eye movements occur whenever we follow a moving object with our eyes. During
pursuit, the visual system transforms the visual information coming from the
eyes to a world-centered reference frame. Physiological and psychophysical
studies had shown that this transformation is achieved via integration of
visual information arriving from the eyes, with an internal motor signal
indicating eye movements. In the present study we constructed a physiologically
based neural network model consisting of three layers of computational units,
simulating V1, MT and MST. The units in the third layer, like some neurons in
area middle-superior-temporal (MST), receive, in addition to visual input, an
extra-retinal signal related to eye movements. The excitatory and inhibitory
connections to the MST layer developed during an unsupervised training process.
The resulting MST units represent a transformation from retinal to real-world
centered coordinates. We compared the response properties of these units with
physiology, studied their relation to the input connectivity patterns, and
studied the sensitivity of the training process to various training parameters.
Next, we studied a pursuit related perceptual phenomena: The alteration of the
perceived path of a moving target during pursuit of a circularly moving target.
We studied the phenomenon experimentally, with a greater number of subjects
then in previous studies, and under various parameter values. We performed
model simulations of the phenomenon, and showed that simulation results match a
variety of related psychophysical findings. By analyzing the model simulations,
the experimental results were explained in terms of underlying neuronal
mechanisms. Finally, we used the model to address the fundamental controversy
regarding the relative weight of the motor signal during pursuit.