Illustration of ON- and OFF-center Receptive Fields
Put electrode either in ganglion cell layer of the retina or in optic nerve. The response of the ganglion cell will depend upon the responses of the cells that feed into the ganglion cell, including the photoreceptors, the bipolar cells and the various lateral interconnections via horizontal cells and amacrine cells. However, what we are really interested in is the relationship between this ganglion cell's activity (firing rate) and the visual stimulus image.
H.K. Hartline of Rockefeller University in New York did the first important experiments on this, working in the frog. Later Stephen Kuffler at Harvard University, and Horace Barlow at Cambridge University in England also did some of the pioneering work on this topic in the early 1950s, working on the visual systems of both frogs and cats. They analyzed the ganglion cells' sensitivity to light by using the following simple technique. A very small spot of light was flashed on the surface of the retina for a brief duration. The position of the spot of light was systematically varied across the retinal surface, and while they did this the response of the ganglion cell was continuously monitored.
Illustration of ON- and OFF-center Receptive Fields
The icons on the left represent the various visual stimuli. Next to that is the signal recorded by the electrode for each stimulus. Spontaneous firing rate of a few spikes/sec measured when no stimulus presented. For most positions on the surface of the retina, flashing a spot of light has absolutely no effect on the cell's response (that is, it continues responding at its spontaneous firing rate). Within a particular region, called the receptive field, flashing the spot affects the ganglion cell's response. Can think of receptive field as either:
Receptive field subregions: The area within the receptive field is subdivided into two regions, center and surround. Two types of ganglion receptive fields:
ON-center and OFF-center ganglion cells is another example of a parallel pathway. Physiologically distinct (as just described above). Anatomically distinct: located in separate retinal sub-layers (ON inner-, OFF outer-plexiform layer). Complete coverage: ON-center and OFF-center receptive fields each completely cover the visual field. ON- and OFF- pathways play off of one another in the retina and the LGN, then merge completely in the complex cells in V1.
Ganglion cell receptive field video
Notion of receptive field is a central one in all of visual neuroscience. Will see many examples of different kinds of receptive fields - not all visual neurons have this simple center-surround arrangement to their receptive fields.
Test Additivity Rule
Data from experiments by Christina Enroth-Cugell. Response to center stimulus alone plus response to surround stimulus alone equals response to simultaneous stimulation of center+surround. That is, obeying additivity rule. Figure doesn't show tests of scalar rule and shift-invariance but those rules hold up well too.
Sinusoid In, Sinusoid Out
Another Enroth-Cugell experiment. Sinusoid in, sinusoid out, as expected of a shift-invariant linear system.
Ganglion cell response is a weighted sum of stimulus intensities, positive weights in ON subregions, negative weights in OFF subregions.
Examples of how to calculate/predict response
Add light intensities at each point weighted by + and - weights corresponding to center and surround subregions of the receptive field. Result is a pattern of activity across the spatial array of ganglion cells.
John Robson coined the term neural image to refer to the distribution across the retina of ganglion cell responses. Since ganglion cells are linear systems, we can predict their responses to any visual stimulus.
Example neural image
Here's an example: a computer simulation of the retinal ganglion cell neural image in response to my kid and her bunny. Mid-gray means no response. White means big response by the ON-center ganglion cells. Black means big response by the OFF-center cells. What's the point of center-surround receptive fields? Emphasizes edges.