So now, after being advised to go read lecture 35 of Richard Feynman's Lectures on Physics, I know something about psychophysics--specifically, how the eye sees colors with three types of cones sensitive to different wavelengths and even though the set of all possible spectral distributions of light is an infinite-dimensional space of non-continuous functions the space of colors we can see is a three-dimensional manifold yadda yadda yadda.
But there is one thing Feynman wrote that is really bothering me. He wrote "yellow is not a greenish-red."
You see, magenta--when the red cones fire and the blue cones fire--is a reddish-blue. Cyan--when the blue cones fire and the green cones fire--is a bluish-green. But yellow--when the green cones fire and the red cones fire--is not a greenish-red.
This bothers me. Why isn't yellow a greenish-red? What is special in the visual cortex about the linkup when green and red cones fire that creates a sensation very different from the color blending we see when blue and red or when blue and green cones fire?
This bothers me a lot.
Posted by DeLong at December 17, 2002 06:09 PM | TrackbackDoes Feynman give a scientific definition for "ishyness"? Does white, or grey for that matter, look like a redish-blue-green?
I always thought magenta was a blue tinged red.
Cyan looks like light blue to me without any suggestion of green.
I could be totally off base here, but this is why I would think that green and red would not produce yellow.
The primary palet colors are red, yellow, and blue. The secondary are green, purple, and orange.
Red and blue make purple,
Red and yellow orange,
Blue and yellow make green,
Any time you take primary and secondary colors where the primary is not part of the composition of the secondary (purple and yellow, blue and orange, and red and green) they clash and it makes a brownish color. red and blue can make magenta because they're both primary. Blue and green can make cyan because green and blue are complimentary. Red and green are clashing.
Now, I know that light colors and palet colors are not the same (sum of all palets is black, sum of all lights is white), but the color combinations for the most part are the same, including those that you describe (magenta and cyan), so I suspect that the rules for one would in fact bleed over to the other when most combinations are formed.
I'm not a physics person, so I could be way off here, but it's a logical explanation to me.
Posted by: R. Alex on December 17, 2002 07:35 PMToo many Davids, to point out which is the main reason I'm adding this comment. Also, wherefore RGB (light) vs. RYB (pigment)? Why am I writing like this? I'm sure eventually you'll get someone who knows what he's talking about, but it sure ain't me.
Posted by: David A on December 17, 2002 07:46 PMThe reason yellow does not appear as greenish-red is that the brain also analyzes information in terms of opponent colors. Some neurons are turned "on" by green and "off" by red, and vice-versa. Since red and green are opposing colors, you cannot simultaneously detect them. The same is true of blue and yellow, and black and white.
For a demonstration of this, try staring at a yellow image and then look at a white sheet of paper. You will notice a blue after-image (yellow's opponent color).
Posted by: YM on December 17, 2002 08:18 PMWell, if you consider traffic lights, perhaps there is something about yellow which makes us think it belongs between the safety of green and the danger of red. Kind of like yellow could go either way.
But to me yellow certainly doesn't look like greenish red like cyan looks like bluish green and magenta looks like bluish red. In old database programming you used to have to specify colors using combinations of RGB, and I noticed it then that G+R = yellow was rather an odd thing unlike the other two combinations. One possible explanation is we are used to subtraction of wavelengths when mixing colors and not used to addition.... we see like we mix paint and wavelengths are absorbed. This makes sense because in nature the source of light is the sun, and from there colors are subtracted. In that sense our brain maps from wavelength addition to wavelength subtraction, which it should. In grade school most of us become familiar with the not quite subtractive primaries red yellow blue, and as a result can easily pick out if an orange is more red or yellow, if a green is more yellow or blue and if a purple is more red or blue.
http://www.beer.org/~tpark/color.html
At the above link, I am having some intuitive trouble with the subtractive mixing of cyan/magenta/yellow. Quite possibly the fact that RYB, but not CMY pigments were readily available from nature also explains why I have evolved to have better intuition of their mixing... does anybody know of a grade school where they teach CMY as primary colors and hand out CMY paint to mix? Might be a good reason for that.
But more generally speaking, there is no absolute reason for any sensation of a color to map to adjacent wavelengths (red and purple look similar to me, but aren't close), or to a continuous sensation between two color receptors. If nature is continuous when it comes to the meaning of colors to an organism, then it would make sense for the organism to have continuous reactions under wavelength subtraction. If we lived in a world where everything gave off it's own light, then perhaps we would see continuously in RGB addition and yellow would just sort of look like reddish green. If there was a particular range of light in the blue spectrum that always meant extreme danger as opposed to water or blueberries, then it is quite possible we would have evolved to experience another color entirely... or maybe instead of blue we would experience a flashing red, or even severe anxiety. The important thing to realize is that in a brain, anything can be mapped onto anything else. The feeling that two colors look like each other don't have to be related to a numerical similarity in physics or to the biology of the eye. Presumably color had some evolutionary usefulness, and sensor impulses are mapped onto an output in some useful way.
as an addendum to my previous message, one can obtain more information about the opponent-process theory at: http://www.yorku.ca/eye/opponent.htm
Feynman's description of color perception is consistent with trichromatic theory; but it only tells half the story. Opponent-process theory explains the rest, ie why we do not see yellow as greenish-red.
Posted by: YM on December 17, 2002 08:58 PMPardon my ignorance but does this have anything to do with the fact that TV beams are yellow, green and red rather than yellow, blue and red? Or do I have it completely wrong?
Posted by: Jean-Philippe Stijns on December 17, 2002 09:05 PMTV beams are RGB, just like the name of the type of computer monitor they used to sell.
Also, I wonder if it could be shown that this opponent process theory somehow helps map from additive to subtractive color.
Posted by: snsterling on December 17, 2002 09:13 PMYellow is very difficult. People tend to see it as either green or orange. It is in some sense always an assumed color. The predominate color we do not need to see. Try some time asking people if the yellow is too green or red. The problem is that colors can be combined in the color wheel but not necessarily perceived by the eye. In physics the combination of all colors is white but in nature the combination of all colors is black That is because of the of reflection.
I don't know if I got that all right but compare Joseph Alpers with Feynman
and you will see what I mean
A physicist is never going to answer this question for you, because it's fundamentally a question about the phenomenology of perception, not its physics or biology.
To put it another way, why do you (and Feynman) say that "yellow is not a greenish red"? What do you think a greenish red looks like? Why not answer "yellow" to that question?
Wittgenstein's "Remarks on Colour" are going to get you thinking much more along the right lines than any attempt at a scientific question to what is fundamentally a philosophical, or even a linguistic question.
I seem to remember that Dennett wrote a few good things on this subject before he decided that it was better to be a very bad evolutionary theorist.
Posted by: dsquared on December 17, 2002 11:05 PMFurther to the above, I'd note that "opponent process" theory can't answer Brad's question at all; it gives a name to what happens in the cortex, but that doesn't get us any closer to explaining why what we see is yellow; there's nothing yellow about the neurons.
Posted by: dsquared on December 17, 2002 11:08 PMD^2 is right. Biology and physics are not the places to look for enlightenment on this. Linguists and anthropologists tend to treat colour naming as their domain. Berlin and Kay's seminal 1969 study of colour suggested a biological basis for colour naming, but their methods and results have come under heavy criticism over the last 30 years, and their conclusions are no longer very widely taken at face value outside of the core nativist community. While colour categories do not appear to ever be completely disjoint, they can span different parts of the spectrum in different languages, and hold varying composite statuses across languages. There is at most a statistically significant tendency to partition the colour spectrum in some ways and not others, but there is no correlation between colour naming and neurobiology, and as Wierzbicka points out, it is easier to explain what universals remain without nativism.
There are languages that categorise yellow as a greenish red, or a redish green.
Paul Kay will no doubt disagree with my conclusions about the non-universal nature of colour naming. He's at Berkeley, so you can call him or look at his website if you want the other side. He regularly publishes defenses of his work, but I don't find them very convincing.
Yellow is a mystery too.
FYI, That "well defined horizontal line" I mentioned before, the one I saw while my eyes were closed and my head was being CAT scanned: I've been thinking it. It was antimagenta--kind of a bright "cosmological green"...
Posted by: Mike on December 18, 2002 04:54 AMMe and Scott appear to be agreeing with each other across the blogs today .... just to mention that the botanists probably have a lot to say about colour differentiation, as it is highly likely that animals' ability to distinguish colours evolved side by side with plants' production of pigments.
Posted by: dsquared on December 18, 2002 05:19 AMHold on, I have a question for Scott: I know that identified colors vary from culture to culture (clod that I am, I often make jokes about the fact that the Japanese call both blue and green by the same name). But surely the cones (rods?) are triggered by certain very specific wavelengths that we can agree are the true Primaries of light (for all humans, anyway); similarly, just because another culture doesn't think yellow is a major color, that doesn't change the fact that it's a pigment primary? Can New Guineans (or whatever) mix 2 colors to get blue?
Posted by: JRoth on December 18, 2002 09:14 AMJRoth: Yes you can mix two colors to get blue. Cyan and magenta if mixed make blue. That's because they are truly primaries and red and blue are only bad approximations. Also... I don't find cyan or magenta to look particularly special or "major". I do however find I look at things based around Red/Yellow/Blue (see my link above which talks about printer's vs. painter's primaries)
Posted by: snsterling on December 18, 2002 09:33 AMFascinating stuff, as always. Maybe someone here knows the mechanism of red-green colorblindness? Is it a physical difference in the structure of the cones, or is it related to the brain structures that do the interpreting?
Posted by: jimbo on December 18, 2002 09:35 AMd-squared,
I have heard you off-handedly disparage Dennett on at least two occasions. As someone who thinks you are both right most of the time, I would be curious to know what, specifically, Dennett argues that you think is so wrong (or if it has to do with his style, please explain why his approach is so irksome). Thanks. Sorry for the OT.
theCoach: no worries. My animus against Dennett is, of course, the zeal of a convert; I used to be a rabid Dennettite on the subject of machine consciousness; I still think that "The Mind's I" is a fantastic book, and "Consciousness Explained" is a classic of its type. But time passes ... and with every passing year, Dennett's offhand treatment of subjective experience just seemed less and less convincing. When arguing, I usually resort to Kripke's modal argument against functionalism (the quality of being painful is a necessary quality of pain, but not of any functional state ...), but when it comes down to it, I just don't believe that "qualia" can be explained away (hence my remarks on colour upthread).
Anyhoo, I started reading more widely in the subject, and found that Dennett had been pretty (no, make that very) badly behaved when it came to falsely attributing strawman positions to John Searle, and not admitting it when corrected. Then Dennett went onto his pop-Darwinist kick, and started saying things about Stephen Jay Gould and Richard Lewontin that I happened to know weren't true (I think I remember rightly that he was the originator of the canard that Gould's work with Eldredge wasn't highly regarded in the field). And that's when the hate developed. (Since then, I've caught him out in a couple of out-and-out fibs about Margaret Mead which are also favourites of Pinker and Dawkins).
I don't like the guy. He's on my "always check" list for any claims of fact. And he has a nasty line in trying to psychoanalyse his opponents by trying to insinuate that the reason they disagree with him is that they are secretly in thrall to religious belief. This looks like projection to me ...
Gosh, that was much more emotionally draining to write than I thought it would be ...
Posted by: dsquared on December 18, 2002 09:58 AMThe most common color anomoly is an apparent lack of one cone class; such a person is called a dichromat, though there appear to be people who have either anomolous photopigments (rather than missing ones) or a smaller density of cones of a particular class than a "normal" observer.
Dichromats either lack long, medium, or short wavelength cones, and are called tritanopes, deuteranopes, and protanopes, respectively. The cones apparently are not actually missing, they just have the "wrong" photopigment (L for M or M for L). I'm not sure what S cones get substituted with. Both protatnopes and deuteranopes have some degree of red-green discrimination difficulty--you have to use a more complicated test to determine whether someone is a protanope or deuteranope.
When I taught computer animation I would use the seemingly paradoxical case of red and green making yellow to test if my students had really done the assigned experimentation with the color systems used in film and in electronic displays. The key point I tried to impress was that, at least at that time, people had far more experience with subratctive color mixing, i.e. pigments, than with additive mixing, thus adding red and green to make yellow was counterintuitive.
Posted by: Thomas Briggs on December 18, 2002 11:09 AMThanks for the reply. I think it was Maynard Smith who said Gould was not highly regarded. Searle's Chinese Room and Dennett were talking past each other, both made points, but Searle's was less relevant (my opinion). Dennett wrote that Gould's PE was either revolutionary, and false or not that interesting, but true. I think he was right, but I would appreciate more info. Any pointers on the Margaret Mead story I would be interested in as well. I am still of the belief that she was probably getting the results she wanted from subjects trying to please, but would love to see information that shows otherwise.
Qualia and Chalmer's ‘hard’ problem of consciousness seem to be the next step. I find find Dennett to be thoroghly convincing in his take down of Chalmer's zombies without consciousness. I guess I do not even see what the problem is. It seems to me the argument on Chalmer’s side is “but it feels like”— a very week argument. Show me where I am wrong, as I would love to understand the argument on the other side.
Posted by: theCoach on December 18, 2002 11:27 AMJRoth:
What's a primary colour? If you define a primary colour as the average maximal response of the finite number of rods and cones in the human eye, then I suppose a New Guinean tribesman would agree with you about which colours are primary, after you'd spent a decade teaching him (or her) to read a major world language and convincing them to read the relevant texts in optics, linguistics, psychology, neurology, biology, and colour theory in the arts. But, I trust you would agree, this exercise would accomplish nothing (except educate some no-doubt deserving impoverished New Guinean.)
When you define the primary colours, what makes them primary except that you say so? Other combinations of paint and light will also produce intermediate colours, as artists know well. Yes, people's rods and cones do maximally respond to certain frequencies that fall within narrow ranges, but there is only the vaguest, most controvertial evidence that people name colours in ways that correspond to their rods and cones. The neurobiology of the eye appears not to correspond strongly to human linguistic categorisation.
The blue-green ambiguity in Japanese is common enough. Latin was the same. But, there are languages where only two terms correspond to hue in general (usually some cognate of warm and cold), and I am hard pressed to see how such people could identify anything as a primary colour.
So, unless you take someone through the science of optometry, there isn't any reason whay they ought to agree on precise shades as being more central than others.
I share D^2 distaste for Dennett, except I have an even more shameful history. I was into Karl Popper and Steven Pinker, until I discovered Lev Vygotsky, Anna Wierzbicka and neural networks.
Posted by: Scott Martens on December 18, 2002 11:28 AMJimbo:
On red-green color blindness: I am not a biologist, but I recall reading that scientists had traced the exact genetic mechanism for this. If I recall correctly, the genes that encode for the "red" cones and the genes that encode for the "green" cones, both on the X-chromosome, differ by only one base pair, the smallest possible difference. The genes for the blue cones are substantially different, and located elsewhere.
People who are red-green color blind -- I believe about 8% of males, which would mean about 2/3 of 1% of females -- don't have this differentiation between the two genes.
By the way, the commonness of this "defect" is the reason that traffic green lights are what we would call "blue-green" and traffic red lights (including car tail lights for reasons I don't really understand) are "red-orange". It's so people with red-green color blindness can still differentiate these by color.
Anyway, don't consider this authoritative -- can anyone out there with a real biology background correct or clarify?
Also, judging from my kids' experiences, young schoolchildren are still being taught that red, yellow, and blue are the primary colors, using subtractive (pigment) examples. I suppose it's a good enough approximation for 1st graders. Heck, I didn't even know about cyan, magenta, and yellow until color printers came out when I was an adult! But I have been thinking of getting some colored party light bulbs to show them additive mixing.
Posted by: Curt Wilson on December 18, 2002 01:47 PM'Further to the above, I'd note that "opponent process" theory can't answer Brad's question at all; it gives a name to what happens in the cortex, but that doesn't get us any closer to explaining why what we see is yellow; there's nothing yellow about the neurons.'
I think it does, actually. A quick spin around Google implies the yellow/magenta difference is the result of something like yellow being located at an overlap in spectra detectable by each cone, at the proper relative intensity while magenta isn't. I think.....
Here's a good summary.
Posted by: Jason McCullough on December 18, 2002 02:13 PMGentlemen! GENTLEMEN! Can we not simply live in a colorblind society, and have no more of these troubling matters?
Posted by: RonK, Seattle on December 18, 2002 04:23 PMBrad,
You should read Oliver Saks on this, particularly
a chapter in "An Anthropologist On Mars" recounting the case of a suddenly color-blinded
artist.
If I remember correctly, he notes (referencing others' work) that it has been demonstrated quite conclusively that color is a construct of the brain having nothing really to do with light wavelengths or any other objective property of the "real" world.
In other words, yellow is what your brain says it is, for its own purposes of information processing and organization and not a whole lot else.
I'm not in the field and am probably not describing this all that well, but the problem Saks describes does sound an awful lot like what you're puzzling over.
It's a great book anyway...
Best,
G
Coach: This is getting a bit long and involved. I'll try to put something up at my own weblog on the subject of l'affaire Margaret Mead, but on the general topic of Dennett on consciousness I'd make two points: 1) when the topic is consciousness "it doesn't feel" is a much stronger argument than one might think, and 2) Understood correctly, Searle's Chinese Room argument is absolutely devastating to any functionalist theory; it makes the point that not only is syntax not sufficient for semantics, but in the absence of a conscious interpreter, machine movements don't even have syntax.
Jason:
>>A quick spin around Google implies the yellow/magenta difference is the result of something like yellow being located at an overlap in spectra detectable by each cone, at the proper relative intensity while magenta isn't. I think.....<<
V. interesting, but somewhat orthogonal? Whatever this might be detecting, there's nothing yellow about it, and nothing which helps you explain why yellow isn't a reddish green.
Posted by: dsquared on December 18, 2002 11:14 PMOkay.
We've established that what we're really talking about here is perception: specifically, how we (or our sensory system) perceives and interprets external physical stimuli--the "physics" of those external stimuli, while interesting, are actually incidental.
That said, I say: the most expeditious way to proceed with this discussion is to think of "yellow" as being analagous to "middle C" in the aural realm.
(Not being a musicologist, I might have chosen the wrong "chord"--I meant to indicate a "harmonious tone" at or near the MIDDLE of our perceptual range which is both distinctive and resonant with other similar "harmonious tones" above AND below it on our musical scale.
There's MY "last words" on the subject. I'll take that blindfold now, thank you. Might as well have the cigarette too: can't dance ;-)
Posted by: Mike on December 19, 2002 04:02 AMThanks, d-squared. I look forward to your posts on your site. 'Understood correctly' are going to be the sticking points aren't they. There is a lot of space there. To me understanding the Chinese Room correctly means understanding that Searle is confusing his analogy – the person in the machine is an automata – it is the entire machine that exhibits consciousness. “Feels like it” ain’t good enough for me unless you give me a reason.
Bottom line, to me anyway, is Dennett is saying it is unlikely that there is any magic going on, where Searle, Chalmers, etc. are saying, “but what about the magic”?
But again I look forward to your posts, here or on your site. Since I have you engaged, would you mind recommending a good book on the basics of Macroeconomics? I am not afraid of math, but it would have to have refreshers in it. Thanks again.
I just re-read Feynman's 35th and couldn't find the reference. In fact eqn 35.1 says the exact opposite:
Y = rR + gG
Yellow defined as some Red and some Green.
What am I missing? or does this solve the mystery?
And here I was pulling out my master's thesis ready to go to war....
Posted by: siliconretina on December 19, 2002 08:07 AMInsn't Brad really confusing two questions?
1. What makes a color on the spectrum.
2. What makes a color in the eye and brain
Isn't it impossible to discuss without separating the two.
Eric M.:
Yes.
It is counter-intuitive though. Steven Pinker has a great book about this aspects of these intuitive ghosts called "The Blank Slate".
(that was for you d-squared ;)
Coach:
"Counterintuitive" is an interesting concept.
Speaking of "intuitive ghosts": Is that like Feynman's idea that an electron is (or, anyway, may be "diagrammed as") a positron which just happens to be "going backwards in time", since it "fits the data"--even if it is an insult to our intelligence ;?)
"Insn't Brad really confusing two questions?
1. What makes a color on the spectrum.
2. What makes a color in the eye and brain
Isn't it impossible to discuss without separating the two."
I don't think so, because a "color on the spectrum" doesn't really exist. Energy at a single wavelength in a certain part of the spectrum happens to stimulate our cones in such a way that we perceive a color and give it a name, but our eyes can't tell the difference between a pure spectral color and a mixture of energy at different wavelengths.
Posted by: Peter MacLeod on December 19, 2002 05:37 PMMike, these ghosts are different but both are interesting.
Peter, on the one hand we have wavelengths of light, on the other we have our brain's perception of it. With our current understanding of how the brain works, we have to separate those two, I think.
Posted by: theCoach on December 19, 2002 07:49 PM(I just had an attack of realism regarding my own disgraceful procrastination and laziness with respect to weblog updates. I might do something in detail on this sometime).
Coach wrote:
>>To me understanding the Chinese Room correctly means understanding that Searle is confusing his analogy – the person in the machine is an automata – it is the entire machine that exhibits consciousness.<<
Ahhh. A number of issues here. First up, the CR isn't really about "consciousness" in the sense in which the arguments about zombies are; it's not part of Searle's concern as to whether the CR has qualia or sensations, or whether it is like anything to be a Chinese Room. Correctly understood (I say this with some arrogance; IMO, Searle has often allowed himself to be sucked in to misrepresenting his own thought-experiment), the Chinese Room is about *meaning*. The matter at issue is; does the Chinese Room mean what it says, or as Searle colloquially puts it, does it speak Chinese? Meaning and consciousness certainly seem to belong together, in that we only want to ascribe meaning to the statements of conscious entities. (Wittgenstein would obviously say that this conflation of our mental states with our use of language is a massive mistake, and at times Dennett seems to be following him). But at base, Searle is concerned with meaning, not consciousness per se. Remember that the original Chinese Room article was published early on in the Artificial Intelligence debate, before consciousness became the shibboleth that it is today.
Your response is what Searle calls the Systems Response in the original article; I don't like his defence of it there, but he fleshes it out well in later works (The Mystery of Consciousness is particularly good). His basic point is that the topological organisation of the Chinese Room is not an important part of the experiment. You could change it to assume that the operator memorises all the syntactic rules in the box, so that the system and the operator are the same thing; he still wouldn't speak Chinese. Or rather, he would appear to speak Chinese, but he wouldn't mean what he said (if someone who had only memorised a set of syntactic rules suddenly voiced the sentence "you're a wanker", you couldn't take offence). I find this convincing.
>> “Feels like it” ain’t good enough for me unless you give me a reason.<<
The reason is that there really isn't any non-circular argument why we should attribute meaning to the utterances of a machine. The point being that not only is syntax not the same as semantics (the original point of the CR article), but that, considered on their own, machines can't even produce syntax. Your computer, for example, is a complicated machine for syntactical manipulation .... or is it? Searle quite correctly points out that what your computer is, is a machine for turning switches on and off and controlling a cathode ray tube or LCD screen. It's your own ability to interpret those switchings which imposes a syntactic structure on them; syntax isn't a physical property.
>>Bottom line, to me anyway, is Dennett is saying it is unlikely that there is any magic going on, where Searle, Chalmers, etc. are saying, “but what about the magic”?<<
"Magic", in context, you'll agree is a pejorative term; Searle, Chalmers and Dennett are all materialists, and Dennett's rough and rude treatment of Searle's hypothesis that consciousness is a biological property was what turned me off the guy in the first place. I'd replace "magic" by "meaning" in your phrase; at base, Dennett, following his Wittgensteinian line, is a sceptic about whether there are such things as meanings.
Dennett thinks that all there is to consciousness is a functional property. Searle thinks that there aren't (can't be) any such things as functional properties unless there is something to interpret them, and that this something appears to be very wrapped up in what it is which differentiates conscious human beings from non-conscious ones. Chalmers, I never read him.
It is my personal belief that there *are* no good introductory macroeconomics texts. Begg, Dornbusch & Fischer isn't too bad.
Posted by: dsquared on December 20, 2002 07:28 AMdsquared- much appreciated. If I recall correctly, I remember Searle was not doing a good job of presenting his actually rather important syntax argument. This thread is probably growing weary, but I think the following is relevant from http://ase.tufts.edu/cogstud/papers/chalmersdeb3dft.htm
Dennett:
A week ago, I heard James Conant give a talk at Tufts, entitled “Two Varieties of Skepticism” in which he distinguished two oft-confounded questions:
Descartes: How is it possible for me to tell whether a thought of mine is true or false, perception or dream?
Kant: How is it possible for something even to be a thought (of mine)? What are the conditions for the possibility of experience (veridical or illusory) at all?
Conant’s excellent point was that in the history of philosophy, up to this very day, we often find philosophers talking past each other because they don’t see the difference between the Cartesian question (or family of questions) and the Kantian question (or family of questions), or because they try to merge the questions. I want to add a third version of the question:
Turing: How could we make a robot that had thoughts, that learned from “experience” (interacting with the world) and used what it learned the way we can do?
There are two main reactions to Turing’s proposal to trade in Kant’s question for his.
(A) Cool! Turing has found a way to actually answer Kant’s question!
(B) Aaaargh! Don’t fall for it! You’re leaving out . . . experience!
"Peter, on the one hand we have wavelengths of light, on the other we have our brain's perception of it. With our current understanding of how the brain works, we have to separate those two, I think."
I think yes and no--nothing becomes "colored" until we perceive it. Our eye/brain can't tell the difference between a single wavelength and a mixture of wavelengths that match in terms of the color matching functions. On the other hand, there is a physical world that we're seeing, and we do want to be able to describe and predict how pigments and dyes mix, how colorant layers behave when they overlay, etc. All of those things, when characterized by color science, are described in terms of wavelength, not in terms of "color." A painter, on the other hand, doesn't know anything about wavelength, but knows how pigments mix based on experience.
d-squared : This is a great discussion. I hope you take this to your blog.
It's all very well dismissing the "systems" answer to the Chinese Room. But if you accept Searle, where do you go from there?
Do you follow Searle into Carbon Chauvanism? And argue that only matter made of biological stuff supports meaning?
Posted by: phil jones on December 20, 2002 11:04 AMI AM therefore I think.
EVERYTHING else is debatable ;!)
Posted by: Mike on December 20, 2002 02:46 PM>>But if you accept Searle, where do you go from there? Do you follow Searle into Carbon Chauvanism? And argue that only matter made of biological stuff supports meaning?<<
Yeah! If you accept Searle, where do you go from there? I mean, suppose we construct a huge Chinese room, with one person for each neuron in the brain, and have them communicate with the other "neurons" in exactly the same way in response to exactly the same signals that neurons do: is "thinking" taking place?
Posted by: Brad DeLong on December 21, 2002 08:30 AM"I mean, suppose we construct a huge Chinese room, with one person for each neuron in the brain, and have them communicate with the other "neurons" in exactly the same way in response to exactly the same signals that neurons do: is "thinking" taking place?"
Please keep in mind that in the past (as well as currently), denials of this have been used to enslave, murder, or deprive of rights various classes of people... animals too. Heck, if the experiences of animals really mattered, we shouldn't be eating them, and if it didn't matter then wouldn't torturing them be ok? Instead most people fortunately believe in some practical middle ground, whether or not it makes sense to.
It's not so surprising that this discussion on perception of color turns into a discussion about invisible "soul/consciousness" type things being overlaid on thought process--invariably when people set out to determine if they have such a thing or not they think about differences between colors, usually blue and green for some reason. Personally, I don't really see how it would help to have one, and if I determined I had one yet could ignore someone else's claim to have one (such as a Chinese Room) I'd have to wonder if I didn't make a mistake somewhere.
Also, just to follow up on green+red=yellow....
The opponent process setup, even if not exact, does a pretty clever job at taking our RGB receptors and turning them into signals useful for the associated subtractive primaries.... that is, it separates out the brightness, and then calculates yellow. This gives the rest of our brains the almost primaries of red blue yellow to work with, and brightness separate is useful too since mixing paint gives opposite results brightnesswise compared to mixing light. Tacking green opposite red is pretty efficient as well if blue is combined with yellow. Anyway, if yellow is pre-calculated early on for the convenience of the rest of the brain, it is no wonder it seems to us a color of its own and not intuitively a mixture.
My, my. Has it come to this?
Must I stoop to "argument from authority" to settle this academic tempest in a teacup about something so obscure, abstract, and theoretical as "color"?
Okay, today's sermon is from the book of Albert:
1. It is a miracle that curiosity survives formal education.
2. Not everything that can be counted counts, and not everything that counts can be counted.
3. As far as the laws of mathematics refer to reality, they are not certain, and as far as they are certain, they do not refer to reality.
4. Reality is merely an illusion, albeit a very persistent one.
5. I am enough of an artist to draw freely upon my imagination. Imagination is more important than knowledge. Knowledge is limited. Imagination encircles the world.
6. I want to know the thoughts of God. Everything else is just details.
7. My religion consists of a humble admiration of the illimitable superior spirit who reveals himself in the slight details we are able to perceive with our frail and feeble mind.
8. Science without religion is lame, religion without science is blind.
Amen
(Wink, wink ;-)
Posted by: Mike on December 22, 2002 04:40 AMach, "Carbon Chauvinism", the very trope that made me begin to actually hate Dennett!
Look guys, Searle never made this claim. He has repeatedly disavowed it and challenged Dennett to substantiate it.
Searle's claim is (in tripartite form):
1. *Anything* could be the substrate for conscious experience, in principle.
2. Because of the CR thought experiment, we know that *whatever* has conscious experiences, *nothing* has them *purely because* of its Turing-machine properties (or any other purely functional properties). This is because a Turing machine is an abstract entity; no physical object is a Turing machine, although some physical objects can be viewed as if they were Turing machines.
3. At present, the only things for which there is any decent evidence at all for consciousness, are brains. THis is an empirical claim about the current state of science.
>>I mean, suppose we construct a huge Chinese room, with one person for each neuron in the brain, and have them communicate with the other "neurons" in exactly the same way in response to exactly the same signals that neurons do: is "thinking" taking place<<
Suppose we construct a huge Chinese room of this thought and have them simulate a gas cooker; will you be able to heat your lunch over it? Maybe, but not because of the functional properties of the abstract syntactic system it instantiates.
You can make the room as big and complicated as you like; fact remains that it _isn't_even_simulating_ anything unless someone chooses to interpret it in that way. As I say above, if such a huge Chinese program happened to call you a wanker, would you take offence?
Posted by: DD on December 22, 2002 08:32 AMOoo, the Dennett-bashing is making me feel all warm and oogly.
A good way to get your loathing on is to go back and reread the outrage Dennett expresses toward Skinner in Brainstorms (which I, too, loved in my salad days), and ask yourself if a similar rant couldn't be written about his late "work." As D2 says, Dennett these days never passes up the opportunity to slyly insinuate that someone who disagrees with him about consciousness or Darwin is in weak-kneed thrall to religion. This sort of sly insinuation is what enraged Dennett about Skinner: "Skinner, then, sees superstition and demonology every time a claim is made on behalf of moral responsibility, and every time a theory seems to be utilizing a homunculus. It all looks the same to him: bad." (57) Also in Brainstorms, Dennett has a brilliant rhetorical analysis of behaviorism's appeal: it doesn't really offer an explanation so much as it suggests that the hidden mechanics are *simple*, and it tends to confuse the smug satisfactions of unmasking the simple for explanation (66). Compare this with Dennett's recent production of simplicities: smug little pop concepts like "skyhooks" and support of little pop concepts like "memes" (At one point Dawkins took such a righteous beating for the incoherence of the gene-meme analogy from philosophers that I think he might have dropped it except for Dennett's lone advocacy as an Actual Philosopher who supported the idea).
For whatever reason, Dennett seems to have become a opportunistic pundit, playing the rhetorical games he used to expose so well in others. The worst part of it is, when he plays them you know that he knows what he's doing. It's the bad faith that's so irksome.
Posted by: T.V. on December 22, 2002 10:52 PMIf you have not read the Dennett piece here, it is worthwhile, at least to the latter parts of this comment thread.
To the anti-Dennett crowd, the argument that he insinuates that those that disagree are relying on religious or quasi-religious beliefs really misses the point. That is the basis of contention, isn't it?
Searle, d-squared, Chalmers, T.V. believe there is some extra something not explained by the physical interactions that we are aware of, and Dennett thinks you are falling into an inherent belief in 'meaning' which is similar to religion. You are not refuting his arguments, just saying that you do not like his technique. Fair enough, but it would go further with some refutation.
To d-squared, would you be upset if a future robot that was created from doing a one to one electronic mapping of a brain scan of Daniel Dennett called you a religious wanker?
>>Searle, d-squared, Chalmers, T.V. believe there is some extra something not explained by the physical interactions that we are aware of<<
Not sure how to parse this sentence. Does the clause "that we are aware of" modify "physical interactions" or "some extra something"? I suspect that a lot depends on this ... I would argue that we are aware of this "extra something" (that it is part of our experience), though I doubt that would convince you. Try this example:
>>You are not refuting his arguments, just saying that you do not like his technique<<
Fair enough, but on this basis, you would have a hard time arguing against someone who was an idealist; who claimed that there was no such thing as a physical world, and that all that existed was sense-data. He would insist that your belief in an objective world was quasi-religious, and basically the whole of philosophy since Berkeley seems to strongly suggest that you would have no better luck coming up with a "refutation" of this argument (in the sense of a logical demonstration of its falsity from agreed premises) than I have with Dennett. At some point, you have to draw a line and refuse to accept explanations which deny the exist of things which seem palpably obviously to exist. To be fair to Dennett, he accepts that this is the case; hence the discussion of "intuition pumps" at the beginning of "Consciousness Explained" (or as it is known to wags, "Consciousness Explained Away").
>>To d-squared, would you be upset if a future robot that was created from doing a one to one electronic mapping of a brain scan of Daniel Dennett called you a religious wanker?<<
No, because I wouldn't think it meant it. Nor would I assign moral guilt to the machine if it punched me in the nose (I might demand that it be dismantled as a dangerous piece of machinery).
A quote giving ambiguous support to both sides on this issue (and perhaps a clue about my own uncomfortableness with Searle's position and sympathy for Dennett's):
"If a lion could speak, we would not be able to understand it" -- Wittgenstein.
Posted by: dsquared on December 23, 2002 09:07 AMd-squared, First, let me say that it is always enjoyable to read your posts, and that I admire your patience, especially with bigoted robots, and with me for coming up with no ideas of my own, but poorly repeating others.
You will agree, I think, that it is in fact possible to replace a neuron in a person with something that simulates that neuron in all functional ways, right? Now lets say we start replacing the neurons in Trent Lott's brain. We start with the ones that light up a brainscan when he is watching Willie Horton attack ads. Eventually, we replace all of the neurons in Trent's head. At what point does he cease 'meaning' his thoughts?
I think at the end of the day all you really have is functionality. If it walks like Trent Lott, talks like Trent Lott, then it means what Trent Lott means.
Regarding the idealist, I simply would not argue with him. My thoughts on that are great enjoy your philosophy, and come back to me when we have a testable hypothesis. I am agnostic if it has no perceivable effect in my world.
Posted by: on December 23, 2002 10:56 AMach, "Carbon Chauvinism", the very trope that made me begin to actually hate Dennett!
Look guys, Searle never made this claim. He has repeatedly disavowed it and challenged Dennett to substantiate it.
Searle's claim is (in tripartite form):
1. *Anything* could be the substrate for conscious experience, in principle.
2. Because of the CR thought experiment, we know that *whatever* has conscious experiences, *nothing* has them *purely because* of its Turing-machine properties (or any other purely functional properties). This is because a Turing machine is an abstract entity; no physical object is a Turing machine, although some physical objects can be viewed as if they were Turing machines.
3. At present, the only things for which there is any decent evidence at all for consciousness, are brains. THis is an empirical claim about the current state of science.
It strikes me that Searle is still being obscurantist. If we accept 1) then his philosophy offers us no help to understand the relation between meaning and material. Nor to help us distinguish in practice which chunks of physical stuff give rise to minds vs. those which only give rise to zombies. Or is he a kind of panpsychist?
2) The CR argues against all functionalism, not just computationalism, so the rest of the stuff about Turing machines is beside the point.
3) But if we reject functionalism we have to return to either Behaviorism or some kind of mapping between brain components and meaning. And isn't this succesfully challanged by examples where different functions are handled by different parts of the brain in different people?
Posted by: phil jones on January 1, 2003 09:45 AM