EFFECT OF PASSIVE EYE POSITION CHANGES ON RETINOGENICULATE TRANSMISSION IN THE CAT

1. Extracellular recordings were made from single neurons in layer A of the left dorsal lateral geniculate nucleus (LGN(d)) of anesthetized and paralyzed adult cats. Responses to retinotopically identical visual stimuli (presented through the right eye) were recorded at several positions of the left eye in its orbit. Visual stimuli consisted of drifting sinusoidal gratings of optimal temporal and spatial frequencies at twice threshold contrast. Visual stimulation of the left eye was blocked by a variety of methods, including intravitreal injection of tetrodotoxin (TTX). The change in position of the left eye was achieved by passive movements in a randomized and interleaved fashion. Of 237 neurons studied, responses were obtained from 143 neuron on 20-100 trials of identical visual stimulation at each of six eye positions. Neurons were classifed as X- or Y- on the basis of a standard battery of physiological tests (primarily linearity of spatial summation and response latency to electrical stimulation of the optic chiasm). 2. The effect of eye position on the visual response of the 143 neurons was analyzed with respect to the number of action potentials elicited and the peak firing rate. Fifty-seven (40%) neurons had a significant effect [by one-factor repeated-measure analysis of variance, (ANOVA), P < 0.05] of eye position on the visual response by either criterion (number of action potentials or peak firing rate). Of these 57 neurons, 47 had a significant effect (P < 0.05) with respect to the number of action potentials and 23 had a significant effect (P < 0.05) by both criteria. Thus the permissive measure by either criterion and the conservative measure by both criteria resulted in 40% and 16%, respectively, of all neurons' visual responses being significantly affected by eye position. 3. For the 47 neurons with a significant effect of eye position (number of action potentials criterion), a trend analysis of eye position versus visual response showed a linear trend (P < 0.05) for 9 neurons, a quadratic trend (P < 0.05) for 32 neurons, and no significant trend for the 6 remaining neurons. The trends were approximated with linear and nonlinear gain fields (range of eye position change over which the visual response was modulated). The gain fields of individual neurons were compared by measuring the normalized gain (change in neuronal response per degree change of eye position). The mean normalized gain for the 47 neurons was 4.3. 4. The nonlinear gain fields were generally symmetric with respect to nasal versus temporal changes in eye position. There were two types of nonlinear gain fields: the inhibitory gain fields (69%, 22/32) had a maximum response within ± 5° of the resting eye position; the facilitatory gain fields (31%, 10/32) had a minimum response within ± 5° of resting eye position. 5. The linear gain fields had a preference for the direction of eye position change such that the maximal response was observed at the extreme temporal or nasal positions studied for each neuron. 6. For 25% (35/143) of the neurons studied, visually elicited activity (visual stimulus contrast = 2x threshold for the individual neuron) and spontaneous activity (visual stimulus contrast = 0) were evaluated at each of six eye positions. Nine of these 35 neurons had a statistically significant (P < 0.05) effect of eye position on the visually elicited response, whereas the effect of eye position on spontaneous activity was not significant (P > 0.05). The relative strength (R) of the eye position effect on the visually elicited response versus spontaneous activity was calculated for each neuron. For most of the neurons (22/35), R was > 3. This suggests that the eye position signal does not interact with the retinal drive to the LGN(d) neurons in a simple linear fashion. 7. The eye position effect is more likely to occur among X- (25/58, 43%) than Y- (18/75, 24%) neurons (χ2 test, P < 0.025). 8. Among X-cells, the eye position effect occurred more frequently with increasing RF eccentricity (χ2 test, P < 0.001). This was not the case for Y-cells (P > 0.05). 9. Passive changes in eye position did not affect retinal ganglion cell activity, and retrobulbar block with local anesthetic eliminated the effect of eye position on LGN(d) neurons. 10. These results suggest that an afferent eye position signal can selectively gate the transfer of visual signals at the retinogeniculate synapse. This gating of retinal signals is stronger and more likely to occur for X-cells than Y-cells and for neurons with receptive fields located outside of central vision. A model is described where a contextual signal for stimulus localization in head-centered space is provided by integration of visual and non-visual (afferent eye position) signals across the geniculostriate pathway.