The author seems unfamiliar with the concept of levels of abstraction, and concludes that information presented in a way other than its rawest, most un-abstracted form isn't really information.
You don't have a photo perfect copy of a one dollar bill stored in your brain? Clearly your brain is not retrieving stored information and making calculations with it.
This author seems to get caught up in overly specific things (your brain doesn't have buffers or models) and misses the forest for the trees (our brains in some way are capable of storing information and the processing it and doing things with the processed information).
Interesting perspective that denies the whole field of cognitive psychology (despite the author being a renowned psychologist himself, albeit in the field of behavioral psychology). But I wonder, if information is going in and information going out, what is it in between if not information processing, aka computing?
Information is nonsense. In our models perfect information and perfect noise are indistinguishable. Usually this is just hand-waved away, but we all know where you go when you start with an incoherent definition –anywhere your imagination can carry you.
One of the major philosophical errors technically educated but philosophically ignorant intelligent persons make is not being aware of the role of the interpretant. Computing "neural networks" have biological neural networks as a necessary precondition. CS Peirce's understanding of semiotic really should be required reading.
Charitably I'd call the tendency to say we are computers (that is to say, we've built machines in our own image) the pathetic fallacy. Less charitably I'd call it idolatry. The dream of building gods is ancient and persistent.
To answer your question, what's going on in the human mind remains a mystery to all honest inquirers.
This is one of the worst straw man arguments I have ever seen. The author makes broad generalizations with no attempt at understanding, and then goes on to make completely ridiculous claims. Honestly made me sick to my stomach to read.
Is it a mistake though to think of people as machines?
Our muscles and joints and bones and tendons and stuff that move us around, even though biological in origin is very much a mechanical machine.
noun:
an apparatus using or applying mechanical power and having several parts, each with a definite function and together performing a particular task.
I don't get why would we suppose our cognitive functions to be any more than the result of physical phenomenon at some level. Especially when we know a little bit about the specifics (i.e drugs, brain surgery etc.).
Just because a machine is complicated or of personal interest doesn't make it any less a machine so I'm not sure this is a mistake.
Perhaps thinking the machine is simple enough to understand or that a frame of reference for objective understanding is possible may be mistakes though.
>Is it a mistake though to think of people as machines?
Yes and no... It is not a mistake because much (and perhaps eventually all) of observable human functionality and behavior can be logically explained and therefore eventually simulated. However, when logic becomes one's "Maslow's hammer" then of course not only are people machines but all of existence is just a giant machine. It's an easy mistake to make because we tend to conflate a very powerful tool used by the mind to operate in reality with reality itself. I think humans are unique (at least on this planet) in their ability to ask this question, but also in their ability to discover that the answer is negative - i.e there is much more to reality than a complex "dead" machine.
The author built a strawman by assuming people who use the metaphor of computation literally mean that the brain has a von Neumann architecture [0] or something similar.
His definition of information is not the information theoretic definition, but one that is narrowly confined to his strawman.
It's just an embarrassingly bad attempt at refutation. I can't imagine how someone of his stature could write something like this when he should be familiar with research like our understanding of how the visual and auditory systems work (literally processing information).
People in technology use the machine metaphor for the human mind literally. See neural networks and even basic terms like Artificial Intelligence (there are no and have never been any intelligent machine or program). This certainly is not a strawman.
Please point me to a technologist who believes the brain to be a von Neumann architecture.
The existence of misunderstandings by lay outsiders does not, in itself, invalidate the use of the metaphor, especially as it's used within the actual research community.
> The author built a strawman by assuming people who use the metaphor of computation literally mean that the brain has a von Neumann architecture [0] or something similar.
But I don't think his target is the neural-net style cognitive science that followed 70s-80s strong AI. It's contemporary neuroscience, which very frequently still models cognition as plain old symbol processing. There are thousands of fMRI-driven papers which do exactly this; neuroscientists still tend to work with a 20th century image of cognition (and a 19th century one of consciousness, but that's another story). This is not a straw man. It's alive and well and driving millions of dollars worth of research.
Exactly. I don't think there is any serious cognitive science who claims that the brain works EXACTLY like a computer. It doesn't and we have understood that for a long time.
But it obviously processes information, or we wouldn't be able to interact with the world, or understand any of our sensory data.
Has this author ever worked with neural networks? It is clear that NNs can store representations, and although they are not the same as data stored on a hard-drive, they are still representational.
This essay seems to ignore the concept of "neural networks" (NN), both biological and artificial.
Now, that doesn't mean that our current artificial representations of neural networks are anywhere close to how biological neural networks work (for instance, as far as I know, we have yet to find anything like "backpropagation" in biological NNs - but if anyone knows differently, I'd be most interested in learning).
But artificial NNs are the closest we've gotten to something that seems to work how brains (biological NNs) work, and how they are structured. We have a multitude (and seemingly ever growing) of models, but there is something lacking - something almost seemingly fundamental.
For instance, we have all of these different models, but none of them are "general" in the sense that there is a single one that can be used for different tasks. Not only that, but there are also a multitude of means for simulating the stimulus/activity/firing of the neurons in the network; for a while the "activation function" was a sigmoid filter, but today best results are found with RELU - then there's the whole concept of "spiking" networks which don't work like such (they seem somewhat "time dependent" from my understanding). There are tons more where those came from.
Now - in this field I don't have much knowledge; I'm no where near an expert, and I am certain it shows. At the same time, based on what I do know, I have this feeling in the back of my head that while what we have built does seem to work ok on certain classes of problems, it isn't the answer.
I think what we've built is overly complex; I think there is an answer out there much simpler. It may be something that needs new hardware. I don't think it will involve GPUs or any other linear algebra methodology (hardware or software). It might likely be one of those "forehead slappers" when we do discover it, whatever it may be.
It's a nagging feeling I have - maybe completely wrong. It just feels like we are in a certain manner trying to build an airplane that flaps its wings to fly, when the answer is ultimately much simpler. Maybe my feelings are false.
> What is the problem? Don’t we have a ‘representation’ of the dollar bill ‘stored’ in a ‘memory register’ in our brains? Can’t we just ‘retrieve’ it and use it to make our drawing?
We can have a representation of a dollar bill, and you can see that the student retrieved that representation: he drew a picture.
Sure, the representation is not exact, and some details have been omitted, but this is true of all representations, even computerized ones. Uber does not store an exact representation of every car; they cannot tell me that there is a dent on the back side of the car picking me up. Yet, my phone relays that a "silver Honda with plates XXXX" is picking me up — a representation of the car, and one good enough to get the job done.
> no image of the dollar bill has in any sense been ‘stored’ in Jinny’s brain.
And how not? The student was, again, able to draw a dollar bill. It wasn't perfect, but neither is a JPEG photo.
> McBeath and his colleagues gave a simpler account: to catch the ball, the player simply needs to keep moving in a way that keeps the ball in a constant visual relationship with respect to home plate and the surrounding scenery (technically, in a ‘linear optical trajectory’).
Perhaps this is how stuff functions, approximately, but it doesn't mean that it is the only means by which the brain could accomplish the task. I can throw and catch objects without looking at them — without a "constant visual relationship with respect to […] the surrounding scenery" — so, how can I do this?
What seems the most likely is that, effectively, memory is like very lossy, probabilistic compression spread with redundancy across a billion analog storage nodes, similar in some ways to a Bloom filter. It’s not binary though, it’s analog. After all, there’s no value in perfectly “unlearning” or destroying a specific memory, just “free” it up by gradual deemphasis of its parts and perhaps it may effectively go away.
Also, the interplay between short-term and long-term memory formation in order to make some approximation of the short-term memory (also an approximation) somewhat permanent.
Granted, recall is imperfect and different every each time.
I enjoyed the article, and always thought it was strange that the people think there is any resemblance between the brain (something beyond comprehension) and some simple man-made tool following basic 1+1=2. I think people are unable to come to terms with their ignorance , indeed, eternal ignorance.
There is plenty of resemblance: the human brain streams data from input devices, transforms the data and produces outputs (in humans, mostly release of chemicals, muscle-flexing and tongue-flapping... oh wait, that's muscle-flexing too).
This is much more obvious in our much simpler-minded cousins, such as insects: turn on a gas lamp and watch those suckers fly straight towards their doom. Basically, they're little more than clever little automatons designed by nature.
If you are of the opinion that behaviors like this don't occur in the oh-so-complex-beyond-all-understanding humans, I remind you of the millions of Americans suffering from disease and early death because there's too much food around the place. Or maybe I'm ignorant and the reason for epidemic obesity shall eternally remain beyond comprehension, I don't know.
This reminds me of a great snippet from one of Raymond Smulyan's books ("This Book Needs No Title") that I think best captures the implied and unfounded fear behind these sort of articles. Here's the quote in full:
“Recently I was with a group of mathematicians and philosophers. One philosopher asked me whether I believed man was a machine. I replied, ‘Do you really think it makes any difference?’ He most earnestly replied, ‘Of course! To me it is the most important question in philosophy.’ I had the following afterthoughts: I imagine that if my friend had finally come to the conclusion that he were a machine, he would be infinitely crestfallen. I think he would think: ‘My God! How horrible! I am only a machine!’ But if I should find out I were a machine, my attitude would be totally different. I would say: ‘How amazing! I never before realized that machines could be so marvelous!’”
Adding my own thoughts to this, I think what generates these sort of articles is an intuitive and well justified resistance to the belief is "life is just a complex machine" but by using completely mistaken arguments.
The essence of life, existence and consciousness can never be reduced to any mechanical or even logical processes (although some would like to believe otherwise), but the fact that it can bloom into amazing creations like the human mind or possibly in the distant future ti some human-made artificial mind doesn't detract from it's magic.
> What is the problem? Don’t we have a ‘representation’ of the dollar bill ‘stored’ in a ‘memory register’ in our brains? Can’t we just ‘retrieve’ it and use it to make our drawing?
This article makes some silly points. Why didn't the student draw a minotaur, or some other thing, when they were asked to draw a dollar bill? There is a clear correspondence between the drawing and the actual dollar bill.
25 comments
[ 3.8 ms ] story [ 69.9 ms ] threadThe author seems unfamiliar with the concept of levels of abstraction, and concludes that information presented in a way other than its rawest, most un-abstracted form isn't really information.
You don't have a photo perfect copy of a one dollar bill stored in your brain? Clearly your brain is not retrieving stored information and making calculations with it.
This author seems to get caught up in overly specific things (your brain doesn't have buffers or models) and misses the forest for the trees (our brains in some way are capable of storing information and the processing it and doing things with the processed information).
Consider me unconvinced.
One of the major philosophical errors technically educated but philosophically ignorant intelligent persons make is not being aware of the role of the interpretant. Computing "neural networks" have biological neural networks as a necessary precondition. CS Peirce's understanding of semiotic really should be required reading.
Charitably I'd call the tendency to say we are computers (that is to say, we've built machines in our own image) the pathetic fallacy. Less charitably I'd call it idolatry. The dream of building gods is ancient and persistent.
To answer your question, what's going on in the human mind remains a mystery to all honest inquirers.
Our muscles and joints and bones and tendons and stuff that move us around, even though biological in origin is very much a mechanical machine.
noun: an apparatus using or applying mechanical power and having several parts, each with a definite function and together performing a particular task.
I don't get why would we suppose our cognitive functions to be any more than the result of physical phenomenon at some level. Especially when we know a little bit about the specifics (i.e drugs, brain surgery etc.).
Just because a machine is complicated or of personal interest doesn't make it any less a machine so I'm not sure this is a mistake.
Perhaps thinking the machine is simple enough to understand or that a frame of reference for objective understanding is possible may be mistakes though.
Yes and no... It is not a mistake because much (and perhaps eventually all) of observable human functionality and behavior can be logically explained and therefore eventually simulated. However, when logic becomes one's "Maslow's hammer" then of course not only are people machines but all of existence is just a giant machine. It's an easy mistake to make because we tend to conflate a very powerful tool used by the mind to operate in reality with reality itself. I think humans are unique (at least on this planet) in their ability to ask this question, but also in their ability to discover that the answer is negative - i.e there is much more to reality than a complex "dead" machine.
His definition of information is not the information theoretic definition, but one that is narrowly confined to his strawman.
It's just an embarrassingly bad attempt at refutation. I can't imagine how someone of his stature could write something like this when he should be familiar with research like our understanding of how the visual and auditory systems work (literally processing information).
[0] https://en.wikipedia.org/wiki/Von_Neumann_architecture
The existence of misunderstandings by lay outsiders does not, in itself, invalidate the use of the metaphor, especially as it's used within the actual research community.
But I don't think his target is the neural-net style cognitive science that followed 70s-80s strong AI. It's contemporary neuroscience, which very frequently still models cognition as plain old symbol processing. There are thousands of fMRI-driven papers which do exactly this; neuroscientists still tend to work with a 20th century image of cognition (and a 19th century one of consciousness, but that's another story). This is not a straw man. It's alive and well and driving millions of dollars worth of research.
But it obviously processes information, or we wouldn't be able to interact with the world, or understand any of our sensory data.
Has this author ever worked with neural networks? It is clear that NNs can store representations, and although they are not the same as data stored on a hard-drive, they are still representational.
Now, that doesn't mean that our current artificial representations of neural networks are anywhere close to how biological neural networks work (for instance, as far as I know, we have yet to find anything like "backpropagation" in biological NNs - but if anyone knows differently, I'd be most interested in learning).
But artificial NNs are the closest we've gotten to something that seems to work how brains (biological NNs) work, and how they are structured. We have a multitude (and seemingly ever growing) of models, but there is something lacking - something almost seemingly fundamental.
For instance, we have all of these different models, but none of them are "general" in the sense that there is a single one that can be used for different tasks. Not only that, but there are also a multitude of means for simulating the stimulus/activity/firing of the neurons in the network; for a while the "activation function" was a sigmoid filter, but today best results are found with RELU - then there's the whole concept of "spiking" networks which don't work like such (they seem somewhat "time dependent" from my understanding). There are tons more where those came from.
Now - in this field I don't have much knowledge; I'm no where near an expert, and I am certain it shows. At the same time, based on what I do know, I have this feeling in the back of my head that while what we have built does seem to work ok on certain classes of problems, it isn't the answer.
I think what we've built is overly complex; I think there is an answer out there much simpler. It may be something that needs new hardware. I don't think it will involve GPUs or any other linear algebra methodology (hardware or software). It might likely be one of those "forehead slappers" when we do discover it, whatever it may be.
It's a nagging feeling I have - maybe completely wrong. It just feels like we are in a certain manner trying to build an airplane that flaps its wings to fly, when the answer is ultimately much simpler. Maybe my feelings are false.
We can have a representation of a dollar bill, and you can see that the student retrieved that representation: he drew a picture.
Sure, the representation is not exact, and some details have been omitted, but this is true of all representations, even computerized ones. Uber does not store an exact representation of every car; they cannot tell me that there is a dent on the back side of the car picking me up. Yet, my phone relays that a "silver Honda with plates XXXX" is picking me up — a representation of the car, and one good enough to get the job done.
> no image of the dollar bill has in any sense been ‘stored’ in Jinny’s brain.
And how not? The student was, again, able to draw a dollar bill. It wasn't perfect, but neither is a JPEG photo.
> McBeath and his colleagues gave a simpler account: to catch the ball, the player simply needs to keep moving in a way that keeps the ball in a constant visual relationship with respect to home plate and the surrounding scenery (technically, in a ‘linear optical trajectory’).
Perhaps this is how stuff functions, approximately, but it doesn't mean that it is the only means by which the brain could accomplish the task. I can throw and catch objects without looking at them — without a "constant visual relationship with respect to […] the surrounding scenery" — so, how can I do this?
Also, the interplay between short-term and long-term memory formation in order to make some approximation of the short-term memory (also an approximation) somewhat permanent.
Granted, recall is imperfect and different every each time.
Still, one can memorize and recall some things with almost perfect accuracy (at least over time), for example phone numbers.
This is much more obvious in our much simpler-minded cousins, such as insects: turn on a gas lamp and watch those suckers fly straight towards their doom. Basically, they're little more than clever little automatons designed by nature.
If you are of the opinion that behaviors like this don't occur in the oh-so-complex-beyond-all-understanding humans, I remind you of the millions of Americans suffering from disease and early death because there's too much food around the place. Or maybe I'm ignorant and the reason for epidemic obesity shall eternally remain beyond comprehension, I don't know.
“Recently I was with a group of mathematicians and philosophers. One philosopher asked me whether I believed man was a machine. I replied, ‘Do you really think it makes any difference?’ He most earnestly replied, ‘Of course! To me it is the most important question in philosophy.’ I had the following afterthoughts: I imagine that if my friend had finally come to the conclusion that he were a machine, he would be infinitely crestfallen. I think he would think: ‘My God! How horrible! I am only a machine!’ But if I should find out I were a machine, my attitude would be totally different. I would say: ‘How amazing! I never before realized that machines could be so marvelous!’”
Adding my own thoughts to this, I think what generates these sort of articles is an intuitive and well justified resistance to the belief is "life is just a complex machine" but by using completely mistaken arguments.
The essence of life, existence and consciousness can never be reduced to any mechanical or even logical processes (although some would like to believe otherwise), but the fact that it can bloom into amazing creations like the human mind or possibly in the distant future ti some human-made artificial mind doesn't detract from it's magic.
The human brain is not a digital computer, but it does process information and store memories.
This article makes some silly points. Why didn't the student draw a minotaur, or some other thing, when they were asked to draw a dollar bill? There is a clear correspondence between the drawing and the actual dollar bill.