Interneurons exhibiting centre-surround antagonism within their recep­tive fields are commonly found in peripheral visual pathways. We propose that this organization enables the visual system to encode spatial detail in a manner that minimizes the deleterious effects of intrinsic noise, by exploiting the spatial correlation that exists within natural scenes. The antagonistic surround takes a weighted mean of the signals in neighbouring receptors to generate a statistical prediction of the signal at the centre. The predicted value is subtracted from the actual centre signal, thus minimizing the range of outputs transmitted by the centre. In this way the entire dynamic range of the interneuron can be devoted to encoding a small range of intensities, thus rendering fine detail detectable against intrinsic noise injected at later stages in processing. This predictive encoding scheme also reduces spatial redundancy, thereby enabling the array of interneurons to transmit a larger number of distinguishable images, taking into account the expected structure of the visual world. The profile of the required inhibitory field is derived from statistical estimation theory. This profile depends strongly upon the signal: noise ratio and weakly upon the extent of lateral spatial correlation. The receptive fields that are quantitatively predicted by the theory resemble those of X-type retinal ganglion cells and show that the inhibitory surround should become weaker and more diffuse at low intensities. The latter property is unequivocally demonstrated in the first-order interneurons of the fly’s compound eye. The theory is extended to the time domain to account for the phasic responses of fly interneurons. These comparisons suggest that, in the early stages of processing, the visual system is concerned primarily with coding the visual image to protect against subsequent intrinsic noise, rather than with reconstructing the scene or extracting specific features from it. The treatment emphasizes that a neuron’s dynamic range should be matched to both its receptive field and the statistical properties of the visual pattern expected within this field. Finally, the analysis is synthetic because it is an extension of the background suppression hypothesis (Barlow & Levick 1976), satisfies the redundancy reduction hypothesis (Barlow 1961 a, b) and is equivalent to deblurring under certain conditions (Ratliff 1965).

1. The impulse-response was used to measure the dynamics of the photoresponse of 8 species of insects from 6 orders in both light-and dark-adapted states.2. The impulse-responses of all cells were well fitted by the two-parameter log-normal curve.3. In the dark-adapted state, the time-to-peak of the response varies from 38 ms in the drone-fly to 55 ms in the locust. Though interspecies variation is small, the house-fly Musca (41 ms) is significantly faster than the locust. In the light-adapted state, there are highly significant variations in the time-to-peak between species. The order is: housefly (12.0ms), drone-fly (16.5ms), dragonfly (17.5 ms), mantid (18.1 ms), locust (21.9 ms) and cricket (22.1 ms). This variation in speed correlates with flight behavior.4. There are significant, though small, differences in the shape of the dark-adapted impulseresponse, with that of the cockroach more symmetrical and the dragonfly more skew than the others. The impulse-response of the fly in the lightadapted state is more symmetrical than that of the other species and results in an even higher frequency response.5. Despite these differences in shape, it is concluded that all species have a similar transduction mechanism. Interspecies differences in time-scale can, at first approximation, be accounted for by the change of a single time-constant.6. The insects' impulse-responses were compared to those of verbrates by using the cascade Present addresses : models of Fuortes and Hodgkin (1964) and Baylor et al. (1974). A large number of stages were required (between 10 and 50) and a greater than 50% variation in the number of stages was needed in order to fit response from different cells within a single species. Furthermore, the basic assumption of Fuortes and Hodgkin (1964) that the timecourse is causally linked to the gain does not hold in the insect. We conclude that no first-order system of chemical cascades can sensibly predict either the time-course of the photoresponse in insects, or the effects of light adaptation and hence that the insect transduction mechanism is fundamentally different to that of vertebrates. Finally, we find that a model using two first order poles, two underdamped second order poles and a pure time delay (French 1980a, b) provides as good a fit to the frequency response as does the log-normal model.

No abstract

No abstract

SUMMARY1. The contrast sensitivity of the optomotor response of the fly Musca domestic was measured using a moving sinusoidal grating as the stimulus. In parallel experiments intracellular recordings were made from photoreceptors and first order visual interneurones to determine their responses to the same threshold stimuli. Measurements of the spatial modulation transfer function for photoreceptors confirm that the optics of the eye were intact during recordings.2. At the lowest intensity at which one can obtain an optomotor response, the photoreceptor signal is a train of discrete depolarizations, or bumps. With constant intensity stimuli, the temporal distribution of bumps follows the Poisson distribution with a mean rate proportional to luminance. The mean bump rate at the threshold intensity for a behavioural response is 1-7 + 0-7 sl (mean+s.D., n = 25).3. Calibrations and the statistical properties of the bump train indicate that a bump represents one effective photon, implying that the bump: photon ratios are quantum capture efficiencies.4. At low intensities the first order interneurones (the large monopolar cells or LMCs) show hyperpolarizing bumps each triggered by a receptor bump. Using a point source stimulus, centred in the field of view, the LMC bump rate is six times that in a single receptor viewing the same stimulus, as expected from the known projection of six receptor axons to each LMC. When using an extended stimulus (the grating), the bump rate is 18-20 times that in receptors. Comparison with earlier work suggests that this increased lateral summation of receptor inputs to LMCs only occurs at very low intensities.5. In both receptors and LMCs the amplitudes and wave forms of bumps depend upon the position of a point source stimulus within the field of view. With the light in the periphery of the field the bumps are smaller and slower than when the light is in the centre. This difference in response suggests that spatial summation is brought about by lateral interactions, possibly between receptors.6. At higher mean intensities the signal-to-noise ratios in receptors responding to the appropriate threshold stimuli increase with intensity. This is suggestive of a decrease in the extent of spatial and/or temporal summation in the optomotor pathway.