Cannot deal with unfamiliar contexts
To pick a motor output (or equivalent), the brain needs to compare the current sensory+motivational state with a long list of stored contexts in the same space. This implies that every possible sensory+motivational context is catered for; the current state is always nearest to “some” stored context so that the brain always has an output pre-programme to deal with that input. Is that feasible?
The sensory+motivational contexts that we actually experience are a very small subset of the theoretically possible set. If, for a period of 5 seconds and without warning, a person’s photoreceptors were stimulated randomly and random input applied to the cochlea, skin receptors and other sensory inputs, most people would probably curl up in a ball with their hands over their head, extremely frightened. Such a default response might apply to the vast majority of possible sensory+motivational space (i.e. it is covered by one very large Voronoi cell.
A more typical example of a ‘unfamiliar context’, one that has been raised several times in discussion of the ideas in these pages, is the case of a person walking into a scene, like a room, that they have never previously visited. How can the person imagine the scene from the other side of the room if they have never been there before? Surely, there are no pre-prepared paths through sensory+motivational space for them to move along that would allow them to ‘think themselves’ to the other side of the room? It might seem obvious that people are able to do this task (imagine correctly what a scene will look like from the other side of a room) and that a scheme based entirely on stored contexts could not achieve this. In fact, there is evidence against both.
Back to objections.