SUMMARY1. This paper introduces a new technique for the analysis of the chromatic properties of neurones, and applies it to cells in the lateral geniculate nucleus (l.g.n.) of macaque. The method exploits the fact that for any cell that combines linearly the signals from cones there is a restricted set of lights to which it is equally sensitive, and whose members can be exchanged for one another without evoking a response.2. Stimuli are represented in a three-dimensional space defined by (a) an axis along which only luminance varies, without change in chromaticity, (b) a 'constant B' axis along which chromaticity varies without changing the excitation of blue-sensitive (B) cones, (c) a 'constant R & G' axis along which chromaticity varies without change in the excitation of red-sensitive (R) or green-sensitive (G) cones. The orthogonal axes intersect at a white point. The isoluminant plane defined by the intersection of the ' constant B' and 'constant R & G' axes contains lights that vary only in chromatic As j In polar coordinates the constant B axis is assigned the azimuth 0-180 deg, and the constant R & G axis the azimuth 90-270 deg. Luminance is expressed as elevation above or below the isoluminant plane (-90 to + 90 deg).3. For any cell that combines cone signals linearly, there is one plane in this space, passing through the white point, that contains all lights that can be exchanged silently. The position of this 'null plane' provides the 'signature' of the cell, and is specified by its azimuth (the direction in which it intersects the isoluminant plane of the stimulus space) and its elevation (its angle of inclination to the isoluminant plane).4. A colour television receiver was used to produce sinusoidal gratings whose chromaticity and luminance could be modulated along any vector passing through the white point in the space described. The spatial and temporal frequencies of modulation could be varied over a large range. A. M. DERRINGTON, J. KRA USKOPF AND P. LENNIE from only R and G cones. These we call R-G cells. The null planes ofthe smaller group were narrowly distributed about an azimuth of 178-4 deg and an elevation of deg, which suggests that these cells receive inputs from B cones almost equally opposed by some combined input from R and G cones. We call these B-(R & G) cells. No cells were found that lacked chromatic opponency.6. By assuming that the spectral sensitivities of the macaque's cones are like those of man's, the azimuths and elevations of the null planes can be transformed by the use of Smith & Pokorny's (1975) fundamental spectral sensitivities to yield the weights attached by each cell to signals from the three classes of cone. This representation shows that cells that receive inputs from B cones have these inputs opposed by varying combinations of inputs from R and G cones.7. Raising the spatial frequency of a grating systematically reduced the elevations but did not systematically alter the azimuths of the null planes of parvocellular units. This change, which was more pronounc...

SUMMARY1. The discharges of single neurones in the parvocellular and magnocellular laminae of the macaque's lateral geniculate nucleus (l.g.n.) were recorded with glass-insulated tungsten micro-electrodes.2. Linearity of spatial summation was examined using the test devised by Hochstein & Shapley (1976). 2 of 272 parvocellular units and 6 of 105 magnocellular units showed clearly non-linear spatial summation. A quantitative index of nonlinearity did not suggest the existence of a distinct 'non-linear' class of magnocellular unit.3. Spatial contrast sensitivity to moving gratings was measured by a tracking procedure in which contrast was adjusted to elicit a reliable modulation of discharge. With the exception of cells that were driven by blue-sensitive cones, measurements of contrast sensitivity did not reveal distinct subgroups of parvocellular units. All had low sensitivity, and those with receptive fields in the fovea could resolve spatial frequencies of up to 40 cycles deg-'. Magnocellular units had substantially higher sensitivity, but poorer spatial resolution.4. The higher sensitivities of magnocellular units led to their giving saturated responses to stimuli of high contrast. Responses of parvocellular units were rarely saturated by any stimulus.5. At any one eccentricity the receptive fields of parvocellular units had smaller centres than did those of magnocellular units. Receptive fields of magnocellular units driven by the ipsilateral eye had larger receptive fields than did those driven by the contralateral eye.6. Parvocellular units were most sensitive to stimuli modulated at temporal frequencies close to 10 Hz; magnocellular units to stimuli modulated at frequencies nearer 20 Hz. The loss of sensitivity as temporal frequency fell below optimum was more marked in magnocellular than parvocellular units.7. Changes in temporal frequency altered the shapes of the spatial contrast sensitivity curves of both parvocellular and magnocellular units. These changes could be explained by supposing that centre and surround have different temporal properties, and that the surround is relatively less sensitive to higher temporal frequencies.

4. Spatial contrast sensitivity curves of X cells and of on-centre Y cells could be described by a model of the receptive field as two concentric Gaussian sensitivity profiles representing the centre and the antagonistic surround.5. Changes in temporal frequency altered the shapes of the spatial contrast sensitivity curves of most units. For X cells sensitivity at the optimum spatial frequency was greater at a temporal frequency of 10-4 Hz than at lower or higher temporal frequencies. The relative sensitivity to low spatial frequencies improved as temporal frequency was raised from 0-16 to 20-8 Hz. The shapes of the contrast sensitivity functions of Y cells were less affected by changes in temporal frequency: at all spatial frequencies sensitivity was greater at 2-6 Hz than at lower or higher frequencies.6. The effect of temporal frequency upon the shape of the spatial contrast sensitivity curve could be explained by assuming that the centre and surround changed their sensitivities without changing their characteristic radii. A simple model, using a temporal R-C filter in the surround pathway, predicted qualitatively similar changes in the shape of contrast sensitivity curves but failed to provide acceptable fits to the observations. A second model, which assumed that surround * Present address:

It is well established that the responses of neurons in the lateral geniculate nucleus (LGN) can be modulated by feedback from visual cortex, but it is still unclear how cortico-geniculate afferents regulate the flow of visual information to the cortex in the primate. Here we report the effects, on the gain of LGN neurons, of differentially stimulating the extraclassical receptive field, with feedback from the striate cortex intact or inactivated in the marmoset monkey, Callithrix jacchus. A horizontally oriented grating of optimal size, spatial frequency, and temporal frequency was presented to the classical receptive field. The grating varied in contrast (range: 0–1) from trial to trial, and was presented alone, or surrounded by a grating of the same or orthogonal orientation, contained within either a larger annular field, or flanks oriented either horizontally or vertically. V1 was ablated to inactivate cortico-geniculate feedback. The maximum firing rate of LGN neurons was greater with V1 intact, but was reduced by visually stimulating beyond the classical receptive field. Large horizontal or vertical annular gratings were most effective in reducing the maximum firing rate of LGN neurons. Magnocellular neurons were most susceptible to this inhibition from beyond the classical receptive field. Extraclassical inhibition was less effective with V1 ablated. We conclude that inhibition from beyond the classical receptive field reduces the excitatory influence of V1 in the LGN. The net balance between cortico-geniculate excitation and inhibition from beyond the classical receptive field is one mechanism by which signals relayed from the retina to V1 are controlled.

SUMMARY1. Extracellular recordings were obtained from units in the dorsal lateral geniculate nucleus of anaesthetized cats.2. Of sixty-nine units, sixty-three could be unambiguously identified as either X (n = 33) or Y (n = 30) by testing the presence of a null response to stationary sine wave gratings presented in different spatial phases.3. In response to stationary gratings flashed on and off, Y cells exhibited bigger, more transient responses than X cells.4. All Y cells but few X cells exhibited a shift effect (modulated periphery effect).5. In response to drifting sine wave gratings of different spatial frequencies, X cells preferred higher spatial frequencies and showed smaller peak contrast sensitivities and somewhat narrower tuning curves than Y cells.6. In response to a sine wave grating of optimal spatial frequency drifting at different velocities, X and Y cells had similar temporal tuning curves. However, Y cells, largely because they preferred lower spatial frequencies, preferred higher drift velocities than X cells.7. Our data suggest that X and Y cells can be differentiated objectively on the basis of a number of discharge parameters. These parameters are compared with similar data collected by others from neurones in the visual cortex.