Minsky spends much of the rest of the chapter considering children's drawing of people, particularly drawings of people that are essentially heads with attached sticks for arms and legs.
"We normally assume that children see the same as we do and only lack our tricky muscle skills. But that doesn't explain why so many children produce this particular kind of drawing, nor why they seem so satisfied with them." Minsky suggests one view might be that the child does not have an image in mind when they draw. Instead, perhaps children have a set of features (propositions!!) in mind that serve as a description of what is to be drawn, and then employ a drawing program that is guided by this description. A simple or unsophisticated drawing program will produce unrealistic results, but can still satisfy the internal description -- which might be why children prefer these kinds of drawings.
(NB: Wow! Shades of the imagery debate here; would Kosslyn's model produce child-like images? Are internal descriptions like sets of propositions? If so, are agents to be viewed as propositions? This would make sense, if they were also viewed as being similar in nature to perceptrons...)
Pearl Street | Society Of Mind Home Page | Dawson Home Page |