The sparsity of phenomenal consciousness, or of cognition, or both

Ned Block gave a Google talk (embedded below) that was ostensibly supposed to be about why AI approaches to cognition won’t work.  However, while he does address this topic briefly, it’s toward the end and he admits he hasn’t really justified it, beyond a vague proposition that while access consciousness involves information processing, maybe phenomenal consciousness doesn’t, or is more dependent on its biological substrate, or something.

But most of the talk is devoted to a discussion about whether phenomenal consciousness is sparse or rich.  Sparseness refers to the fact that if you look at a detailed picture or diagram, then after it has been removed from view, you are asked to write down what you remember from it, your recollections are going to be sparse.

We have an impression that the initial sensory experience was rich, that we took in all the detail, and that it’s only our memory of it that is sparse.  However, many researchers think that the initial impression or richness is an illusion, that the initial conscious perception is far sparser than we think it is.

We can take in a complex scene and quickly get a gist of it, make a prediction based on sparse information.  Which, as the history of witness trials show, can often be wrong.  We see what we expect to see.  We might then focus on one or two details in the scene, and then come away with the impression that we experienced the entire scene, when in reality we only experienced the broad outline of it followed by the details we focused on.

Block disagrees.  He’s the philosopher who originally made the distinction between phenomenal consciousness and access consciousness.  Phenomenal consciousness is raw experience, the ineffable what-it’s-like aspect of consciousness.  Access consciousness is our ability to access this raw experience, to use it in reasoning, and to make self reports about it.

It’s worth noting that not everyone is convinced that this is a meaningful distinction.  Philosophers like Daniel Dennett think it’s a mistaken distinction, that there is no phenomenal consciousness without access consciousness.  I tend to agree with this, although from a purely subjective point of view the distinction still seems useful.

There are aspects of consciousness that seem more primal, ineffable, impossible to describe, such as the redness of red or the painfulness of pain.  I personally don’t have an issue with giving a name to this category of experience, while realizing that without access consciousness, we couldn’t even have a concept of phenomenal consciousness.

Anyway, Block’s contention is that phenomenal consciousness actually is rich in detail, but that this richness is lost in access consciousness, that it’s cognition which is sparse.  He provides ostensible evidence in the talk, which I personally didn’t find convincing.

For example, he cites eye tracking studies, asserting that eyes focusing on something indicates  conscious awareness of what the eye is pointed toward, but this doesn’t square with what I’ve read, where a lot of eye movement is driven by the superior colliculus below the level of consciousness.  And it’s worth noting that we only have high resolution reception in a small part of our retina.  Our impression of a scene is built on the eye moving around, movements of which we’re generally not aware.

Block notes that a lot of his opponents do think there is richness in unconscious perceptions, but that what makes it into consciousness is sparse.  He thinks the richness does make it to consciousness, but not in a way that we can remember or self report.

To a large degree, this seems to get into which definition of consciousness we’re using.  His opponents appear to be using a definition of what is accessible by introspection, which in humans feels like the default definition.  Block seems to be saying that there is conscious content that we don’t access, which seems a bit strange.  On the other hand, we often have perceptions that we don’t remember.  The question is what we can say about these ephemeral experiences if we can’t introspectively capture them.

This is a scenario where the concept of consciousness may be inviting people to argue about nothing.  If we make the divide between perceptions that are available for self report versus perceptions that aren’t, both parties might be in agreement.  Arguing about whether we are conscious of things we can’t introspect seems like the proverbial dilemma of whether the refrigerator light stays on when the door is closed.

Block’s talk is about an hour long.  I put it at the bottom to make it optional for this post.  Block comes across as very personable, someone I’d enjoy having a conversation with, but I found the talk itself to be a bit scattershot.  Still, despite disagreeing with his positions, it was interesting.  You might find it worth the time, and I know some of you will agree with him.

Block’s citation of his opponents also reminded me that Stanislas Dehaene, a proponent of global workspace theory, has a book on consciousness I need to read.

h/t Gregg Caruso

31 thoughts on “The sparsity of phenomenal consciousness, or of cognition, or both

  1. I’ll watch the video later (maybe during a ballgame — they’re good for that; I often watch with the sound off while doing something else).

    “Daniel Dennett think it’s a mistaken distinction, that there is no phenomenal consciousness without access consciousness.”

    So do animals lack PC? Or is it agreed they have some form of AC? (I am not going to type “phenomenal consciousness” and “access consciousness” over and over!)

    To me, that AC is the only way we can acknowledge our PC doesn’t in any way damage the putative reality of PC.

    “Anyway, Block’s contention is that phenomenal consciousness actually is rich in detail, but that this richness is lost in access consciousness, that it’s cognition which is sparse.”

    I would very much agree. Memory is sparse, and isn’t that where AC has to come from? When I report an experience (even to myself), aren’t I recalling that phenomenal experience?

    You know why phone numbers are seven digits, right? 🙂

    “The question is what we can say about these ephemeral experiences if we can’t introspectively capture them.”

    They very likely are part of the mix of our PC. Its content forms from everything that impinges on it. Those off-the-wall intuitions that turn out to be right. (Some studies suggest our first impressions are surprisingly accurate.) To the extent our consciousness is (mathematically) chaotic, even the tiniest of inputs can matter.

    Our PC is a combination of sight (hugely), sound (muchly), smell, touch, and taste, those last three applying to various degrees depending. (A lot of food “flavor” is actually smell, for instance.) That makes for a pretty rich blend of inputs to be processed.

    I think our PC also contains the “state of the (consciousness) system” as a result of recent history. For example, while driving, my PC is maintaining a model of what’s going on right around me. That model seems shared between a PC-like “in the moment” feeling as well as in my ability to access parts of it (e.g. think about the car behind me, glance in the rear-view to see if it still roughly matches the model, because some part of the PC “timed out” and suggested a refresh of that part of the model).

    So I think our PC is very rich, indeed. But our AC does kinda suck.

    Liked by 2 people

    1. “So do animals lack PC? Or is it agreed they have some form of AC?”

      As I understand it, animals do have AC, although for most of them it doesn’t include metacognition. Block considers any use of the contents of consciousness for reasoning to be AC.

      ” For example, while driving, my PC is maintaining a model of what’s going on right around me.”

      I think Block would say you’re lumping AC in with your PC here. But I think this is Dennett’s point, that the distinction is far blurrier than it might initially appear.

      I think you’d enjoy the talk. Block’s understanding of consciousness is far more robust than the deflated version I or most illusionists hold, and I suspect he’d agree with your intuition about it being irreducible.


      1. “I think Block would say you’re lumping AC in with your PC here.”

        Sure, or (as I said) shared. No reason it has to be the exclusive property of one or the other.

        “But I think this is Dennett’s point, that the distinction is far blurrier than it might initially appear.”

        I’ve never been one to think the mind is highly compartmented. As you mention, I see it more as an holistic mechanism.

        Liked by 1 person

    2. “To me, that AC is the only way we can acknowledge our PC doesn’t in any way damage the putative reality of PC.”

      Hear hear!

      As for sparsity, we all agree that AC is more sparse than the information that gets unconsciously processed. I’d favor the hypothesis that (there is a PC level and) the information loss occurs at both transitions, unconscious to PC and PC to AC.

      Liked by 1 person

  2. A brilliant book o this topic is The User Illusion. The author compares how many bits of information are brought in by the senses and them points out there is way too much to solve it all, so we parse it and actually dump a great deal of raw (rich?) information. This partially explains why we keep a simulacrum of our surroundings in memory, that is rather crudely constructed, occasionally updated, and not very accurate. It is because we do not have the capacity to store all of that extra info. It may well explain why our eyes see well only in a narrow cone and poorly outside of that (again, couldn’t retain the info if we had it).


    1. That book sounds interesting. It also gets into why we have attention, the need to prioritize what we process and react to.

      Does the author take a position on whether the initial load of information is conscious or unconscious?


  3. I can see how some might want to eliminate the distinction between the two varieties of experience (which I think are better characterized as pre-reflective and reflective consciousness).
    Reflective consciousness is just a kind of metacognition. But the distinction helps when we start to talk about how metacognition and systematic representation work.
    That way we don’t have to keep saying meta-meta-cognition – though as I directly experience the sound of “meta-meta-cognition” I must, on other grounds, reconsider my aversion to the term, and I now think that maybe we ought to be saying it after all.

    Liked by 1 person

    1. I probably didn’t do Block’s conception of access consciousness justice in the post. He considers any use of the contents of phenomenal consciousness for reasoning or thought to be access consciousness, so it’s broader than just metacognition or self reflection.

      One of the reasons I’m dubious about the distinction (while admitting it does exist subjectively) is that separating phenomenal experience from what we do with it seems artificial. On the other hand, we can conceptually separate, say, our voice from spoken language, and that distinction might be useful for some purposes.


  4. As far as I can tell, he is just rehashing the language of phenomenology.
    It does seem to confuse things in a couple of important ways.
    First, if the phenomena are being accessed – accessed by what and on what basis? I sense a homunculus lurking nearby.
    Second, as in reflective consciousness, it leaves the impression that our direct experience is unconditioned – non-theoretical, if you will – and that is almost certainly not the case (witness the activities of the colliculus).


    1. I suspect Block would deny he’s positing a homunculus, but with consciousness discussions, I often find people posit things implicitly, often unwittingly so.

      “it leaves the impression that our direct experience is unconditioned – non-theoretical, if you will – and that is almost certainly not the case”

      I agree completely. There are a number of philosophers who try real hard to have pre-theoretical observation of the world, but our perceptual systems just don’t work that way. Perception is prediction, and prediction can only happen when we have mental concepts. Of course, most of those concepts have to be built, but they’re built on top of a core of innate ones we’re born with. Even babies have theories of how the world works.

      Liked by 1 person

  5. Mike, do you think small children (less than 3) perceive color? That was the most interesting thing I learned from the talk. Small children don’t conceptualize color. They don’t seem to be able to make decisions on whether two colors are the same or different even when they can make such decisions based on shape.

    And what do you think is going on in the 3×4 grid experiment?
    [if you didn’t watch the video: they show a grid of letters, 3 rows of 4 letters, and then take it away, with 4 variations in what happens next:
    1. They ask which letters they remember: subject usually gets 3 to 4 (out of 12)
    2. They play a high pitch tone, which means the subject should try to remember the top row. Result: usually 3 to 4 (out of the four)
    3. They play a medium pitch tone, which means recall the middle row. Result: 3 to 4
    4. Low pitch tone means lower row. Result: 3 to 4]

    This suggests that the entire grid was experienced, so, rich experience, even though you can only recall 3 to 4 letters. Otherwise you would have to ask – what if there was no question afterwards? Was the grid perceived/experienced?

    [and again I assume you noticed the thalamus reference?]


    1. James,
      The stuff on children and color was interesting. And I can see how Block might interpret that to validate his ideas. But to me it only demonstrates that young brains have cognitive limitations that they eventually mature out of. The fact that they can discern and act on shapes tells me that they have both PC and AC of it, it’s just that their systems have limited metacognitive abilities.

      My take on the grid experiment is that we quickly get the gist of it when we first see the grid, and then maybe we look at some of the numbers, then take ourselves to have absorbed the whole thing. But if the contention is that we are conscious of every individual letter at some point, then just forget it, I can’t see that it’s demonstrated. The tone thing just focuses people’s attention on a particular row before it’s removed. Or am I missing something?

      I did catch the thalamus reference. As you know, I think it’s a critical structure, the switchboard of the cortex, but whether it by itself is conscious is another matter, and depends on which definition of consciousness we’re using.


      1. You did miss something. The tone happens after the image is removed.

        What I want to know is how much eye movement happens while the image is still available. I assume it is enough to cover the whole image.

        That was another thing about the change blindness. Especially in the pictures with people, our eyes have a fairly restricted pattern of movement, but I wonder if it has to do with the whole conceptualization thing. Do our eyes move to the objects we recognize as distinct and mostly don’t notice the changes in things we see as background? Can we only track what we have “conceptualized” from flash to flash? Can we only “access” what we have first “conceptualized”.



        1. You’re right, I did miss that. Thanks! Digging around for more info on the experiment, I found a Wikipedia article on “Iconic memory”:

          Iconic memory appears to be a full visual memory of a perception, that last less than a second. If we equate iconic memory with phenomenal consciousness, then I can see where Block’s coming from. Interesting. I’ll have to give this some thought.

          I’m still not sure if this proves his thesis. Since this has been around so long (1960), I’d like to see what the sparseness proponents have to say about it. But it is interesting.

          On the change blindness, since iconic memory fades quickly, I wonder if the blackout period, if it’s short enough, returns our ability to detect changes. Obviously if there’s no blackout period, we can.


          1. Ha! I think I was getting a little cozy in Block’s camp, but that little tidbit threw me straight out. Thanks for doing the homework.

            If there’s an iconic memory that lasts less than a second, that suggests, to me, that the only things that become conscious are those which get “conceptualized”. That might be my new slogan: “No Experience Without Conceptualization!” Maybe it’s all access consciousness, and phenomenal consciousness is just a subset of that.

            That would mean those small children are not conscious of color, even though they can see and distinguish color.

            Actually, it may be that we’re not conscious of color until we conceptualize it. That was the point of the slow change demonstration. I actually figured out what was changing because I knew something was changing (he said so) and color seemed like a likely candidate. But even then, at the end I was only 60-75% confidant, and I could not have told you what the beginning color was.

            [fortunately, this all fits with my paradigm, so I’m good]


          2. Have to admit it had me wondering if Block was right. But I think you’re right. Iconic memory strikes me as basically the neural image map in early sensory areas. It’s probably best thought of as being available for conscious processing, but not part of it.

            What drew me back was considering the effect of daydreaming while having my eyes pointed at a particular scene. The light from that scene impinges on my retina and creates patterns in the early sensory areas. But I can be imagining something completely different, that is, not be conscious of what’s in front of me.

            But then something in that scene triggers a reflex that bottom up focuses my attention on it. Now I’m imagining what I’m seeing, I’m actively running predictions on the sensory data coming in. I can intuitively be thought of as being conscious of it. In the language you’re using, I’m conceptualizing it.


          3. “Iconic memory strikes me as basically the neural image map in early sensory areas. It’s probably best thought of as being available for conscious processing, but not part of it.”

            That seems unfair, to use “available for conscious processing” to dismiss iconic memory from phenomenal consciousness. The whole point of positing phenomenal consciousness is to explain things that are available for (further, linguistically expressible) conscious processing.


          4. So are you saying everything coming into iconic memory should be considered part of phenomenal consciousness, even when we’re not paying attention to it? Would you consider subliminal messages part of phenomenal consciousness? That seems like something that would flash in iconic memory, be registered unconsciously, but for which we don’t have introspective access.


          5. Maybe I misunderstood what “iconic memory” is, but I doubt that subliminal messages are even briefly available to Access Consciousness. I doubt you could play musical tones (or whatever) after the flash and alert the person so that they can recognize the subliminal message.


          6. I have to admit that currently my only source of information on iconic memory are Wikipedia articles. Searching several of my books, the concept doesn’t come up. But in those articles, visual memory appear to be split between iconic (maintained for fractions of a second), visual short term memory, and long term memory. The article on VSTM describes it as pre-theoretical, or theory neutral, but it seems like it would have to be less theory neutral than iconic memory (otherwise what’s the distinction?).

            The iconic memory article’s description of Sperling’s cued experiments discuss that the effect diminishes rapidly. By 1000 ms (one second), performance is back to the uncued versions.

            All of which is to say, it’d be interesting to see if anyone has tested subliminal message recall with cueing. I suspect it would depend on what is presented immediately after the subliminal message. If a detail picture, then the message is probably obliterated. If blankness, the person might have a stronger chance of retrieving what was there from the after image.


  6. Where you fall on interpretation of psychedelic experience? Here’s an article that might provide an introduction to the topic.

    The issue seems to be that psychedelics seem to provide a richer and deeper experience than normal consciousness. More pixels is how you have analogized it. The problem is that for fMRI scans there seems to be less brain activity going on, rather than more compared to normal consciousness. Some have argued this indicates consciousness is more than neurons. To quote the article:

    “As predicted, profound changes in consciousness were observed after psilocybin, but surprisingly, only decreases in cerebral blood flow and BOLD signal were seen, and these were maximal in hub regions, such as the thalamus and anterior and posterior cingulate cortex (ACC and PCC). Decreased activity in the ACC/medial prefrontal cortex (mPFC) was a consistent finding and the magnitude of this decrease predicted the intensity of the subjective effects. “


    1. I haven’t done a lot of reading in this area. My concern about any drug induced altered states of consciousness is that our ability to assess those states is compromised while we’re in them. People have tried to capture their thoughts while in them and the result is usually far less significant than they remember.

      But from what I understand, psychedelics compromise the brain’s ability to quickly categorize what’s being seen. It makes us like a newborn, like we’re seeing things for the first time, which causes us to notice things we might not have noticed in a long time. I read something a while back that indicated that babies may be in a perpetual psychedelic state by our standards, because they’re still learning associations and categories, so their ability to quickly predict what they’re seeing doesn’t exist yet. There was also a TED talk by a neuroscientist describing a stroke she had, where the effect seemed similar.


      1. Our ability to assess our own states is always somewhat compromised. With psychedelics maybe in a different way than normal consciousness.

        Sometimes it seems to me that psychedelics cause the brain to run away with what is being seen – overextending patterns, colors, and lines – sort of like when you imagine faces in random cloud formations but on steroids when the dose the high enough.

        The sense of time slowed down which is common does align somewhat with reduced brain activity if the internal clock is calibrated by brain activity to some extent. It could be the over extension is a product of dropped inputs, which might also occur if parts of the brain slow down with respect to other parts.The brain elaborates patterns and connections to fill in the gaps in the dropped input. This might also partly explain hallucinations during sensory deprivation and dreaming.

        If we take the experience at face value, wouldn’t it either be enhanced PC or AC getting greater access to PC?


          1. Yes. I agreed. However, doesn’t that lead to some interesting thoughts about interpretation of fMRI scans and blood flow?

            It would mean that a lot of what we associating with consciousness on scans is actually suppression of information, not coalescing or analysis of it.


        1. “Our ability to assess our own states is always somewhat compromised.”

          Definitely. That’s probably the most important lesson when learning about how the mind works. Introspection is unreliable. When it’s our only source of information, we should be cautious.

          “If we take the experience at face value, wouldn’t it either be enhanced PC or AC getting greater access to PC?”

          At face value, sure. The question is which.


  7. Really interesting discussion here, Mike. And a great post as always. The discussion just above of normal consciousness involving suppression was particularly interesting to me. Particularly when it is combined with the idea that without conceptualization there can be no experience. I wish I had more to say about why exactly–it just tickled me in a sense. Which suggests it resonates with ideas I carry around, I suppose.

    I think it is that I am one who endorses the notion that concepts shape our perceptions and experiences to a greater degree than we would care to admit. A concept is not the thing perceived, of course. It is one of those mental models. And as I happen to think that our conceptual underpinnings act like a filter on our encounters with whatever it is we are actually encountering, this idea of suppression makes some sense to me. I do think there is a greater data set available in PC that we seldom access in AC, but the definition of consciousness makes this paradoxical. How we call it consciousness if it is something we’re not aware of?

    Truth is, we sometimes experience things that we don’t understand, and only a refined AC can actually interpret those events fully. But still… something came through. As an example, having played competitive sports in my younger days, I know that I acted on a knowing I didn’t consciously access at the time. Even in hindsight there was very little that was consciously conceptual about it, but once in a rare while I made decisions that reflected an awareness of multiple players and their movements that only made logical sense afterwards. There are countless examples of this–a guitarist skilled enough to let go of conscious control, and able to explore an emotional-sensory terrain that is free form and extemporaneous.

    Last year or the year before, I read the book Opening the Doors of Perception by Anthony Peake. He had an interesting take on certain forms of brain injury. It’s been too long since i read it, and I don’t recall the details well enough to present them to you. And even if I did, I don’t expect his central thesis will necessarily resonate with you, but a part of it was this idea that our brains actually perceive a much greater slice of reality, and that the broader perceptions of reality (by the right hemisphere, I believe) are squelched by the more analytical functions of the other hemispere (by the left hemisphere, I think). This is of course a good thing–so we can function as we do. But Peake suggests that in certain types of autism, epilepsy, and in other “openings” that we might medically call degenerative–the reins are loosened and a fuller picture emerges.

    It seems consistent with the notion that there is a broader range available to us, that is not altogether missed, but not altogether accessed either. Is that liminal range consciousness? I don’t know. Depends who you ask, of course.


    Liked by 2 people

    1. Thanks Michael. Interesting points, as always.

      “How we call it consciousness if it is something we’re not aware of?”

      That’s the rub. And I think we’re talking about stuff right on the liminal boundary, so debating too much about whether it’s subliminal or supraliminal (i.e. conscious) may not be productive. I wouldn’t be surprised if the liminal boundary isn’t amorphous and shifty in reality, context dependent.

      On your sports navigation, there’s been a lot of research in recent years on the role of the hippocampus and entorhinal cortex in spatial navigation. I suspect a lot of what you’re describing happens there, well below the threshold of consciousness. If we’re aware of it in the moment, it’s only as intuitions or feelings. And as you describe, we’re only able to analyze it afterward, because the systems that do that are much slower than the ones using habitual or instinctive action programs.

      On Peake, you know me well. 🙂 His views don’t resonate too much with my understanding. In general, based on what I’ve read, the pursuit of a pre-categorical or pre-theoretical perception may be misguided. Perception by its very nature seems theoretical and categorizing. Which is why all observation is theory laden. We might be able to achieve a more primal level of categorization, but to eliminate it entirely is, from what I understand, to eliminate perception.

      Put another way, the neural image maps that form from sensory data are not consciousness. But our conscious perception is built from them. Put a third way, the conceptual stuff Block consigns to access consciousness is the very stuff of us apprehending phenomenal consciousness. At least according to my current understanding. (Which may be different tomorrow.)


    2. I can attest to the phenomenal aspect of “being in the zone” playing music. It’s definitely not AC, because too much AC tends to take you out of that zone. And there is what both athletes and musicians call “muscle memory” and spend many hours of practice acquiring.


Your thoughts?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.