Recognition of a familiar place by the honeybee (Apis mellifera).

Recent work shows that at any one place bees detect a limited variety of simple cues in parallel. At each choice point, they recognize a few cues in the range of positions where the cues occurred during the learning process. There is no need to postulate that they re-assemble the surrounding panorama in memory; only that they retain memories of the coincidences of cues in the expected retinotopic directions. The cues could be stimuli that excite groups of peripheral visual neurons. All the experimentally known cues are described, including modulation of the receptors, the locations of areas of black or colour, the nearness, size, averaged edge orientation, and radial and tangential edges. Cues of each type are separately summed within large fields, the size of which varies with the cue. Local orientation cues from edges at right angles cancel each other within each field, which also suggests that the discrimination of shape and texture is limited. Resolution depends on lateral interactions and the number of ommatidia required for each cue. To identify a new place, a few sparse cues, together with their directions, are learned in orientation flights. When the bee returns, the cues in the panorama are progressively matched as they coincide with the cues in memory. The limited number of cues, though economical for memory, may restrict the foraging behaviour and lead to flower constancy. This kind of a visual system is a candidate model for other animals or machines with economical processing systems.

The effect of complexity on the discrimination of oriented bars by the honeybee (Apis mellifera).

Visual discrimination of black bars by honeybees was studied in a Y-choice apparatus with fixed vertical patterns at constant range. The problem is to discover how bees remember different degrees of complexity of the orientation cue. Previous conclusions with parallel gratings and single bars disagree. With broad bars versus orthogonal bars, the bees learn the orientation cue if the bars are centred at the same place, but they learn the position cue in the vertical direction when the bars are at different places on the two targets. With several bars on each target, the bees learn their orientation and positions. As fixed patterns increase in complexity, the bees follow a simple rule, to look only at the range of places where the cues were displayed. The frame of reference is disrupted when a black spot is added to the training pattern. There is abundant evidence that the bees do not re-assemble the pattern or learn shapes. The filters that detect the position and orientation cues are coarsely tuned, so that they respond in a graded way, but the memory of the range of directions of the cue, as seen from the point of choice, is more exact.

Pattern discrimination by the honeybee (Apis mellifera): training on two pairs of patterns alternately.

Pattern discrimination in the honeybee was studied by training alternately with two different pairs of patterns. Individually marked bees made a forced choice from a fixed distance in a standard Y-choice maze for a reward of sugar solution. Bees were trained, first on one pair of patterns for 10min then on a second pair, and so on, alternately between the two pairs. The pairs of patterns were selected to test the hypothesis that bees have a limited number of parallel mechanisms for the detection and discrimination of certain generalized global features. If this is so, it might be expected that each channel could process one pair of patterns simultaneously, but two pairs of patterns that are processed by the same channel would interfere with each other during the learning process. Features tested were: average orientation of edges, radial and tangential edges based on a symmetry of three or six, the position of a black spot, and the exchange of black and white. The bees fail to learn when the two alternated pairs of patterns offer the same feature, and they discriminate when the pairs offer two different features.

Spatial coincidence of cues in visual learning by the honeybee (Apis mellifera).

The discrimination of patterns was studied in a Y-choice chamber fitted with a transparent baffle in each arm, through which the bees had a choice of two targets via openings 5cm wide. The bees see the positive (rewarded) and the negative (unrewarded) targets from a fixed distance. The patterns were bars (subtending 22 degrees x5.4 degrees at the point of choice) presented in one-quarter of each target. The bars were moved to a different quarter of the target every 5min, to make the location of black useless as a cue. A coincident presentation is when the bar on the left target is on the same side of the target as the bar on the right target. The bees learn the orientation cue when the presentation is coincident but otherwise cannot learn it. This experiment shows that bees do not centre their attention on the individual bars, otherwise they would always discriminate the orientation. Centring the target as a whole precedes learning. Having learned with the bar on one side of the targets, bees do not recognize the same cue presented on the other side. A separate orientation cue can be learned on each side. A radial/tangential cue is preferred to a conflicting orientation cue.

Vision of the honeybee Apis mellifera for patterns with one pair of equal orthogonal bars.

The visual discrimination of patterns of two equal orthogonal black bars by honeybees has been studied in a Y-choice apparatus with the patterns presented vertically at a fixed range. Previous work shows that bees can discriminate the locations of one, or possibly more, contrasts in targets that are in the same position throughout the training. Therefore, in critical experiments, the locations of areas of black were regularly shuffled to make them useless as cues. The bees discriminate consistent radial and tangential cues irrespective of their location on the target during learning and testing. Orientation cues, to be discriminated, must be presented on corresponding sides of the two targets. When orientation, radial and tangential cues are omitted or made useless by alternating them, discrimination is impossible, although the patterns may look quite different to us. The shape or the layout of local cues is not re-assembled from the locations of the bars, even when there are only two bars in the pattern, as if the bees cannot locate the individual bars within the large spatial fields of their global filters.

On the existence of 'fast' and 'slow' directionally sensitive motion detector neurons in insects.

In a fly, butterfly, locust and dragonfly we examined the responses of a variety of directional motion-sensitive neurons which run from the brain down the ventral cord. The stimulus was a sinusoidally modulated moving pattern of regular stripes presented at a range of velocities in random order for either 0.1 s or 2.0 s. The response was measured as the total number of spikes to each stimulus. The neurons fall into two groups, 'fast' and 'slow'. The responses of the fast type rise progressively to a peak contrast frequency at 15-20 Hz for all four insects, and decline at higher contrast frequencies. The responses of slow neurons rise rapidly to a peak at 1-10 Hz and then decline more slowly across the range where the fast neurons are at their peak. The existence of two groups of neurons with overlapping response ranges to different velocities of the same pattern, presented in exactly the same way, provides the insect with a means of measuring angular velocity irrespective of contrast, spatial frequency or intensity. As an input mechanism it is proposed that there are two types of unit motion detector, fast and slow, the latter being the main input to the optomotor system. It is also argued that even these inputs are not sufficient to provide a mechanism for the whole repertoire of normal insect vision.

Insect motion perception.

The first step in this work of reconstruction of a theory of insect vision was to demonstrate that visual behaviour relies on scanning by self-motion and apparently involves measurement of angular velocities of contrasts moving across the eye. The next step was to demonstrate that parallax is also significant as a way of segmenting the visual scene into separate objects. There followed a series of experiments to rule out the existing theory that motion perception depends on autocorrelation, and at the same time an alternative theory was developed. The new theory assumes that at the level of the optic medulla there are numerous parallel channels on each visual axis, representing different neurons, all looking out for their specific combination of signals. The combinations are formed by positive, negative or no-change temporal contrasts at two adjacent visual axes at two successive times, forming 3(4) = 81 possible templates. Simulation of this highly parallel system shows that it can represent the moving image in a compact form that would be adequate to explain what is known for motion and form vision (but not colour vision) in insects. Form, like colour, would be seen as the ratio of numbers of responses of particular templates, in the same way that colours are seen as ratios of responses of receptors for different wavelengths.(ABSTRACT TRUNCATED AT 250 WORDS)

Ratios of template responses as the basis of semivision.

The template model starts with a layer of receptors that in the case of vision are leaky detectors or counters of photons. In many animals, the ratio of the responses of a few spectral types is the basis of colour vision irrespective of intensity. Ratios of template responses are now introduced as the basis of form discrimination. In insects, the second-order neurons on the visual pathway appear to detect temporal contrast at the spatial resolution of the retina. At the next level, in the optic medulla, we find a large number of small local neurons in a column on each visual axis. The template theory is a hypothesis about how the above system functions. All possible combinations of positive, indeterminate or negative temporal contrast are considered, at two adjacent visual axes at two successive instants, giving 81 possible local templates. These templates are therefore phasic detectors of all the possible spatiotemporal contrast combinations. Some of the template responses indicate polarity of edge, flicker, or direction of motion and other abstracted features of the stimulus pattern with the maximum spatial and temporal resolution. The ratios of numbers of template responses, in higher fields at a higher level, yield quantitative measures of the qualities of edges independently of the number of edges, but taking ratios causes a corresponding loss of the spatiotemporal resolution and the pattern within each field. Templates respond to transients without computation, are readily modified or selected in evolution and can be simulated in artificial vision.

Implementation of the template model of vision.

Adopting principles learnt from insect vision we have constructed model of a general-purpose front-end visual system for motion detection that is designed to operate in parallel along each photoreceptor axis with only local connections. The model is also designed to assist electrophysiological analysis of visual processing because it puts the response to a moving scene into sets of template responses similar to the distribution of activity among different neurons. An earlier template model divided the visual image into the fields of adjacent receptors, measured as intensity or receptor modulation at small increments of time. As soon as we used this model with natural scenes, however, we found that we had to look at changes in intensity, not intensity itself. Running the new model also generated new insights into the effects of very fast motion, of blurring the image, and the value of lateral inhibition. We also experimented with ways of measuring the angular velocity of the image moving across the eye. The camera eye is moved at a known speed and the range to objects is calculated from the angular velocity of contrasts moving across the receptor array. The original template model is modified so that contrast is saturated in a new representation of the original image data. This reduces the 8-bit grey-scale image to a log, 3 = 1.6-bit image, which becomes the input to a look-up table of templates. The output consists of groups of responding templates in specific ratios that define the input features, and these ratios lead into types of invariance at a higher level of further logic. At any stage, there can be persistent parallel inputs from all earlier stages. This design would enable groups of templates to be tuned to different expected situations, such as different velocities, different directions and different types of edges.

A template theory to relate visual processing to digital circuitry.

Simple stimulus patterns, in this case visual, are represented by spatiotemporal Boolean functions that can be summarized in a 4 x 4 look-up table of 16 templates behind each sensory neuron. These groups of templates correspond to groups of neurons in columns behind each receptor. They abstract specific combinations of input in simple combinations and include two successive states in time. A template is like a neuron field at threshold, and responds as the field is convolved with the stimulus pattern. The same structure can be repeated in successive layers to make progressive categorization and to reject inappropriate combinations. At any level, the templates act in groups, so providing a very large number of combinations that can represent more complex stimulus patterns at deeper levels.

The evolution of visual processing and the construction of seeing systems.

This paper is concerned with the evolution of visual mechanisms and the possibility of copying their principles at different levels of sophistication. It is an old question how the complex interaction between eye and brain evolved when each needs the other as a test-bed for successive improvements. I propose that the primitive mechanism for the separation of stationary objects relies on their relative movement against a background, normally caused by the animal's own movement. Apparently insects and many lower animals use little more than this for negotiating through a three-dimensional world, making adequate responses to individual objects which they 'see' without a cortical system or even without a large brain. In the development of higher animals such as birds or man, additional circuits store memories of the forms of objects that have been frequently inspected from all angles or handled. Simple visual systems, however, are tuned to a feature of the world by which objects separate themselves by movement relative to the eye. In making simple artificial visual systems which 'see', as distinct from merely projecting the image, it is more hopeful to copy the 'ambient' vision of lower animals than the cortical systems of birds or mammals.

The eye of the soldier beetle Chauliognathus pulchellus (Cantharidae).

The soldier beetle eye is unusual in having large optically isotropic corneal cones which project inwards from a thick isotropic cornea. Refraction is mainly at the corneal surface. Calculation shows that the first focal plane is near the tip of the cone, from which the optical pathway continues as a crystalline tract. At the distal end of the crystalline tract, 3 micrometer in diameter, the four cone cells enclose the proximal tip of the corneal cone; at the proximal end they enclose the distal tip of a long fused rhabdom rod. The eye is remarkable in that there are two classes of retinula cells; four cells contribute to the long thin axial rhabdom, 2 micrometer in diameter and 120 micrometer long, and the other four cells form two rounded rhabdoms, 10 x 4 micrometer in cross-section and 20 micrometer deep, which lie to one side of the optical axis. The physiological properties of individual retinula cells were measured by intracellular recording. The retinula cells are of three spectral types with peaks near 360, 450 and 520--530 nm. Except by the criterion of spectral sensitivity, the retinula cells sampled could not be sorted into more than one class. The measured value of the acceptance angle, near 3 degrees in the dark-adapted state, is consistent with the hypothesis that all sampled cells were of the anatomical type that participate in the central rhabdom rod. A calculation of the theoretical field size of individual retinula cells from measurments of refractive index and lens dimensions predicts that cells which participate in the central rhabdom will have acceptance angles near 3 degrees. The conclusion, therefore, is that only one anatomical type of cell has so far been sampled.

The distribution of bumps in the tail of the locust photoreceptor afterpotential.

An extended tail or prolonged depolarizing afterpotential (PDA) follows the receptor potential of a locust retinula cell when the stimulating light is in the intensity range that saturates the receptor potential. The amplitude and duration of this afterpotential depend on the intensity and duration of the stimulus. As the afterpotential decays, apparently exponentially, it becomes resolved into bumps, which we call light-induced dark bumps (LID bumps). The intervals between light-induced dark bumps are distributed in a way that is indistinguishable from a random (Poisson) distribution. As previously demonstrated, LID bumps are indistinguishable from bumps directly induced by low intensity light in light-adapted cells, which in turn grade into the slightly larger bumps produced, each by a single photon, in dark-adapted cells. The light-induced dark bumps continue for up to an hour in darkness, slowly becoming like dark-adapted bumps in amplitude and shape. To account for the random occurrence and discrete features of bumps after so long a latency, we propose that intense light generates a significant amount of an intermediate molecule or packet which decays slowly to start the same process that normally generates bumps with a short delay.

The path and rate of growth of regenerating motor neurons in the cockroach.

Nerve root 5 that supplies the coxal depressor muscles from the metathoracic ganglion in the cockroach was crushed. Regeneration of the motor neurons was studied by cutting the nerve at several distances from the crush point and introducing cobalt chloride into the cut end. The operation was followed by a lag period of about 13 days after which the axons regenerated at a rate of 0.9 mm per day. After regeneration had been completed a pattern of axon distribution was established among the nerve branches that was very similar to that found in intact, normal cockroaches. This pattern was established through an apparent directed growth of certain axons from identified cells into branch 5rl, their normal pathway. However, at the same time, radom or increased branching of other unidentified motor neurons produced some errors in the distribution of axons among the nerve branches. Eventually these errors were corrected by the degeneration of neuronal processes that did not synapse with correct target muscles. These results demonstrate the requirement for a highly specific intercellular recognition process between individual, identified motor neurons and the appropriate muscles they innervate in order to reform the original innervation pattern during regneration.

Structural changes in light- and dark-adapted compound eyes of the Australian earwig Labidura riparia truncata (Dermaptera).

The apposition acone eye of Labidura is relatively small--550-600 facets--with a thick corneal lens and shallow retina. The retina cell columns are each formed of six peripheral cells plus two central cells, a partially fused rhabdom, and dense pigment in two or three cell types. Upon adaptation from light to dark, the most striking photomechanical response is a proximal broadening of the cone cells, which results in a 38-fold increase in cross-sectional area of the aperture. While longitudinal rhabdom movement is small, microvillar diameters swell in response to light and contract in the dark. Irregularities of facet pattern and shape, and in ommatidial alignment were found, particularly towards eye margins. Three types of interommatidial sense organs on the eye surface are described, one of which has not been previously reported. An argument is presented to explain how the field of view and sensitivity are both apparently decreased in the acone eye by exposure to light.