Feature Extraction by Optic Lobe Units in Flies

Much basic early work (anatomy, neurophysiology, optomotor studies) was been done on the visual systems of flies by DeVoe and Ockleford (1976), DeVoe (1980), Mimura (1970; 1972), Bishop et al. (1968), etc. As will be seen, rapidly flying insects face special challenges to their visual systems.

In addition to having to avoid obstacles, a flying insect faces the additional burden of having to provide its CNS with visually driven, flight stabilization information in order that the flight motor system can compensate for yaw, pitch, and roll. These are extremely complex operations to understand in terms of muscle forces and instantaneous aerodynamic forces and torques. It stands to reason that the more agile the flier, the more sophisticated will be the control. Thus, one should not expect sessile or crawling insects to have the visual motion-sensing sophistication that can be found in agile fliers such as dragonflies, moths, and flies. Several classes of flying insects have additional, nonvisual aids for flight stabilization. For example, the diptera (flies, mosquitos, etc.) have vibrating gyroscopes called halteres (see Section 2.7) to sense roll and yaw, and locusts have hair patches on their heads that sense wind velocity; their antennae have the same role and may sense yaw and roll from differential wind forces (Gewecke, 1970).

Flying insects also face the problem of separating the visual background motion due to flight from the (relative) motion of other flying insects. Wide-field DS units have been found in fly OLs that have a rear-to-front (anteriad) preferred direction (PD); others have a posteriad PD. A wide-field DS unit responds well to stripes, as well as to spots and square-wave gratings. When an object is moved in its RF in the PD, it fires at a rate given approximately by r(v) = k log(1 + v/vo), where k and vo are positive constants and v is the speed of the object (Collett and King, 1974).

From an engineering viewpoint, it is easy to see how a flying fly might use the difference between the firing rates of left and right DS units with posteriad PDs to correct for yaw in flight (assume stationary, vertical contrasting objects on both sides of the fly's flight path). Anteriad DS units could also be stimulated if a flying fly, for example, turns sharply to the left. The posteriad DS units of the right eye would be stimulated strongly, but if the object of the left eye has a relative, forward tangential velocity that exceeds the forward flight velocity, the object will appear to move anteriad, and stimulate the left eye DS units with anteriad PDs. (See Collett and King, 1974, for an excellent review of the problem of directional vision during flight.) Note that DS units with anteriad PDs would be stimulated if the fly is sitting, and some object, e.g., another fly, flies past it from the rear.

Collett and King (1974) have recorded from the so-called small-field DS (SFDS) units in the OLs of the hoverfly with microelectrodes located in the external chiasma, the medulla, and the lobula. Mimura (1974a) reported that SFDS units were most commonly found in the region between the medulla and the lobula (inner chiasma). The SFDS unit RF was ipsilateral, contralateral, or bilateral to the recording site. The RFs were ~20° in diameter. SFDS units gave little or no response at ON and OFF of general illumination. Many but not all SFDS units were directionally sensitive with posteriad-moving object preferred directions. SFDS units were not spontaneously active; they responded best to high-spatial-frequency objects (2 to 4° diameter black spots) and not at all to moving long bars, gratings, or single contrasting edges. A curious property of the SFDS units was that their optimum spot object size was velocity dependent. For example, a 2.5° diameter black spot moving in the preferred direction (posteriad) at 70°/s gave a strong response, but a 7° diameter spot moving at the same velocity in the PD gave little response. When the 7° spot was moved at 430°/s in the PD, it gave a strong burst while in the RF; the 2.5° spot given the same motion gave no response.

Collett and King (1974) found in the hoverfly that the surround of an SFDS unit strongly inhibited the center response to a moving spot when it was stimulated by a complex pattern moving at a lower or equal velocity than the spot, in the spot PD. Thus, it is probable that most of the SFDS units would be "turned off" (inhibited by background motion) during flight.

Mimura (1970) reported the finding of LUs in the fly Boettscherisca peregrina that gave complex directional responses. One such unit studied responded to a 1.15° diameter light spot moved on the left front of the fly in a plane perpendicular to the center (flight) axis of the fly. (See Figure 5.4-1.) The unit gave a positive response to a linear spot motion moving to the left in the fly's upper left quadrant, also to a linear downward motion in the left upper and lower quadrants, and to a left to right linear motion in the lower left quadrant. It was evident that the unit responded maximally to a counterclockwise rotation of the spot in the left hemifield. Another LU responded to counterclockwise light spot rotation over the entire frontal field; a clockwise rotation unit was also found. In the same paper, Mimura reports having found the more common, linear DS units, and units that responded similarly to the LGMD/DCMD units in grasshoppers. No control was made to find the exact recording sites.

It is tempting to speculate that the "feature" extracted by Mimura's rotation units is in fact roll (while the animal is flying), and that such units could have direct inputs to the fly's flight stabilization control system.

A comprehensive study of fly OL units was reported by Bishop et al. (1968) in which they classified units according to their response properties. Nine classes (including subclasses) of OL units were described, as well as three subclasses of visual unit recorded in the brain corpora pedunculata region. Bishop et al. worked with the flies Calliphora phaenicia and Musca domestica. Metal extracellular micro-electrodes were used. Their Class I unit was nondirectionally sensitive. It was found in the anterior medulla tracts; the RF was ipsilateral and relatively small for a CE, 15 to 40° in diameter. Class I units responded to:

FIGURE 5.4-1 Schematic of a fly viewing a light spot rotated counterclockwise on its left side. Mimura (1970) found a lobula unit giving such a complex response. See Section 5.4.2 for discussion.

"patterns fixed in space with transient illumination, or to moving patterns. Their response to moving patterns appeared independent of the direction of motion."

Information was not given whether the response to transient illumination was at ON or OFF, or both. Also, no tests for object size and contrast optimality or adaptation to object motion were reported.

The Class II units were true directionally sensitive units, recorded from the lobula-lobula plate region. Class II units responded to ON of general illumination with an ON burst; there was a weak or no OFF burst. In the SSD, Class II units fired randomly at 1 to 3 spikes/s. In SSL with no pattern motion, the firing rate increased to 5 to 20 spikes/s. Stimuli were circular spots with contrasting stripes appearing within them. The stripes could be moved within the stationary spot, or the spot with fixed stripes could be moved as a whole. Spot diameter, d, and stripe period, X, could be varied independently. Stripes were always moved perpendicular to their long dimension. For spots about 22° in diameter and less, the measured DS response fell off as the cosine of the angle a between the PD and the spot/stripe vector (up to ±90°). The Class Ila group had four subclasses, dependent on the PD: Class IIa units had contralateral, full-eye receptive fields. Type IIal had an approximately horizontal PD directed toward the anterior of the animal (anteriad), the type IIa2 PD was vertical downward (ventrad), the type IIa3 PD was horizontal toward the posterior (posteriad), and the type IIa4 PD was vertical upward (dorsad). Class Ilb units had ipsilateral, full-eye RFs. They had horizontal PDs; the type IIbl PD was anteriad, and the type IIb2 PD was posteriad.

Class III units were found in the brain corpora pedunculata region. Class IIIa units had monocular contralateral RFs, and had similar properties to Class IIa; Class IIIb units had monocular ipsilateral RFs, and had properties similar to Class IIb. Class IIIc units were binocular with contralateral response dominance. Their firing appeared to be the summations of paired combinations of Class IIa and b unit responses.

DeVoe and Ockleford (1976) recorded intracellulary from single units in the optic medulla of the fly Calliphora erythrocephala. Because intracellular recording was used, slow potential shifts could be seen as well as spikes when units fired.

Thus, more pieces of the structure/function puzzle were available to gain understanding of how the various types of DS units in the dipteran visual system work. In addition to several types of DS units, DeVoe and Ockelford found a new type of unit that they named the change of direction (CoD) unit. The CoD unit fired as long as the test object was moving, regardless of its direction. When the object changed direction, there was a pause in the firing rate of the CoD unit. Firing resumed when object speed resumed in the null direction. Unlike grasshopper multimodal units (Northrop, 1974), and LGMD units (O'Shea and Rowell, 1976), CoD units did not adapt. The purpose of the CoD unit is unknown.

Was this article helpful?

0 0
Peripheral Neuropathy Natural Treatment Options

Peripheral Neuropathy Natural Treatment Options

This guide will help millions of people understand this condition so that they can take control of their lives and make informed decisions. The ebook covers information on a vast number of different types of neuropathy. In addition, it will be a useful resource for their families, caregivers, and health care providers.

Get My Free Ebook

Post a comment