“When words become unclear, I shall focus with photographs. When images become inadequate, I shall be content with silence.”
— Ansel Adams
Last time, we introduced Dan Dennett’s new paper “Facing up to the hard question of consciousness” 2018 Phil. Trans. R. Soc. B 373: 20170342. http://dx.doi.org/10.1098/rstb.2017.0342 See “Do Neurons Cause Consciousness” https://better-questions-than-answers.blog/2018/09/02/do-neurons-produce-consciousness/
It has proved thought provoking, as a good philosophy paper should. Let’s talk it over with BQTA contributor Rick Lenon, M.D.
GNS: It’s not easy to appreciate fully how radical Dennett’s view is or what the commitments required are to adopt it.
RL: I think even Dennett would permit us the kind of model/representation we’ve been talking about.
GNS: No, I don’t think he would. This is the ambiguity of the terms “model/representation”. By “model”, you don’t mean just the spike trains. You mean some kind of an image or picture that corresponds to the outside world, or that occurs in imagination or dreams. Neurons produce consciousness by creating the picture. What else could a dream be? This is what Koch is talking about. There must be a causal relation between the NCC— which are neurons and their activity—and the conscious events they are correlated with.
Dennett denies this because it would require a second transduction. “There is no central arena or depot where these spike trains become recipes for a second transduction that restores the properties transduced at the periphery, or translates them into some sort of counterpart properties of a privileged medium.”
By “restores the properties” he means that the dog in the Koch diagram, for example, had a certain shape. That shape was transduced “at the periphery” (the eyes) into spike trains. There is no further transduction that restores the shape of the dog in a mental image. By “translates them into some sort of counterpart properties”, he is referring to properties like color. Color cannot be “restored” in the second transduction, because it was never there is the first place. Instead, color would be the counterpart property of the wavelength information that made it into the spike trains.
Dennett is saying that this does not happen:
Colour vision is accomplished by a sophisticated system of information processing conducted entirely in spike trains, where colours are ‘represented’ by physical patterns of differences in spike trains that are not themselves colours.
True that wavelength information is represented in spike trains that are not themselves colors. What happens next? Dennett says that the sophisticated system
is designed by evolution to deliver useful information about the affordances that matter to the organism in a form that is readily usable or consumable by the specialized circuits that modulate the behaviours of systems external and internal.
That’s true too. Light yields valuable information the organism needs about objects. The system can deliver information about ripe fruit, for example, which reflects wavelengths different from the surrounding vegetation. Dennett seems to be referring to color, when he mentions “a form readily useable”. But how do spike trains “that are not themselves colors” turn into colors? Doesn’t that require the dreaded second transduction?
RL: His insistence that the real problem is what happens next is also in accord with the questions we’ve been struggling with.
GNS: I’m not clear what he means by the real problem or the hard question. “What happens next” is… nothing, on Dennett’s view. Nothing other than further neural activity that processes the information and creates a response. Maybe he simply means the real problem is just the hard scientific work necessary to uncover the details of the neural activity.
RL: I do think that some significant part of that has already happened by the time it’s part of the model; the model is not just raw data, waiting to be interpreted. Grossly, we’ve tested everything we see for whether or not it’s within expectation by the time it becomes conscious. Anything unusual will demand attention, maybe even before we “see” it.
GNS: Lots of processing must take place before an image “becomes conscious”. Why do you put “see” in quotes? Because you think of “becoming conscious” as a process of creating a visual mental image? I am forming a mental image of a eucalyptus tree right now, even though none are in my vicinity. It’s not a real tree, but it is a real mental image.
RL: I’ve been making much of the idea that what we see is has already been interpreted; snake or stick, optical illusions, etc. Admittedly, “What’s that?” also happens. Leslie and I were speculating about a white patch at the top of a eucalyptus. Binoculars, and it was leaves on a dead branch. It tried to be a bird when we first saw it.
GNS: You saw the white patch as a bird. You had a model in your mind that presented a picture of a bird high in the eucalyptus. When you wrote the story in the email later, you probably had a memory that included a mental image of the tree and the white patch. The problem is that all of these mental images require a second transduction, and believing in that is like believing in leprechauns.
RL: Must admit though that the above still has observer and observed, which leads me into the homuculus fallacy.
GNS: You’re probably right, though I’ve never been able make up my mind about whether that is a fatal flaw. Maybe we can avoid the homunculus. What if we said that you are the mental images. The self is consciousness itself. So there would be no need for a separate observer. I think this would be something like Hume’s view of the self.
Dennett thinks that the main problem is that our concept of mental imagery would require a second transduction into an immaterial mental realm. Spike trains are information, but not images.
RL: The philosopher’s perspective in this seems impossible to avoid. NCC’s are all very well, but only what we would expect. The scientific perspective seems to need… well, augmentation from philosophy, much as I hate to admit it.
GNS: If all the neuroscientists were to get together and agree on a perspective that turned out to make perfect sense, all the philosophers would leave and find new jobs. This, however, is the heart of the mind/body problem. There is no completely satisfactory perspective. If there were, we would not be talking about it. We would leave and find new jobs too.
Dennett thinks he occupies the scientific perspective and he is very popular among scientists. But they don’t understand what he is really saying, imho. That’s the hard work we have to do. My suspicion is that working scientists just want to do their work and not get involved in the squabbles underlying their conceptions. Why risk alienating a funding source?
Dennett does not believe in NCC, which are the basis of cognitive science. In his system, there are only functional/behavioral states. Neurons cannot be correlated with those because they are basically all the same thing. There is no consciousness, other than functional/behavioral states.
Dennett’s view means that we are all zombies. Galen Strawson made the point convincingly earlier this year, which is worth revisiting in light of our current thinking.
https://www.nybooks.com/daily/2018/03/13/the-consciousness-deniers/ and especially the follow-up exchange https://www.nybooks.com/daily/2018/04/03/magic-illusions-and-zombies-an-exchange/ These guys pull no punches.
RL: He does allow for virtual representation in brain software…
GNS: Sometimes Dennett fudges things. That’s what Strawson was complaining about. The phrase “virtual representation” does not occur, but “virtual” about everything else does. Virtual machine, virtual paint, virtual glue, virtual hotel, virtual clay, etc. I found his discussion on this confusing.
Do we get mental images or not in Dennett?
If we do, because “virtual representation” means creating a mental image, then Dennett’s complaints about no second transduction are empty. If neural activity produces mental images with color, sound and smell, that’s all the second transduction we need.
If we don’t, because virtual representation is not a mental image, then Dennett seems to be denying the manifest. That’s the charge Strawson lays on him. I think Strawson has a point. Of course, Strawson has no adequate solution of his own, and neither does anyone else.
RL: I think he is objecting to instantiation, not representation..
GNS: Yes, nobody objects to the concept of representation of information in the spike trains. The light captured from the reflections off the dog must be represented as spike trains in the optic nerve. However, in order to experience color or sound or smells, the information has to be more than represented. It has to be instantiated in experience as color, sound, smell. A string of neural impulses is not enough. It’s not color, sound or smell. Dennett himself emphasizes this point.
RL: I think what he’s complaining about is the depiction of qualia as innate in the representation we are looking at, rather than a product of the processes that yield the seeing.
GNS: I think you are on to something. But it also shows how slippery the term “representation” can be. We cannot be looking at a representation, but maybe we can experience one. Is the representation a picture of the visual scene created by the processes that yield seeing? That sounds like the second transduction: the neural processes create an image in the mind. Visual perception is a product of brain processes. The point of neuroscience is to discover the neural correlates of that image.
RL: It seems that the major point here is the difference between a representation and an instantiation, where the latter means to represent by means of an example of the thing represented. If the second transduction means that spike trains are transduced into something other that spike trains (or a standing wave like phenomenon), I doubt that many would defend that.
The representation would not be a simple projection of the world outside that the homunculus would then see. Qualia would not be a phenomenon present in the world, but rather a representation. By counterpart property, he seems to intend something substituting for a property of the thing observed. He wants to insist on it being spike trains, all the way down.
GNS: Here’s the nub of the problem: if it’s spike trains all the way down, then nothing exists but spike trains. We would have eliminated qualia and consciousness.
That’s what Strawson accuses Dennett of doing. Dennett always denies that he is eliminating anything and says “it’s just not what you think it is.” Then what consciousness really turns out to be is nothing but neural events and behavior. Those meager materials are not sufficient. Consciousness cannot be merely neural events and consciousness.
Spike trains must carry encoded information about the outside world. The information is in the pattern of the spikes. Call that pattern a representaion if you like. I see nothing wrong with that. But that pattern alone is not color, sound and smell. Not by itself.
Representations of phenomenal properties need not be, and usually are not, the phenomenal properties themselves. This word “red” right here on your screen, for example, represents red, but it is not red. It’s black.
How do we get the actual experience of color out of spike trains? Color is never instantiated? There must be something more going on than simply neural processes leading to further neural processes leading to “the specialized circuits that modulate the behaviours of systems external and internal.” None of that is color.
You say that you doubt that many would defend that spike trains are transduced into something else. Indeed, that sounds impossible, ridiculous, and a wholesale retreat from science. That’s Dennett’s clever move. Who would say that neural impulses turn into something immaterial?
RL: The Mona Lisa example, quoted in Dennett’s article from his buddy Richard Power, helps here: “the medium is a painting that hangs in the Louvre; the content is an Italian woman who modeled for the artist centuries ago.” By contrast, the medium of human perception is in the neural processes, not directly accessible; we can’t see the paint, only the content.
What Dennett is telling us is that we must get past the very seductive idea that there is a second presentation of what’s out there, somewhere inside your head, and that that’s what we look at. That qualia are present in that second presentation.
A Very Big Problem (as Dennett would say) with this presentation inside your head is, what happens next? More spike trains, to transmit that presentation to yet another screen?
I’m sympathetic with Dennett’s intentions, and agree that the Important Question is how the spike train becomes conscious experience. But I do think there is a model created inside our heads. He reminds us in this article that acuity is much sharper centrally than at the periphery, and that there is no color perception at the periphery; but that’s not what we see. Then there are all those optical illusions, that affect what we actually see, not what we think. When we spin rapidly, long enough, the world spins when we stop. So we are looking at a representation of the world, not the world itself. Hume said something like that, way back in the 18th century, did he not?
Given that I think there is a model.. all in spike trains of course… it has to be represented somehow. So why not as conscious experience?
GNS: By saying conscious experience is “all in spike trains” you have reduced consciousness to neural impulses. How are patterns of neural impulses color, sound, smell? Those sorts of phenomenal experiences are part of consciousness, but not part of spike trains.
As for Mona Lisa, the medium of the painting in the Louvre is paint, but the content is the image, not the lady who modeled in the 16th century. That some content can be presented in other media, such as pixels, like this:
This pixel image has the same content as the original painting. The lady who modeled, thought to be Lisa Gherardini, the wife of Francesco del Giocondo, is the person the painting is about—the “intentional object.”
Both the painting in Louvre and the pixels you are looking at now are straight-forward physical objects. So are your spike trains. The paint and the pixels give off light—generated in the case of pixels, reflected by paint. The image is in the light. Neural processes do not give off light. So where is the image in the spike train? According to Power,
we conceptualize the medium of our internal representations by abstracting some features from the content, and attributing them to some kind of spiritual or ghostly substance. That is the best we can do, because actually we cannot experience the medium at all and have to look for analogies in the external world. The idea that the medium is some state of the brain seems intuitively absurd, so powerful is the illusion that we are dealing with an iconic representation in a medium of spirit.
The medium can be the brain, but where is the conscious mental image. It does not consist of light, so what does it consist of? Power is right that there are dis-analogies between physical images and mental images. However, both involve an image. Power and Dennett would say that I’m looking for “some kind of spiritual or ghostly substance… a medium of spirit.”
RL: If there is a model… and I think there has to be… it has to take some form. Relative position has to be represented, so why not a map? There’s advantage to distinguishing wavelengths… so why not “green?” What leaves the retina is a spike trains, and everything that happens after that is made outa spike trains.
One explanation for afterimages is that if we gaze long enough at a red object, our red receptors get fatigued, relative to the green. There is constant input from the green, so shift your gaze, and green is what’s there. The afterimage from really bright light is usually red, if I recall. Not so with the LED’s in our kitchen… just tried it, and they eventually get blue when I close my eyes… not right away. I think maybe I’m not supposed to do that.
GNS: Take it easy with those home experiments. But… afterimages are made outa spike trains? You’ll have to explain that to me next time.