A rhythm recognition computer program to advocate interactivist perception

This paper advocates the main ideas of the interactive model of representation of Mark Bickhard and the assimilation/accommodation framework of Jean Piaget, through a rhythm recognition demonstration program. Although completely unsupervised, the program progressively learns to recognize more and more complex rhythms struck on the user’s keyboard. It does so without any recording of the input flow, and without any pattern matching in the usual sense. On the contrary, internal processes are dynamically constructed to follow and anticipate the user’s actions. We show that these processes are representations of the rhythms in the interactivist sense, and that they emerge from non representational grounds, avoiding the symbol-grounding problem. They are not copies or transductions of reality, but ideal internal constructions of the agent, avoiding the circularity pointed out by Piaget. In practice, the active nature of this recognition process allows it to work even with noisy and complex input flows. © 2003 Cognitive Science Society, Inc. All rights reserved.