Brains like Computers?

Posted on April 10, 2008
Filed Under Philosophy | 3 Comments

Are brains like computers? Over at Developing Intelligence Chris Chatham has two interesting posts on this. The first is about differentiating brains from computers. The second is responding to a set of criticisms over at Conscious Entities.

Now most of these I don’t disagree with. (Of course I’m anything but a cognitive scientist) I want to take a slightly different approach to criticizing a few points though. (Although perhaps nuancing rather than criticizing is a better term)

First difference #1. This is making a distinction about computers being digital and brains analogue. While I agree with the basic thrust I think he pushes this too far. Consider his point about the rate of neuron firing. Yet, as anyone who has done device drivers or networking knowns, digital doesn’t mean equal speed. Try doing any interfacing between two computers and you run into many of these same issues. Networking code can take into consideration speed differences and react differently.

Yes there still are more differences due to neurons and digital circuits being different. However one should note that digital systems don’t always act digitally in the sense of the simplistic models non-programmers often think of computers in. Obviously one example are neural nets which attempt to emulate neural systems. We have fuzzy logic. More importantly people can use quite good pseudo-random number generators to tweak random behavior. There are reasons to do this in some cases.

My point isn’t to dispute the larger point that brains aren’t digital. Just to argue that in practice with complex systems the difference is much less than it first appears.

Next difference #2. I do a lot of pattern recognition, search indexes, and so forth. While computer memory is, of course, as Chris points out. When we talk about memory in a more abstract sense so as to include databases, hash tables, and so forth, things get more complex. I think Chris recognizes this as he talks about a “built in Google” as an analogy to the brain. However that metaphor is kind of ironic given Google is being used as an example of how the brain works. (Something he acknowledges)

While Chris’ point is true for most code I suspect that it highlights a big difference. Are we talking about computers in a low level sense? (i.e. roughly the way we thought about computers in the 1980′s) Or are we talking about computing in a more complex sense? (The way most run today) This problem highlights what I see as a problem. The brain is an amazingly complex system. Yet, as computers become more complex and distributed then the dis-analogy breaks down. Doesn’t this then suggest brains are like complex computers? (I don’t think they are, for the record. I just point out the problem as we draw analogies)

Next difference #3. Modules. The problem is that while some systems use pretty independent modules in practice there is tons of feedback between the module and what is calling the module. The idea that many people have of modules is modules as fairly independent. That’s just not the case. (It’s certainly rarely the case in code I’ve worked on – although there often are good reasons to try and keep modules more independent. Yet most bugs appear because they aren’t.)

The bigger problem is that we here are conflating functional modules from implementation modules. Now I do agree that happens when folks talk about the mind or brain. So people talk about language (a fairly complex function) and then conflate this with parts of the brain contributing to the function. It may well be that a part of the brain dealing with verb recall and manipulation is very module like even if the function of using verbs is anything but. This aspect of holistic functions vs. individual modules is a big issue in computing. Especially once you end up with a complex system often written by dozens or even hundreds of people over the years.

My ultimate critique of computing metaphors of the mind (not brain) though is really this conflation between real world function and the idea of modules realizing function. Even folks who should know better make this mistake. Part of the problem is that function is always function in terms of my world. There is a holistic sense to function that goes well beyond my parts that might contribute to that function. That is even if there were a part of the brain purely dealing with language (and I don’t think there is) it wouldn’t be a language function module simply because language is social (and thus as a function involves multiple people) and is referential (and thus involves the things in my environment in a fashion beyond their representations.

Most of the rest of his comments I agree with. I just think that often when the brain as a computer is criticized it’s with perhaps a too narrow sense of what computers are and how software works.

Comments

3 Responses to “Brains like Computers?”

1 Clark on April 10th, 2008 1:46 pm

BTW – Chris’ number 10 was my favorite. Not just because of my Heideggarian biases. But because it really gets at how the brain uses its environment for processing. Typically we make a huge distinction between my memory as stored in mnemonics around me (say a string tied to my finger); between stuff I record on a pad of paper; and what goes on in my brain. Yet in a real sense this is a matter of degree rather than kind.

2 Jacob J on April 10th, 2008 6:47 pm

However one should note that digital systems don’t always act digitally in the sense of the simplistic models non-programmers often think of computers in.

It’s funny you said non-programmers since most programmers don’t know how computer circuitry works any better than the average computer user. As a person who does microprocesor circuit design for a living, I agree with your points. The distinctions being made between “digital” computers and “analog” brains don’t hold up as far as I’m concerned. Everything is analog in its fundamental behavior. “Digital” is a way of interpreting analog circuitry, treating every signal as a 1 or 0, which is sort of like saying a neuron has either fired or it has not (unless I am mistaken about neurons, since I am not a neurologist). At which point, we don’t even need to bring up the fact that computers have lots of strictly analog circuitry…

3 Lincoln Cannon on April 10th, 2008 8:01 pm

Clark, I’ve posted some thoughts on reconciling these differences, here:

http://transfigurism.org/community/blogs/lincoln_cannon/archive/2008/04/10/4283.aspx

Leave a Reply