About perception

In a recent post, I suggested that perception is an important part of perception. One commenter disagreed, and wrote “Perception is simply raw sensory impressions”. That’s actually a common view. A lot of the literature on empiricism suggests that it starts with impressions.

I’m actually not quite sure what people mean by “impression” when they make such statements. I expect that they are probably thinking of conscious impressions — something that we can think about. However, it seems to me that much of perception is prior to conscious experience.

The expression “raw sensory impression” could also mean the forming of an image (in the visual case). I’m doubtful that we form images. I cannot see any good reason why an ability to form images would evolve, because it is hard to see much benefit from having such an ability. What really matters to an organism, is getting useful information about the immediate environment. So my own thinking about perception has tended to emphasize that information problem.

Motion

One of the problems for visual perception, is that we are moving. We move our bodies. We also move our heads relative to our bodies. And our eyes move (in saccades) relative to the position of our heads.

If you have ever tried taking a picture with a moving camera, you probably know that this can result in a blurry picture. A high shutter speed can reduce the problem. However, the response of retinal sensory cells is slower than high speed shutters, so that method is not available to the visual system.

If we were designing a robot, we would use a gyroscope to give us a reference direction. And then we would switch the input information to cancel out the effects of motion. However, we do not appear to have a gyroscope in our heads, and the switching speed of neurons would be too slow to solve the problem this way.

For the camera, we can mostly avoid the problem of motion by holding it as steady as possible while taking a photograph. If our perceptual system used this method, then the hunter would have to stop and hold steady to get a good glimpse of what he is hunting. This would not work. The hunter needs good vision while running in pursuit of prey. Clearly, the visual system needs a better method.

Categorization

This is where categorization can be useful. A crude categorization of the visual field gives it some structure. And that structure, in turn, can be used to stabilize the visual field. In order to categorize, you need to find boundaries. And boundaries can be located more readily than other details. The motion of the eye in saccades can actually be useful in locating boundaries.

This is part of why I see categorization as important. It gives the visual system a way of dealing with and compensating for our motion.

The other reason for emphasizing categorization, is that our information is always information about categories. And because I see the main role of perception as providing us with information about our environment, then it must categorize to do this.

Note that the latter point is not restricted to visual perception. When we talk about taste, we may describe it with terms such as “sweet” or “sour” or “bitter”. Those are category terms that we use in our descriptions of taste. We need the perceptual system to determine, or at least to suggest, these categories to us.

This is why I see categorization as central to perception.