5
$\begingroup$

I am a computer programmer or computer engineer, and am interested in comparing the brain to a classical computer in some way. How well does this comparison hold up?

This is a general introduction to the computer/brain metaphor and its practical limitations. Similar questions on this forum include:

$\endgroup$
  • $\begingroup$ I understand what you are trying to do here, Arnon, but as is it does not seem to fit the Q&A format very well. One option could be to add this overarching documentation to tag info, although this might be hard to find (don't know the current SE stance on this). Another would be to make the question much broader (something that typically would be closed, but in this case it would have merit, since we can close all similar questions as a duplicate of this), and include all the questions you link to in the answer, possibly categorizing them. $\endgroup$ – Steven Jeuris Jun 18 '18 at 10:51
  • 1
    $\begingroup$ I like how you summarized this. More selective and stable than a tag would be and yet organized in one place. Here is a way chemistry SE did a summary of resources for learning chemistry: chemistry.stackexchange.com/q/37303/64817 $\endgroup$ – Frank Hubeny Jun 18 '18 at 13:40
  • 1
    $\begingroup$ @StevenJeuris let's discuss further in meta psychology.meta.stackexchange.com/q/2369/4397 $\endgroup$ – Seanny123 Jun 18 '18 at 16:36
2
$\begingroup$

This question's reference to a classical computer refers to a "Turing Machine" style of computation, also known as a knowledge system, in which decisions and possible results are pre-programmed using if-statements, loops, and other logical constructs.

However, most modern computer programmers and engineers are at least somewhat familiar with neural networks, used for machine learning. Artificial neural networks (ANNs) are loosely based on the way the brain (a biological neural network) works. If you want to use your computing background to understand the brain, then I recommend learning about neural networks instead. If you already have some understanding of neural networks, then think about how Siri or Watson work.

While neural networks can certainly be thought of as storing data, executing algorithms, parallel processing, and having memory and speed measurements, they are conceptually very different from classical computers, and generally uninstructive to compare. If you are still interested in such a comparison, then many of the questions listed above already have great answers, so check them out.

If you are interested in modeling the brain using a classical computing methodology, then check out ACT-R, a popular architecture for modeling the brain using a modular, knowledge-based system (also see this forum question).

Artificial neural networks are not a perfect model for biological neural networks either, and if you are interested in the difference, then see this forum question.

$\endgroup$
  • $\begingroup$ I doubt that the question refers to a 'Turing Machine' style of computation (i.e. infinite tape with DFA control structure). I suspect programmers are thinking in terms of the von Neumann architecture instead. Of course, the point of CS is that these models (including ANNs) are equivalent & many of the questions we might ask about one translate over (upto some simulation factors) to the others. I would also be skeptical of encouraging engineers to think they'll learn something about the brain (nevermind mind) from looking at CS ANNs. $\endgroup$ – Artem Kaznatcheev Jul 8 '18 at 12:19
1
$\begingroup$

This line of thought actually represents the current state of the art in scientific approaches to the brain. Before we discovered electricity and that (peripheral) neurons conduct electricity, one analogy or explanation suggested that the mind worked through hydraulic action.

It is therefore possible that we have to abandon this analogy, if we find a better one. For now, we give it explanatory power. However, we do know that the brain circuitry is very different from a computer. At the same time, neurons have discrete outputs, which can be represented in binary (all-or-none). (But there is also non-spiking neurons with a more continuous "analog" response.)

Suggested readings:

Cognitive Science

Philosohpy of Mind: Computationalism

A classical argument against computationalism: The Chinese Room Argument

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.