1.12 – Learning and Representation

Behavioral evidence implies that even insect brains compute representations of behaviorally relevant aspects of the world. For example, they compute the animal’s position in the word by integrating its velocity with respect to time. Other examples are the learning of the solar ephemeris, the construction of a cognitive map, and episodic memory in food caching. Representations require a symbolic read-write memory that carries information extracted from experience forward in time in a computationally accessible form. The close analogy between the architecture required to access memory in a computer and the architecture of the molecular machinery that accesses genetic information suggests the sort of memory structure to be looked for in the nervous system.