LOCALIZING OBJECTS IN SPACE AND TIME: A COMPUTATIONAL MODEL BASED ON A PREDICTIVE MECHANISM
A computational model is presented, which can account for spatial mislocalizations. In fact, there are small but consistent errors in position judgement for a moving object at both its starting point (both in- and opposite to- the direction of motion) and vanishing point (in the direction of motion). Our model is based on motion extrapolation, that we consider as a process, wherein each step of information processing contributes in minimizing the spatial error due to the neural delay. The model presented here mimics the first step of the extrapolation process, where an early mechanism is supposed to work at a very first level of the information processing, (i.e. at retinal and sub-cortical level). Our model can reproduce all the SP and VP mislocalizations, as a consequence of its described strategy, as well as their dependence on velocity.