Neural computation of motion in the fly visual system: quadratic nonlinearity of responses induced by picrotoxin in the HS and CH cells

Abstract

1. A computational model accounting for motion detection in the fly was examined by comparing responses in motion-sensitive horizontal system (HS) and centrifugal horizontal (CH) cells in the fly's lobula plate with a computer simulation implemented on a motion detector of the correlation type, the Reichardt detector. First-order (linear) and second-order (quadratic nonlinear) Wiener kernels from intracellularly recorded responses to moving patterns were computed by cross correlating with the time-dependent position of the stimulus, and were used to characterize response to motion in those cells. 2. When the fly was stimulated with moving vertical stripes with a spatial wavelength of 5-40 degrees, the HS and CH cells showed basically a biphasic first-order kernel, having an initial depolarization that was followed by hyperpolarization. The linear model matched well with the actual response, with a mean square error of 27% at best, indicating that the linear component comprises a major part of responses in these cells. The second-order nonlinearity was insignificant. When stimulated at a spatial wavelength of 2.5 degrees, the first-order kernel showed a significant decrease in amplitude, and was initially hyperpolarized; the second-order kernel was, on the other hand, well defined, having two hyperpolarizing valleys on the diagonal with two off-diagonal peaks. 3. The blockage of inhibitory interactions in the visual system by application of 10-4 M picrotoxin, however, evoked a nonlinear response that could be decomposed into the sum of the first-order (linear) and second-order (quadratic nonlinear) terms with a mean square error of 30-50%. The first-order term, comprising 10-20% of the picrotoxin-evoked response, is characterized by a differentiating first-order kernel. It thus codes the velocity of motion. The second-order term, comprising 30-40% of the response, is defined by a second-order kernel with two depolarizing peaks on the diagonal and two off-diagonal hyperpolarizing valleys, suggesting that the nonlinear component represents the power of motion. 4. Responses in the Reichardt detector, consisting of two mirror-image subunits with spatiotemporal low-pass filters followed by a multiplication stage, were computer simulated and then analyzed by the Wiener kernel method. The simulated responses were linearly related to the pattern velocity (with a mean square error of 13% for the linear model) and matched well with the observed responses in the HS and CH cells. After the multiplication stage, the linear component comprised 15-25% and the quadratic nonlinear component comprised 60-70% of the simulated response, which was similar to the picrotoxin-induced response in the HS cells. The quadratic nonlinear components were balanced between the right and left sides, and could be eliminated completely by their contralateral counterpart via a subtraction process. On the other hand, the linear component on one side was the mirror image of that on the other side, as expected from the kernel configurations. 5. These r...

Dendritic Computation of Direction Selectivity and Gain Control in Visual Interneurons

“…Our finding of an increased PD response together with a decreased change of input resistance only agrees with model 1, i.e., the assumption of weak directional tuning of EMDS. This is in accordance with previous suggestions Kondoh et al, 1995) and with recordings from T4-cells (Douglass and Strausfeld, 1996), a columnar cell type for which synaptic contacts onto L P TC s have been demonstrated and that might represent the output component of an EMD (Strausfeld and Lee, 1991). Thus, as is proposed for other motion-sensitive cells in vertebrates (Levick et al, 1969;Snowden et al, 1991), direction selectivity is significantly enhanced through the opponent action of local input signals on the dendrites in fly L P TC s.…”

Mutation of theDrosophilavesicular GABA transporter disrupts visual figure detection

“…Both cell ablation and pharmacologic studies using picrotoxin indicate that GABA release from CH cells is required for the small field tuning of the figure detection cell FD1 Warzecha et al, 1993). Application of picrotoxin also disrupts the function of HS, VS and other LPTCs (Egelhaaf et al, 1990;Gilbert, 1990;Kondoh et al, 1995;Schmid and Bülthoff, 1988). More generally, injection of picrotoxin into the abdomen of Drosophila inverts the polarity of both the cellular and behavioral response to movement (Bülthoff and Bülthoff, 1987).…”

Two classes of visual motion sensitive interneurons differ in direction and velocity dependency ofin vivo calcium dynamics

“…Thus, they represent an example of how differing voltagedependencies of calcium influx contribute to transform an input signal, which cannot be distinguished on the systems level (Kondoh et al, 1995), into a chemical signal with an altered stimulus-dependency, such as a different velocity tuning. To our knowledge, the present study of CH-and HS-cells is the first example of a voltage-[Ca 2ϩ ] i relationship measured in vivo and in response to sensory stimulation.…”

“…Kondoh et al, 1995) suggest that they receive similar input by pooling retinotopic information, possibly from the same set of ipsilateral elementary motion detectors . However, their anatomical vicinity, receptive field, directional selectivity, and local input-output characteristics (e.g.…”

“…However, their anatomical vicinity, receptive field, directional selectivity, and local input-output characteristics (e.g. Kondoh et al, 1995) suggest that they receive similar input by pooling retinotopic information, possibly from the same set of ipsilateral elementary motion detectors . Therefore, dendritic calcium accumulation can be studied under identical stimulus conditions, and differences in the cellular responses are likely to reflect different cellular properties.…”