README.txt

Binocular receptive field and responses to random-dot stereogram of a complex cell based on disparity energy model

Authors: Kota S. Sasaki and Izumi Ohzawa
Graduate School of Frontier Biosciences, Osaka University
kota@fbs.osaka-u.ac.jp, ohzawa@fbs.osaka-u.ac.jp

License: BSD license, http://creativecommons.org/licenses/BSD/

Date: 2007-12-20

------------------------------------------

RDSscript.m:

This Matlab script generates a pair of random-dot stereograms with
a central vertical strip sticking out (closer to you) from the background on each side.

DisparityEnergyModelscript.m:

This Matlab script computes binocular receptive fields of a complex cell
and its linear subunits (simple cells).
Computed results are displayed in 3 windows.  Captured version of these
are in results/DisparityEnergyModelscript_1.png, ..2.png, and ..3.png.

Figure 1: Binocular and monocular receptive fields of a pair of subunits.
Figure 2: Binocular and monocular receptive fields of a pair of subunits which are in quadrature relationship with the others shown in Figure 1.
Figure 3: Binocular receptive field of the complex cell, tuned to non-zero disparity.

DisparityDetectionscript.m:

This Matlab script computes the responses of a complex cell to a random-dot stereogram.
The model disparity model is implemented by 1-d profiles monocularly, which requires
only one scan-line worth of data.  Therefore, the plot represents
the sum of activities from all rows (scan lines) of the random-dot stereogram as viewed by
this 1-d complex cell model.  What this model does is not that different from that of
an actual complex cell.  If we take the 1-d profile we modeled as representing the
X-dimension of the receptive fields, the Y-dimension (direction parallel to the orientation)
of a receptive field is generally a Gaussian.  Real cells weight the Y-dimension according
to a Gaussian.  The simplistic model sums the Y-dimension with equal weight (rectangular
windowing).  Each time one runs this script, a different random-dot stereogram is
generated, and one can observe the variability in the responses due to stimulus variations.
In spite of these variations, it is clear that the responses of the complex cell in the
central strip region is consistently higher than the cell's response when viewing
the 0-disparity background regions.






