Continuous attractors and oculomotor control

摘要: A recurrent neural network can possess multiple stable states, a property that many brain theories have implicated in learning and memory. There is good evidence for such multistability the brainstem controls eye position. Because states are arranged continuous dynamical attractor, store memory of position with analog encoding. Continuous attractors model networks depend on precisely tuned positive feedback, their robust maintenance requires mechanisms synaptic plasticity. These ideas may wider scope than just oculomotor system. More generally, internal models postulated by biological motor control be attractors.