Doors, Computers, Tigers

Friday, 29 September 2006, 22:06 | Category : Philosophy
Tags :

Here’s where this mental exercise started… Loren got me interested in reading Space and Place by Y-Fu Tuan. More on that book later, when I’ve read further. My thought process began with a door, then ran through computers, and ended up at a tiger. As I read tonight, I happened to glance across the room. Not at anything in particular, just one of those random glances you sometimes make in the midst of a prolonged ponder. I focused on the door, about 12 feet away, the bookshelf beside it, then another door nearby. I thought about how my brain seemed to synthesize several responses to the scene in a package: the door is about 12 feet away, the wall is white, there are about 6 books on each shelf (it’s a narrow bookshelf). Then I thought about how a computer would “see” that scene, if it had a camera attached. I can grasp how it would characterize color. I can understand how it might deduct the presence of multiple books on a shelf, by an algorithm that would note the different roughly parallel rectangles formed by the book spines. But then I thought about distance. How would a computer know distance? Certainly, it could use a laser range finder – but in the absence of a direct measurement tool, how would it do this? Our brains, I’m sure, use a combination of experience and knowledge to perform a quick calculation – but even when there aren’t reference objects available, like the futon between me and the door, our brains can do a pretty good distance calculation in most situations. I don’t know how a computer would do this type of intellectual gymnastics. Then I thought about tigers. Tigers are very good at leaping. Along with many other things, I learned this from Winnie The Pooh. And the Discovery Channel. And Wild Kingdom. Tigers leap very well. But I don’t think they calculate distance. Tigers don’t say to themselves “Let’s see, the deer is on top of that earthen bank, the bank is about 13 feet away, I have to clear a 3 foot shrub…”. I think tigers think effort. They don’t know how far they need to jump, they just know how hard they need to jump. I’ll admit I have no scientific basis for making such a statement, it just seems to me that distance is a concept that tiger brains don’t handle. They don’t experience the world that way, in the same way that we don’t calculate the distance to the basketball goal, we just know we need to shoot the ball that hard. We can, however, stand on the basketball court, look at the goal, and estimate that we are 18 feet away. Same physical setting, 2 different ways of experiencing it. When we shoot the ball, we’re tigers. When we mentally calculate the distance, we’re people. But neither of these experience sets is applicable to a computer. So I go back to wondering how a computer would devise a value for a distance. If you have knowledge of how this is done, feel free to enlighten me. Just remember, no laser range finders.

5 Comments for “Doors, Computers, Tigers”

  1. 1loren

    Whoaa. I’m not sure Loren got THIS started.

    Sounds to me like you’re exploring the mind and the various ways it perceives, not just Space and Place.

    Though I’ll admit that it is a fascinating comparison. It reminds me of what a mistake it was to start thinking in a basketball game. Once you started that, you might as well just quit because you couldn’t seem to hit even the easiest baskets.

  2. 2Billunit

    I submit that without reference objects or memory of an object’s size, our brain could not calculate, but only guess at the distance. We could be easily fooled in a controlled environment.

  3. 3Harry

    That’s why I said “Our brains, I’m sure, use a combination of experience and knowledge to perform a quick calculation”. Although if I’m standing in an open space looking at a lone flagpole, I can make a reasonable guess at the distance. The accuracy falls apart rapidly as the distance grows, but for useful distances it’s still reasonably accurate. I still believe we experience distance in a different way than a tiger, and replicating that with a computer would be a major challenge.

  4. 4Dave

    To demystify things a bit, the way a computer would do it is actually the same way a human does it. It’s primarily due to stereoscopic vision, especially for things in close to medium range from the viewer. That is, if you have two eyes (or cameras), the picture from each will be slightly different. By seeing how far a reference point (a corner, boundary between colors or shadow and light, etc.) moves from image to image, you can tell how far away it is.

    The brain does this almost exactly the same way a computer would, by first extracting points and lines and then cross-referencing the two images to get depth information. The brain then does additional inference based on – among other things – gradations of shadow (used for fine contours), relative motion of objects (useful when you or the target is moving) and light scattering (useful for very different objects, which naturally appear hazy).

    The correspondence between wanting to jump to a particular point and the amount of force required is a learned one, and one we pick up at a relatively young age through practice. One could train a neural network in a robot to do a similar sort of thing. But you’re right that you don’t necessarily *think* about it – the calculation is done at a level that’s below your conscious mind, like many other things people do.

  5. 5Harry

    And I think that’s the point I was making, but you put it more elegantly. And it’s because it’s done below our concious thought level that will, I believe, make it so difficult to have it done by a computer.