Fig. 3.

Simulation of pursuit of a partially occluded target. This figure reports the posterior or conditional expectations about hidden states and causes during the simulation of oculomotor pursuit movements over one cycle of target motion. The position of the occluder is illustrated in all panels by the parallel grey lines – these do not always coincide with changes in state estimates because the latter are calculated in 16 ms time steps. The upper left panel shows the proprioceptive predictions (blue line) and prediction errors (dotted red lines). The upper middle panel shows the predicted retinal input – the red, cyan and purple lines correspond to the middle three of an array of 17 photoreceptors (the target is centrally fixated) – and the dotted red lines are prediction errors. The sensory predictions are based upon the expectations of hidden oculomotor (blue line) and target (red line) angular displacements shown on the lower left; the corresponding velocities are shown as the green (eye) and cyan (target) lines on the lower middle graph. The grey regions correspond to 90% Bayesian confidence intervals. Note the increase in uncertainty about the location of the target during periods of occlusion. The hidden cause of these displacements (broken black line) is shown with its conditional expectation (blue line) in the lower right panel. The true cause and action are shown on the upper right. The action (blue line) is responsible for oculomotor displacements and is driven by proprioceptive prediction errors.