Saturday, January 28, 2012

Metaphor needed

Simone Schnall writes about embodied cognition at Edge. It seems I cannot link directly to her comments.
Philosophers and psychologists grappled with a fundamental question for quite some time: How does the brain derive meaning? If thoughts consist of the manipulation of abstract symbols, just like computers are processing 0s and 1s, then how are such abstract symbols translated into meaningful cognitive representations? This so-called "symbol grounding problem" has now been largely overcome because many findings from cognitive science suggest that the brain does not really translate incoming information into abstract symbols in the first place. Instead, sensory and perceptual inputs from every-day experience are taken in their modality-specific form, and they provide the building blocks of thoughts.
I am not sure I understand what she means.

What does she mean by "modality-specific"? We know that our senses are not isolated from one another. It is not only by observing synaesthetics that we know this. The recent work on hacking our senses has shown us that we can train one sense to use the inputs which would normally feed another, and evoke a real, useful "image" (whether visual, aural, or otherwise) which we would normally not be able to experience. For example, the article I've linked to describes how scientists have been able to teach visually impaired people to see with their tongues(!). We know that they are actually "seeing" the image because they have been able to replicate the results with people who can see normally, and can compare the experience with normal vision.

She describes experiments which suggest that people associate the word "Up" with virtue, power, and God, while they associate the word "Down" with vice, weakness, and the Devil; that people who had just moved up on a mall escalator were more likely to give to charity than those who had moved in the other direction. These are all alright, but I find them unsatisfying.

Surely there have been more convincing experiments which show that the body is much, much more than merely an appendage to the brain?

We already know that the body can do a lot without the brain being involved. It is not only headless chickens which can run but, as this old and terrible experiment shows, decerebrate cats can too.

What is more, it seems that nature uses some very simple tricks to enable the body to work without close supervision from the brain. We know this because now even we can create things which walk without central processing.

I had previously blogged about some scientists who were able to build a mechanical spider which could walk across a wire mesh with large, gaping holes. The key to building a body which "naturally" does what you want it to do, without having to refer everything to a central computational model of the world and the body in it, is to use the right materials and shapes. Wisdom without knowledge. Behold the Octopus!

I came across Simone Schnall's piece via Janet Kwasniak's blog post. As she puts it:
They point out that computers can simulate brains (or more or less anything). But there is a difference between being able to simulate the weather in order to predict it and treating a computer as the model of how the weather engines work. The same applies to brains. Computers are very useful tools for studying various simulations of the brain, but the brain is not a computer.
She is absolutely right, of course, but by that standard, only computers are computers. The question is rather: in what ways is the brain like a computer, and in what ways is it not?

I agree with what they say: my point is that Schnall has chosen some very weak examples to illustrate the idea behind embodied cognition. I understand that the Edge Annual Question is not the place to look for deep discussions, but this may leave readers with the idea that Embodied Cognition is an airy-fairy topic. On the contrary, it is a sea-change in how we think about the mind, the brain, and the body, and how those relate to each other.

We are very far from understanding all of this, but these various experiments, and the little machines we are building, are clues to a much more interesting relationship between brain and body.

As Andrew Wilson wrote, Embodied Cognition is not what you think.

No comments: