To know how boxes keep things in, geometry is not enough. You also have to know how moving works. "The diagram belwo depcits an agency that represents the several ways an arm can move inside a rectangle ... If we connect each of these sub-agents to the corresponding side of our four-sided box frame, each agent will be able to test whether the arm can move in the corresponding direction."
(NB:This section is important for a number of reasons. First, Minsky provides an explict example of agencies representing -- different agents detect motion in different directions (or, more correctly, the ability to move in different directions), and also are connected to properties in the world (which must also be represented by agencies). Second, Minsky introduces another functional primitive -- inhibitory connections between agents. This feels more and more like connectionist-style modeling to me. Third, we get a sense here of bridging of a different sort -- bridging between agencies that represent structural descriptions (i.e., properties of the arch) and agents that represent functional affordances (i.e., ability to move).)
Pearl Street | Society Of Mind Home Page | Dawson Home Page |