Philosopher Colin McGinn makes a real hash of reviewing How to Create a Mind: The Secret of Human Thought Revealed, by by Ray Kurzweil. I can’t comment on the book itself, because I’ve not read it. But there’s enough wrong with the content of McGinn’s to know that he isn’t the person who should be reviewing it. Well, not for a fair critical review. If the New York Review of Books merely wanted clueless philosopher to stir up some hyperbole, then fair enough.
But McGinn is one of those irritating philosophers that seems wilfully ignorant. So I’ll respond to some of the content of the review, which is here.
Let’s start with the obviously philosophically suspect:
“However, that claim seems obviously false.”
It doesn’t matter what the specific claim is of Kurzweil. When you hear that from a philosopher, that something is obvious, think ‘mind already made up’. Though his review may be critical, it contains little critical thinking. I always thought it was obvious that philosophers challenged the obvious. I’m obviously wrong in that assumption, and I’ve been shown to be wrong by several philosophers, thereby showing again that the obvious isn’t always so.
McGinn’s main gripe seems to rest on his misunderstanding of patterns and pattern recognition. From the computer science perspective I can see where Kurzweil might be coming from. McGinn starts off badly and never picks up.
“Pattern recognition pertains to perception specifically”
No it does not. Pattern recognition is about matching patterns, and they don’t even have to be in the same encoding. Patterns of visual perception need be nothing like the corresponding neural content; but nevertheless, similar visual patterns can stimulate similar neural firing patterns. There can be a correspondence between visual patterns as perceived and patterns in neural circuits, synaptic connections, firing. It’s not a simple one-to-one correspondence, because the brain has history and merges many sources of pattern.
The pattern of an image taken by a digital camera will be persisted in the camera in states of transistors, and they need have nothing like the geometric spatial correspondence of the original 2D representation of the 3D scene. In transmitting the image over the internet the image data may be compressed and encrypted beyond all recognition (by humans). The compressed encrypted pattern is still a pattern with a direct correspondence to the original geometric image (in lossless compression). And at the other end the data may be decrypted, decompressed and displayed in the same geometric pattern as the original. This would not be possible if there was no correspondence, no pattern, being preserved along the way.
The internal brain representations need not have the same lossless fidelity I’ve just described. It appears that brain memories and perceptions are very dynamic and not at all the high fidelity representations we subjectively feel they are. Nevertheless, patterns are what it’s about to a great extent.
The same can apply to audio signals. And vibrations from earthquake monitors. Things we humans can’t detect can be transformed into patterns we do recognise, so we see patterns in graphs of data that we couldn’t see just observing the original source. So, though stock market data comes in from all over the world to a computer system, and though there is no way humans could directly detect patterns in that data, it can be transformed into stock market graphs that we understand the meaning of very well: patterns.
Pattern recognition is so much at the heart of how we put contextual meaning to visual stimuli that we have the phenomenon of Pareidolia. We ascribe meaning where there is none when our pattern recognition is overly active.
You only have to watch this TED video to see the importance of patterns.
McGinn misses the big picture. In the above video the context of a pattern is in itself a greater pattern. What we perceive moment by moment is only a fraction of what we experience and learn over time. The momentary visual perceptual patterns are not the full extent of the patterns being recognised. In the video the different meanings of two identical geometric shapes only acquire their respective meaning in the context of the surrounding patterns.
“In what way does thinking involve processing a stimulus and categorizing it?”
Thinking isn’t limited to that. Thinking is the dynamic on-going processes in which many partly autonomous regions of the brain are doing their own thing, and we perceive that as thinking. The fact that I have to use the phrase “we perceive that as” is the unfortunate teleological language we have become accustomed to, which may even be evolved as part of our evolutionary development of language and its use to identify a self. That we have this limitation in language use is no excuse for McGinn to make naive claims as to what is obvious – claims to the obvious discredit the whole of philosophy.
“When I am thinking about London while in Miami I am not recognizing any presented stimulus as London-since I am not perceiving London with my senses.”
But McGinn is way off the mark here. The stimulus in this case is not coming from the eyes, but from the brain. When his brain starts to think of London (whether in London or not) it reconstructs in the visual cortex the patterns of neural activity that are activated when he sees London. When he thinks of London, and in particular Big Ben, say, then the various experiences of Big Ben he has had, directly, or on the TV News, or on post cards, have all contributed to his brain’s learning of what London means to his brain, so that just thinking about London conjures up these internal experiences again, some as pseudo-visual images.
“There is no perceptual recognition going on at all in thinking about an absent object.”
There is both perceptual reconstruction, and subsequent perceptual recognition.
“This point seems totally obvious and quite devastating…”
For heaven’s sake, stop with the ridiculous claims to the obvious. If all this was obvious there would be no debate! You want to witness devastation of an idea? Try Dennett’s 1991 review of McGinn – see later.
“The notion of “pattern” has lost its moorings in the geometric models of letters and faces: Are we seriously to suppose that dreams and thoughts have geometrical shape?”
No! This is dreadful. Geometric patterns (by which McGinn means spatially geometric) are only one type of pattern. There are non-spatial geometries, patterns.
“At best the word “pattern” is now being used loosely and metaphorically…”
No. It’s just that McGinn doesn’t get the actual broader meaning of ‘pattern’. It is McGinn’s limited knowledge that is holding him back in this respect.
“Why is remembering that I have to feed the cat itself some kind of pattern?”
Because it is a pattern of neural activity that has evoked conscious awareness of something that must be done, and with it are associated patterns of various kinds, only some of which may be related to visual experiences, of cat and food. When McGinn remembers again the next day many of the same neurons will fire again, in similar patterns!
“What has happened is that he has switched from patterns as stimuli in the external environment to patterns as mental entities, without acknowledging the switch”
Maybe because it is well known that patterns are everywhere in all this and the external patterns on the retina are just one of the many patterns. Without reading Kurzweil I couldn’t say what his intention was, but my guess would be that he didn’t feel the distinction would be necessary to make, or that he made it but McGinn missed it.
“… blithely proceeding as if everything mental involves perception”
Well yes and no. It’s not all visual perception; and it’s not even all conscious perception. Are we restricting the use of the term perception to only the external events on the retina? Or beyond and down the optic fibres? Or through the parts of the visual cortex? Or do we include the reconstructions and merging of immediately acquired and stimulated reconstructions? It’s all so complicated that McGinn’s use of the term perception is very limiting; so showing how little McGinn actually knows about any of this.
“When I see an apple as red, do I recognize the color as a pattern? No, because the color is not a geometric arrangement of shapes or anything analogous to that…”
What? Is he serious? Geometric? Colour is not a pattern system? Does he not know what spectra are? This is unbelievably ignorant. My guess is that his understanding here is based on ancient philosophical notions that he’s not been able to shake off. It’s as if 200 years of science have passed McGinn by.
“When I see an apple as red …it is simply a homogeneous sensory quality”
He’s even confusing himself now. When he sees a red apple he isn’t merely seeing ‘red’, as if the old philosophical ‘redness’ is what determines colour perception. He should watch the Beau Lotto TED talk.
“Is the sweetness of sugar or the smell of a rose a pattern?”
Yes! The nose and tongue are chemical pattern recognition components. When they detect chemical patterns they fire off neural signals. The pattern of the periodic table of elements is a representation, a model, of the physical patterns of the different elements. And more complex atomic patterns, as molecules, stimulate specific receptor patterns in nose and tongue. The whole of biology is about patterns. DNA is about patterns, pattern recognition, pattern building.
“Then there are such mental phenomena as emotion, imagination, reasoning, willing, intending, calculating, silently talking to oneself, feeling pain and pleasure, itches, and moods-the full panoply of the mind.”
Emotion is very much pattern recognition. We recognise patterns in the world and they stimulate our emotions. It’s because the stimulating patterns are different that we cry at sad events and laugh at funny ones. Imagination is the construction of patterns, of worlds, images, ideas. Reasoning is the attempt to discover and build patterns of logic. Willing (don’t get me started on free will) is the causal effect of internal patterns that drive behaviour, that make motor neurons drive muscles. Talking to one’s self is the use of language patterns internally.
This is all getting to be too much ignorance. I’m beginning to suspect wilful denialism.
“In what useful sense do all these count as “pattern recognition”?”
It’s not just recognition. It’s also pattern creation and reconstruction. We learn by both experiencing the world, but also by probing it, and it is in this way we build patterns that on the whole start to have a related contextual significance; and therein lies meaning.
Eventually McGinn moves on to the teleological languages used in neuroscience.
“There is another glaring problem with Kurzweil’s book: the relentless and unapologetic use of homunculus language.”
Three points here. First, that seems to be the language we are stuck with. Maybe it’s evolved – evolving with us, so that it is a very part of our nature to think teleologically. It certainly seems very difficult to escape. Second, coming to us so easily it’s very easy to use, and therefore convenient and efficient: evolutionary biologists that think we are physical systems and that DNA has no intellect or purpose of any kind that we might associate with agency, they still use teleological language. Third, the first two points explain why people like McgGinn find it difficult to shake off the notion of teleology, particularly with regards to humans, and why when they see teleological language used as metaphor they take it to be literal. In this case McGinn thinks Kurzweil’s use of such language is literal. There is no homunculus!
“Presumably (I am not entirely sure) Kurzweil would agree that such descriptions cannot be taken literally: individual neurons don’t say things or predict things or see things-though it is perhaps as if they do.”
He’s not entirely sure? Then he really doesn’t get it at all. He is missing the whole point of the absence of teleology that physics, chemistry, biology, evolution has been hinting at.
“People say and predict and see, not little bunches of neurons, still less bits of machines.”
Well, since people are big bunches of neurons then it is the neurons doing it, but not as some thinking homunculi. It’s the whole mass of neurons and their behaviour that gives the impression of a person, a self. This is the mental illusion that hides the physicalism of what we are! The very ‘self’, ‘I’, that we feel we are is a living metaphor – we are our own homunculus.
“First, homunculus talk can give rise to the illusion that one is nearer to accounting for the mind…”
It’s a metaphor!
“But if we strip our theoretical language of psychological content, restricting ourselves to the physics and chemistry of cells, we are far from accounting for the mental phenomena we wish to explain.”
Now I agree we are far from accounting for much. But if only the likes of McGinn, and Nagel, Chalmers and Tallis, could apply their critical scepticism to their own ideas of what conscious is; or even more, apply it to their certainty of what it cannot be, then they’d be on reasonable ground. But that isn’t what this is about. For these guys it isn’t about a fair appraisal of how far science is getting on in the understanding of consciousness, because they don’t seem interested in the science. They seem more intent on maintaining the special nature of humans as non-animals, or at least animals with minds that are some way free of the very physical constraints of the brain. Oh yes, they admit the mind has something to do with the brain, and while not being able to say what the mind has to do with the brain they are very keen to say to what extent the brain is not the mind. These guys are philosophers. They don’t spend the time covering the science to appreciate how far it goes in not finding a mind and how far it sees only evidence consistent with physicalism.
“Neurons simply emit electrical impulses when caused to do so by impinging stimuli; they don’t recognize anything in the literal sense.”
They do! In a literal sense. The problem here is that McGinn is restricting ‘recognise’ to the teleological homunculus kind that has traditionally been associated with mind. When a component, be it a neuron or computer chip, responds to some patterns and not to others then there is a real and literal physical sense in which the input pattern has been recognised.
“Recognizing is a conscious mental act.”
Only if you define it as such. And here we find the limitations of McGinn. He defines recognition as something that only conscious systems can do; and surprise: only conscious system can do recognition, when so defined! He is making it so by his own definition, his own restriction of the use of the word. Now that human teleological meaning may be the source of the word, but it is now a well-used word in many sciences, and especially in computer science. Much language has been adopted and does not retain its teleological significance. I don’t suppose McGinn has much problem with the term ‘memory’ when applied to computers.
The last bit of interest is McGinn’s understanding of signal processing and information processing. He lacks understanding.
” in sober neuroscience textbooks we are routinely told that bits of the brain “process information”, “send signals”, and “receive messages” – as if this were as uncontroversial as electrical and chemical processes occurring in the brain. “
It is uncontroversial. Well, maybe it’s as controversial to McGinn as is Evolution to Young Earth Creationists.
“It is a collection of biological cells like any bodily organ, much like the liver or the heart, which are not apt to be described in informational terms.”
You’ve got to be kidding me! First, the liver. It does process signals. Mostly chemical ones. The heart? What on earth does he think drives heart beats? And neurons? Please! Neurons are all about signal processing. Even a simplistic description has them summing input signals before they decide to fire; transmitting neurochemicals in accordance with other factors in compex signalling from one neuron to another; they process signals to learn, to habituate, inhibit, etc. Please, McGinn, learn some biology.
“It can hardly be claimed that we have observed information transmission in the brain, as we have observed certain chemicals; this is a purely theoretical description of what is going on.”
Where has McGinn been this last hundred years? Try looking up Santiago Ramón y Cajal. Better still read this.
Still not convicned? Try this list. Seriously, go down this list and just look at all the indicators of signal processing, information processing. And don’t forget it is chemical too, not just electrical.
“The answer must surely be that the brain is causally connected to the mind and the mind contains and processes information.”
What mind?!! Where is it? What is this mind he is so stuck on? He dismisses all evidence of the pattern recognition features of brains, and of their signal processing operations, but with no evidence whatsoever insists there is a mind?
“That is, a conscious subject has knowledge, memory, perception, and the power of reason – I have various kinds of information at my disposal. No doubt I have this information because of activity in my brain, but it doesn’t follow that my brain also has such information”
What is this I, this conscious subject? He looks for it but never demonstrates it.
“To create a mind one needs at a minimum to create consciousness, but Kurzweil doesn’t even attempt to describe a way for doing that.”
Well, first, that’s not true. It appears the minimum we think is necessary is a brain. We have not yet seen consciousness outside a brain that we know of. It’s quite possible that we don’t understand what consciousness is; so that it may turn out to be nothing at all but behaviour of a complex brain. There is nothing else on offer. So, it’s quite respectable to suppose that as we come to understand more about the brain we will be able to figure out how consciousness comes about, and then build non-wet conscious systems. We already attribute consciousness to many non-human animals that aren’t even close to having some of the capabilities McGinn points out in humans. The trouble for McGinn is that he doesn’t understand or cannot define consciousness either, and so is in no position to decide how close scientists have come to it. This next statement demonstrates his ignorance:
“Clearly, unconscious processes of so-called “pattern recognition” in the neocortex will not suffice for consciousness, being precisely unconscious.”
I could write another whole post on why McGinn is wrong in his assessment of physical laws and the technological limits computing power. But I’ve had enough.
Is McGinn truly a philosopher? Has he no critical skills at all? Is he so poorly out of date, still sticking to the same thinking he had way back when? Here’s Dan Dennett on McGinn back in 1991:
“I find his thesis not just incredible and ludicrous. As a fellow philosopher, I find it embarrassing. It is not that I disagree with McGinn about the possibility in principle that there are phenomena that will forever defy human understanding, but just that I find him arriving at his pessimistic verdict about consciousness after such a paltry canvassing of the opportunities.”
What was Dennett disappointed in? McGinn’s claim that human brains were never meant to understand consciousness. Well, they were never meant to understand chemistry, maths of any kind, or for that matter, philosophy. The intellectual capabilities of our brains seem to go way beyond what they were ‘meant’ to do (as if they were ‘meant’ to do anything – be as wary of ID language as of teleological language).
McGinn thinks we do not have the capacity to understand consciousness? Well then how does McGinn assure himself that he knows enough about consciousness, or human brains, to be able to make that claim. It’s a self-refuting claim. Not, note, that this makes the opposite affirmed, that we will understand consciousness. Just because McGinn’s brain is not wired for understanding consciousness, or some basic principles of critical thinking, does not mean that the subject cannot be understood.
Really, you should read the rest of Dennett’s review. I’ll give up on McGinn as a lost cause.
12 thoughts on “Colin McGinn – Hopeless”
I decided to subscribe to your blog.
In any case, what you discuss is way over my head, but some philosophers have been very critical of the New York Review of Books treatment of philosophical topics in general.
Here is one of Brian Leiter’s comments.
Leiter has criticized that publication on other occasions, which you can find with a Google search, I’m sure.
Congratulations on an excellent blog.
Thanks for the link. I had seen complaints about NYRB but hadn’t taken much notice. It was only that I saw this one was by McGinn, and knowing some of his thoughts on consciousness gave it a read. I soon lost any hope of reading a good critique of Kurzweil’s book.
Nothing to add but a round of applause. Well done.
Welcome. Just popped over to Dead Voles and caught your posts on Nagel and “Current State of Play”. Damn! Another must-read blog to add to my list.
Thanks! I haven’t been able to blog much lately, but my big dream is to write a full-fledged guide to neural networks aimed at philosophers and non-computer people. I’m hopeful that the chasm of understanding you so usefully described in this post can be bridged.
I’m not a Kurzweil fan. I tend to agree with McGinn, that Kurzweil overhypes a lot. However, apart from that point, McGinn get most of it wrong. I agree with most of your criticism of McGinn.
On the other hand, I can’t say that I am surprised. McGinn has a reputation as a mysterian, so I expected him to find fault, even if fallaciously, with attempts to explain what McGinnwould rather keep as mysterious.
I sort of know Kurzweil from the software world, and Google of course. But his other stuff does sound a bit pie in the sky. Maybe I misjudged NYRB and McGinn was just the man for the job.
I have a lot of interest in this topic. Have just bought the book and have not started it yet.
I think the process has already started and will only continue to improve, much in the same digital technology has evolved, where originally it was very crude and pixelated and now the products of digital technology (picture/sound,etc) it virtually undiscernable from its analog & real life counterpart.
Todays robots and computers appear quite artificial and “pixelated” when it comes to trying to mimic a conscious entity…but I believe the progress will be unstoppable and quite impressive, until it gets to the point where an AI machine will seem “conscious” and “self-aware” to the average layman observer and will also appear to have free will. I have some co-workers that seem to be zombie like and half brain-dead..so of what use is it that they are the real thing??
“much in the same way that digital technology has evolved”—
First, I happened to drop into the McGinn critique and was shocked. Then I found the counter-critique, and my day was saved.
btw, I found it very funny to compare implicitly a (wired) telephone line with the brain, both being just passive “things”! It’s funny to compare wire with the complex system “brain” (in total contrast to his article), and, last not least, today phone lines are also computers. And they know very well the information they transport (see NSA),
Hi Walter, welcome.
A neuron is indeed something like a wire. It carries a signal. It has many end points that connect to other neurons, as do wire systems. A neuron is more complex than a single wire, and a brain more complex than any computer system we’ve managed to build so far.
The active/passive dichotomy is a tricky one that depends on one’s perspective.
A wire is passive in the sense that it merely carries any signal pushed down it. But a transistor is considered an active device in electronics, in that it can take a small input signal and amplify it. In some respects a single neuron is active in this sense too, and can be modelled to some extent by logical gates, which in turn can be modelled (and implemented using) transistors.
A computer system is passive in the sense that a dead computer does nothing other than respond to what is put into it: power, data. And of course a program; but a program is nothing more than a specific set of data that the computer responds to. So, a passive computer can be considered an active system once it is up and running and doing things.
A brain consists of passive elements and molecules. Unlike a computer a brain doesn’t have to wait to be programmed, since it’s programming begins as the brain develops and continues as the brain takes in data and learns. The passive chemistry becomes active biology, becomes an active brain.
The active/passive dichotomy is one that has crept up on us for millennia, and possibly from way back deep into our evolutionary past. We are just a bunch of passive molecules.
But we and other animals have evolved to recognise different sorts of bunches of passive molecules. Some bunches of molecules we think of as part of the surroundings, as inanimate objects. But stuff that seems to move with its own motor action we see as special. Most animals will recoil at seeing a moving object, because it might be a danger. Humans can even mistake a stick among the leaves on the ground for a stationary snake, so effective is our detection mechanism for other special bunches of molecules.
Humans have conceptualised this automatic evolved classification of bunches of molecules. We think of living things as being active, and non-living things as passive. Plants are sort of considered somewhere in between, mainly because they don’t move about so much I guess.
It’s only been a few centuries that humans have had a chance to play with automata that are not life forms. We started building mechanistic automata, but now we have computers and rudimentary robots. The distinction between life and non-live, passive and active, is now quite a vague one. And it confuses the likes of McGinn.