Is my brain analog or digital?
Letters The suggestion that we think in analog, not digital, inspired you to one of the most interesting postbags of the year. An emergent type at Cornell University, Dr Spivey, suggests that the mind "should be thought of more as working the way biological organisms do: as a dynamic continuum, cascading through shades of grey." Neuroscientist Dr Bill Softky was on hand to pour a little scorn on the claims.
Was Spivey talking about the conscious mind, or the brain, you ask. It's another example where the Tower of Babel we call the internet isn't shy about offering an opinion. So leaving aside the "don't knows" and the Lizard People, here are a selection of your views.
All right, let's end this moronic debate over analog vs digital right now: The world is NOT analog. It is digital. Ask any quantum physicist if there is anything in this universe that cannot be reduced to a collection of specific quantum particles. To better illustrate this point, let us turn to that champion of analog examples, the magnetic tape. This is always put forth as the "analog" data storage system. But what is magnetic tape? It is a ribbon of material along which magnetic _particles_ are arranged to store data. Can you store data on this tape that is represented by 1/2 a particle? No, the "analog" magnetic tape is actually digital, its "digit" being the magnetic particle. So-called analog systems are actually digital systems whose intrinsic level of precision is beyond that which we can (or wish to) control or observe. That does not make them "not digital", it makes us "not capable".
The Siegelmann stuff, at least as presented, is just plain embarrassing.
The argument that real-valued dynamical systems can exhibit behaviour that cannot be simulated accurately (chaos etc etc) by a digital computer is
b) long way from being a convincing argument that cognitive processes must be such dynamical systems.
c) Very shaky ground if you're talking about physical systems - quantum mechanics: physical systems are NOT classical (real-valued) dynamical systems.
Give me the quantum computing folks any day. At least their physics and mathematics is solid even if I haven't seen a convincing analogs to the Turing notions of machine composition / universal machine in their work (but, hey, I haven't looked that hard).
Emergency exitYou were almost able to find the truth in all this, but then you got distracted. The resolution to the analog vs digital problem is simple: stop trying to expect a device to be either one or the other. Especially when using these words more liberally to express the concepts of "discreteness" vs "continuity," you must realize that a device can easily operate in one way on one scale, and in another on a different scale. '
Take the computer. Each transistor and circuit behaves discretely. Yet the transistors are made of parts--seas of electrons--which are actually very analog. In fact, they're as non-discrete as you can get. Yet, the emergent behavior of this set of continuous entities is, overall, discrete. Similarly, if you use these circuits to build a program, you can create something that ends up behaving continuously. The mass behavior of these discrete elements combines to create an emergent behavior that is continuous (on a certain scale). We see both of these occurances--continuous things attaining discrete emergent properties, and discrete things having continuous behaviors--all the time: in nature, in our inventions, in everything. So it is with our brain. Its neurons may be discreet (although, just because they fire pulses, doesn't mean they are. A lot of probabilistic mixing occurs in the lead up to the decision on whether or not to fire), but its overall behavior may be continuous and analog.
To me, at least, the continuous analog-like behavior of the mind seems rather obvious. This is also why Michael Spivey uses the word "mind" and not "brain." He is specifically trying to discuss the higher-level behavior of the brain, not its operation on the scale of neurons. In summary, to say that an object is either all analog or all digital is almost always inaccurate. In fact, the duality of discreetness and continuity is inherent in the very physics that governs the behavior of all things.
But lastly, i must point out that technical (or maybe just original) meaning of the word "analog" does not at all mean non-discrete. Instead, it means that the device works by mimicking (being an analog) of something else. For example, a cassette tape records a magnetic resemblence to a sound wave. This definition actually poses a more interesting question--whether the mental patterns of the brain resemble what they are processing. I mean, obviously on some level they will, but at issue is how much encoding and abstraction is applied. This question has very important implications for both our understanding and for any mind-interfacing technology we may wish to produce.
What I think your article highlighted well is that the cognitive and neural sciences need more understanding of each other's fields. To be honest I don't know what all the fuss is about; if you consider the brain 'the hardware' and the mind 'the software' the computer analogy still stands up. There is plenty of software running on digital hardware that simulates analog behaviour.
Hi Andrew "Dr Softky says he agrees with much of what Professor Spivey values" comes as no great surprise. Dr Softky says the brain is digital, which is obvious, Prof Spivey is talking about the mind being analogue, which, since it spends its time interpretting and interacting with an analogue world, is hardly surprising either. Both propositions are reasonable
Fascinating topic, and you have dug up some excellent comment. I especially liked Softky's "accumulation of bullshit" aphorism, which has much wider applications.
(It pretty much takes care of religion, economics, sociology...)
The meaning of life
I have just read your review of Spivey’s recent PNAS paper. To disclose my interests upfront, I was a graduate student of Spivey’s and am now a cognitive psychologist working at Stanford University.
I entirely agree with your sentiment that many large scale theories of the brain are put forward nowadays with little regard for the constraints of empirical evidence or logical parsimony. And you make a good point that any form of computer analogy (digital or analogue) has limitations that may cloud our understanding of brain function. I am entirely baffled, however, by your virulent tone you take against Professor Spivey. You make the very unpleasant implication that Spivey is not a “real” scientist, and sneer at his empirical methods.
However, your only actual criticism of his experiment is that in your opinion 42 subjects is not enough. You do not say why you take issue with the statistical standards of this paper that are common to almost all behavioural research papers. The experiment ran 42 subjects and, following all the accepted statistical criteria of our field, rejected the null hypothesis. This evidence was enough to satisfy the peer reviewers of the Proceedings of the National Academy of Sciences (and of Science and other journals Spivey has published in). But you weigh the opinions of all those scientists and the statistical standards of the field against your own feeling that 42 is kinda not enough people, and conclude that on balance, you must be correct, and therefore Spivey is not a real scientist.
Your other criticisms appear to be that the conclusion that ‘the mind is not like a computer’ is not sufficiently supported by the evidence, and that Spivey does not explain terms such like ‘dynamic representations’ clearly. These are interesting points to discuss. But to damn Spivey research because it doesn’t explain every concept or offer irrefutable proof for a metaphorical understanding of the mind, is to seriously misunderstand what a scientific paper is. This evidence is only an incremental piece of a large body of evidence and theory: that’s why these papers have citations. If you want to understand what ‘dynamic representations’ are you have to do some reading. Similarly, the bold claim that the mind is not like a computer is an interpretation of a large amount of research, not the unassailable conclusion one draws from the behaviour of some Cornell undergraduates. This is disappointing, since there many interesting comments and ideas raised in your article.
Why didn’t you simply present an argument about these issues? Your piece contains ill-informed, nastily ad hominem comments that have little or no scientific content. Frankly, it appears as if you only read the press release about the article. Judging people by what PR officers write and your own opinions seems like pretty shoddy journalism, and is certainly nowhere near “real science”. Yours, Dr. Daniel Richardson.
Seems to me that the research proves nothing about how brains work, but more about the fact that the human brain detects differences by searching for the obvious first, then the subtle later, meaning subtle differences take longer to detect using the brain's algorithms. Whether those are implemented in an analogue or digital fashion can't really be determined just by working out how the brain works out differences, because even if it were digital, we already know humans are good at approximations. Personally, I think you'll only ever get algorithms from asking people to do things. Analogue / digital is going to require someone playing at "brains in jars", although where we'll find the volunteers from is anyone's guess... Simon Collins.
Many thanks all. In the instant poll at the end of the original article we asked if you wanted more analog. 97 per cent said yes - so consider us bidden. ®
Sponsored: Benefits from the lessons learned in HPC