Sunday, 1 July 2012

Dan Dennett: A Phenomenal Confusion About Access and Consciousness

    Abstract: Many researchers on consciousness have adopted Ned Block's purported distinction between "access" consciousness and 'phenomenal' consciousness (Block, 1995, 2005, 2007), but in spite of its evident appeal, it is not a defensible distinction. Earlier critiques (Dennett, 1994, 1995, Cohen and Dennett, 2012) have not deterred those who favor the distinction, but perhaps one more exposition of the problems will break through.

Block, N. (2005) Two neural correlates of consciousness. Trends Cogn. Sci. 9, 46–52 7
    Block, N. (1995) On a confusion about the function of consciousness. Behav. Brain Sci. 18, 227–287
    Block, N. (2007) Consciousness, accessibility, and the mesh between psychology and neuroscience. Behav. Brain Sci. 30, 481–499
    Cohen and Dennett  2012, "Consciousness cannot be separated from function," Michael A. Cohen and Daniel C. Dennett, Trends in Cognitive Sciences, August 2011, Vol. 15, No. 8, pp. 358-64.
    Dennett, 1994, 'Get Real,' Philosophical Topics, special double issue on Dennett's philosophy.
    Dennett, 1995, 'The Path Not Taken,' commentary on Block 1995, Behavioral and Brain Sciences

Comments invited


  1. Dan asks: "What is phenomenality?"
    Answer: Feeling. We all know what it feels like to feel.

    Dan is right that there are not 2 consciousnesses, "access consciousness" and "phenomenal consciousness".

    But there is certainly one. And it is the (cartesian, indubitable) fact that we feel (when and what we feel when we are feeling), and unconscious entities lack that, we are here to discuss.

    It is the causal role of *that* that this Turing Institute is about.

    Dan is talking about ontology: what is the (felt) vision of the red stripe?

    But the hard problem is explaining how/why, not ontics.

    1. Stevan rightly says that the hard problem is that of explaining how/why there is phenomenal consciousness (or "feelings" - if we prefer).

      However, asking this question in such terms already presupposes some sort of ontological stance, namely dualism or at the very best some sort of conservative reductionism. The hard question - as Chalmers puts it - is how/why we have (specific sorts of) phenomenal consciousness on top of "mere" neural/physical/bodily activity.

      I acknowledge that there are very strong reasons (based on introspection) to hold that there is something it is like to "feel" the redness of the flag stripe - regardless of whether that is an illusion or not. To me, it feels red. What makes the character of any conscious experience (such as "feeling" the redness of the stripe) peculiar, is its subjective character, the fact that it feels something to me.

      And that - phenomenal experience - is something that could - at least logically - be absent (in a zombie world) without there being any functional difference in how that world would work.

      As Stevan says, how and why conscious experience "is there" is what needs an explanation.

      Now, the problem is: what kind of explanation are we looking for? What kind of reply would satisfy us? Do we need empirical studies showing us "how" and "why" we feel redness - what causes us to feel it? Or is it sufficient to give a developmental/evolutionary answer such as: conscious experience is there (it has been preserved in natural selection) because it plays a causal role that an unconscious system could not achieve in an equally efficient (economic) way?

      If the latter is the way to go, of course, we would have to focus on understanding what that causal role is, and we would have to focus on an even harder question, namely that of explaining how conscious experience (the subjective "feeling") could ever be causal - since by definition we ontologically separate it (in some way or another) from the physical domain. Are we ready to drop the causal closure of the physical domain just to fit in conscious experience?

      There is still quite a lot of work for philosophers here, and I think that the ontological question should always remain in the background if we do not want to fall into a regression...

      Pietro Snider (Université de Lausanne, Switzerland)

    2. Your points are all valid -- except the problem is not, I think, ontic, but epistemic. It is *explaining" the causal function of feeling that we seem to be unable to do (on the overwhelmingly reasonable assumption that psychokinetic dualism is false).

    3. I liked this distinction between "access consciousness" and "phenomenal consciousness". I agree with you, that are not two types of consciousness, but only one with two stages. It is possible to not arrive as human in the second stage.
      But I can't say, that in the second stage, we are not able to feel, but the feeling experience is basic. This is the degree difference between the two stages.
      The example that I have is the case of autism. Of course, it is depending what autistic level. But I wonder if it is not possible to say that in some autistic cases, we have only "access consciousness" and not phenomenal.
      I need to understand that I am thinking correctly.
      In some high-functioning autistic people, I will have also a phenomenal consciousness.


      "Maybe Access-consciousness should be name "attention" instead of consciousness and "consciousness" could be kept for feeling and experiencing ?"

      "The problem is not what you call it; the problem is that access is not conscious unless it's felt. Ditto for attention (which, if unfelt, is simply information detection or selection)."

    5. Why can't we legitimately say that:

      a. Feeling is subjectivity.
      b. Subjectivity is realized by a particular system of neuronal mechanisms in the brain.
      c. The brain mechanisms that constitute subjectivity are a relatively recent evolutionary adaptation.

      This theoretical stance regarding consciousness/feeling/subjectivity is presented in the following publications:

  2. From the standpoint of a basic neurobiologist currently working in the retina - I am finding it difficult to process Dan's supposition that a concrete, material flag image falls into a different realm than the after-image of the red stripe.
    The afterimage is well understood to be compensation by photoreceptors after colour bleaching. Following that this cone-produced signal activates retinal ganglion cells in the same way as the image originating from the external world - and in subsequent processing our visual cortex is entirely unable to make the distinction between a real image and an after-image, then how are these cognitively appraised in a different way? Dan seemed to indicate that it was the 'immateriality' of the red-stripe after-image which negated its value as a conscious experience - BUT, if the two are the same at the level of the retina and upwards, where does the difference lie?
    Does Dan suppose that it is the access to the knowledge that the image does not actually exist which creates a dissociation between the two?
    Any clarification here?

    Adele (facebook-less) Tufford, McGill

  3. I think your point is valid, so I cannot answer for Dan. The only difference between the feeling of seeing red when photons from a red surface hit my retina and seeing red after my cones have been bleached by photons from a green surface is the input and neural substrate. What does that tell us about feeling one way or the other?

  4. Dan's position can be summarized in one sentence: Once you've explained how and why organisms can do everything they can do (as Turing urged), there's nothing left to be explained.

    The trouble is that feeling is not something we do, it's something we feel. And even when all's said and done about doing, it remains to be explained how and why we feel.

    1. Prof. Dennett argues further that Phenomenal Consciousness, what we perceive as experience and causation of consciousness as a whole, is in fact just a property of it. Dennett argues that experience lies in the ability to "access": he says that our world and conscious experience are both founded upon affordances, e.g. a cup affording holding or drinking (as Gibson would say).

      I tend to agree with him here. If we create, perhaps by way of evolutionary robotics, an artificial mind of the type in which we are here interested, it won't be that we finally understood the basis of phenomenal experience and proceeded to implement it. When we finally deem some artificial mind intelligent, with phenomenal capacities, it will be that it has crossed some subjective threshold beyond which we cannot any longer differentiate man from machine. Phenomenal consciousness will have "appeared" to us at this point, after having been engineered in every capability that allows it to perform something we perceive as innately human.

      These will have been our expectations being displayed in our creation. At this point, our AI will be able to encode new information, analyze it, access it later. It will have motivations and analogs of our mental/emotional states, and it will use words like "feel" to describe them to us. Once the robot provides output that indicates presence of phenomenality, we can't hope to discredit it, and we still won't really know the true origin of the subjective experience we refer to as "feeling" or "consciousness".

    2. Stevan: "The trouble is that feeling is not something we do, it's something we feel."

      If you agree that the brain causes conscious experience, why wouldn't you agree that feeling is what the brain does?

  5. Unless it is felt, "awareness of," "consciousness of" etc. all just mean access to some information. And information is just data, unless it is felt.

    What (IMHO) Dan should have said about "access consciousness" vs. "phenomenal consciousness" was that there is no "access consciousness": there is only access to information. That information can either be felt or unfelt. If unfelt, then it's just mindless information-processing (i.e., doing). If it is felt, then we have the only consciousness there is, which is feeling (so why bother with the long and redundant synonyms "phenomenological consciousness"?)

    But of course that's not what Dan said. Rather, what he said was that there is no phenomenological consciousness! Rather, what we really mean by consciousness (and feeling) is just access (i.e., just doing, no feeling -- or rather what we mean by feeling is just certain doings -- the "accessible" ones we can talk about...)

    1. The question is, however: Does the value of emotion in evolutionary terms depend on emotions’ being felt? In other words, perhaps it is the very fact that emotions feel like something which yields the functional, adaptive value of emotions. Perhaps emotion is the limit-case at which feeling and function coincide. What I am suggesting is that for representations of internal, homeostatic states, maybe access really just amounts to feeling. Perhaps feeling is the organism’s means of access to its own internal state.

    2. Stevan, I like the idea that there is only access to information, and that information can either be felt (ie. what we call conscious) or "mindless information-processing" (maybe "sensing" would be a better term?) I'm convinced. Any ideas as to how Dennett would rebuke this idea?

      According to this view, consciousness is a process for us to access information. Now we need to speculate on why this particular process has been advantageous.

    3. Perhaps it is advantageous to feel, because such feeling-states are precisely the most adaptive way to represent internal homeostatic states?

    4. Since access to information can be either felt or unfelt, the question is how and why felt access is advantageous.

    5. Maybe the question "how and why felt access is advantageous" has to be taken in terms of timing : mainly, when are we feeling and when are we doing? are we feeling and doing at the same time, or are we feeling and then doing, or are we doing and then feeling? when does the doing need the feeling to be done? to which extend the doing needs the feeling? why do we do things without feeling and why do we do things that cannot be done without feelings? I know, there's the argument that in theory, everything we do could be done without consciousness, but let's have a look in a species context. As humans for instance, if we don't feel the pain because we are anesthetized, if we are poked by a needle, we won't feel anything so we won't do anything. So, in that case, the doing need the feeling. So if the doing need the feeling for some specific kind of doing, the feeling has to have its own purpose.
      To understand this purpose we have to compare what is felt and what is unfelt in relation to what is done and what isn't done in different situations. This can be done by looking at "abnormal" (or altered) states of consciousness compare to a "normal" state of consciousness.

    6. Good questions, Pauline, but no answers. (And the right comparison is not an anaesthetized and unresponsive human but an unfeeling yet responsive robot.)

  6. Thus, we can say that the distinction between many consciousness is did by the feelings, but finaly all creatures (or almost, human persons at least, even with disabilities) can have this consciousness which is understood the access to information?
    But this is also a little complicate, because there is people who don't have access to information, very low mental impairments?
    Thus? How we can relate "acces information" to the concept of consciousness?
    Camelia Dascalu (Paris 3 University)

    1. Yeah, this is a really good point! What about people with mental disabilities that do not allow them to access lots of information . . . do you deny that they feel things? This is a difficulty for Dennett I think because he seems to deny that there can be feeling (phenomenal consciousness in his terms) without access to information (or access consciousness). I guess you could argue that people with low mental capabilities are still getting some information from the external world and therefore are feeling as a result of the information that they are getting. Or it could be as Maxwell said earlier - that feeling is the organisms means of access to its own internal state (but is your internal state independent of the external world?).

  7. "I acknowledge that there are very strong reasons (based on introspection) to hold that there is something it is like to "feel" the redness of the flag stripe - regardless of whether that is an illusion or not."

    I don't get the qualia language. Why refer to "what it is like to feel something is red?" Instead of just asking what is the color red (the sensation not the wavelength of light) and what causes it? Perhaps this language is only intended to clarify that one is referring to the sensation of color or sound, and not photons and physical sound waves. But how could the color red be an illusion? Even if it were in any sense an illusion, this would be beside the point. I know the color red exists because I see it. The question is what exactly is and is causing the sensation of color , whether or not it is an "illusion."

    1. Good remark, I find it very difficult to talk about illusion/or not, feeling/or not when we "feel" or "see" an epiphenomenon. After all, all that we have access through consciousness is related to our body that is sensible to other bodies. In my eyes, that kind of "epi red stripes" are red because some way the red nervous configuration have been activated. By the way, for me Dennett have made his point convincing, it's clear that the distinction p/a is unclear.


      The problem is not whether the feeling of seeing red is real or an illusion.

      Of course the feeling is real, if we are feeling it.

      Nor is the problem whether there is really a red object out there, when we are seeing a red object.

      The problem is the causal role of the fact that the red is felt, rather than merely detected and acted upon (doing).

    3. How do reality, illusion and causation apply to various sleep states? What is happening when we dream of the color red? Are we feeling?

  8. It's not entirely clear to me how Professor Dennett's argument against Ned Block's distinction between P-consciousness and A-consciousness is supposed to go. How does his talk about our propensity to get causal links backwards (e.g., our propensity to think that we like sweet food because it is intrinsically sweet) helps him to undermine Block's distinction? Here's how I understand it (tell me if I am wrong): 1- Just like we are sometimes lead to postulate that some external objects have intrinsic properties that are not in fact intrinsic (e.g., sweetness, cuteness), we are lead to postulate that some internal objects have intrinsic properties (e.g., phenomenality) that are not in fact intrinsic. 2- So, the idea of phenomenality (or phenomenal consciousness) as an intrinsic property of some mental objects is confused. 3- The only way to maintain a workable notion of 'consciousness' in cognitive science is to go back to Block's notion of A-consciousness, which is the only one we can have. If my interpretation is correct (and I'm not sure it is), then, why should we accept premise (1)? Dennett has given us no ground to believe that the analogy is valid. (Originally posted on FB.)


      Access to information (data) is access.

      Consciousness of information is consciousness.

      We are conscious of information only if and when it feels like something to access it.

      Block distinction between "access consciousness" and "phenomenal (i.e., felt) consciousness is incoherent.

      Access can be felt or unfelt.

      If access is not felt, it is no kind of consciousness.


    "Dan's Darwinian explanation is about *doing* what it takes to survive and reproduce: it doesn't explain why it *feels* like something to do those things."

      "In the context of an evolutionary explanation, a sort of standard response here would be to contend that we do not need an explanation of why doing those things feels like something, unless it can be demonstrated that feeling has an adaptive function. Of course, one could argue against this idea by pointing out that there IS a feeling associated to doing which needs to be explained, but then we'd encounter the "consciousness is a spandrel" argument.

      However, Damasio's view that feeling adds a layer of homeostatic control could provide some solid grounds on which to elaborate an account of the adaptive function of consciousness."

      "Also, it could be that organisms that need to be able to attribute value to various and variable outcomes in a complex environment and so on, need a sort of "feeling" capacity so as to make these valuation procedures efficient given task and cost constraints."

      ""Valuation" is doing too much hermeneutic work for you. Value is not value if not felt; it's just functional, or formal. And the hard problem is: why is it all not just formal/functional?"

      "The question is, however: Does the value of emotion in evolutionary terms depend on emotions’ being felt? In other words, perhaps it is the very fact that emotions feel like something which yields the functional, adaptive value of emotions. Perhaps emotion is the limit-case at which feeling and function coincide."

  10. The adaptive (hence causal) value of emotions is in what they dispose us to do or not do. The adative (hence causal) value of the fact that they are felt is another matter (and a hard one...)

    1. Why can't the adaptive (hence causal) value of the fact that emotions are felt be that they motivate/dispose us to do certain things?

    2. Because disposing to do is doing. (And motivation is a weasel word: felt motivation or just unfelt homeostasis, as in a thermostat?)

  11. I was wondering about the strange inversion argument Dennett presented.
    If I understood correctly, the point is about the wrong causal attributions we make about things of value in the world.
    Applied to consciousness, the point would be that causally attributing our doings to our feelings is another such error? Is there another strange inversion there: "We feel because we do" and not "we do because we feel"? Or... is there simply no causal relationship there. They are independent processes only one of which we can explain: doing.

    There are things in the world that are relevant to us and our survival - e.g. food, mates,shelter.

    But our expectations about these things also matter to us, what kind of value we assign to the multitudes of things we sense in our environment is extremely important to mobilize our resources to get said things.

    1. Whether we do because we feel or we feel because we do, the causality has to be cashed out, because, on the face of it, doing seems enough, biologically and functionally.

      In other words. neither of the "because's" is a causal explanation.

    2. Stevan: "Whether we do because we feel or we feel because we do, the causality has to be cashed out, because, on the face of it, doing seems enough, biologically and functionally."

      The retinoid model of consciousness "cashes out" the causality of feeling. Stevan, do you know of a competing brain model that cashes out the causality of feeling? Do you have a principled objection to the retinoid model of feeling?

  12. Daniel Dennett mentioned people's reactions to Cog's humanoid actions. Here's an intro to some interesting work in that area:

  13. What's striking about all these comments - and about the philosophy of consciousness overall, in my view, - is that no one yet seems to be able to state clearly and uncontroversially what is MEANT by the idea of consciousness ("access", "phenomenal" or whatever). Which is perhaps why people like Block can so plausibly split it into two: if the basic idea remains vague and ill-defined, how can one argue convincingly against such a step? Indeed, what's to stop someone from saying that consciousness has three or four parts - or more? Arguments against would be just as difficult.

    1. What's vague or ill-defined about stating the obvious? "Consciousness" means feeling. If an organism feels, it is conscious. If it does not feel, it is not conscious. And feeling means any state it feels like something to be in, whether it's seeing something, hearing something, touching something, smelling something, doing something, wanting something, believing something, knowing something: All mental states are felt states. Access is just conscious if it is felt access; otherwise it's just access, such as a computer's access to data.

  14. Hi Steve

    Consciousness is anything but obvious. It is, in my view, one of the most – if not the most – elusive of ideas to define. Simply calling it feeling does not get us anywhere, I’m afraid. It’s perfectly true (presumably) that if one does not feel one is not conscious, but equally, one does not run, or add up, or believe one is a great philosopher, or a million other things, if one is not conscious.

    And to say that “feeling means any state it feels like something to be in” sounds like an appeal to the rickety, old Nagel formula. Do you really want to go there? Isn’t the Nagel thing dead and buried? It should be.


  15. Hard problem of consciousness solved by bridging the explanatory gap