where X and Z have the same restrictions as just described and W is a location. Again we see an object's changing location -- a common occurrence in movement and a building block Moran identified.
As we present Moran with more complex inputs and situations other than those that are described by movement verbs, we predict that Moran will identify more building blocks that can be related to subparts of Schank's conceptual dependencies.
Moran is capable of inferring structural building blocks. The parallel problem of inferring and associating semantics with the inferred structural building blocks is not addressed. Learning the procedures that are the semantics of structural units appears to be a more difficult problem than learning the structural units. This is a problem that must be addressed if truly acquisitive schemes are to be developed. It is a fertile and challenging research area.
As a cognitive model, inferred building blocks may account for the effects of "bad teaching" -- that is, an unfortunate sequence of examples of a new concept. If examples are so disparate that few building blocks exist, or so unrepresentative that the derived building blocks are useless for future imputs, then the inferred building blocks will impede efficient representation. The knowledge organization will not tie together what we have experienced in the past or predict what we will experience in the future. Although the learning program could infer more useful building blocks at a later time, that process is expensive, timeconsuming, and may be unable to replace information lost because of poor building blocks chosen earlier. In general, however, we must assume that our world is described at a level appropriate to how we must process it. If this is the case, then inferring a set of building blocks is an advantageous strategy.
I am indebted to David Warren, David Maier, Mark Jones, and Raphael Finkel for their criticisms, comments, and lively discussions, and to Judy Wagner for implementing Moran's building-block component.
This research is supported by National Science Foundation Grant ENG-7907994.
Minsky M. A framework for representing knowledge. In P. Winston (Ed.), The psychology of computer vision. New York: McGraw-Hill, 1975.