Multi STF MT model

V1 normalization
The V1 normalization is determined by the <norm> parameter object within the <v1> object. An example of this is shown below, followed by a description of the parameters.
  <norm>
    type     local_stf  # Type of normalization: 'local_stf'
    alpha_1    0.99     # Tuned normalization
    alpha_2    0.01     # Untuned normalization
    alpha_3    0.014    # Stimulus-independent
  </norm>
The normalized response, r^*, in a given channel is computed from the raw response, r, in that channel as follows:
  r^* = r / d
where d depends on the type of normalization.
If the normalization type is 'local_stf', then the alpha parameters (shortened to 'a' below), are used as follows:
  d  =  a1 * r  +  a2 * sum/m  +  a3,
where the sum is taken over all SF, TF and directions at the local region of the receptive field of the V1 signal being normalized. In other words, the normalization is spatially local. The value m is the number of STF channels times the number of directions, thus the number of signals added in sum.