Dynamics of context modulations in the "reach" network

Marina Brozovic

The pro-reach/anti-reach task is well suited for studying the way in
which context configures cortical operations.  The sensory inputs are
the same, only the context/state of the animal is different.  Yet these
context differences lead to completely different neural processes and
responses.  Our neural network study is based on the experimental
evidence which shows that the neurons in the parietal reach region (PRR)
predominantly encode the position of motor goal as opposed to the
position of the visual cue. Furthermore, the motor goal representation
in PRR is delayed by ~60 ms in anti- compared to pro-reaches, implying
the need for an additional step of cortical processing.  Based on our
recurrent network models, we propose that the basic reach network
(visual areas ->PRR->PMd) has mainly feed-forward nature for the
instructed pro-reaches, because the context information is not
necessary. On the contrary, the reach network strongly depends on
the feed-back projections from the PMd to the PRR for the case of
anti-reaches, because these pathways bring in the context
information for the anti-reaches.