Thursday, November 13, 2014

Are we really conscious?

I've done a post on Graziano's 'attention schema' theory of conscious, and thought I would pass on some clips from his recent exposition of the model in a NYTimes piece. In the title, and in the text below, a few qualifying phrases might have avoided the strong responses to Graziano in the letters to the editor that I also note below. I've added some qualifiers in brackets [ ]. For the inflammatory title "Are we really conscious?" what about adding [in the way we commonly suppose?].
...the argument here is that there is no subjective impression [in the way we commonly suppose]; there is only information in a data-processing device. When we look at a red apple, the brain computes information about color. It also computes information about the self and about a (physically incoherent) property of subjective experience. The brain’s cognitive machinery accesses that interlinked information and derives several conclusions: There is a self, a me; there is a red thing nearby; there is such a thing as subjective experience; and I have an experience of that red thing. Cognition is captive to those internal models. Such a brain would inescapably conclude it has subjective experience.
I concede that this approach is counterintuitive. One reason is that it seems to leave a gap in the logic: Why would the brain waste energy computing information about subjective awareness and attributing that property to itself, if the brain doesn’t in fact have this property?
This is where my own work comes in. In my lab at Princeton, my colleagues and I have been developing the “attention schema” theory of consciousness, which may explain why that computation is useful and would evolve in any complex brain. Here’s the gist of it:
Take again the case of color and wavelength. Wavelength is a real, physical phenomenon; color is the brain’s approximate, slightly incorrect model of it. In the attention schema theory, attention is the physical phenomenon and awareness is the brain’s approximate, slightly incorrect model of it. In neuroscience, attention is a process of enhancing some signals at the expense of others. It’s a way of focusing resources. Attention: a real, mechanistic phenomenon that can be programmed into a computer chip. Awareness: a cartoonish reconstruction of attention that is as physically inaccurate as the brain’s internal model of color.
In this theory, awareness is not an illusion. It’s a caricature. Something — attention — really does exist, and awareness is a distorted accounting of it.
One reason that the brain needs an approximate model of attention is that to be able to control something efficiently, a system needs at least a rough model of the thing to be controlled. Another reason is that to predict the behavior of other creatures, the brain needs to model their brain states, including their attention. This theory pulls together evidence from social neuroscience, attention research, control theory and elsewhere.
Almost all other theories of consciousness are rooted in our intuitions about awareness. Like the intuition that white light is pure [when it is in fact a spectrum of all colors], our intuitions about awareness come from information computed deep in the brain. But the brain computes models that are caricatures of real things. And as with color, so with consciousness: It’s best to be skeptical of intuition.
The letters in response to the above take Graziano to task for "explaining consciousness away" when what he is trying to do is not discount our experience of awareness or selfhood, but make a description of the physical process that actually constitutes them, a description that is counter-intuitive, but I think more likely to be correct than anything I've see thus far.


  1. It's not clear that awareness and self are of the same type of thing. The self is something that develops over time is malleable (unique to humans?) and yes we treat it as something real but it is fundamentally illusory, a mental construction. Awareness on the other hand seems to be more fundamental; something contemplative/meditative practices point toward. This seems to be preverbal and more universal. Eg Tononi's work

  2. Graziano could have made more clear that awareness and awareness specifically of 'having a self' are totally not the same thing. Awareness can be attributed to animals that do not pass the mirror test (recognizing your own image in mirror). His model is for the more general kind of awareness or consciousness, saying it is a model that melds just the useful components of various more direct and fundamental sensory and motor inputs.

  3. Anonymous6:02 AM

    Not convincing at all in any way. So if I write a few codes of line in my matlab program that say to the matlab program "you exist and are conscious" (and how the hell would I do that anyway) then the program gains a self and consciousness? Or no wait, we just couple attention (a signal amplifier) to it, and now yes clearly, I've just created a conscious program.
    This stuff is disappointing in many ways. 1) It shows a basic lack of understanding of what the mystery of consciousness is by a famous professor. 2) The sheepish cheering on of whatever anyone says because this person is an "authority". 3) The lack of genuine interest by many people, or to go beyond simple one liners and go into critical thinking.
    Really, is it so easy to create consciousness? Interesting, please tell me how I have to program my computer to get a conscious entity with actual experiences. Or any other mechanical device. Please tell me how I can compute what my computer experiences right now. Please let me know why it is 'obvious' that all mammals have experiences.

    Alright, I'm pretty sure that I'll basically reach no one with these comments. Please go on with the self congratulations and the strong conviction that you can easily solve any mystery with feel good stories, as long as this story gets the stamp of approval of an "authority".

  4. My, my. quite the tirade…someone’s feathers have gotten thoroughly ruffled up! - to the point of issuing a fairly invective screed. I’m pleased to have such strong sentiments expressed in a comment, because this blog doesn’t usually draw comments as interesting as yours. I don’t think anybody is claiming they have the final word, there is mainly jockeying over what might be the best guess at the moment. You might have indicated, apart from an essential Cartesian dualistic approach, to state what “the mystery of consciousness is” that you take the famous professor to lack. And, yes, the extrapolation of the basic approach at Graziano, Tononi, Metzinger, and others take is that it should be possible to create a mechanical device with consciousness, but extremely improbable anytime soon, because of the immense computational problem presented by the current impossibility of separating our consciousness from its embodiment.

  5. Anonymous7:54 AM

    Yes, no you are certainly right, an emotional response on my part :) Thanks for your response anyway, I really enjoy your blog, so maybe my anger is misdirected. Just to give some personal background, I'm also active in the field, and I routinely run into people who dismiss consciousness research as uninteresting, or ill-defined (yawning through presentations on this topic), while they spend their lives figuring out details of the attentional blink.
    I'm not saying that other work is uninteresting, I'm just really dismayed by the off-handedness, and yes, arrogance, with which this topic is treated.

    Mechanistic accounts are great, but they should be well thought out and serious. Theories that essentially say "what is all this broo ha ha about consciousness, it is really just all attention/higher order thoughts/recurrency/global workspace/integrated information/etc" seem to miss the mark in my opinion.

    I'm not claiming that the "hard problem" is insolvable per se, however, I think we have to acknowledge that WE can NOT solve it now. The hard problem is currently simply a description of the state of affairs. If I give you a set of mechanistic rules/computer program, then you can calculate how the machine will react given certain input, however, it is impossible to calculate what it experiences.

    Perhaps in the end we will somehow be able to derive experiences from mechanisms, just like we can calculate output based on input + mechanisms. However, this seems like a tall order.

    My personal opinion is that this problem is still the neglected child of science. Behaviorism is officially dead, but unofficially alive and kicking. Reaction times, ERP's fMRI scans, MEG time-frequency analyses: that is all real stuff. But you: the experiencer, and your experiences: well I cannot see you, or your experiences like I can see fMRI scans, so I can just revert to fairy tales when it comes to that topic.

  6. I find it curious that those who claim they do serious research on consciousness never state one thing: the field of research is what is apparently "out there". It's the world as we see, hear and touch it; what seems to be out there, "The World", it's actually a very intricate virtual reality show generated by consciousness. It's actually just like a dream but one difference: when awake, this dream is fed with external input from senses. It's still a dream, nonetheless.

    Therefore research that overlooks how this virtual reality show is *generated* will not be able to answer even the most basic questions about the hard problem of consciousness. And, I have never ever seen any explanation on how ordinary matter (what our brains are made of) can generate subjective virtual reality shows.

  7. I think that Graziano's response would be that your virtual reality show is not generated by consciousness, it is what we are experiencing as consciousness.

  8. Anonymous7:22 AM

    Jani, I think you touch upon an interesting point: what we are talking about is really: why is there a subjective world. Which is almost the same as saying: why is there a world?
    The way I think about this: if there would be no experiences, there could be a world, but no one would know, so effectively, this would (for any observer) be indistinguishable from no world at all.

    Deric is also right of course: consciousness = your subjective world.

    So I think the topic at hand does matter. To put it into really down to earth terms: visual consciousness = visual experiences. Don't we want to know how to restore visual experiences for someone who has gone blind?

    Therefore I wholeheartedly disagree with the current neo-behaviorism, which basically handwaives at consciousness, and without any real evidence declares it to be insignificant, ill-defined, or easily solvable.

    Let's not be so hung up on ourselves and our theories, but let's try to really get to the bottom of this challenging mystery. And, yes, let's call out the current neo-behaviorism. IMO it's fueled by arrogance and sloppy thinking, but at any rate, let's be realistic: we are nowhere close to actually understanding why and how electrons and atoms moving around can create consciousness.

  9. Fair point. One of the issues, actually, is that we don't even have anything to conclude whether the show is generated by consciousness, is experienced as consciousness, is a manifestation of consciousness or is the consciouness itself. It might be even something else.