Information Technology Reference
In-Depth Information
to the retinal inputs, geniculate relay cells are innervated by non-retinal inputs
that account for approximately 90%-95% of the total synaptic inputs. They come
from local GABAergic neurons, reticular cells, interneurons, and also from two
sources of extrinsic inputs, i.e a feedback projection from cortical layer 6 and an
ascending projection from various scattered cell groups in the brainstem retic-
ular formation. This complex circuitry seems to play a role that could be more
relevant that just being a relay step. Besides the complicate schema of the in-
puts, we should take into account that the number of relay cells in the dLGN
of the cat doubles the number of retinal ganglion cells [1]. Moreover, inhibition
in the dLGN does not come from the retina, but locally from the interneurons
that receive inputs from retinal ganglion cells of the opposite sign, and it was re-
ported that the number of interneurons is half of the number of retinal ganglion
cells [1]. The issue about how many connections are established by each relay
cell with the retina and how many with the local interneurons, was addressed
by different authors (Pei et al.,1991; Hirsch et al.,1995; Martinez et al.,2005),
who carried out whole cell recordings in the dLGN of the cat while stimulating
with sparse-noise protocol (Jones and Palmer,1987). With this kind of record-
ings, ON and OFF maps(receptive field maps when stimulated with a white and
dark square) were obtained. By comparing both maps, two main mismatches
were detected. First, the center of the ON map is bigger than the center of the
OFF map, this difference being measured with the size index(SI). Second, their
peaks are displaced, this feature being measured with the overlap index (OI).
The differences between the ON and the OFF map represent a good opportunity
to know more about the connectivity of the system as the two maps come from
different pathways. While the ON map come directly from the retinal ganglion
cells, the inhibition comes through the interneurons. In order to know which
are the connectivity configurations (convergence/divergence) that best fit the
data obtained in whole cell recordings, Molano and Martinez (2009) [14] built a
retino-thalamic computational model similar to [2] from which we can obtain the
number of connections between system elements that best fit the experimental
data. The values obtained are consistent with those deduced by other authors
using a different methodology [3].
2TheProbem
The goal of this paper is to explore the type of spatio-temporal processing that
is taking place in the dLGN, with the convergence/divergence data of the push-
pull circuit derived from the computational model developed by Molano and
Martinez (2010) [14]. Concretely, we face the following questions: 1. Center-
surround receptive fields from channel ON are suciently effective in edge de-
tection? 2. What makes the asymmetry of the ON and OFF channels and the
divergence/convergence of information processing that the retina sends to the
dLGN? 3. Could the push-pull circuitry increase the spatial resolution?.
Search WWH ::
Custom Search