ABSTRACT
Electrophysiological recordings from the sub-oesophageal ganglion and cervical connective of Musca domestica have revealed the existence of units which integrate visual stimuli that elicit landing behaviour (light-off stimulus or expansion of a spatial pattern) and those which cause inhibition of a landing behaviour already in progress (light-on stimulus or contracting pattern).
INTRODUCTION
The landing response is exhibited by flying insects just before they make contact with a suitable landing surface. It consists of a lowering of the legs, which are folded up beneath the body during flight, as the insect approaches the surface. As soon as the first and second pair of legs make contact with the surface, the wings stop beating and the body is bent round (Goodman, 1960), more or less depending on the angle between the direction of the flight and the landing surface (Hyzer, 1962; Taddei-Ferretti et al. 1977).
The visual stimuli required to elicit the landing reaction are either a general decrease in the light intensity, if there is no movement in the visual field or, with a constant level of illumination, the expansion of a pattern from the centre to the periphery of the binocular visual field (Diptera: Goodman, 1960, 1964; Braitenberg&Taddei-Ferretti, 1966; Taddei-Ferretti & Fernandez, 1973a, b; Zanforlin, 1974; Taddei-Ferretti et al. 1977; Saviolo & Zanforlin, 1980; Heteroptera: Coggshall, 1972). Such expansion can be simulated by the rotation of a spiral pattern in a direction which creates the subjective impression of an expanding visual target (see Materials and Methods) (Braiten-berg & Taddei-Ferretti, 1966). Indeed, only that part of the spiral pattern which appears through a slit cut out of a mask which completely covers the pattern, and through which radial motion is seen along one and any direction from the centre to the periphery, is an adequate stimulus for the elicitation of the landing reaction (Fernandez & Taddei-Ferretti, 1970; Taddei-Ferretti & Fernandez, 1973b). Conversely, rotation of a spiral pattern so as to produce the impression of a contraction of the target (i.e. movement from the periphery to the centre of the binocular visual field) does not elicit the landing reaction (Braitenberg & Taddei-Ferretti, 1966). The contraction stimulus or an abrupt increase in light intensity both exert an inhibitory effect on the landing reaction (Taddei-Ferretti & Fernandez, 1973 a). An algebraic summation of the effects of these two types of stimuli (i.e. luminous variation and motion) is seen in behavioural experiments. For example, the decrease in light intensity required to elicit the landing reaction is lower when a simultaneous expansion at subthreshold velocity is present and is higher when a contraction is present (Taddei-Ferretti & Fernandez, 1973 a). In addition, a reaction already in progress, elicited by an expansion, can be stopped by an increase of the light intensity, and a reaction elicited by a light intensity decrease can be stopped by the presentation of a contraction (Taddei-Ferretti & Fernandez, 1973 a). The inhibitory effect of contracting radial motion has been confirmed in behavioural experiments by Eckert, Fligge & Hamdorf (1979) and by Eckert & Hamdorf (1980).
Since the landing reaction may be graded, the threshold value of a stimulus for the reaction is defined as the minimum value required to produce a full landing response for each stimulus trial in a single animal (Goodman, 1960). The threshold value of a change in light intensity for the landing reaction depends on the combined effects of the initial steady light intensity level and the value of any simultaneous motion stimulus (Taddei-Ferretti & Fernandez, 1973 a). The threshold value for landing response elicited by the expanding motion stimulus depends only on the steady intensity level of the light falling on the compound eye (Taddei-Ferretti & Fernandez, 1973 a), but not on the light falling on the ocelli (Goodman, 1960 ; Saviolo & Zanforlin, 1980). All these observations suggest the presence in the central nervous system of high-level integration units which summate various aspects of the visual information until a threshold is reached, at which point the reaction is triggered.
In the optic lobes many units are known which respond to light changes or to movement : simple or sophisticated stimuli are necessary for the responses of different units, depending on the location of each unit in the different zones of the optic lobes (lamina, medulla, lobula, lobula plate). The more complex units, in which a higher amount of integration is accomplished, are found in the less peripheral zones. Motion detection units, direction-non-selective or-selective, with narrow or wide visual fields, as well as on, off, and on-off units are found in the insect medulla (Bishop & Keehn, 1967; Bishop, Keehn & McCann, 1968; McCann & Dill, 1969; Collett, 1970; Mimura, 1971; McCann & Foster, 1971; DeVoe & Ockleford, 1976). The preferred directions of motion-sensitive units are in some cases horizontal or vertical (e.g. Bishop et al. 1968), while in others the preference could be for any given direction (e.g. Mimura, 1971). An alerting unit was reported by Mimura (1971), weakly responding to movements in any direction, rapidly adapting and probably useful for escape or startle behaviour. A unit sensitive to change of direction was found by DeVoe & Ockleford (1976). The site where motion detection occurs, which is the basis for optokinetic reaction (Reichardt, 1961; Fermi & Reichardt, 1963; Gotz, 1964; McCann & Mac-Ginitie, 1965 ; Kirschfeld, 1972), seems to be the medulla. In the lobula also, directionally selective movement detectors are found (Mimura, 1971; Zaagman et al. 1977). Complex semi-integrative and integrative units were found by Mimura (1971) predominantly in the medulla and in the lobula respectively. Such units seem to be devoted to the recognition of complex spatial and temporal patterns probably utilized during complex behaviour such as equilibrium maintenance during flight. A limited number of giant neurones are found in the Diptera lobula plate (Pierantoni, 1974; Hausen, 1976a; Strausfeld, 1976): the activity of such units was detected by Dvorak, Bishop & Eckert (1975) and exhaustively analysed by Hausen (1976b), Hengstenberg (1977), Eckert (1978, 1980) and Eckert & Bishop (1978). Each one of them responds to motion either in a horizontal or a vertical direction (although some of them have an additional sensitivity to motion in the orthogonal direction). Some of these giant neurones show action potential activity, while some of them only respond with slow depolarizations to light stimuli and with slow hyperpolarizations or depolarizations to vertical movements, but all have very large horizontal or vertical visual fields. These units may be used for simple operations mediating fast movements during flight navigation and probably during escape and landing reactions. A unique lobula giant neurone in each optic lobe receives information from the frontal zone of the retina and interacts with the horizontal and vertical giant neurones of the contralateral lobula plate, where other different species of large-field motion-sensitive neurones are located. The unique lobula giant neurones probably modulate the output of the horizontal and vertical giant neurones towards the thoracic ganglia in response to stimuli occurring in the zone of binocular vision (Strausfeld, 1976).
It has also been found that a certain amount of the total nerve activity travelling along the cervical connective is due to a stimulus (monolateral progressive motion of a striped pattern) (Hengstenberg, 1973) that is also capable of eliciting the landing reaction (Taddei-Ferretti & Fernandez, 1973b).
We report here the results of our studies of the activity of units downstream from the large horizontal and vertical movement detectors of the lobula plates and upstream from the motoneurones responsible for the precise movements of the legs in the landing reaction.
MATERIALS AND METHODS
Specimens of male and female Musca domestica were fastened with a drop of beeswax and colophonium to a small rod, on the dorsal surface of the thorax, anterior to the wings. Head movements were restricted by means of a bridge of beeswax and colophonium glued to the head and the thorax.
The stimuli used to elicit the landing reaction were light-off stimuli (‘off stimuli’), or circular or sectorial expanding patterns. Stimuli used to inhibit a landing reaction in progress were light-on stimuli (‘on stimuli’), or circular or sectorial contracting patterns. The expanding or contracting patterns were generated by a disc of 300 mm diameter, rotating in front of the fly at a distance (d) of 160 mm, or by a similar disc hidden by a stationary mask having a sector cut out. Two arithmetical spirals of pitch (p) = 80 mm, and dephased by π one from the other, were drawn on the disc and the two zones between the spirals were painted, one in white and the other in black. While rotating, the pattern simulated an expansion or a contraction, depending on the direction of the rotation, without variation of the total light flux arriving at the fly (Fig. 1). The animal’s long axis pointed towards the centre of the disc, from which the spirals originated. The velocity of expansion or contraction, as seen by the fly, hereafter referred to as ‘perceived velocity’ Vp, was proportional to the rotation speed of the disc by a factor p/d (Fernandez & Taddei-Ferretti, 1970).
Disc for the generation of the expanding or contracting patterns, when rotating clock-wise or counter-clockwise respectively.
Signals were recorded from the suboesophageal ganglion or from the cervical connective, using glass micropipettes filled with 2 M-KCl (40–100 MΩ). The recording electrode was advanced by a Haer hydraulic microdrive and the indifferent electrode was placed in the dissected region. Signals were fed to a × 5 gain, capacitance-compensated WPI M-701 amplifier, displayed on a Tektronix 5031 oscilloscope and stored for subsequent analysis on an Ampex PR 500 FM tape recorder with a frequency response of 0–2500 Hz. Some intracellular and many ‘semi-intracellular’ and extracellular recordings 01 units were obtained.
RESULTS
- Some units responded to on, off, or on-off light stimuli; some responded to expansion (Fig. 2), or contraction movement stimuli.Fig. 2.
Intracellular recording of a unit of the sub-oesophageal ganglion, the spike frequency of which increases on the presentation of a circular expanding pattern (indicated by a bar). Illumination level = 120 lx. Perceived radial expansion speed of the moving pattern = 2 ·5 rad s−1. Calibration: time = 1 s; voltage = 25 mV.
Fig. 2.Intracellular recording of a unit of the sub-oesophageal ganglion, the spike frequency of which increases on the presentation of a circular expanding pattern (indicated by a bar). Illumination level = 120 lx. Perceived radial expansion speed of the moving pattern = 2 ·5 rad s−1. Calibration: time = 1 s; voltage = 25 mV.
- Other units responded with an increase of spike frequency to two types of stimuli. Units of one class, which sometimes were silent when unstimulated, responded with an increase of spike frequency to an expansion and also to an off stimulus, but did not respond either to a contraction of the same velocity as the expansion or to an on stimulus (Fig. 3). Units of another class responded with a spike frequency increase (from a silent or ionically firing pre-stimulus state) to a contraction and also to an on stimulus, but did not respond either to an expansion or to an off stimulus (Fig. 4). As these two classes of units respond respectively to exactly the same stimuli that elicit or inhibit the landing reaction, we will refer to the units of the two classes respectively as ‘landing’ and ‘anti-landing’ units, without as yet giving any specific indication of the role of these units in the landing reaction.Fig. 3.
Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a unit, responding either to a circular expanding pattern (a) or to an off stimulus (b). The unit does not respond either to a circular contracting pattern (c) nor to an on stimulus (d). I = illumination level. Vp = perceived radial expansion or contraction speed of the moving pattern; negative values indicate contraction. Recording from the cervical connective. In this figure, as well as in the later ones except Figs. 7 and 8, the moving stimulus accelerates during the first 200 ms of stimulation, due to the inertia of the system.
Fig. 3.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a unit, responding either to a circular expanding pattern (a) or to an off stimulus (b). The unit does not respond either to a circular contracting pattern (c) nor to an on stimulus (d). I = illumination level. Vp = perceived radial expansion or contraction speed of the moving pattern; negative values indicate contraction. Recording from the cervical connective. In this figure, as well as in the later ones except Figs. 7 and 8, the moving stimulus accelerates during the first 200 ms of stimulation, due to the inertia of the system.
Fig. 4.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a unit, responding either to a circular contracting pattern (a) or to an on stimulus (b). The unit does not respond either to a circular expanding pattern (c) or to an off stimulus (d). I = illumination level. Vp = perceived radial contraction or expansion speed of the moving pattern; negative values indicate contraction. Recording from the sub-oesophageal ganglion.
Fig. 4.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a unit, responding either to a circular contracting pattern (a) or to an on stimulus (b). The unit does not respond either to a circular expanding pattern (c) or to an off stimulus (d). I = illumination level. Vp = perceived radial contraction or expansion speed of the moving pattern; negative values indicate contraction. Recording from the sub-oesophageal ganglion.
- Usually the response of landing and anti-landing units to moving stimuli was highest during the first 0·5 s and then fell to a steady lower level for the remainder of the duration of the 10 s stimulus (Fig. 5 a, b). Slow adaptation is seen during very long periods of stimulation (e.g. 60 s) (Fig. 6a, b, c, d). Adaptation was more rapid if the perceived expanding or contracting movement was slower. Sometimes, at the end of the stimulation, the response persisted for some seconds, the period becoming longer as the perceived expanding or contracting movement was made faster (Figs. 5c, 6e). The response to light stimuli was almost always of the tonic sustained type for the on stimulus in anti-landing units (e.g. Fig. 46, d); and usually a strong phasic response was followed by a slowly adapting weaker tonic one for the off stimulus in landing units (e.g. Fig. 3b, d).Fig. 5.
Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of an anti-landing unit, in response to a circular contracting pattern, during different phases and for two different radial contraction velocities of the stimulating pattern. The unit, if unstimulated, is silent; it returns silent after the last activity shown here. Stimulus duration: 12 s when high contraction velocity, 10s when low contraction velocity. (a) Before and at the beginning of the stimulus. (b) During the 5th second of the stimulus, (c) At the end of and after the stimulus. Illumination level = 120 lx. Vp = perceived radial contraction speed of the moving pattern. Recording from the sub-oesophageal ganglion.
Fig. 5.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of an anti-landing unit, in response to a circular contracting pattern, during different phases and for two different radial contraction velocities of the stimulating pattern. The unit, if unstimulated, is silent; it returns silent after the last activity shown here. Stimulus duration: 12 s when high contraction velocity, 10s when low contraction velocity. (a) Before and at the beginning of the stimulus. (b) During the 5th second of the stimulus, (c) At the end of and after the stimulus. Illumination level = 120 lx. Vp = perceived radial contraction speed of the moving pattern. Recording from the sub-oesophageal ganglion.
- The threshold values of the perceived expansion and contraction speed necessary to produce a spike frequency change were evaluated by stimulating either with movements of continuously increasing velocity, or with movements of different steady velocity values reached abruptly. In the first case it was found that, both for landing (Fig. 7a) and anti-landing (Fig. 8 a) units, during an acceleration phase of the stimulus from o to ±1 · 25 rad s − 1 (negative values signifying contractions) a very small transient spike frequency increase occurred, while during an acceleration phase from ± 1 · 25 to ± 2 · 5 rad s − 1 the spike frequency increase was much more noticeable and persisted, Wthough at a value less than the initial increase. In the second case, the spike frequency of both landing (Fig. 7b) and anti-landing (Fig. 8b) units increased with the stimulus velocity increase, the response being low at ± 1 · 25 rad s − 1 and appreciable at ±2 · 5 rad s − 1 ; with anti-landing units it was occasionally found that at still higher contraction velocities the response decreased.Fig. 6.
Example of the temporal distribution of the mean frequency, of spikes (200 ms bins) of the same anti-landing unit as that of Fig. 5, in response to a circular contracting pattern, during different phases of a stimulus of long duration (31 s). The unit, if unstimulated, is silent ; it returns silent after the last activity shown here, (a) Before and at the beginning of the stimulus. (b) During the 7th second of the stimulus, (c) During the 17th second of the stimulus, (rf) During the 27th second of the stimulus, (e) At the end of and after the stimulus. Illumination level = 120 lx. Vp= perceived radial contraction speed of the moving pattern. Recording from the sub-oesophageal ganglion.
Fig. 6.Example of the temporal distribution of the mean frequency, of spikes (200 ms bins) of the same anti-landing unit as that of Fig. 5, in response to a circular contracting pattern, during different phases of a stimulus of long duration (31 s). The unit, if unstimulated, is silent ; it returns silent after the last activity shown here, (a) Before and at the beginning of the stimulus. (b) During the 7th second of the stimulus, (c) During the 17th second of the stimulus, (rf) During the 27th second of the stimulus, (e) At the end of and after the stimulus. Illumination level = 120 lx. Vp= perceived radial contraction speed of the moving pattern. Recording from the sub-oesophageal ganglion.
Fig. 7.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a landing unit, in response to a circular pattern expanding at different velocities, (a) The radial expansion velocity of the pattern increases gradually during 2 s. (b) The radial expansion velocity has different steady values and the responses are shown from 5 s after the onset of each stimulus. Illumination level = 120 lx. Vp = perceived radial expansion speed of the moving pattern. Recording from the cervical connective.
Fig. 7.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a landing unit, in response to a circular pattern expanding at different velocities, (a) The radial expansion velocity of the pattern increases gradually during 2 s. (b) The radial expansion velocity has different steady values and the responses are shown from 5 s after the onset of each stimulus. Illumination level = 120 lx. Vp = perceived radial expansion speed of the moving pattern. Recording from the cervical connective.
Fig. 8.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of an anti-landing unit, in response to a circular pattern contracting at different velocities, (a) The radial contraction velocity of the pattern increases gradually during 2 s. (b) The radial contraction velocity has different steady values and the responses are shown from 5 s after the onset of each stimulus. Illumination level = 120 lx. Vp = perceived radial contraction speed of the moving pattern. Recording from the cervical connective.
Fig. 8.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of an anti-landing unit, in response to a circular pattern contracting at different velocities, (a) The radial contraction velocity of the pattern increases gradually during 2 s. (b) The radial contraction velocity has different steady values and the responses are shown from 5 s after the onset of each stimulus. Illumination level = 120 lx. Vp = perceived radial contraction speed of the moving pattern. Recording from the cervical connective.
- Sometimes the recordings were made while masking all the circular moving pattern except for a sector, thus drastically reducing the moving stimulus area. The sector was changed consecutively in four perpendicular zones of the visual field. In these experiments units were found showing an increase in spike frequency at the off but not the on stimulus (landing units) or at the on but not the off stimulus (antilanding units) and during a sectional expansion but not a contraction (landing units) or a sectional contraction but not an expansion (anti-landing units). Thus, typical landing and anti-landing unit responses were obtained in each of the four positions of the moving pattern (Fig. 9).Fig. 9.
Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a unit, responding either to an expanding sector (see Materials and Methods) (a) or to an off stimulus (b). The unit responds only slightly and occasionally to a contracting sector (c) and does not respond to an on stimulus (d). The sector, of 45 °, can be placed in any of the four zones of the visual field specified in the figure, without a change in the response. I = illumination level. Vp = perceived radial expansion or contraction speed of the moving pattern; negative values indicate contraction. Recording from the sub-oesophageal ganglion.
Fig. 9.Example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a unit, responding either to an expanding sector (see Materials and Methods) (a) or to an off stimulus (b). The unit responds only slightly and occasionally to a contracting sector (c) and does not respond to an on stimulus (d). The sector, of 45 °, can be placed in any of the four zones of the visual field specified in the figure, without a change in the response. I = illumination level. Vp = perceived radial expansion or contraction speed of the moving pattern; negative values indicate contraction. Recording from the sub-oesophageal ganglion.
- In a few experiments we used different temporal combinations of landing and anti-landing stimuli, presented either simultaneously or in rapid succession. A single successful record from a landing unit showed that the initial values of the spike frequency increase due to an expansion were lower if such an expansion was presented simultaneously with an on stimulus than if the expansion was presented during a constant level of illumination (Fig. 10a). The spike frequency increase in response tq an off stimulus was higher when the stimulus was preceded by an expansion and was lower when it was preceded by a contraction than the value when no motion occurred before the light stimulus (Fig. 10b).Fig. 10.
An occasional example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a landing unit, in response to different combinations of stimuli, (a) A circular expanding pattern is presented either alone or stimultaneously with an on stimulus. (b) An off stimulus is presented either alone or preceded by a circular expanding or contracting pattern. I = illumination level. Vp = perceived radial expansion or contraction speed of the moving pattern ; negative values indicate contraction. Recording from the cervical connective.
Fig. 10.An occasional example of the temporal distribution of the mean frequency, , of spikes (200 ms bins) of a landing unit, in response to different combinations of stimuli, (a) A circular expanding pattern is presented either alone or stimultaneously with an on stimulus. (b) An off stimulus is presented either alone or preceded by a circular expanding or contracting pattern. I = illumination level. Vp = perceived radial expansion or contraction speed of the moving pattern ; negative values indicate contraction. Recording from the cervical connective.
DISCUSSION
Units responding to on, off, or on-off stimuli are already well known. Medulla units responding to expansions or contractions were apparently found by Bishop & Keehn (1966). A few units were found by Collett & Blest (1966) in the optic lobes of Sphynx ligustri responding to a movement from front to back (or vice versa) in the visual field of both eyes. The units to which Mimura (1971) refers as semi-integrative units could be considered as responding to a partial zone of our circular moving patterns. The vertical giant neurones of the lobula plate examined by Eckert & Bishop (1978) each show a maximum sensitivity to movement along a different direction (only vertical; horizontal and mainly vertical, i.e. oblique; vertical and mainly horizontal, i.e. more oblique). Hengstenberg (1977) and Eckert & Bishop (1978) found lobula plate giant neurones showing depolarization to light stimuli in addition to the response to motion. We report here the behaviour of units more complex than all the preceding ones.
Units such as those described in Result 2 respond to both the types of stimuli (movement and light-level variation) that either elicit the landing reaction (expansion or light-level decrease) or inhibit it (contraction or light-level increase). We therefore suppose that these units are in some way involved in the production of the landing reaction (Taddei-Ferretti, Chillemi & Cotugno, 1980). Also, the observation that the adaptation to prolonged expansion or contraction was slower the higher the velocity of the moving stimulus parallels the observation (C. Taddei-Ferretti, unpublished results) made during behaviour experiments that, for a prolonged presentation of an expansion, the landing response lasted longer the higher the velocity of the expanding stimulus or the intensity of the off stimulus.
Our landing and anti-landing units are not simply units which respond to one type of light-change stimulus and to a movement along one direction in a limited zone of the binocular visual field. We have shown that the units respond to movement along all of the radial directions of an expansion or a contraction in the total zone of the visual field occupied by the rotating disc (Fig. 9). Taken together, our experiments confirm that the so-called landing and anti-landing units respond to the same stimuli that elicit or fail to elicit (or inhibit) the landing reaction (Result 5, in agreement with behavioural results: Taddei-Ferretti & Fernandez, 1973b, page 588). Such units cannot be considered analogous to the alerting units of Mimura (1971), which respond to movements in any direction and are fast-adapting, because they show a much more consistent response which does not adapt for a long time, and because they do not respond to the reverse stimuli (e.g. contraction vs. expansion). This last consideration speaks against their identification as units similar to the change-of-direction-sensitive units of DeVoe & Ockleford (1976).
The range of the threshold values of the perceived expansion or contraction velocities necessary to produce a change of the spike frequency of landing and anti-tanding units (Result 4) compares well with the values of the threshold perceived expansion velocity necessary to produce the full behavioural landing reaction for each trial stimulus. For the same illumination level as that used during experiments of Result 4, the velocity required for the landing reaction found in behavioural experiments is about 0 · 75 rad s − 1 (Taddei-Ferretti & Fernandez, 1973 a). At a perceived steady velocity of ± 2 · 5 rad s − 1 of expansion or contraction, a considerable response of landing or anti-landing units is observed, and these responses are already detectable at ± 1 · 25 rad s − 1 (Figs. 7b, 86). With stimuli which are accelerating towards ± 2 · 5 rad s − 1 the threshold for a considerable and persistent response is usually found to be 1 · 6 – 1 · 8 rad s − 1. This higher value, as compared with 0 · 75 rad s − 1 obtained for expansions in behaviour experiments, could be due to adaptation, i.e. to the fact that expansion and contraction stimuli were allowed to accelerate slowly to the maximum speed over a period of 2 s in the experiments reported in Figs. 7,a, 8a, while in behaviour experiments the moving stimulus was presented to the fly having already achieved the maximum speed. A further possibility is that convergence of landing units in the thoracic ganglion would result in a lower threshold for the landing reaction behaviour than for the generation of easily detectable responses in each landing unit. In behavioural experiments in which a flying fly was displaced towards the centre of a fixed stationary disc with painted spirals, the threshold value of the perceived radial movement of the contrast boundaries of the disc at the moment of the landing reaction was higher than that observed if the fly was fixed and the disc rotating (Taddei-Ferretti et al. 1977). This indicates that probably an air stimulus has an inhibitory effect on the production of the landing reaction (cf. effects of air currents on movement detectors: Mimura, 1975). This suggests that it would be interesting to repeat our investigations of landing and anti-landing units while also supplying air stimuli. Finally, higher values (and quite uniform values obtained with many individual flies) of the threshold of the perceived radial movement of the surface on which the fly lands are found during landing of a freely flying fly, probably because of differences in the proprioceptive information available by the fly in laboratory and free flight conditions (Taddei-Ferretti et al. 1977). If this is the case, a higher spike-frequency change of landing units or the convergence of a number of such units might be necessary to elicit the landing reaction in a freely flying fly. The small transient spike-frequency changes already observable with a perceived moving stimulus speed of less than ± 0 · 5 rad s − 1 (Figs. 7a, 8 a) could be responsible, in the case of expansion, for the incomplete, weaker responses of shorter duration and not elicited by each stimulus trial which are observed during behavioural experiments (Goodman, 1960; Zanforlin, 1974; Taddei-Ferretti, unpublished results) with perceived expansion speeds lower than the threshold value as defined by Goodman (1960).
Fig. 10 shows that integration between excitatory landing stimuli and inhibitory anti-landing stimuli can occur at the neurone level. Such integration has been observed at the behavioural level (Taddei-Ferretti & Fernandez, 1973 a). In fact, the effect of a landing movement stimulus is decreased by the simultaneous presentation of an antilanding light intensity stimulus (Fig. 10a), while a landing movement stimulus increases the effect of a landing light intensity stimulus given just as the preceding movement stimulus stops (as if the effect of the first stimulus persists for some time after the stimulus ends). Conversely an anti-landing movement stimulus decreases the effect of a landing light intensity stimulus given just at the moment at which the preceding movement stimulus stops (cf. Fig. 4 of Taddei-Ferretti & Fernandez, 1973 a) (Fig. 10b). Such integration would be characteristic of an interneurone to which information is conveyed from wide-field horizontal and vertical movement detectors of the optic lobes and by the on and off detectors of the medulla.
The first step of integration (i.e. the interaction occurring either between light intensity and movement-excitatory landing stimuli, or between light intensity and movement-inhibitory anti-landing stimuli respectively in the landing and antilanding units) in the case of the landing units needs the convergence of fibres coming from off-detector units and from neurones which process all the information arriving from movement detectors having preferred directions and visual fields such as to include all the possible components of an expansion in front of the fly. In the case of anti-landing units, convergence is required of on-detector neurones and of neurones which collect the information of frontal contractions. The subsequent step would require the existence of a command neurone or neurones, to which the output of a landing unit should have an excitatory effect, while the output of an anti-landing unit should have an inhibitory effect. To account for the behavioural results which show that the threshold of the perceived expansion velocity necessary to elicit the landing reaction is dependent on the steady illumination level (up to a level above which the light-level value has no more effect) (Taddei-Ferretti & Fernandez, 1973 a), such a command neurone should also have a threshold dependent on the average level of illumination. This threshold would be determined by input from luminosity units separate from the off- and on-detector units converging to the landing and antilanding units described here.
However, more complex interactions could occur than in the model presented above : in fact, the preliminary results on the combination of the effects of landing and anti-landing stimuli, given simultaneously or in rapid succession (Fig. 10) are obtained from a unit that, at the separate presentation of only one or the other of these two types of stimuli after the recovery from the effect of the preceding stimulus has taken place, behaves like a landing unit, i.e. does not change its spike frequency at the presentation of anti-landing stimuli alone.
The next stage of the research will be to try to confirm, by observing the landing behaviour of tethered flies while recording the activity of landing and anti-landing units, the role of such units in the production of the landing reaction, as well as to investigate the effects of the simultaneous landing and anti-landing stimuli in both landing and anti-landing units.
ACKNOWLEDGEMENTS
We are indebted to Dr L. J. Goodman (Queen Mary College, University of London) for fruitful discussions and suggestions; to Dr J. A. Patterson (St Bartholomew’s Hospital Medical College, University of London) and Dr R. G. Guy (City University, London) for their help during the initial phase of the work and many suggestions and constructive criticism of the results; to Mr A. Cotugno for continuous valuable assistance; to Mr F. Forte for the preparation of some electronic circuits; to Ing. M
Briggs-Smith for the correction of the English manuscript; to Dr J. A. Patterson for the critical reading of the final manuscript; to Professor S. Bettini (Istituto Superiore di Sanità, Roma) for supplies of flies; to Mrs M. Izzo for typing the manuscript.
