each subject that you test. and While I realize that this sounds easier than it would be in actuality, this is not so big a problem that the whole process could not be made to work, imho. --Lonnie ___________________ It is a common fate of all knowledge to begin as heresy and end as orthodoxy. -Thomas Huxley Lonnie A Nelson Department of Psychology Human Energy Systems Laboratory University of Arizona lan@u.arizona.edu From lachter@u.arizona.edu Mon Sep 27 16:06:31 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id QAA19051 for ; Mon, 27 Sep 1999 16:06:31 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGGYOESHSWA63QBN@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Mon, 27 Sep 1999 16:06:41 MST Received: from trifid.u.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGGYODBC8WB8W6LE@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Mon, 27 Sep 1999 16:06:39 -0700 (MST) Received: from [128.196.99.98] ([128.196.99.98]) by trifid.u.arizona.edu (8.8.8/8.8.8) with ESMTP id QAA17456; Mon, 27 Sep 1999 16:06:37 -0700 Date: Mon, 27 Sep 1999 16:06:38 -0700 From: Joel Lachter Subject: Re: Considering conditions C In-reply-to: X-Sender: lachter@pop.u.arizona.edu To: Lonnie A Nelson Cc: David Chalmers , scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: text/plain; charset=us-ascii References: Status: RO At 2:00 PM -0700 9/27/99, Lonnie A Nelson wrote: >I was aware that the threshold was a statistical construct, however, it >should not take all that long to test for each subjects individual >threshold (perhaps 30 trials, to assume a random distribution). and you >would still have a difference even if there were only 30% that the subject >was "not seeing" you could compare imaging on that 30% to the trials in >which they "saw" the stimuli I don't know how much it is worth belaboring this point but... The problem is that you can't pick out the trials on which subjects saw the stimulus from the trials on which they did not. When you are performing such a task you feel like you are guessing on every trial, and the computer is often telling you you are wrong on those few trials you feel somewhat confident about. The correct trials are contaminated by ones where you were just guessing. The wrong trials are contaminated by ones where you thought you saw something but answered the other way because the last three times you responded that you saw it when you were not completely sure you got it wrong. It is not the case that you either see these stimuli or you don't. There is a gray area. When you have stimuli near threshold you are in that gray area. >One would expect >that the largest difference should be able to be found in the actual NCC >itself, as it "got new information" and is no longer "just hanging out". I would not expect it to turn out that way. Some might say that that almost guarantees that it would turn out that way:-) However, it seems to me that any plausible NCC is going to be in some gray area, reflecting how well the observer can actually see the stimulus. Areas farther downstream, however are going to be acting differently because they are not providing a graded output. Downstream areas will act in qualitatively different ways depending on how much information they have (eg., little information = frustration, thoughts of breaking computer; midlevel = interest, pride when you get it right, anger when you get it wrong; higher level = boredom, can do in sleep, thoughts of lunch). In all probability the NCC does not think "Did I see something?" It just represents the available information. Stuff farther downstream thinks about this information and thinks and acts in entirely different ways depending on the amount of information available. Joel From lan@U.Arizona.EDU Mon Sep 27 18:56:55 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id SAA19565 for ; Mon, 27 Sep 1999 18:56:54 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGH4MOPOW0B8WJRL@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Mon, 27 Sep 1999 18:57:05 MST Received: from orion.U.Arizona.EDU by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGH4MN7NZKB8WB2T@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Mon, 27 Sep 1999 18:57:04 -0700 (MST) Received: from localhost (lan@localhost) by orion.U.Arizona.EDU (8.8.6 (PHNE_17190)/8.8.6) with ESMTP id SAA13335; Mon, 27 Sep 1999 18:57:02 -0700 (MST) Date: Mon, 27 Sep 1999 18:57:02 -0700 (MST) From: Lonnie A Nelson Subject: Re: Considering conditions C In-reply-to: To: Joel Lachter Cc: David Chalmers , scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO On Mon, 27 Sep 1999, Joel Lachter wrote: > I don't know how much it is worth belaboring this point but... I am relatively sure that this will not have any far reaching impact on the state of the science, but... > The problem is that you can't pick out the trials on which subjects saw the > stimulus from the trials on which they did not. When you are performing > such a task you feel like you are guessing on every trial, and the computer > is often telling you you are wrong on those few trials you feel somewhat > confident about. The correct trials are contaminated by ones where you were > just guessing. The wrong trials are contaminated by ones where you thought > you saw something but answered the other way because the last three times > you responded that you saw it when you were not completely sure you got it > wrong. It is not the case that you either see these stimuli or you don't. > There is a gray area. When you have stimuli near threshold you are in that > gray area. There should, then be a place on either side of that gray area, one of which is above, and one of which is below or within that gray area. > >One would expect > >that the largest difference should be able to be found in the actual NCC > >itself, as it "got new information" and is no longer "just hanging out". > > I would not expect it to turn out that way. Some might say that that almost > guarantees that it would turn out that way:-) However, it seems to me that > any plausible NCC is going to be in some gray area, reflecting how well the > observer can actually see the stimulus. Which is why the Above condition would have to be "above" and the other would have to be within the grey or below it. > Areas farther downstream, however > are going to be acting differently because they are not providing a graded > output. Downstream areas will act in qualitatively different ways depending > on how much information they have (eg., little information = frustration, > thoughts of breaking computer; midlevel = interest, pride when you get it > right, anger when you get it wrong; higher level = boredom, can do in > sleep, thoughts of lunch). THis could undoubtedly be made at least a little bit interesting, the vast majority of cognitive research is really boring for the subject, but people participate and effects are found. If there is sufficient difference between trials that can be seen with high accuraccy (~100% +/- 3%) vs. slightly below threshold (~chance) the task would not necessarily be that difficult or frustrating, (though boredom will definitely be a factor, but hopefully thoughts of lunch would be randomly distributed across your sample:-) and the imaging should still tell us the differences in the reactivity of the brain. THough we would likely get a picture of the processing going on afterward, this is arguably part of the reaction of the NCC. > In all probability the NCC does not think "Did I > see something?" It just represents the available information. Stuff farther > downstream thinks about this information and thinks and acts in entirely > different ways depending on the amount of information available. > Maybe we are drawing the line in different places for an NCC, I have no desire to cut off the "higher order processes" of consciousness, only to separate out the difference between conscious percept and *not*. Lonnie ___________________ It is a common fate of all knowledge to begin as heresy and end as orthodoxy. -Thomas Huxley Lonnie A Nelson Department of Psychology Human Energy Systems Laboratory University of Arizona lan@u.arizona.edu From lachter@u.arizona.edu Mon Sep 27 21:32:03 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id VAA19710 for ; Mon, 27 Sep 1999 21:32:03 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGHA21BZ80B8W05N@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Mon, 27 Sep 1999 21:32:14 MST Received: from trifid.u.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGHA2077XCB8VX3H@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Mon, 27 Sep 1999 21:32:13 -0700 (MST) Received: from [10.0.2.15] (tec3.Psych.arizona.edu [128.196.98.11]) by trifid.u.arizona.edu (8.8.8/8.8.8) with ESMTP id VAA15192; Mon, 27 Sep 1999 21:32:09 -0700 Date: Mon, 27 Sep 1999 21:31:48 -0700 From: Joel Lachter Subject: Re: Considering conditions C In-reply-to: X-Sender: lachter@pop.u.arizona.edu To: Lonnie A Nelson Cc: David Chalmers , scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: text/plain; charset=us-ascii References: Status: RO At 6:57 PM -0700 9/27/99, Lonnie A Nelson wrote: >There should, then be a place on either side of that gray area, one of >which is above, and one of which is below or within that gray area. Yeah. But now you are not talking about "just above one JND" and "just below one JND". I take it the force of your original proposal was that you could have two stimuli which looked very similar to the retina, but which look quite different consciously and thus presumably quite different at the NCC. Once you move things out of the gray area you are talking about things that look quite different on the retina. >Maybe we are drawing the line in different places for an NCC, I have no >desire to cut off the "higher order processes" of consciousness, only to >separate out the difference between conscious percept and *not*. It seems to me (tell me where I am wrong) that you are proposing running an experiment with two conditions. In condition A the subject can see the stimuli. In condition B the subject cannot see the stimuli. We then do a subtraction to see where "seeing" is occurring. If I have reconstructed your proposal correctly I see two problems with it. One is that you cannot make condition A and condition B comparable on the low level visual stuff (and still keep your clean distinction between being able to see the stimuli and not being able to see the stimuli). As a result, low level things will light up in your subtraction. Similarly, you cannot keep condition A and condition B comparable with respect to high level stuff. Doing a task when you can see the stimuli from doing it when you cannot see the stimuli in more ways than simply being able to see the stimuli. These conditions also differ in cognitive and emotional respects which are likely to cascade throughout the system. So the problem is, before running the experiment, I don't have any strong intuition telling me that the neural correlate of seeing these stimuli is going to light up more strongly than, say, the neural correlate of attending to the banging of the magnet, or the neural correlate of fidgeting. Thus I don't see how you can tell after running the experiment whether you have found the neural correlate of seeing vs a myriad of other possibilities. Joel From chalmers@paradox.soc-sci.arizona.edu Tue Sep 28 23:31:14 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id XAA22099 for ; Tue, 28 Sep 1999 23:31:14 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGISI6JUHCB8WY8J@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Tue, 28 Sep 1999 23:31:27 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGISI5BJ00B8WO2F@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Tue, 28 Sep 1999 23:31:25 -0700 (MST) Received: (from chalmers@localhost) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) id XAA22086 for scicon; Tue, 28 Sep 1999 23:30:40 -0700 Date: Tue, 28 Sep 1999 23:30:40 -0700 From: David Chalmers Subject: NCC postings To: scicon@paradox.soc-sci.arizona.edu Message-id: <199909290630.XAA22086@paradox.soc-sci.arizona.edu> MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=US-ASCII Status: R Hi all, good to see the spate of interesting NCC postings. Here are a few brief comments: Overall comment: we have to distinguish the three questions: (i) what is it to be an NCC, (ii) is there likely to be an NCC, and (iii) what will an NCC tell us about consciousness? In my own paper, I was mostly trying to give a plausible answer to (i) based on the way the notion is used in the field. I also said a few words about (ii), suggesting that it wasn't unreasonable to expect that there may turn out to be NCCs so defined. I said very little about (iii). As I said in the paper, I think it is strategically useful for people in the field to define NCC in terms of correlation alone, without any additional claims such as, the NCC is the basis of consciousness, the NCC generates consciousness, the NCC explains consciousness, the NCC is the key to a theory of consciousness. That way, the search for NCCs is a common, relatively "objective" project that almost everyone can participate in without too much ideological baggage. Of course at the end of the day, that does leave open the question of what, if anything, the NCC will tell us about consciousness. LIS: I take it that Lis is mostly concerned with (iii). The worry is that physiology alone doesn't tell us much; it will only tell us something in the context of a theory. I think there's a lot to be said for that; on the other hand, it is quite possible that isolating an NCC will itself help in the development of a theory. To take Li's example of binocularly rivalry, learning that IT is an NCC doesn't tell us anything much interesting about binocularly rivalry per se. But once we know this, we can look at how and when IT is activated, and how it related to other brain areas, and perhaps reach some very interesting conclusions about the place of conscious vision in binocular rivalry. I do agree with a lot of what Lis says about Crick & Koch and Logothetis's approach being based on assumptions (re temporality, re connections to prefrontal cortex) that may not be justified. The trouble is that everyone needs to make some assumptions to get these things off the ground. Some assumptions are more reasonable than others; when they seem weak, one had better hope that the assumption itself will ultimately be testable. Of course I think that at the end of the day there are some pre-experimental background assumptions that everyone needs to rely on, but these had better be pretty straightforward. (E.g., the connection between report and consciousness.) Lis points that consciousness might somehow be grounded in neural areas independent of content, and that content just "feeds into" consciousness, is interesting. I guess there still has to be a distinction between those contents that make it into consciousness and those that don't; it's not as if merely representing a content anywhere in the brain gets it into consciousness. So one will stil have a relevant distinct between content-related neural systems here. Then the further question is whether the content has to "go somewhere" (e.g. to Lis's base system) to make it into consciousness. If that's the case, then content representation in the base system will be relevant after all. If that's not the case, then (as Simon points out) it's hard to see just why some contents make it into consciousness and not others. NICK: Nick says that a potential NCC N may not be sufficient on its own for consciousness, since one needs the right kind of activity in N. I think that's right, and strictly speaking an NCC should be characterized also in terms of the sort of activity. But the sort of general point I was making will still hold -- one can imagine that exactly that sort of activity could be happening inside N in a petri dish, and one presumably wouldn't get consciousness. So even this activity in N isn't sufficient for consciousness, except against the background of conditions C. The relations to other brain areas that Nick talks about might in effect be part of conditions C (normal functioning brain). JURAJ: Juraj makes the excellent point that the sort of questions I raise (re lesions etc.) about NCCs also come up for neural correlates of other sorts of mental states and conditions, such as prosopagnosia (and presumably learning, memory, language, etc, etc, etc). In fact they arguably come up throughout biology, engineering, etc, when we are concerned with finding the "area for X", the "mechanism of X", and so on. I don't know if there is much literature on the general problem, but there ought to be. And maybe some conditions, such as prosopagnosia, will raise special problems (for the reasons Juraj suggests), so one will only be able to identify correlates in a coarse-grained way. Maybe the thing to say would be that in these cases, there is just no fact of the matter about what the "real correlate" is -- there's just a complex causal chain. LOGAN: Logan raises the interesting possibility that early visual areas may produce "low-level" qualia while later areas produce "high-level qualia". A normal visual experience has elements of each and so has correlates in both places. Cases such as intermediate blindsight may have the info in the later areas but not the early areas, yielding high-level qualia without the low-level qualia, which would be understandably strange! I guess, if the high-level qualia are "cognitive qualia", we have interpretation (3) (I think) of intermediate blindsight from today, ie. conscious visual judgment but no sensory phenomenology. On the other hand, if the high-level qualia are more like high-level sensory phenomenology, something like the grouping of the pixels, or the "seeing" of the light patterns as a dalmatian, then we'd have something closer to (i) or (ii). Understandably, this could be very strange. PETER: Peter asks, how can IT and not V1 be the NCC, but IT gets input from V1? I guess the answer is that the NCC could come at an intermediate stage (because e.g. it takes processing for info to become conscious), in which case, it's only to be expected that it will get input from areas that aren't themselves part of the NCC. Of course there may well be multiple NCCs, as Peter says. I think this actually answers Peter's other question, of why we only require sufficiency and not necessity of an NCC. If there are multiple NCCs, any one won't be necessary. But sufficiency alone will mean that it's "important" for consciousness in some sense -- activate this system, and we get consciousness! Of course, one still has to understand how and why the NCC correlates with consciousness, but sufficiency might be taken as a sign that we're at least looking in the right ballpark. JOEL P.: Joel suggests that split brain studies give reasons to believe in multiple NCCs. These people might be interpreted as having two streams of consciousness, and two NCCs. Normal people like us presumably have both neural systems, so why don't we have two consciousnesses two? I guess one could get the beginnings of an answer by saying that the difference in normal functioning between ourselves and split brain cases is such that there's no reason to believe that the NCC for one will also be the NCC for the other -- it's just the sort of case where we don't want to draw conclusions about our own NCC from lesion studies! Personally I'm not at all sure that split brain cases involve two streams of consciousness. But let's say they do, and that they have two corresponding NCC's. My own guess would be that nevertheless, we don't have two corresponding NCCs, but rather one corresponding NCC that goes across both hemispheres. Most of the time, the info in the two hemispheres is in synch, so we get a coherent consciousness; if the info were to come radically apart, we might have a weird consciousness, but still within a single consciousness. Basically, the reason would be that streams of consciousness might go with something like loci of integration and integrability. In the split brain cases, the info in the two NCCs is not integrable with each other, so we get two streams. In our case, info is integrable (e.g. across the corpus callosum), so we get one stream. SIMON: Simon makes interesting points both re what Lis says about content (as above) and about lesions. The weaker interpretation of lesion studies will basically be that if lesioning area N affects visual consciousness, then area N will be involved in the causal chain underlying visual consciousness. That seems like a reasonable conclusion. Though of course for all we've said it could be a fairly distant and indirect part of the causal chain, in the way that the heart is! SARAH: Sarah worries about how downstream lesions could change the location of the NCC. I admit my remarks here were a bit brief and cryptic. But here's one way things might work. Let's say that the NCC will be the locus of direct availability for global control, as I suggest, or the mechanism of the global workspace, as Baars suggests. Then it may be that if one lesions downstream from the NCC, this system will no longer be the locus of direct availability for global control (it won't play the control role any more), and it won't be the global workspace (as it won't transmit the info any more). Maybe other areas will take over and fill the void instead, they'll send forward the info that controls behavior, etc. If so, the original system will no longer be the NCC, and other systems will become the NCC. Something like this will happen if we think that what makes the NCC an NCC is in part its functional role, or at least its downstream functional role (whatever that role is). Downstream lesions may kill off that role, and so may kill off the fact that the system is an NCC. Of course, if one doesn't think that downstream functional role is relevant to making the NCC an NCC, this won't follow. ALLEN: Allen suggests, as does Logan, that maybe weak (less than 100%) correlation might be enough to qualify a system as an NCC. My main worry here is that this way, too many brain systems might qualify as NCCs. E.g., the retina state, V1 state, etc, might all fairly reliably correlate with consciousness, and come apart only in special circumstances (e.g. binocular rivalry). It seems that in such cases we want to say there's something special NCC-wise about the system that still correlates as opposed to dissociating. So I think maybe there's a case for requiring full correlation. If a system sometimes dissociates from consciousness, then we should be looking either elsewhere or to a broader system for an NCC. Allen also raises the very interesting point that artificial brain stimulation might make another system M take on the functional role definitive of the NCC, where system N had played it before. In this case, we'll dissociate N from consciousness, suggesting that N isn't really the NCC; but maybe we want to say is that in this case we've actually moved the NCC from N to M, because of the stimulation. Maybe this is another reason for being suspicious of artificial brain stimulation as a criterion, just as we are suspicious of lesions. At least, for being suspicious of too much stimulation. If it's just a little stimulation, making M active in a normal way when it hadn't been a moment ago, it's hard to see how this could make M play a functional role that it couldnn't normally play. What would be required for Allen's scenario is stimulation that makes M active in an abnormal way, thus making it play an abnormal functional role. Maybe we should have some sort of restriction to stimulation of "normal" patterns of activity, if that makes any sense. Allen also raises the worry that if consciousness lags behind NCC activity, the temporal matching methodology may go wrong. That's a good point, and in synch with what Lis says about the temporal methodology being based on assumptions. One question is whether the assumption could itself be tested, e.g. by finding correlations based wholly on non-temporal aspects, and then seeing how the temporal aspects line up. If they don't, then we'll have to be careful! In advance of doing that, presumably any conclusions based only on temporal matching will have to be fairly tentative. LONNIE AND JOEL L.: Lonnie suggests that one route to finding the NCC might involve imaging the brain across closely related conditions of conscious and unconscious perception, separated by a JND or some such. Joel points out that around the "threshold" one is really almost arbitrarily guessing that one saw it or not, so it may not correspond to a deep conscious/unconscious distinction. One potential way to avoid this problem is to give up on the JND idea, and to image relevantly similar episodes of clearly conscious vs. clearly unconscious perception. E.g., that is what Weiskrantz, Zeki, and others have done with blindsight patients such as G.Y. Image him in "blind" seeing, image him in standard seeing, and see what brain areas are different. The result is that a lot of different brain areas show up as different, as Joel predicted. Presumably one reason is that a difference in consciousness will also be associated with a lot of differences "downstream" from consciousness, e.g. in language, planning, thinking, etc, and related areas. Zeki does claim to find some fairly consistent relevance of V5, though. ADAM: Adam worries that the whole NCC approach is premature, since we need to get a good phenomenological method off the ground first. Not just to gather the first-person data, but also to gather the third-person data from neuroscience, etc. My worry here is that the same presumably goes for third-person data in any science at all, e g. physics, chemistry, biology. But if Newton had waited for phenomenology to be properly developed, we would have been waiting a long time! My own view is that a developed phenomenology is vital for gathering first-person data, and may end up helping us gather third-person data, but that at least the third-person part of the story ought to be able to get off the ground without it. It's not clear why the NCC search is any worse off here than any other area of neuroscience, or of science in general. --Dave. From rachaelp@U.Arizona.EDU Thu Sep 30 23:29:25 1999 Return-Path: Received: from Arizona.EDU (Penny.Telcom.arizona.edu [128.196.128.217]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id XAA25995 for ; Thu, 30 Sep 1999 23:29:25 -0700 Received: from DIRECTORY-DAEMON by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) id <01JGLL0OI0DCB8XB9X@Telcom.Arizona.EDU> for chalmers@paradox.soc-sci.arizona.edu; Thu, 30 Sep 1999 23:29:40 MST Received: from paradox.soc-sci.arizona.edu by Telcom.Arizona.EDU (PMDF V5.1-12 #24137) with ESMTP id <01JGLL0MKWPCB8XE8K@Telcom.Arizona.EDU> for chalmers@Arizona.EDU; Thu, 30 Sep 1999 23:29:38 -0700 (MST) Received: from pavo.U.Arizona.EDU (pavo-2.U.Arizona.EDU [128.196.137.196]) by paradox.soc-sci.arizona.edu (8.9.3/8.9.3) with ESMTP id XAA25990 for ; Thu, 30 Sep 1999 23:29:12 -0700 Received: from localhost (rachaelp@localhost) by pavo.U.Arizona.EDU (8.8.6 (PHNE_17190)/8.8.6) with ESMTP id XAA26869; Thu, 30 Sep 1999 23:29:29 -0700 (MST) Date: Thu, 30 Sep 1999 23:29:29 -0700 (MST) From: Rachael J Parkinson Subject: more on NCCs In-reply-to: <19990930231157.11402.qmail@hotmail.com> To: chalmers@U.Arizona.EDU Cc: scicon@paradox.soc-sci.arizona.edu Message-id: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII Status: RO Just one more thing on NCCs... In "What is the Neural Correlate of Consciousness?" Chalmers discusses what sorts of conscious states would be relevant to determining the neural correlate of consciousness. He explores three classes of states that we should consider: that of being conscious (or unconscious), background states of consciousness (including sleep, hypnosis, dreaming, etc.) and contents of consciousness. It is the latter, contents of consciousness, that Chalmers argues may be the most interesting states of consciousness. It is these that I am concerned with in this post. It seems to me that, with the right methodology and experimentation, we may come to discover neural correlates of vision, for example. Certainly the work of Logothetis and others points to something positive in that direction. Suppose then, that we were to discover a correlate between particular 'horizontal' neurons and first-person 'horizontal' experience. Though I agree that this would be a significant step in developing a Chalmersian science of consciousness, I would be reluctant to identify such a finding as the 'neural correlate of consciousness.' This is because it is inadequate to the data of our phenomenal experience which rarely, if ever, consists of the single experience of a horizontal line. Our phenomenal experience is rich and diverse and it somehow comes together in our unified subjective experience. It seems to me that a neural correlate of consciousness then, must at least point to what unifies our phenomenal experience. If it does not, then it seems that what we are looking at is merely the 'neural correlate of vision' or the 'neural correlate of audition.' Note that I am not advocating any sort of a priori claims about a Cartesian theater. It may well be that the unity of consciousness can not be attributed to any particular set of neurons or brain processes. But it seems that what we should be looking for in our search for the neural correlate of consciousness is something that, though not offering a reductive explanation, can account for, or point to, that thing which correlates with our *unified* phenomenal, conscious experience. What do others think on this matter? Best, Rachael