Vision in toads

The neural basis of prey detection, recognition, and orientation was studied in depth by Jörg-Peter Ewert in a series of experiments that made the toad visual system a model system^[1] in neuroethology (neural basis of natural behavior).^[2] He began by observing the natural prey catching behavior of the common European toad (Bufo bufo).

Ewert's work studying toads^{[3][4][5][6][7]} yielded several important discoveries. In general, his research revealed the specific neural circuits for recognition of complex visual stimuli. Specifically, he identified two main regions of the brain, the optic tectum and the pretectal caudal thhalamic region, that were responsible for discriminating prey from non-prey and revealed the neural pathways that connect them. Furthermore, he found that the neural mechanisms are plastic and adaptable to varying environments and conditions (e.g., discussed by Carew^[8] and Zupanc^[9]).

Vision in toads is a complex matter. Toads respond to objects depending on: a) what they see (recognizing the object; e.g., prey?), b) where the object is located in the visual field (localizing the object; e.g., choice of capture strategy?), and c) the toad's motivation, i.e. its interest, which can be influenced by endogeneous factors such as hunger, the time of day, and season, as well as event-related individual experiences.

Common toads Bufo bufo bufo (L.) are mainly active at dusk and dawn. Their annual life cycle is divided into different seasonal periods: winter for hibernation, spring for mating, summer for prey hunting as well as threat avoiding, and fall for migration to wintering grounds.

In spring, when the nighttime temperature no longer drops below 6 °C, the time has come for multisensory perception that allows common toads to interact socially in search of mates. They leave the soft forest floor, where they had burrowed up to 50 cm deep to hibernate, and then embark on a migration of almost 600 m per day over up to 5 km to their spawning grounds, where they grew up. During migration, the smaller males grab the heavier females as they pass by and cling to them, riding on their backs to reach the pond where mating takes place after several males have fought over each female.

In the context of a main season, behavioral objectives and their underlying motivations are commensurate. However, a high level of motivation can substantially influence the attribution of a visual object to a particular meaning category and vice versa. For example, during the mating season, male common toads are attracted to the large swollen bodies of the females, which are suitable for grasping and carrying a male in piggyback mode. In the event of a male, driven by testosterone eager to mate cannot find a female in the pond quickly, it will even utilize an oversize piece of floating tree bark as a substitude by intensely clasping it. Their sensory organs now focus on tactile sensations of their body skin and on communication through mating calls, while vision does not yet play a dominant role.

The term arousal characterizes the behavior of toads during the mating season and refers to a physiological state that promotes 'sensory vigilance.' The migration to their spawning grounds is the result of a complex interplay of various factors, not all of which are yet understood, such as how they do navigate at night in the rain. A special brain structure, the reticular formation, plays a prominent role. It integrates sensory information provided, for example, through the mesencephalic optic tectal and the somatosensory subtectal and tegmental layers. Olfactory information collected by the main olfactory bulb, associated and stored in the medial pallium of the telencephalon, is used for local orientation and pathfinding. In this scenario, the pituitary/hypothalamic system exerts significant functions by linking endocrine, sensory, and behavioral signals to the seasonal environmental events.

In the laboratory, this multisensory event is reflected during microelectrode recordings of individual neurons from the layers of the mesencephalic optic tectum, subtectum, and tegmentum:^[10] The input from the retina that arrives in the superficial layers of the optic tectum enables rough visual orientation without specific information processing at spring time. During the mating season, common toads do not feed, so the T5.2 neurons that recognize prey show only a weak response or no response at all. In the deeper tectal layers, there are neurons whose visual receptive fields are expanded, enabling them to obtain information from the entire field of vision of one or both eyes. This indicates enormous visual convergence, for which T4-type neurons, whose dendritic trees extend far into the tectal layers, form the appropriate neuroanatomical substrate.^[11] These wide-field neurons — recorded both in frogs and toads — are more sensitive than technical motion sensors, as they fire even when the experimenter moves something, such as one of his fingers, at a distance of 10 m. Deeper in the optic tectum, toward the tegmental areas, such a visual wide-field neuron receives additional somatosensory input from the entire ipsilateral body skin, so that tactile and/or vibratory stimuli can lower the response threshold for a potentially expected visual stimulus. In the laboratory, vibrations caused by footsteps on the floor trigger strong neuronal activity even at a distance of 10 m. This explains why, in the wild, frogs and toads jump into nearby ditches when someone tries to approach them cautiously. As the recording electrode advances laterally towards the semicircularis nucleus, auditory neurons contribute to the 'multisensory concert.'

After pairing, this concert reaches a moderate volume, and common toads behave like loners. They migrate back to their hunting grounds near the forests, which is accompanied by a neurophysiological 'return of precise vision,' in which visual information is processed appropriately and behaviors are visually triggered and controlled, for example when catching prey.

During the hunting season, the common toad responds to a moving insect or worm with a series of prey-catching reactions:^[12]

1. orienting towards prey
2. stalking up to prey
3. binocular fixation
4. snapping
5. swallowing
6. mouth-wiping with forelimb

The sequence of reactions (1) to (4) forms a kind of stimulus-response chain^[13][14] in which the prey-signal, in combination with a new location-signal, determines the stimulus constellation for the next type of reaction: When an object is recognized as prey and thus attracts attention, and the prey is in the peripheral field of vision, the toad turns its head and body toward the prey. If the prey is far away, the toad begins to pursue it. After reaching the prey, the toad fixes its gaze on it intensely with both eyes. Finally, it snaps at prey from short distance with its tongue or jaws. These stimulus responses form a series of clearly defined behavioral patterns.^[15] One reason for this type of stimulus-response chain is that, unlike humans, toads do not have involuntary saccadic eye movements,^[16] and they also cannot perform 'tracking eye movements.' Their visual system, therefore, focuses on the perception of retinal images of moving visual objects.^[17][18]

The lack of saccadic eye movements forces the toad to hold its eyes in rigid positions. Since the stimulus parameter movement is an essential prey feature, the toad must decide whether an object is 'prey' or 'non-prey' before moving itself. As the toad orients towards a moving prey object, it has already decided prey. Upon recognition, the toad indicates its 'prey arousal' often by a characteristic trembling of the long toes on its hind limbs. In the event that the prey suddenly disappeared, the toad may snap at nothing. This means that the prey information obtained can be stored for a certain period of time so that it can be retrieved shortly afterwards and converted into a catching response.

This is consistent with the frequency time histogram of the discharges of a prey detecting T5.2 neuron for orienting toward prey recorded from optic tectum in a freely moving common toad: A warming-up of the discharge frequency announces the toad's orienting turning movement of head and body that starts shortly after the warm-up peak and lasts for the duration of neuronal firing. The difference between neuronal discharge frequencies in response to prey vs. non-prey is maintained even below the threshold of prey capture.

When stimulus conditions allow for snapping at prey, the histogram of the firing rate of a corresponding T5.2 neuron over time reveals another interesting phenomenon: A strong "warm-up peak of the discharge frequency" —after prey has been detected and the trigger for snap activated — is terminated during snapping. Apparently, this "snap correlated inhibition" terminates visual perception and separates it from the subsequent consummatory reaction. This ensures that the toad's head is correctly aligned during the precise snapping attack with jaw and tongue: (1) opening the mouth, (2) projecting the tongue toward prey, (3) grasping the prey with the tongue, (4) retracting the tongue with prey, (5) closing the mouth and swallowing prey.^[19][20]

In summary, both response types of T5.2 neurons — characterized by "orient sustaining firing" vs. "snap correlated inhibition of firing"— exhibit a typical premotor warm-up phase that correlates with the trigger threshold for prey capture. T5.2 neurons, whose receptive fields are located in the binocular visual field, can take on either response type, e.g., orienting toward prey after a failed snap. This shows that T5.2 neurons in the tectal representation of the binocular visual field can be shared by releasing systems for orienting and snapping. In satiated toads, both the warm-up phase of T5.2 neurons and the prey capture behavior fail to occur.

In each case, the data were collected from a freely moving common toad using a sophisticated stimulus presentation and microelectrode recording system.^[21]

Among potential prey, common toads prefer small invertebrates that move in the direction of their body axis, such as ants, slugs, beetles, millipedes, and worms. They avoid large, moving objects such as the shadows of birds of prey or objects whose body parts are aligned perpendicular to the direction of movement, such as snakes or certain caterpillars. When a toad is presented with a moving stimulus, it generally may react with one of two responses: Depending on the size and the configuration of the stimulus, it will either engage in prey orienting (prey-catching) behavior or predator avoidance (escape) behavior, which consists of ducking-down postures or jumping away in panic. Snakes are considered the arch enemies of toads. Snakes move in a way that toads perceive as a warning: the snake raises its head and moves its body coils sideways, hence creating images that run transverse to the direction of motion. This dynamic gestalt (see feature detection (nervous system)) contrasts with that of an earthworm, for example, and is therefore referred to in laboratory jargon as anti-worm configuration vs. worm configuration. When confronted by a horseshoe whip snake,^[22] the toad will crouch down and try to dig in, jump away, or remain motionless. Sitting still, it employs various defense strategies while maintaining eye contact and secreting poison from its skin glands. Its body swells, causing it to assume a defensive stance with its head lowered and legs extended. At the same time, its head and back serve as a protective shield. This posture puts the toad in a position that makes it difficult for a snake to attack or grab it.

In determining the overall size of a stimulus, a toad will consider both the angular size, which is measured in degrees of visual angle, and the absolute size, which takes into consideration the distance between the toad and the object. This second ability, to judge absolute size by estimating distance, is known as size constancy.^[23][24][25]

To study behavioral responses of toads to varying types of stimuli, Ewert conducted experiments by placing the toad in the center of a little cylindrical glass vessel. He then rotated a small stripe (bar) of contrasting cardboard (acting as a visual 'dummy') around the vessel to mimic either prey-like or threat-like stimuli; see Video. The rate of turning was recorded as a measure of orienting behavior (prey-catching activity). By changing characteristics of the visual stimulus in a methodical manner, Ewert was able to comprehensively study the key features that determine behavior.^[26][27]

Up to a certain size, squares rotating around the toad successfully elicited prey-catching responses. Toads avoided large squares. Vertical bars nearly never elicited prey-catching behavior, and they were increasingly ineffective with increasing height. Horizontal bars, in contrast, were very successful at eliciting prey-catching behavior and their effectiveness increased with increasing length, to a certain degree. Additional vertical segments on top of horizontal bars significantly decreased prey-catching responses. By means of a different experimental setup it was shown that the worm vs. anti-worm discrimination is independent (invariant^[28] ) of the direction the object moves in the toad's visual field. Taking into account motion direction invariance, a more general formulation is: alignment of a bar parallel to vs. transverse (across) its direction of motion. [Note: The terms horizontal bar and vertical bar refer specifically to motion in the horizontal plane; see also: feature detection (nervous system).] The movement specificity of the toad's visual system is instrumental for classifying moving visual objects as prey or threads. This classification is based on a calculation of the alignment of the longitudinal axes of the objects in relation to the direction of their motion, as opposed to gravity as a reference value (cf. vertical or horizontal orientation): A millipede running horizontally along the ground or climbing up a vertical blade of grass falls into the toad's prey category. For models of prey vs. threat recognition, see: concept of 'neural filter interaction'^[29]; 'systems theoretical analysis'^[30]; 'schema theoretical approach'.^[31]

A stationary object, such as a barrier, elicits evasive detour behavior of the toad.^[32] However, there are examples showing that stationary objects under certain conditions trigger prey capture: When a toad moves its head with its immobile eyes, the retinal image of an object moves in the opposite direction due to induced movement. It is recognized as prey and responded to with prey capture, provided the configuration of the image matches the prey schema of the analyzing central visual system. This applies when the background of the object is untextured, e.g. a black object in front of a white background. However, if the background is textured, such as by a Julesz random-dot texture or a natural view of a forest floor with little branches and leaf structures, the object is ignored by the toad due to visual masking: The object is masked by the moving background texture. The inhibitory phenomenon between the target stimulus and the masking stimulus was investigated in toads quantitatively using various experimental procedures.^{[33][34][35][36]}

Problems with purpose-oriented interpretations?

A change in the contrast direction between a moving object and its background — black on white or white on black – has no fundamental influence on the algorithm for distinguishing configurations, for example, when the stimuli are worm-like or antiworm-like, as shown by the stimulus-response lines (a) and (b) in the figure. In the case of antiworm (b) and square (c), the corresponding contrast boundaries transverse to the direction of movement are decisive and thus support the predator's preference for worms.

Small square objects that are white on a black background trigger significantly stronger prey-catching activity than black objects of the same size on a white background. At first glance, we are familiar with a comparable phenomenon from visual perception, namely the irradiation illusion: a white object on a black background appears larger than a black object of the same size on a white background. An evolutionary advantage to consider could be that small, light-colored objects are easier to see at dusk or in low light conditions, as white color reflects more photons than black. The contrast phenomenon even reverses seasonally.^[37]

However, a disadvantage could arise when a hunting toad is confronted with worm-like prey. If the prey is black and the background is white, the toad snaps successfully at the edge that points in the direction of movement (head preference), but if it encounters a worm with reverse contrast, it chooses the rear edge of the white worm (tail preference) and often misses the prey when attempting to catch it by snapping into the void. Neurophysiological 'off' effects (rapid change in luminance from light to dark) caused by the moving contrast boundary play an important role here, which could increase the survival rate of a light-colored worm.

Selected functions of the toad‘s visual system:

1.) The neural circuitry mediating between configurational features of the visual prey stimulus and prey capture behavior consists of retina-fed tectal pathways and corresponding rhombencephalic/spinal nuclei.

2.) The distinction between prey (food) and non-prey (e.g., threat) is based on network interaction of the retino-tectal pathways of the mesencephalon with retino-pretectal thalamic structures of the diencephalon.

3.) The localization of moving visual objects takes advantage of the retinotopic map in the optic tectum in connection with monocular and binocular depth calculations and retino-tectal/tegmental patways.

4.) Targeted attention, i.e., the translation of visual perception into motor action, is based on disinhibitory modulating forebrain loops involving the telencephalic striatum.

5.) Modification of configurational prey recognition in the course of visual or olfactory associative learning are based on a modulating disinhibitory forebrain loop involving the telencephalic posterior ventromedial (hippocampal) pallium, the anterior thalamus, the pretectal caudal dorsolateral thalamus and the optic tectum.

6.) Dopaminergic modulations in the brain's macronet (e.g., after systemic administration of the dopamine agonist apomorphine) increase firing rates of retinal ganglion cells and the preying motivation and alter capture strategies with regard to pursuing and hunting prey versus waiting for prey and snapping.

To understand the neural mechanisms underlying the toad's behavioral responses, Ewert performed a series of recording and visual stimulation experiments. First and foremost, the results allowed him to understand the way the visual system is constructed and connected to the central nervous system. Secondly, he discovered areas of the brain that were responsible for differential analysis of visual stimuli.^[39][40][41]

The neurophysiological analysis of vision in anuran amphibians was introduced in the 19fifties by Jerome Lettvin, Horace Barlow, and colleagues, who were interested in: What does the eye communicate to its brain in terms of the parameters of visual stimuli, and how is the brain involved in visual pattern recognition?^{[42][43][44][45]} Their investigations in frogs focused on the response characteristics of retinal ganglion cells, which process the information provided by the photoreceptor cells via intermediate interneurons (amacrine, horizontal, bipolar cells), and transmit the results to the brain via their axons in the optic nerve.

In both common toads and frogs, the retina is connected to the optic tectum of the midbrain by at least three types of ganglion cells.^[46] The responses of a retinal ganglion cell can be recorded from its axon terminal in the superficial tectum using an electrolytically sharpened and lacquer-coated metal microelectrode. Each of the ganglion cells has an excitatory receptive field and a surrounding inhibitory receptive field, but they differ in the diameters of their central excitatory receptive fields, ERF, and the inhibitory receptive fields, IRF. The ERF-diameters in class II (R2) ganglion cells are approximately four degrees visual angle. Those in class III (R3) cells are about eight degrees, and ERFs of class IV (R4) ganglion cells range from twelve to fifteen degrees. As stimuli traverse the toad's visual field, information is sent to the optic tectum. The visual field of each eye is projected via the axons of the R cells in the optic nerve through an almost complete optic chiasm to the contralateral optic tectum. The field of vision is 'scanned,' and visual stimuli are pre-processed by the ganglion cell classes in terms of visual angular size, visual angular velocity, stimulus background contrast, and changes in diffuse illumination. The three classes respond to moving visual objects whose size correlates with the diameter of their ERF. The neurons of class R2 exhibit a peculiarity in that they not only discharge while a small object is moving, but also for a period of time after the object has suddenly stopped moving within the ERF. This inspires a comparison with the scenario of a fly after landing: it walks, stands still, and simply remains standing.

The optic tectum exists as an ordered localization system, in the form of a topographical map.^[47] Each point on the map corresponds to a particular region of the toad's retina and thus its entire visual field. Likewise, when a spot on the tectum was electrically stimulated, the toad would turn toward a corresponding part of its visual field, as if this part was stimulated by a prey object, hence providing further evidence of the direct spatial connections.

Among Ewert's numerous experimental findings is the identification of feature detectors, i.e., neurons in the visual system that selectively respond to behaviorally relevant features of a visual stimulus.

At the retinal level, however, the quantitative experimental results in common toads showed no 'fly detectors', 'bug detectors', 'beetle perceivers', 'worm detectors', or 'enemy detectors' in terms of their discharge characteristics in response to configuration defining features of moving objects: Retinal ganglion cells of the classes R2 or R3 discharge moderately when, for example, a Carabus beetle traverses their excitatory receptive field, but fire even more strongly when the same beetle is experimentally presented in a nonprey configuration — „anti-bug“ — that causes a toad to ignore or avoid it.

Instead, he found that the optic tectum and the pretectal caudal thalamic region play significant roles in the analysis and interpretation of visual stimuli (e.g., summarized in Ewert^[48][49]).

Electrical stimulation experiments demonstrated that the optic tectum initiates orienting and snapping behaviors.^[50][51] It contains many different visually sensitive neurons, among these type I and type II neurons (later named T5.1 and T5.2, respectively). Type T5.1 neurons are activated when an object traversing the toad's visual field is extended (aligned) parallel to the direction of movement; type T5.2 neurons, too, but they will fire less when the object is extended across the direction of movement. Those T5.2 neurons display prey-selective properties; see prey feature detectors. The discharge activity of T5.2 neurons in response to changes in the prey defining configuration features of a moving object correlates with corresponding prey capture activity. Furthermore, in freely moving toads, T5.2 neurons exhibit the phenomenon of size constancy, i.e., by estimating the distance between a visual object and the eye, they become sensitive to the actual size of that object. There are several methods for determining object distance: lens accommodation, triangulation, or binocular disparity. The discharge patterns of T5.2 neurons — recorded in freely moving toads^[52] — precede and thus predict prey-catching reactions, e.g., the tongue strike of snapping. Their axons project down to the bulbar/spinal motor systems, such as the hypoglossal nucleus which harbors the motor neurons of the tongue muscles. In combination with additional projection neurons, prey-selective T5.2 cells contribute to the ability of the tectum to initiate orienting behavior and snapping, respectively.

The caudal thalamic region initiates avoidance behaviors in the toad. More specifically, electrical triggering this region initiates a variety of protective movements, such as eyelid closing, ducking, and turning away. Various types of neurons in this region^[53] are responsible for the avoidance behaviors, and they are all sensitive to different types of visual stimuli: One type of neurons (TH3) is activated by large, moving objects and — in the case of bar stimuli — those that are aligned across the direction of movement. Another type (TH6) is activated by a looming object moving toward the toad. Still other types (TH10) respond to large stationary obstacles,^[54][55] and there are also neurons responding to stimulation of the balance sensors in the toad's ear. Stimulation of (a combination of) such types of neurons would cause the toad to display different kinds of protective behaviors.

Lesioning experiments led to the discovery of pathways extending between the optic tectum and the pretectal thalamic region.^[56] When the tectum was removed, prey-orienting behavior disappeared. When the pretectal thalamic region was removed (optic tectum intact), avoidance behavior was entirely absent while prey-orienting behavior was enhanced even to predator objects. Prey-selective properties were impaired both in prey-selective T5.2 neurons and in prey-catching behavior.^[57] Finally, when one side of the pretectal thalamic region was removed, the disinhibition of prey-catching applied to the entire visual field of the opposite eye. These and other experiments suggest that pathways, involving axons of type TH3 threat-feature detectors —which are also retinotopically mapped— extend from the pretectal thalamus to the map of the optic tectum, suitable to modulate tectal responses to visual stimuli and to determine prey-selective T5.2 properties due to inhibitory influences.^[58][59][60] During intracellular recordings from frog’s tectal neurons in response to pretectal stimulation, about 98% of the implated cells showed inhibitory responses.^[61]

The present hypothesis on prey recognition in toads is rooted in a causal-analytic experimental approach to visual feature processing, based on interactions between retina-fed optic tectal and retina-fed pretectal thalamic networks. Current experimental data refute a competing purpose-oriented motor hypothesis which claims that the motor actions of prey catching are essential in the process of prey recognition. One of the main arguments against the motor hypothesis is the fact that the visual responses of prey-selective T5.2 neurons to a prey object – in a freely moving toad – precede the subsequent motor responses for catching the prey.

Having analyzed neuronal processing streams in brain structures (pretectal, thalamic, tectal, medullary) that mediate between visual stimuli and adequate behavioral responses in toads, Ewert and coworkers examined various neural loops that—in connection with certain forebrain structures (striatal, pallial, thalamic)—can initiate, modulate or modify stimulus-response mediation.^[62][63] For example, in the course of associative learning, the toad's visual prey schema can be modified to include non-prey objects.^[64] The application of the [14C]-2DG method showed that the telencephalic posterior ventral medial pallium vMP revealed a significant increase in metabolic activity in the conditioned toads during prey capture toward large moving objects they had avoided prior to associative conditioning. After lesions to vMP, the conditioning effects of the toads failed, and their recognition of prey demonstrated species-specificity with the selectivity presumed to be phylogenetically conserved (see: Ewert J.-P., Dinges A.W., Finkenstädt T. (1994) Species-universal stimulus responses, modified through conditioning, reappear after telencephalic lesions in toads. In: Naturwissenschaften 81, 317-320.)

The posterior part of the ventromedial pallium is homologous to the hippocampus of mammals which is also involved in learning processes. Both in anuran amphibians and mammals striatal efferents^[65][66] are involved in directed attention, i.e. gating an orienting response towards a sensory stimulus. The anuran striatum is homologous to a portion of the amniote basal ganglia.

From an evolutionary point of view, it is important to note that the tetrapod vertebrates share a common pattern of homologous forebrain and brainstem structures.^[67][68][69] Neuroethological, neuroanatomical, and neurochemical investigations suggest that the neural networks underlying essential functions—such as attention, orienting, approaching, avoidance, associative or non-associative learning, and basic motor skills—have, so to speak, a phylogenetic origin in homologous structures of the amphibian brain. Detlev Ploog and Peter Gottwald open a discussion in evolutionary psychology whether heading towards something or heading away from something—in a general sense— might have their primordial roots in diencephalic (thalamic) and mesencephalic (tectal, tegmental) nuclei of amphibians.^[70]

From a neural network approach, it is reasonable to ask how the toad's ability to classify moving objects by special configuration cues develops, and whether this is unique in the animal kingdom. Developmental studies suggest that the detection principle is an adaptation in amphibians to their biotope and life style.

After hatching from the egg, the tadpole of Bufo bufo is programmed for a life in water and the subsequent postmetamorphic transition to life on land. To achieve this, hormonal, neural, neurosensory, and sensorimotor adaptations take place. The feeding behavior of the tadpole—which consumes bacteria and algae on aquatic plants, as well as carrion and debris—shifts during metamorphosis to a predatory, carnivorous lifestyle. This change goes hand in hand with a selectivity for visual prey objects that ‘move’ and thus show that they are alive and different from the stationary surround. An important discovery, therefore, is that common toads are predisposed to hunt worm-like moving objects in a configuration whose extension parallel to the direction of motion is typically greater than the extension transverse to it, and to avoid objects as threat that occur in anti-worm configuration. The ‘worm preference‘ is established after metamorphosis—during the toad's first steps on land—without it having to learn, and this preference matures over time.

The discrimination between prey and predator and the hunting behavior of toads towards the end of metamorphosis therefore require a redesign of visual perception with regard to the necessary neural and sensorimotor functions: The motion specificity of retinal R2 (and R3) neurons contributes to the excitability of tectal T5.1 neurons in response to spatiotemporal features of contrast borders of objects aligned with their direction of motion. Since the neural discharge rates increase with the increasing edge length of an object, its trigger values for activating prey capture also increase. This, in turn, would mean the risk of neuronal avalanche excitation^[71] due to intratectal recurrent excitation if there were no inhibitory neuronal counteraction. Based on the brain development of Discoglossus frog tadpoles, it is assumed that during metamorphosis at Taylor&Kollros-stage-X optical morphogenetic influences from the optic tectum to the adjacent caudal dorsomedial thalamus initiate a parcellation process^[72] of segregation.^[73][74] The new aggregate is called the dorsolateral area. It is thought to contain retino-receptive neurons, which—like TH3 cells in adult toads—are known to control the excitation of the optic tectum through massive inhibitory influences along their axons,^[75] similar to what is observed in frogs.^[76] The thalamic TH3 cells with input of retinal R3- (and R4-) neurons of the toad are sensitive to contrast borders that are aligned transversely to their direction of movement, thereby limiting tectal excitation in a finely tuned manner that simultaneously determines the prey selectivity of tectal T5.2 neurons.^[77]

The neuroanatomical differentiation of the dorsolateral region of the caudal thalamus, which continues beyond the end of metamorphosis, suggests a correlation with the ontogenetic maturation of object size constancy and the improvement of acuity in configurative prey selection.

The tectal avalanche hypothesis in common toads Bufo bufo bufo (L.) is substantiated by results of a lesion in the pretectal caudal dorsal thalamus (dorsolateral area). Experimentally, when the surfaces of the various standard worm (W), anti-worm (A), and square (S) test objects were stepped up in size, the tectal neuronal T5.1;2 responses of the brain lesioned toad and its prey-catching reactivity increased progressively all in the same way.

This lesion eliminated the toad's ability to distinguish between objects of different configurations and thus to correctly assign moving visual objects to a category of meaning, either prey or threat. Every moving object (or an induced moving retinal image) triggered a sometimes intense reflexive mouth behavior in the toad — a kind of prey object tracking and prey-oriented empty snapping — even in the direction of its moving hind feet.

If the retina-fed caudal dorsolateral thalamic region, which is tiny compared to the volume of the retina and the optic tectum, loses its control over retino-tectal information processing, the toad’s visual system collapses—in the broadest sense, comparable to visual agnosia.

Large threatening objects stimulate pretectal thalamic TH3 and tectal T5.1 neurons to activate the threat avoidance channel. The question arises: What controls the pretectal dorsal thalamus? We recall the experiments on conditioning with hand feeding: The toad’s species-specific ability to recognize prey, can be modified by individual experience. A simplified working hypothesis suggests why toads after associative conditioning classify a large moving object as prey,^[78] much like following a pretectal thalamic lesion.

During the training phase, the experimenter presented a mealworm (prey) with hand (threat) to the toad once a week for two weeks. It is assumed that at the same time, a subliminal prey signal from the neuronal prey-catching channel AND a subliminal threat signal from the threat-avoidance channel were paired in hypothetical neurons of the telencephalic posterior pallium, vMP. Such pallial cells become sensitized after a period of the prey-threat pairing and thus can be activated by a stimulus from one of the two channels: prey OR threat. The pallial cells (vMP) — via the anterior thalamus (aTH) — are suggested to inhibit threat-detecting TH3 cells of the pretectal dorsolateral thalamus, thereby interrupting the threat channel, and disinhibiting prey-detecting T5.2 cells, thereby channeling the prey channel:

Disinhibitory pathways [excitation–>, inhibitory influences–/ ]

vMP–> aTH–/ TH3–/ T5.2–> disinhibiting prey capture

alternatively: vMP–/ aTH–> TH3–/ T5.2–>

The T5.2 cells expand their configurative response spectrum and include visually threatening objects as prey, which is reflected in corresponding prey capture activity. Experimental test of the pallial influence: After vMP lesion, the species-specific prey selectivity in toad's prey capture behavior is re-established. Disinhibition in neuronal networks^[79] is defined as a temporary inhibition brake that promotes excitation. This evolutionary conserved circuit is implicated in different functions, including sensory processing, learning, and memory.

Experiments, in which the olfactory non-prey stimulus cineol was combined with a visual prey stimulus during prey capture, showed that the toad's configurative prey selectivity for the moving test objects was expanded in the presence of the conditioned stimulus cineol. That kind of expansion resulted from an olfactory stimulus transmitted by the main olfactory bulb, associated with a visual signal in the medial pallium, passing the ventral hypothalamus, and arriving back in the optic tectum. This influenced the level of prey motivation as it served to decrease a visual threshold to the readiness of preying. The prey-oriented behavior exhibited towards the test objects was found to be equally unselective as the behavior that followed visual threat/prey conditioning or a lesion of the pretectal thalamus.

As for the recovery effects^[80] the species specificity of visual pattern recognition was fully restored immediately after a vMP lesion. This contrasts with the restoration of species-specific visual pattern recognition following a pretectal thalamic lesion, which recovered to a certain extent after 5 days but did not reach its original selectivity in the following 10 days. The observation of a partial functional recovery of visual pattern recognition after a pretectal thalamic lesion suggests corresponding compensatory neurophysiological processes. This leads to the assumption of an 'intrinsic law' that strives for a physiologically acceptable imbalance between neural excitation and inhibition, which is necessary to meet the criteria for visual pattern recognition by tectal and pretectal networks. In order to understand the underlying functions, the question arises as to what happens when the visual output of the retinal R2 and R3 neurons to both networks is significantly increased in terms of their discharge frequencies. This is possible after systemic administration of the dopamine D1,D2 receptor agonist apomorphine.^[81] Although Apo-induced enlargements of the ERFs of R2 and R3 neurons do not result in changes of size preference in prey capture, recordings from T5.2 neurons demonstrate that the basis for species-specific pattern discrimination is preserved.^[82] This is consistent with studies measuring regional energy metabolism in the brain. The [14C]-2DG method shows^[83] a corresponding pattern between strong retino-receptive tectal activities and strong retino-receptive pretectal activities whose influences 'compete' in the deep tectal efferent layers. With regard to tectomotor output, there are qualitative differences in the choice of prey capture strategy, depending on the duration of exposure and the dose of apomorphine. This includes a decrease in the rates of orienting toward prey in favor of prey snapping.

The significant role of pretecto-tectal inhibitory connections for visual configurational pattern recognition^[84][85] raises the question of the kind of neurotransmission. Various authors suggest that pretectally released neuropeptide Y via Y2 receptor has an evolutionarily conserved function in the modulation of visual information processing in tetrapods^{[86][87][88][89]} and may exhibit plasticity with regard to neural restructuring during phylogenetic development, for example, with regard to differences in visual pattern discrimination by modifiable (innate) release mechanisms.

Behavioral studies showed^[90] that the fire-bellied toad (Bombina orientalis) disregarded a worm-like prey dummy of black cardboard when the object traversed the visual field of toad‘s tectal lobe, on the surface of which porcine NPY had been administered, soaked in an agarose gel-pad. As this prey object passed through the visual field of the untreated tectal lobe, which was covered with a gel-pad soaked with frog-Ringer without NPY, it released prey capture. Comparable experiments measuring metabolic activity^[91] confirmed a difference in [14C]-2DG uptake in dorsal tectal layers between the NPY-treated and untreated tectal lobes. That is consistent with studies in cane toads in which excitatory tectal field potentials in response to electrical stimulation of the contralateral optic nerve were attenuated by NPY-application to the tectal surface.^[92] Fact is also that pretecto-tectal projections in frogs – originating ipsilaterally from pretectal dorsomedial and dorsolateral thalamic nuclei – release NPY^[93] which in turn suggests that NPY exerts an inhibitory influence on retino-tectal transmission: If electrical stimulation of the ipsilateral pretectal dorsomedial/lateral thalamus (in the absence of NPY) preceded an optic nerve stimulation by a delay of 15 ms, the excitatory tectal field potential to optic nerve stimulation was strongly attenuated. This allows the conclusion that NPY — released from pretecto-tectal projecting fiber terminals and signaling through Y2 receptor — acts presynaptically on retino-tectal terminals to inhibit glutamate mediated retino-tectal transmission (see also feature detection (nervous system)). In fact, spike amplitudes of retino-tectal projecting R2 and R3 ganglion cells in response to moving visual objects were reduced under NPY influence to the superficial tectal layers, thus leading to a reduced release of glutamate.

The pretectal thalamic TH3 and TH4 neurons of the toad's visual system are attributed the following main functions: i) Detection of a large moving threat object and, specifically, a contrast boundary aligned transverse to the direction of its motion; ii) adaptive, configuration-dependent scaling and categorization of moving visual objects through computations and in interaction^[94] with T5.1 and T5.2 neurons, as well as producing behavioral trigger thresholds that result in sensorimotor decisions; iii) a finely tuned modulation of inhibitory processes within the optic tectum.

An important property of configurational object perception in toad vision is the “implicit computation”^[95] of contrast borders in relation to the direction of their movement. This can be regarded as a ‘trick‘ by which anuran brain evolution circumvents the need of a huge number of visual neurons with ‘asymmetric‘ receptive fields in order to “explicitly compute“ various possible contrast boundaries, as known from the visual cortex. In toads, an apparently equivalent solution is achieved by T5.2 and TH3 cells, whose excitatory receptive fields are ‘radially symmetric‘, thus enabling the calculation of contrast borders implicitly with respect to any direction of motion. That too shows how a small brain with limited neuroarchitecture can solve a hardware problem with the help of its intelligent software.

Comparable detection principles^[96] are discovered in amphibious fish (Periophthalmus koelreuteri) and in insects (Sphodromantis lineola). In mammals, erect body postures, for example, may address a threat signal to a rival. In humans, the gesture of a vertically oriented, horizontally waving index finger is worth mentioning. Depending on the context, this gesture can be interpreted as either a threatening or warning gesture in the biological sense. For instance, it can precede an attack or serve as a deterrent. The ritualization of behaviors is defined^[97] as stereotyped actions or sequences of actions which are either re-enacted within an individual's lifetime or transmitted by repetition from multiple individuals.

This suggests that the configuration-algorithm responsible for the distinction between profitable (e.g., prey-like) vs. dangerous (e.g., threat-like) may be implemented by quite different neural networks. Experiments with artificial neuronal nets support this presumption.

Previously thought to be characteristic of solely primates and mammals with front-facing eyes, the ability to process depth information from multiple visual points in space has been determined to be possessed by most amphibians, namely, frogs and toads. From an evolutionary point of view, there has been substantial support for the idea that stereo vision has evolved as a natural progression for animals with binocular vision, meaning, a substantial distance between the two eyes.^[98] This theory would mean that stereopsis has evolved independently at least four times to account for stereo vision being present in mammals, birds, amphibians, and some invertebrates.

For toads in particular, stereopsis would prove massively advantageous, as depth perception is particularly useful in calculating distance and aiding in catching prey.

In common toads, two visual information processing streams are involved in the recognition and localization of prey.^[99] The visual data is transmitted via tectofugal pathways that originate in the topographic map of the retinotectal system. The processing streams of both visual systems^[100] — one conveying information about the characteristics of the prey, the other about the current distance between the toad and the prey — are not completely separate from each other. Stereopsis and lens accommodation enable correspondence between the two.

Prey recognition uses calculations in retino-tectal and retino-pretecto-tectal/thalamic networks, the results of which are transmitted via tecto-bulbar/spinal projecting axons^{[101][102][103]} —ultimately to the hypoglossal nucleus, which triggers the motor coordination of snapping. The strategy of catching when determining the respective distance to the prey involves a tecto-tegmental relay circuit,^[104][105] which mainly uses monocular depth calculations. The information obtained from the accommodation of the eye lens^[106] makes it possible to estimate distances between the toad and its prey in a range between 5 and 20 cm, data that is also used to determine the absolute size of a prey object in order to ensure size constancy.^[107][108] Results from recording studies of toad's medullary neurons support the assumption that prey detection and local sign carrying tectal command elements converge on bulbar motor pattern generating nuclei.^[109][110]

As an equivalent of an innate releasing mechanism (IRM),^[111] it is proposed that a command releasing system for triggering prey capture includes visual command elements for prey 'recognition' (T5.2-neurons) AND visual command elements for prey 'localization' (T5.1- and T4-neurons), which control the 'orient' command:

• T5.2: recognized as prey AND T4 and T5.1: prey located in the peripheral visual field ——> ORIENT !

• T5.2: recognized as prey AND T1.1 and T2.2: prey located in the frontal visual field far afield ——> APPROACH !

• T5.2: recognized as prey AND T1.2: prey located binocularly at short distance ——>FIXATE !

• T5.2: recognized as prey AND T1.3 and T3: prey located binocularly in striking distance ——> SNAP !

Both processing streams in this multiple action system converge when the toad has its prey within striking distance. If the head end of an earthworm stretches in a way that is difficult to handle, the toad is prompted to tilt its head accordingly and choose a practical catching method, either by flicking its tongue or grabbing with its jaws. This binocular fixation process^[112] requires sensorimotor correspondence between configurative prey selection and object distance estimation. Since the experimental relaxation of the lens muscles with atropine had no significant effect on the accuracy of snapping in binocular toads^[113] —while one-eyed toads missed the target—it is assumed that stereopsis enables a finely tuned perception/action cycle during fixation.

The study on the interaction of spatial and spatio-temporal variations in contrast boundaries in the visual recognition of prey and threat objects raises the question of whether toads can distinguish and memorize visual details of moving stimuli. Yes, they do. Toads perceive different black-and-white patterns within the external dimensions of a configurative worm or bug shape (e.g., of 5 mm x 20 mm). They demonstrate this through stimulus habituation, a type of non-associative learning.^{[114][115][116]} When a dummy prey of a pattern A circles around a toad in the direction of its longitudinal axis in a special experimental setup, the toad's orientation turning reactions to visually catch the prey unsuccessfully decrease due to stimulus-specific habituation until A remaines unaddressed. But if, immediately after habituation to A, another prey dummy with a different pattern B triggers (dishabituates) prey orienting, it can be concluded that the toad distinguished between A and B. If in the reverse type of experiment, the pattern A following habituation to B elicits no prey orienting, toads preferred B over A. If in both habituation series the second pattern dishabituates the prey-orienting activity to the same extent, both were discriminated and equally attractive to the toad.

Based on paired tests of different patterns, a hierarchical order (dishabituation hierarchy)^[117] of more or less salient visual cues emerged within the dimensions of a prey schema, such as: surface shapes, leading edges, isolated dots, and stripes.

While a prey dummy circles around the toad in the rotating stimulation set-up, stimulus-specific aftereffects (memory traces) apparently accumulate, causing the latency periods of the toad's successive turning reactions to become longer, which reduces the turning frequency until stimulus-specific habituation occurs. The temporal course of recovery of stimulus responses after habituation to a prey dummy suggests two types of storage: short-term in the first 5–10 minutes and long-term up to 24 hours. Dishabituation through other stimuli^[118] during recovery (such as visual threats or somatosensory stimulation of the body skin) is most effective in the short-term phase.

The brain of the common toad can distinguish and store different visual stimuli through habituation based on the functions of brain structures, for example: a) for the specificity of stimulus patterns: the optic tectum in connection with caudal dorsal thalamic structures, b) for pattern-related aftereffects and memory: the ventral medial pallium, and c) for locus specificity in relation to areas of the visual field: the retino-tectal map.

Computer simulations demonstrate a match between a model^[119][120] and the original experimental data^[121] on which the dishabituation hierarchy is based. Model predictions concern the processes of habituation and the function of the ventral medial pallium, vMP, and explain why habituation is abolished after experimental lesions^[122] to vMP.

Stimulus specific habituation belongs to the evolutionary conserved neural circuits, from molluscs to mammals.^{[123][124][125]} The Aplysia gill and siphon withdrawal reflex of the marine snail Aplysia californica protects the gill and siphon from potentially threatening stimuli. To repeated release of this behavior with the same stimulus, the animal habituates in a stimulus-specific fashion in order to avoid wasted effort and be visually perceptive for new stimulus situations. Multidisciplinary experimental analyses of the complex molecular memory processes in sensori-motor integration of Aplysia result in the identification of a valuable model system with developments for medical application.^[126][127] Eric Kandel is a recipient of the 2000 Nobel Prize in Physiology/Medicine for his research on the physiological basis of memory storage in neurons.

• Feature detection (nervous system)
• Jörg-Peter Ewert
• Neuroethology
• Frogs portal

• Details and references on toad vision—see also video: [1] [2] [3]
• Feature detection by artificial neuronal net