The camera analogy(2) – acquiring knowledge (learning)

This continues the series that I began with “The camera analogy(1) – introduction“.  In this post, I wish to use the analogy to discuss the question of knowledge, and how we acquire it.

Typical discussions of knowledge acquisition and philosophy of science take for granted that there are many facts that we can readily pick up by observation.  They concentrate on the question of how we acquire more facts that are not so easy to find.  Let’s refer to those not-so-easy facts as “enhanced facts.”  These enhanced facts are often said to be derived by inductive inference from the easier facts, as in “All the many crows I have seen are black; therefore all crows are black.”  There is also a long history of skepticism with regard to the use of induction, though I won’t be discussing that here.  The behaviorists from psychology often describe learning in terms of conditioning.  And since conditioning is said to result from repetition and reinforcement, that is similar to the use of induction.

Using our camera analogy, the easy to find facts are like the photographs that we take with our camera.  Inductive inference is then about the same as taking a large collection of photographs, and searching them to try to find patterns that are common to many of the pictures.  Any patterns found could be said to be found by pattern induction, and would be counted as enhanced facts.

There is an entirely different way that we can obtain advanced facts with a camera.  And that would be to change the camera itself, giving us an enhanced camera.  For example, we could use an infrared sensitive film, together with filters that only allow infrared to pass through the camera lens.  That would give us infrared photography, and would allow us to capture images that we could not see with visible light.  Or we could replace the camera lens with one of greater resolving power, and use a finer grain film or a higher resolution digital imaging, to obtain pictures with more detail.  That is to say, instead of finding enhanced facts by searching for patterns (induction) in the easy to find facts, we could enhance the apparatus that we use to gather facts, so that the modified apparatus delivers facts not previously available.

The idea that we could gain knowledge by enhancing the apparatus, instead of by using induction, is what psychologists refer to as “perceptual learning.”  Eleanor Gibson wrote a 1969 book, “Principles of Perceptual Learning and Development”, on the topic.  The research paper, “Linguistic Experience Alters Phonetic Perception in Infants by 6 Months of Age”, Kuhl et al., Science 255 (1992), seems to point to the importance of perceptual learning in language acquisition.

When we consider science, we should notice that science itself creates a lot of apparatus to enable us to pick up facts that were not previously available to us.  So learning by changing the apparatus is not limited to sensory perception, but may indeed be a widespread part of our ability to acquire enhanced facts.

There are important philosophical consequences of the possibility idea that acquiring knowledge might partly depend on changing the apparatus.  I shall be examining some of those in future posts in this series.  For one example, consider the computationalist assumptions of AI (Artificial Intelligence).  Those are based on the idea that a fixed apparatus produces representations of the world, and that everything cognitively important amounts to computation using those representations obtained using a fixed apparatus.

There are biological reasons why I favor the idea of perceptual learning.  It seems implausible that human DNA could carry full specifications of the perceptual system.  For one thing, the capacity of the DNA seems to be too small.  This suggests that the perceptual apparatus might, instead, be constructed during the period of a child’s development, with continual enhancement through childhood, and thus might be adapted to be optimal for the kind of world that we happen to live in.  And then, in addition, notice that the apparatus does change during childhood.  For example, visual depth perception depends on the use of two eyes, and the actual details depend on the distance between the two eyes.  This distance increases in the growing child, and the apparatus is thus changing and other changes must be going on to adjust for this change in distance.

In summary, the main aim of this post was to suggest that perceptual learning is important, and that there is more to learning than simply acquiring facts.