Towards a Model of Grounded Concept Formation

In most research on concept formation within machine learning and cognitive psychology, the features from which concepts are built are assumed to be provided as elementary vocabulary. In this paper, we argue that this is an unnecessarily limited paradigm within which to examine concept formation. Based on evidence from psychology and machine learning, we contend that a principled account of the origin of features can only be given with a grounded model of concept formation, i.e., with a model that incorporates direct access to the world via sensors and manipulators. We discuss the domain of process control as a suitable framework for research into such models, and present a first approach to the problem of developing elementary vocabularies from perceptual sensor data. 1 Introduction The goal of human concept formation is to arrive at a conceptual system that partitions the encountered objects and events in a way that enables us to effectively deal with our environment. As such, the task of concept formation can be split up into two subtasks, concept aggregation and concept characterization. Concept ag-gregation is the decision about which entities are to be grouped together into a concept, concept characterization means finding an intensional description of the proposed concept based on its extension. The latter task is often called concept learning (from examples). Consequently, what separates concept formation from concept learning is the difficult question of deciding which objects to aggregate, i.e., how to carve up the world into different concepts. Most operational models of concept formation, most notably conceptual clustering Gennari et a/., 1989] base their answer to this question on an assumption that was nicely formulated by Rosch and her colleagues [1976; 1978]: The "correlated feature" view assumes that features in the world occur as clusters , and that the best concepts are those that maximize intra-concept correlations, and minimize inter-concept correlations, i.e. reflect the presumed cluster structure of the world. There is a significant amount of empirical evidence that people detect and use feature correlations [Medin et a/., 1982; Younger and Cohen, 1984], and the existing clustering systems based on the correlated feature principle and extensions thereof 1 have been very successful, both in modeling psychological data [Fisher, 1987], and in solving application problems. None of the existing concept formation models, however , examine the fundamental question of where the building blocks of correlations, namely features, come from. Invariably, features are assumed to be …