A new way of linking information theory with cognitive science

The relationship between the notion of *information* in information theory, and the notion of *information processing* in cognitive science, has long been controversial. But as the present paper shows, part of the disagreement arises from conflating different formulations of measurement. Clarifying distinctions reveals it is the context-free nature of Shannon's information average that is particular problematic from the cognitive point of view. Context-sensitive evaluation is then shown to be a way of addressing the problems that arise.