September 18, 2003

Dennett's Stance

So I'm halfway through The Intentional Stance by Daniel Dennett and I wanted to get some thoughts down before I lose them. I'll try to update this when I finish the book.

I read Freedom Evolves earlier in the summer. I enjoyed it but felt that while jumping from interesting point to interesting point, it didn't tie everything together in a way I would have liked (I'll try to post more on it later). Since it references many of his earlier works, I decided to read a few in order to see if I was just missing some fundamental thing that would bring his argument together. So I picked up this book. In it, Dennett tries to give his view of belief, desire, intentions and how we understand them.

The book, as you'd guess, is about the intentional stance. Dennett defines this as a strategy that one adopts to try to predict the behavior of a system. It is one of several different stances that we humans are wont to adopt. In the physical stance we predict behavior based on the laws of physics: gravity, inertia, friction, etc. This obviously works well, but is extremely cumbersome, if not impossible, with complicated systems. In the design stance, we abstract things and assume that components of the system will act "as they were designed to". In this stance, we can predict the behavior of a car when we turn the steering wheel (without doing complicated equations of motions) because we know that that's what a steering wheel is for. Likewise, we can predict that setting an alarm clock will make a noise at the correct time without knowing the diagram of the circuit board and whether it has a quartz mechanism or uses atomic resonance.

In the intentional stance we assume that the system (e.g. person, animal, artifact) that we're trying to predict has beliefs and will act rationally to achieve its desires. What does it believe? It believes true beliefs, i.e. beliefs it ought to believe. What does it desire? It desires what it ought to desire. This may sound a bit circular but it's the way we do a lot of reasoning about interactions we have everyday. With other people, with our pets, with our anthropomorphized computers.

In my mind (though I'm stretching a bit here) it's an extension of the principle of charity. The principle of charity purports to explain how we can bootstrap communication with other people without a universal, shared set of meanings. We basically have to assume that we mean the same things by certain signifiers (e.g. words), before we can understand enough to know that we mean different things by some of them. Likewise with intentions, we have to assume that someone else believes the true and desires the good (for her) or we have no common ground to start from. Any false beliefs or irrational desires that we attribute need to have "special explanations" — a reason or a cause for them to exist.

With this definition, Dennett calls any system whose behavior can be predicted from the intentional stance an "intentional system", and humans, he claims, are the best examples around.

So now is where he gets controversial. He claims that all it means (all it can mean) to say that something or someone (let's call them X) believes something is that X's behavior is predictable using the intentional stance and attributing that belief to X. This is obviously an instrumentalist position to take. He's claiming that there is no more meaning in the concept of belief than the fact that it's useful in predicting behavior.

He stakes all of this out in "True Believers", one of the first essays in the book. He then spends an extremely long time trying show that this assertion doesn't back him into two corners that he finds unattractive: subjectivism and anti-realism.

The first problem is showing that saying that X has belief P is an objective statement. By defining the belief by its usefulness in prediction, haven't we put belief in the eye of the beholder? Dennett wants to squirm away from that relativism and say that objectively X either believes that P or she doesn't. For me, this is the least compelling part of his argument (though he, as usual, makes some interesting points that I find hard to do justice to in this space). In the end, he clings to his objectivism tightly.

The other thing he clings to is his materialism. Everything that goes on in the human mind happens because of something in the human brain, which happens because of material circumstances and physical processes. So, since he's spent several chapters arguing that beliefs are only instrumental, he wants to assure that he's still a realist. He believes (or at least you can predict his behavior if you attribute this belief to him) that there are "real" things going on in our heads when we believe things.... it's just not proper to call them beliefs. This is where he threads the needle and distances himself from the capital-R Realists who think that we actually have structures, patterns, whatever in our brains that are our beliefs. Dennett holds that there is no reason to assume that what goes on in our head is homologous or has a one-to-one mapping with beliefs. But, not willing to join the anti-realist camp, he also claims that something happens in our brains that explains the fact tht we have beliefs (are predictable when attributed beliefs).

Anyway, that's where I am now, both in the book and my understanding of what he's saying. I'll update this post when I make more progress. Re-reading this, I'm not sure I'm doing his arguments justice. For someone who hasn't read the book, it may not be that useful. For those who have, comments, clarifications, and corrections would be appreciated.

Posted by richard at September 18, 2003 11:55 PM