Is the idea that neurons perform ‘computations’ in any way meaningful?

I wrote this up two months ago and then forgot to post it. Since then, two different arguments about ‘computation’ have flared up on Twitter. For instance:

I figured that meant I should finally post it to help clarify some things. I will have more comments on the general question tomorrow.

Note that I am pasting twitter conversations into wordpress and hoping that it will convert it appropriately. If you read this via an RSS reader, it might be better to see the original page.

The word ‘computation’, when used to refer to neurons, has started to bother me. It often seems to be thrown out as a meaningless buzzword, as if using the word computation makes scientific research seem more technical and more interesting. Computation is interesting and important but most of the time it is used to mean ‘neurons do stuff’.

In The Future of the Brain (review here), Gary Marcus gives a nice encapsulation of what I am talking about:

“In my view progress has been stymied in part by an oft-repeated canard — that the brain is not “a computer” — and in part by too slavish a devotion to the one really good idea that neuroscience has had thus far, which is the idea that cascades of simple low level “feature detectors” tuned to perceptual properties, like difference in luminance and orientation, percolate upward in a hierarchy, toward progressively more abstract elements such as lines, letters, and faces.”

Which pretty much sums up how I feel: either brains aren’t computers, or they are computing stuff but let’s not really think about what we mean by computation too deeply, shall we?

So I asked about all this on twitter then I went to my Thanksgiving meal, forgot about it, and ended up getting a flood of discussion that I haven’t been able to digest until now:

(I will apologize to the participants for butchering this and reordering some things slightly for clarity. I hope I did not misrepresent anyone’s opinion.)

The question

Let’s first remember that the very term ‘computation’ is almost uselessly broad.

Neurons do compute stuff, but we don’t actually think about them like we do computers

Just because it ‘computes’, does that tell us anything worthwhile?

The idea helps distinguish them from properties of other cells

Perhaps we just mean a way of thinking about the problem

There are, after all, good examples in the literature of computation

We need to remember that there are plenty of journals that cover this: Neural Computation, Biological Cybernetics and PLoS Computational Biology.

I have always had a soft spot for this paper (how do we explain what computations a neuron is performing in the standard framework used in neuroscience?).

What do we mean when we say it?

Let’s be rigorous here: what should we mean?

A billiard ball can compute. A series of billiard balls can compute even better. But does “intent” matter?

Computation=information transformation

Alright, let’s be pragmatic here.

BAM!

Michael Hendricks hands me my next clickbait post on a silver platter.

Coming to a twitter/RSS feed near you in January 2015…

 

The bigger problem with throwing the word ‘computation’ around like margaritas at happy hour is it adds weight to

Advertisements

5 thoughts on “Is the idea that neurons perform ‘computations’ in any way meaningful?

  1. I seem to lose my understanding of what computing is supposed to mean to people. As I said back then and as I said in that tweet, to me if you take a number or a signal, and transform them in any meaningful way, I’d call it a computation.

    However, the problem for people to me appears to be that transformations of any sort in the neuron are basically passively caused by physical deterministic mechanisms (which btw are dynamically adjusted in real time) which seems not to be what people think computation can be.

    Now I wonder, if you say neurons don’t compute, but we use neurons to calculate a + b = c – this can mean we are not actually computing anything, either, and we use deterministic processes in a complex way to make things appear as if they were computed. It could also mean ‘computation’ is one of these things that emerge when we put simple things together to form complex systems that have ‘new’ properties and we think about it like we think about concepts of consciousness and life.

    If we say, computations don’t take place in the brain at all, because all those neurons only respond in a physically deterministic way, then computation in silico – by microchips – aren’t actual computations either. Because of course they also have deterministic reactions to inputs.

    In total I am getting the idea that either ‘computation’ as such doesn’t really exist, or it is a metaphysical thing that emerges from ‘sufficiently complex’ systems. I am not happy with either and keep calling any meaningful transformation of a signal a computation, disregarding the apparent ‘complexity’ of the system. Biological systems are always complex, anyways.

  2. This discussion sort of explains why I’ve never really understood these discussions whenever they cropped up (and they always do, repeatedly, predictably): people mean different things with ‘computation’ all the time and seem to be unaware that they do and everybody else as well, lol 🙂

    Seriously, though, a lot has been written here and elsewhere about neurons generally computing I/O functions. The example with the myocytes made already clear that this is a generalization that’s not admissible. But it becomes even more clear when you look at Fig. 4B in this paper:

    http://journal.frontiersin.org/Journal/10.3389/fnins.2012.00113/full

    here the figure: http://www.frontiersin.org/files/Articles/24467/fnins-06-00113-HTML/image_m/fnins-06-00113-g004.jpg

    B63 is a neuron that initiates behavior. The trace you see is what it does when it is depolarized by a constant current. This is what animals do, encapsulated in a single neuron: animals do stuff to find out how the environment reacts such that they can survive. In many cases, they need to compute certain things in order to do evaluate the response of the environment to their actions, but that’s just a consequence of what they actually do.

    Metaphors are useful, but they only go so far. Using metaphors of our engineered world will get you the engineering-like properties of the system under study. Given the plasticity and versatility of nervous systems, you can push them to behave almost like any engineered system, as long as you don’t look to closely (see, e.g. the stimulus-response metaphor of old http://ecx.images-amazon.com/images/I/81Gu0NJ7iTL.jpg ). However, if you look without an engineering metaphor, but with the realization that brains evolved, you quickly find that, yes, brains do sometimes compute certain things, but that’s neither what they do nor is it what they evolved to do. Brains chose behavior to make their organism survive. There are a lot of things they need to do to accomplish that and sometimes computations are part of that.

    To say that ‘brains compute’, or even ‘neurons compute’ is like saying ‘people have sex in cars’ – yes you can, in some better than in others, some look like that’s what they were designed to do, but if you look closely enough, you’ll realize that cars primarily do something else entirely. Even their seats, while at this particular moment they may seem to fit the purpose perfectly, are generally not well adapted to that particular task…

  3. I have been thinking about this a lot, and my insight is that information transformation is not all there is to computation. There is another key ingredient that separates “computation” from transformation, and that is the use of MEMORY. There are many algorithms or IO transforms that can be performed without memory (a filter, a feedback controller) and a computer can implement these algorithms. However, there are many more algorithms that require memory (a sorting algorithm), and I think that this class is what truly separates something that computes from something that is just a machine. The brain and neurons compute, because they do information transformations and they use memory. A system without memory cannot perform true computation. A machine can execute an algorithm that requires no memory, but only a computer can execute an algorithm which must use memory.

  4. Pingback: Why I think neurons are computers and why it matters. | Eckmeier.De|nnis

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s