"What's so special about moving left or right or up or down? At first one might suppose taht these ideas work only for motions in a two-dimensional space. But we can also use this square-like frame for many other realms of thought, to represent how pairs of causes interact." An example of such an interaction is how a combination of horizontal and vertical movement produces diagonal movement.
How do we learn to control movements mediated by our joints? Minsky suggests that we use interaction-square arrays. "An interaction-square array provides a convenient way to represent all the possible combinations...If square-arrays can represent how pairs of causes interact, could similar schemes be used with three or more causes?" No, too complicated to use.
(NB: Two points. First, Braitenberg -- with his use of networks to represent hyperspace -- does not shy away from multidimensional representations. Second, is the interaction-square another primitive? Or is it merely a convenient description of how different agents interact, as might be the case for the example given in Chapter 14.8?)
Pearl Street | Society Of Mind Home Page | Dawson Home Page |