A mechanism for directional selectivity

In sensory systems, there are many examples of neurons that act as filters, responding most vigorously to one particular stimulus configuration. The selectivity of these neurons for their preferred stimulus must arise from the way in which their presynaptic neurons are arranged, and study of fly lobula plate neurons has enabled insights into the way in which one particular kind of selectivity - directional selectivity - arises. The neurons are considered to be driven by an array of local motion-detecting units, each of which detects the movement of images in a particular direction over a small part of the eye. These units are commonly referred to as elementary motion detectors. Their properties are known in some detail, although the neurons of which they are composed have not been identified. Many of the properties of elementary motion detectors were first discovered in the 1950s from experiments on the optomotor behaviour of beetles and flies, well before methods had been developed for recording signals from single neurons in the fly brain. More recently, their properties have been inferred by recording the responses by lobula plate neurons to moving stimuli that are seen by small regions of the eye, when the number of local motion detectors stimulated would be very small.

When the image of a moving object travels across an eye, different photoreceptors are stimulated in a particular order. Information about the direction of movement could, therefore, be obtained from the sequence in which the photoreceptors are stimulated. The open rhabdom design of the fly eye has enabled Nicholas Franceschini and colleagues (Franceschini, Riehle & Le Nestour, 1989) to perform a remarkable experiment in which direction-ally selective responses from a lobula plate neuron were elicited by stimulating just two photoreceptors in sequence (Fig. 5.7a). The neuron they chose to study was H1 (see Fig. 5.6b), which is the easiest lobula plate neuron to record from. A special optical instrument enabled an experimenter to view a single ommatidium and to direct a tiny spot of light onto each of two individual photoreceptors which view adjacent points in space. H1 is excited when the posterior receptor is stimulated just before the anterior receptor, and it is inhibited when the two receptors are stimulated in the reverse order (Fig. 5.7b). This corresponds with the directional selectivity by H1 for moving objects, which is forwards over its own eye.

Illumination of either receptor alone, or both at the same time, does not cause any response in H1;there must be a delay between stimulation of one and stimulation of the other. Illumination of the posterior receptor facilitates a response by H1 to stimulation of the anterior receptor. This facilitation does not start immediately when the posterior receptor is stimulated, but after a

Figure 5.7 Directional selectivity in the housefly (Musca) H1 neuron to stimulation of two individual photoreceptors. (a) A fly ommatidium viewed through its lens, showing the seven individual rhabdomeres that are visible. The anterior (a) and posterior (p) photoreceptors that were stimulated in the experiment are indicated. (b) Responses by H1 to different sequences of stimulating the two photoreceptors. The times when a pinpoint of light was directed at each photoreceptor are indicated. There was a vigorous response when a stimulus to the posterior receptor preceded that to the anterior, but almost no response when the two receptors were stimulated simultaneously. Note that the light-on and light-off stimuli generated separate peaks in the response. In another experiment (shown in the bottom trace), stimulation of the anterior receptor before the posterior one (the null direction) caused inhibition of H1. Each recording was an average from 100 stimulus repetitions, and shows the response of H1 as instantaneous spike frequency. (c) Diagram of an elementary motion detector. Open triangles represent the two photoreceptors. (b modified from Franceschini etal., 1989; copyright © 1989 Springer-Verlag.)

delay of at least 10 ms. It reaches maximum strength after 50 ms, and lasts for up to 250 ms. This means that H1 is most sensitive to movements that stimulate the anterior receptor 50 ms after the posterior receptor (these movements are produced by images travelling over the surface of the eye at a speed of 72 degrees per second). The delay in facilitation ensures that H1 is not excited when overall illumination of the eyes alters or if light intensity flickers, which would occur when the insect flies into or out of shade.

The wiring diagram in Fig. 5.7c summarises how a directionally selective motion detector, with a preferred direction from posterior (p) to anterior (a), might be organised. The diagram summarises how information is processed. The inputs are elements that respond to light-on or light-off stimuli, and each probably includes a photoreceptor and various neurons in the lamina and medulla. The signal from the posterior photoreceptor is delayed before it is combined with that from the anterior photoreceptor. The combination is indicated as a switch that enables one photoreceptor to regulate the output of the other, although in reality the way the outputs from the two photoreceptors is combined is more sophisticated than a simple on-off device. The two signals are thought to be multiplied together, which ensures that the output is very small when only one of the two receptors is stimulated (multiplying by zero gives a result of zero). The result is that a strong output from the motion detector is only produced when stimulation of the second photoreceptor follows stimulation of the first photo-receptor with a certain delay.

It would be reasonable to assume the medulla contains many such circuits, perhaps one per column, so that H1 is excited by movement in its preferred direction over any part of its eye. In fact, at each location there must be four copies of similar circuits. The first duplication is necessary to account for inhibition of H1 by stimuli moving backwards over the eye, the null direction. Inhibition is apparent in experiments in which H1 shows quite a high rate of spontaneous spike discharge in the absence of any movement stimulation, and sequential stimulation of the two photorecep-tors in the null direction briefly inhibits H1, reducing its spike rate. This inhibition shows the same time-dependent properties of facilitation as excitation by movement in the preferred direction. To account for this, there must be two mirror-image, motion-detecting circuits working together, one exciting H1 and the other inhibiting it. The second duplication, in which there are two of each excitatory and inhibitory circuit, is

Through slit

Through slit

Null Preferred

Figure 5.8 Responses by elementary motion detectors. (a) When a fly viewed a moving pattern through a vertical slit cut in a mask, only a narrow band of ommatidia and elementary motion detectors was stimulated. (b) A wide-field HS neuron responded to each lightening or darkening of this band, as shown in the intracellular recording. When the stimulus moved in the null direction, each response by the strip of motion detectors caused hyperpolarising potentials; and when the stimulus moved in the preferred direction, another set of motion detectors responded, causing depolarising potentials. When the mask was removed so the eye saw the whole moving pattern, responses from different strips of motion detectors summed so that sustained excitatory or inhibitory responses were recorded from the HS neuron. (b modified after Egelhaaf & Borst, 1993.)

Null Preferred

Figure 5.8 Responses by elementary motion detectors. (a) When a fly viewed a moving pattern through a vertical slit cut in a mask, only a narrow band of ommatidia and elementary motion detectors was stimulated. (b) A wide-field HS neuron responded to each lightening or darkening of this band, as shown in the intracellular recording. When the stimulus moved in the null direction, each response by the strip of motion detectors caused hyperpolarising potentials; and when the stimulus moved in the preferred direction, another set of motion detectors responded, causing depolarising potentials. When the mask was removed so the eye saw the whole moving pattern, responses from different strips of motion detectors summed so that sustained excitatory or inhibitory responses were recorded from the HS neuron. (b modified after Egelhaaf & Borst, 1993.)

necessary because light-on stimuli are processed separately from light-off stimuli. Although photoreceptors are excited by light-on stimuli and inhibited by light-off stimuli, neurons such as H1 respond to moving light or dark objects, so signals from photoreceptors must be channelled through different, parallel pathways before they reach the lobula plate. Microstimulation of single pairs of receptors shows that light onto the first receptor facilitates responses to light on, but not to light off, delivered to the second. This duplication ensures that a local motion detector will respond only when there is correspondence in the stimulus that the two photore-ceptors see.

In another type of experiment, Martin Egelhaaf and Alexander Borst have made intracellular responses from an HS neuron (Egelhaaf & Borst, 1993). They stimulated the eye with a pattern of vertically oriented stripes that drifted backwards or forwards over it (Fig. 5.8a). The darkness of each stripe varied sinusoidally, so there were no sharp borders between light and dark in the pattern. A mask with a narrow slit cut into it was placed between the moving pattern and the eye so that only a narrow, vertically oriented band of ommatidia was stimulated. When the stimulus moved, the photorecep-tors in this band experienced sinusoidal changes in light intensity, rather than the abrupt switches in light that occurred during the microstimulation experiments. When the stimulus moved in the preferred direction, the response by the HS neuron consisted of a series of discrete depolarising potentials in which a large potential alternated with a smaller one (Fig. 5.8b). Each large potential corresponded with a darkening of the photoreceptors in the band of stimulated ommatidia, and each small potential corresponded with a lightening of this band. When the stimulus moved in the null direction, the neuron responded with a series of alternating large and small hyperpolarising potentials. When the mask was removed so that a large part of the eye saw the moving pattern, the intracellular recording showed a sustained depolarisation of the neuron, indicating excitation, or a sustained hyperpolarising potential, indicating inhibition, depending on which direction the stripes moved. These results are consistent with the hypothesis that motion is detected by local circuits, similar to Fig. 5.7c. When the eye sees a large, extended pattern rather than a narrow slit, an HS neuron will receive signals from many vertical bands of ommatidia. Because the outputs from the different bands are out of phase with each other, the summed signal in the HS neuron will be a steady depolarising or hyperpolarising potential in which the fluctuations in output from individual motion detectors are ironed out.

Neurons like H1 and the HS neurons are excited most strongly by movement of large-field stimuli in a particular direction. The amount of excitation also depends on the speed and repeat pattern of the stimulus. These two stimulus features, speed and repeat pattern, cannot be distinguished from each other by an elementary motion detector because the response to a narrow series of stripes moving quite slowly over the eye is the same as the response to stripes that are twice as broad but moving twice as rapidly (Fig. 5.9). The strength of excitation, therefore, depends on the frequency with which the detector is stimulated with changes in the contrast of light. Slowly moving, closely spaced stripes will stimulate a detector with the same contrast frequency as more rapidly moving, broader stripes. The fact that the responses of H1 and HS neurons also depend on contrast

120 Stimulus filtering: vision and motion detection -►

Figure 5.9 An elementary motion-detector circuit cannot distinguish between narrow stripes moving slowly and wider stripes moving more quickly. In this circuit, each receptor is excited whenever a light-dark edge moves over it. The amount by which the detector is excited depends on the delay between excitation of the two receptors, which is short if either the speed of movement is fast or if the stripes are spaced close together.

frequency indicates strongly that they are driven by this type of elementary motion detector. The strengths of optomotor turning responses by a tethered fly also depend on contrast frequency of stimuli, rather than speed or spatial pattern, and this lends support to the hypothesis that the HS neurons are responsible for controlling these behaviours. Some movements by insects, however, are controlled in a way that suggests the insects can measure stimulus speed independently of image structure (Box 5.2). Visual systems channel information about movements through a number of independent pathways, each focusing on particular aspects of the stimulus. Ambiguities, such as that between stimulus velocity and pattern, can be removed at later stages by combining the outputs of different channels.

Box 5.2. Bees can measure image speed to fly a straight course

Bees can be trained to fly along a tunnel if it is part of the route between their hive and a good food source. They tend to maintain a course straight along the centre of the tunnel, and they could do this by balancing the relative speed of motion that is detected by the left and right eyes. Kirchner & Srinivasan (1989; see also Srinivasan,1992) showed that bees detect the speed of motion over each eye by observing flight paths along tunnels with side walls decorated with vertical stripes. No matter how broad the stripes on each side are, the bee flew straight down the middle (a). If the stripes on one wall moved (short arrow) in the same direction as the bee, the bee flew closer to that wall (b), but if the stripes moved in the opposite direction, the bee flew closer to the other wall (c). These results suggest quite strongly that bees have neurons that can compare the speeds of the images of the two walls, something that would be difficult to achieve with the optokinetic neurons of the fly lobula plate.

Essentials of Human Physiology

Essentials of Human Physiology

This ebook provides an introductory explanation of the workings of the human body, with an effort to draw connections between the body systems and explain their interdependencies. A framework for the book is homeostasis and how the body maintains balance within each system. This is intended as a first introduction to physiology for a college-level course.

Get My Free Ebook


Post a comment