Placegraph and equivalent viewgraph from Gillner and Mallot (1998)
There is a pervasive assumption in neuroscience that 'sensory' representations are built up and then 'sensori-motor transformations' allow these to be used by the motor system. These transformations are supposed to be 3D rotations and translations of coordinate frames. Nothing of that kind is proposed here. Instead, the motor system provides a method of moving across a sensory+motivational manifold. It is sometimes said that the sensory system is only there to let the motor system know that it has achieved its goal. One could equally say that the motor system is only there to change the sensory state. These statements are two sides of the same coin. The important thing is that we need only posit one representation and a method of moving within it.