Visual Perception

Dr. Robert G. Cook, Tufts University

Comparative Psychology: A handbook (1998)


Any inventory of the animal world quickly reveals a bewildering assortment of visual systems evolved for the purposes of detecting and using information from reflected light. These range from elementary photoreceptors that only discriminate light from dark, to the considerably more complex interactions of eye and brain responsible for visual perception in birds and mammals. This ability of nervous systems to construct internal visual representations of the outside world represents one of the most important milestones in the evolution of animal behavior and cognition. "Seeing" has the great advantage of allowing animals to obtain information concerning the nature and location of objects in their environment without the need for direct or close physical contact, as required by more proximal senses like touch, taste and smell. Because of this, visual information has become crucial to many animals for locating and identifying food, suitable habitats, predators, and conspecifics, as well as functioning to orient animals in their overall surroundings.

The direct physical stimulus for visual perception is reflected light of differing wavelengths. It is important to keep in mind that the resulting internal perception of this stimulation is not only a reflection of its physical properties, but also the changes induced by its transduction, filtering, and transformation by the animal's nervous system. Perhaps because of our strong visual predisposition as a species, the psychological difficulties inherent in using light to discern the structure of the external world are not widely appreciated and easily overlooked. For example, how are three-dimensional perceptual relations reconstructed from just two-dimensional retinal information, or how are an object's boundaries properly determined given all of the luminance contrasts present in any visual scene? It is the rapid and apparently effortless resolution of such computational problems that make the brain, and not the eye, the true organ of visual perception. Given the brain's very important interpretive role in the construction of any complex visual impression, it is far more important to be cognizance of an animal's perceived environment than its physical environment when trying to understand any visually-guided behavior. In a closely related point, it is important to recognize that each species also possesses a distinctive combination of sensory equipment and perceptual capacities tuned to the particular demands of its niche. The term "umvelt" has been employed by ethologists to refer to these different constellations of perceptual abilities and priorities across species.

Humans perceive light wavelengths of 400 to 700 nanometers, for instance, with a peak sensitivity near 555 nanometers. It is this physical stimulation that eventually results in our psychological impression of the color spectrum ranging from the blues to the reds respectively. Our umvelt or filter for "visible" light often causes us to overlook the fact that other animals are able to sense wavelengths outside of this range and into the ultraviolet or infrared regions of the spectrum. Bees can detect ultraviolet light, for example, which allows them to see the distinct ultraviolet patterns reflected by many flowers which act as visual guides to help the bees locate the flower's nectar. Variations in the umvelt of different species extend to other visual features besides the perception of color. The visual acuity of birds of prey, such as the falcon, easily exceeds our own, allowing these aerial hunters to detect prey at considerable distances. Pigeons can see patterns of polarized light in the daytime sky which are invisible to us, providing yet another possible source of information for the remarkable homing abilities of these birds. These examples offer only a glimpse of the kind of visual information available to various animals, but hopefully demonstrate that our own visual experience is only a rough guide, at best, to visualizing the perceptual world of other animals.

With such cautions in mind, comparative psychologists have made significant experimental progress over the last several decades towards understanding visual perception and its underlying mechanisms in animals. In this pursuit, these scientists have focused on three broad and related sets of questions. The first set of questions have been directed at determining the basic visual faculties of animals. The second set of questions have been more functional in nature, devoted to asking about the role of different forms of visual information in an animal's daily survival, and more specifically, the identity of the effective stimuli controlling these behaviors. The third set have focused on identifying and analyzing the mechanisms underlying these perceptions. Explorations of this latter question have ranged from studies of the anatomy and physiology of single nerve cells in the visual cortex to investigations of visual discrimination behavior in individual animals. Because of the vast wealth of information in the visual sciences, this essay by necessity must be limited in scope. As such, its goal is to provide a brief overview of how these different questions have been advanced by explorations of the visual stimulus control of behavior.

Identifying the effective stimulus controlling an animal's behavior is among the oldest and most fundamental of concerns in comparative psychology. Pursued in both the field and laboratory, the answer to this question not only advances the functional analysis of behavior, but indirectly furthers our understanding of an animal's visual capacities and priorities. The almost universal tactic in this approach is to appraise how animals react to variations of the visual input governing a particular behavior in order to isolate and identify the critical controlling features. This stimulus-analytic strategy typically involves a series of tests in which the complex original stimuli associated with a behavior are decomposed into their simpler constituent features or configurations to see which are still capable of maintaining the behavior of interest.

This strategy can be seen in Tinbergen's classic research on the begging responses of young Hering gulls (Tinbergen, 1951). Soon after hatching, gull chicks peck at their parent's bill in order to obtain regurgitated food. Tinbergen evaluated the effective stimuli controlling this begging response by presenting the chicks a graded series of cardboard models that mimicked the parent's head and bill in a variety of ways. The number of responses elicited from the young gulls by the different models revealed that the size, action, and position of the bill were all involved, but perhaps most important was the presence of the contrasting red spot at the tip of the adult's bill. Importantly, he also established that not all features of the adult's head were critical, as neither the color of the model's head or its bill influenced the strength of the chick's response. The prey-catching behavior of toads has been similarly subjected to this same type of analysis by measuring the vigor and number of responses elicited by systematically varying models of worm-like stimuli (Ewert, 1987). These studies have revealed that this animal's visual recognition of prey entails a conjunction of attributes involving the model's size, shape, and direction of motion; such that thin elongated stimuli moving along an extended worm-like axis are considerably more preferred by toads than the same shape moving perpendicular to this axis.

The selective responsiveness of the toads and young gulls to particular stimulus features in these functionally-oriented analyses of behavior provide an important, but limited, picture of these animals' perceptual aptitudes. Limited to readily summoned natural behaviors and their associated stimuli, these types of analyses reveal little about the range and sensitivity of animals to different forms of visual stimulation. Such analyses of the psychophysics of perception, which map out in detail the relations between a subject's psychological reaction to highly specified sets of physical stimuli, are best conducted with specially trained animals in the laboratory. This setting allows for a greater variety of stimuli to be tested, an increased precision in their description and method of presentation, and the opportunity to meticulously and repeatedly measure the animal's response to these stimuli.

Two types of visual discrimination procedures have been traditionally employed with animals for these psychophysical examinations of perception. The first involves teaching the animal a response to a single stimulus, which is then followed by a series of stimulus-analytic transfer tests examining their reaction to variations of the original signal. A pigeon might be trained to respond to a pecking key illuminated with a particular wavelength, for example, and then presented a variety of other colors to see how far this pecking response will generalize. The same tactic can been employed using habituation procedures. After an animal's response to a particular stimulus has been habituated through its repeated presentation, the animal's internal representation of the repeated stimulus is examined by measuring the amount of dishabituation produced by other stimuli. Again, the degree of stimulus control maintained by the transfer stimuli indexes the perceptual and conceptual similarity of these stimuli to the original. This type of habituation procedure has been extensively used to study the perceptual world of human infants, for instance. A second and superior discrimination procedure involves teaching the animal to differentially behave in the presence of two or more stimuli in an operant setting; either by making a response or not, as in a go/no-go procedure, or better yet requiring the animal to make a choice among two or more distinct response alternatives associated with these stimuli, as in a matching-to-sample procedure.

Since the 1950s, these types of learned discriminations have been used with animals to productively study their spectral sensitivity, visual acuity, and capacities to detect and discriminate fundamental visual dimensions such as hue, size, orientation, and brightness (Berkeley & Stebbins, 1990). While assembling these standardized measures of basic visual performance have been an essential step towards understanding any animal's perceptual world, they leave unanswered many of the most important and intriguing problems of visual perception and cognition. Consider for a moment the highly variable and constantly changing nature of the light falling upon the eye. Despite the numerous ambiguities and limitations in this changing input, our brain is still able to reconstruct a stable, unitary, and three dimensional visual impression of the world. Thus as any object moves it continues to be perceived and recognized as the same "thing," despite the continual transforming and different patterns of light produced by this movement. The exact computational processes by which the brain solves this "stimulus equivalence" or "many-to-one" problem remains a puzzle.

Judging from much of their behavior, however, complex animals such as birds and mammals act as if they too perceive a stable perceptual world, where associated collections of form and color attributes are also consistently interpreted and recognized as invariant "objects". The mechanisms underlying these more complicated aspects of visual cognition have become of increasing interest to animal researchers (Stebbins & Berkeley, 1990). One important catalyst for this has been the recent research on natural concept formation by pigeons (e.g., Herrnstein, Loveland & Cable, 1976). In such experiments, the animals learned to discriminate among realistic color photographs of different classes of objects, such as trees, humans, fish, and water. Not only were these discriminations between categories easily and rapidly learned by the pigeons, they also generalized to novel examples of these categories as well, suggesting a form of rudimentary conceptual behavior and raising the suspicion that the birds were perceiving the "objects" depicted in the slides. These results stimulated a great deal of new research into the perception and categorization of all types of complex visual stimuli by animals, and most especially, the pigeon.

One important by-product of this new look at complex stimulus perception in animals has been its inevitable comparison to human perception and performance with similar stimuli. Recent studies, for example, have provided experimental evidence that pigeons, monkeys, and humans similarly perceive some types of visual stimuli. Evidence for this shared perception comes from analyses of the discrimination errors made by each species when distinguishing among the same stimuli. An example of this can be found in Sands, Lincoln & Wright's (1982) experiments testing pictorial perception in rhesus monkeys. In this experiment the monkey had to judge whether two separate pictures were identical or not. Testing many pictures of flowers, fruits, monkey and human faces, the monkeys consistently found images from the same categories harder to discriminate than those of different categories, suggesting their perceptual categorization of these stimuli matched our own grouping of them. Given our shared primate heritage and similar brain organizations, this similarity is perhaps not too surprising.

Of greater interest is that pigeons have been shown to produce comparable results in strategically similar analyses. Blough (1982) required pigeons to discriminate among different letters of the alphabet. He found that the pigeons exhibited a pattern of discrimination errors very much like our own, confusing similar letters like "O", "Q" and "D" for instance. Despite the considerable differences in the organization, size, and natural history of the mammalian and avian brain, this behavioral similarity suggests a corresponding internal representation of these particular stimuli. If so, it raises the interesting question of whether these shared impressions are the product of analogous or common psychological algorithms as embodied by different neural architectures, or are instead generated by different computational processes that function to the same visual end.

This issue of mechanism has been explored using a variety of stimuli and behavioral procedures that try to isolate and measure the different portions of the perceptual process. One way is to test whether animals experience our visual illusions, since the "misperceptions" invoked by such stimuli help to directly reveal the visual system's active contribution to perception. Pigeons seem to suffer from some of the same geometric illusions as humans, such as the Ponzo and Mueller-Lyer illusions (Fujita, Blough, & Blough, 1991; Malott & Malott, 1970). Besides indicating a common perception, this type of similarity suggests even further that some of the underlying processes involved are also the same. In an attempt to isolate the early visual mechanisms responsible for registering and discriminating object surfaces and edges, my colleagues and I have been investigating the phenomenon of perceptual grouping in pigeons and humans by testing them with different types of multi-element visual stimuli. The results thus far have encouraged the view that these visual grouping processes are similarly organized in both species (Cook, 1992). Other visual discrimination experiments comparing pigeons and humans, however, have suggested that important process differences also exist. Humans differ in how quickly they can find a particular "target" element in a display depending upon the elements surrounding it, making it is easier, for example, to locate a "Q" in a field of many "O"s than vice versa. These type of asymmetries in the speed of human visual search help reveal the structure and organization of the elementary features employed in form perception. Testing pigeons with similar combinations of Qs and Os, Allan and Blough (1989) found no evidence of comparable asymmetries in the search behavior of these animals, raising the possibility that a different sets of visual features may be emphasized in the avian and mammalian perception of form.

This essay has tried to weave together something of the questions, findings, history, methods, and strategies employed in the comparative investigation of animal visual perception. Through the various experimental approaches outlined, we have been gaining an increasingly better understanding and insight of their internal world. Many of our answers, however, remain speculative and tentative. As a consequence, they offer an exciting and open invitation to all students to join in the scientific search for a better bird's eye view.

Allan, S. E., & Blough, D. S. (1989). Feature-based search asymmetries in pigeons and humans. Perception & Psychophysics, 46, 456-464.

Berkeley, M.A., & Stebbins W.C. (1990). Comparative perception: Complex signals. John Wiley: New York.

Blough, D.S. (1982). Pigeon perception of letters of the alphabet. Science, 218, 397-398.

Cook, R. G. (1992). Dimensional organization and texture discrimination in pigeons. Journal of Experimental Psychology: Animal Behavior Processes, 18, 354-363.

Ewert, J.-P. (1987). Neuroethology of releasing mechanisms: Prey-catching in toads. Behavioral and Brain Sciences, 10, 337-405.

Fujita, K., Blough, D.S., & Blough, P.M. (1991). Pigeons see the Ponzo illusion. Animal Learning & Behavior, 19, 283-293.

Herrnstein, R. J., Loveland, D. H., & Cable, C. (1976). Natural concepts in pigeons. Journal of Experimental Psychology: Animal Behavior Processes, 2, 285-311.

Malott, R. W., & Malott, M. K. (1970). Perception and stimulus generalization. In W.C. Stebbins (Ed.), Animal psychophysics: The design and conduct of sensory experiments (pp. 363-400). New York: Plenum.

Sands, S. F., Lincoln, C. E., & Wright, A. A. (1982). Pictorial similarity judgments and organization of visual memory in the rhesus monkey. Journal of Experimental Psychology: Animal Behavior Processes, 4, 369-389.

Stebbins, W. C., & Berkeley, M.A. (1990). Comparative perception: Basic mechanisms. John Wiley: New York.

Tinbergen, N. (1951). The study of instinct. Clarendon Press, Oxford.