Orbeli's Experimental Work on Color Discrimination in Dogs GEORGE WlNDHOLZ

University of North Carolina at Charlotte Abstract--In the 1900's, L. A. Orbeli, on I. P. Pavlov's request, attempted to establish color discrimination in dogs. Previous experiments used the motor discrimination method and gave inconsistent results. In contrast, Orbeli used the salivary conditional reflex method, which he considered to be more precise than the method that relied on erratic movements of a dog. After experimentation that lasted about one and a half years, Orbeli failed to establish color discrimination. When subsequent experiments by Russians and Germans yielded positive results with the motor discrimination method, Orbeli switched to this method and also obtained positive results. These findings were confirmed by most subsequent experiments on color discrimination in dogs. The utility of Pavlovian conditioning in sensory experimentation was not universally justified and its importance was greater in the study of learning.

PROBABLY IN 1905, I. P. Pavlov received in his laboratory at the Imperial Institute of Experimental Medicine two distinguished visitors. One of them was Pavlov's patron, Prince A. P. Oldenburgski~, and the other was Sir Arthur Nicholson, England's Ambassador to the Czar's Court in St. Petersburg. Pavlov, explaining his conditional reflex experiments, told the visitors that dogs can be conditioned to any stimulus. OldenburgskiT interpreted Pavlov's words to Nicholson. The Ambassador must have fully grasped Pavlov's explanation because he asked whether this meant that a dog could be conditioned to a queen of spades. Oldenburgski'~ laughed and suggested that Pavlov try it. Pavlov said "good" and urged his coworkers to determine whether or not dogs could be conditioned to respond to colors. Pavlov's disciple L. A. Orbeli complained years later of having to spend one and a half years studying color discrimination in dogs (Orbeli 1966, pp. 48 49). Leon A. Orbeli (1882-1958) first heard Pavlov lecture in 1900, while a medical student at the Imperial Military-Medical Academy. After 1904, Orbeli served in the Baltic Fleet and was stationed near St. Petersburg. This allowed him to work at the Physiological Department at the Imperial Institute of Experimental Medicine under Pavlov's direction. Orbeli's initial research dealt with digestion, and then, following Pavlov's own change of interest, on conditional reflexes in dogs. Orbeli's

work on color discrimination in dogs was done as part of the requirements for the MD degree (Kvasov and Fedorova-Grot 1967).

Previous Experimental Work on Color Discrimination in Dogs Orbeli's attempt to determine whether dogs could discriminate colors was by no means original. Such studies were conducted as early as 1884 by Vitus Graber and by Sir John Lubbock in 1888. Of methodological importance was Graber's reliance on the animal's motor movements in a choice situation and the distinction between the quantitative and qualitative effects of light waves. The quantitative aspect of light involves the perceived degree of brightness, whereas the qualitative aspect refers to perception of chromacity. As the experiments involved discriminative stimuli that humans perceive as colored, Graber performed experiments of brightness and color discrimination. The dog was presented with two adjacent chambers and given the choice of entering 1) a darkened or a lighted chamber and 2) a light-red or a darkblue chamber. The dog entered the lighted chamber and the dark-blue chamber. In Lubbock's experiment, a poodle named Van had to discriminate between pairs of colors. Lubbock reports: 1 prepared six cards about ten inches by three, and coloured in pairs--two yellow, two blue, and

133

134

WINDHOLZ

two orange. I put one card of each colour on the floor, and then, holding up one of the others, endeavoured to teach Van to bring me the duplicate. That is to say, that if the blue was held up, he should fetch the corresponding colour from the floor; if yellow, he should fetch the yellow, and so on. When he brought the wrong card he was made to drop it and return for another, until he brought the right one, when he was rewarded with a little food (1888, p. 280). The dog was unable to make correct choices even after a ten-week training period. However, an experiment on color discrimination in dogs by Himstedt had a different outcome. A poodle was trained to bring a red ball on the command "bring red." Then the dog was held on a leash while red, green, or other colored balls were thrown in front of him. To control for smell, the balls were dyed with the same paint, impregnated with paraffin, or halves of the balls were covered with a green and the other halves with a red material. The balls were dropped about 4 to 6 meters from the dog to further minimize the influence of smell. The dog was given the command "find red" and released. The frequency of correct choices made by the dog led to the conclusion that dogs do distinguish colors (Himstedt and Nagel 1902). In view of existing knowledge of color recognition by dogs, Orbeli's research project required justification. It is believed that the assumption of Pavlov's decision to experiment with color in response to Prince Oldenburgskn s suggestion is alluring but not necessarily correct in view of Pavlov's own intellectual curiosity. During the first decade of his systematic investigation of conditional reflexes, Pavlov was still testing the possibilities and limits of the method. Orbeli stated that he planned to determine color discrimination in dogs by conditioning of salivary reflexes, whereas previous research relied on a motor discrimination method, that is, the movement of a dog in a choice situation. The conditioning method seemed to have the advantage that the flow of saliva could be precisely measured, whereas a dog's erratic movements were often confusing. Hence the salivary response was a reliable indicator of underlying processes. Scientists, in general, had explained dogs' movements in a manner that was dubious at best, whereas the production of saliva was a completely objective observation. Orbeli's Experiments on Color Discrimination in Dogs

Orbeli's first experiments on color discrimination in dogs were performed in August 1905 and continued until March 1907. By that time, Pavlov's research with conditional reflex was already in an

Pay, J. Biol, Sci.

October December1989

advanced stage. Seven years had passed since Pavlov's student, S. G. Vul'fson (1898), found in laboratory experiments that dogs salivated to foods presented at a distance. In 1903, Pavlov's coworker I. F. Tolochinov had described the principles of reinforcement and extinction, and in 1906 Pavlov's student V. N. Boldyrev established that dogs not only salivate to foods shown at a distance, but also to extraneous stimuli--auditory, olfactory and visual--provided that these stimuli are presented simultaneously with food. Another of Pavlov's disciples, N. A. Kashereninova (1906), formulated the principles of generalization and discrimination for tactile sensory modalities. Kashereninova's research was of crucial importance to Orbeli's work for determining whether or not dogs could discriminate colors. Orbeli's (1909) subjects were three dogs, a mongrel, a crossbred, and a Gordon setter. The saliva came from the parotid gland and mucous membranes. The salivary reaction was measured during time intervals lasting 1, 2, and 3 minutes. A dog was experimented on only once in a day with an interval of one to three days between trials. The experiments were performed in a darkened room, sufficiently illuminated to allow the experimenter to observe the salivary response. The stimuli were projected on a screen in front of the dog. Initially Orbeli hypothesized that dogs discriminate colors provided that, when conditioned to respond to one color, they did not respond to other colors. To test the hypothesis, the dog Belka was placed in a darkened room in front of a screen on which was projected a red square. After each successive presentation, the dog was reinforced with meat-powder. The conditional reflex was established between the 65th and 70th trial. After the conditional reflex was well established, namely after 170 reinforced trials, identical squares of green, blue, violet, and white colors were projected. The dog responded with almost equal amounts of saliva to each color, suggesting that the dog failed to discriminate colors. Because stimuli presented without reinforcement tend to evoke less and less saliva, Orbeli projected violet, blue, green, and red colors in that order. The premise was that if the dog does not discriminate colors, salivation in response to all colors will diminish progressively. This was the case since even the red color, which was originally reinforced, failed to evoke salivation. In the next experiment, Orbeli decided to control for brightness by reinforcing the red but not the green squares. Within one and a half months, 78 stimulations with the red square were reinforced by food, while 82 stimulations with the green

Volume 24 Number 4

COLOR DISCRIMINATION

squares were not. Orbeli found that when the green square was presented for a number of times, the salivary response diminished, but so did the flow of saliva in response to the projections of the red square. When the presentation of the red squares was reinforced again, the conditional reflex to red stimuli was restored, but so was the salivary reaction to the green squares. Since no discrimination was obtained, Orbeli surmised that dogs did not react to colors per se. Orbeli decided that a more stringent control for brightness was necessary. The dog was presented for a longer period with a bright green square. Occasionally, a red square was substituted for that stimulus and then was followed by feeding. Orbeli hypothesized that if dogs react to stimuli of different wavelengths the red square should evoke a conditional salivafy~ response. It did so with difficulty, only after about 140 reinforced trials, and even then the amount of saliva was small and unstable. Orbeli then projected violet, yellow, and blue squares, which also evoked similar amounts of saliva. Orbeli surmised that the stimuli that actually evoked the salivary responses were fluctuations in light intensity that came about when color squares were changed9 To control for the constancy of the light intensity, a new series of experiments were started, using the dogs Gordon and Kikimora as subjects. A white square was projected on the screen and on it either the red or green squares were superimposed in such a manner that the degree of brightness was lower yet equal for both colors. The presentation of the red stimuli, but not that of the green, was reinforced with food. The dog Gordon received 95 reinforced and 215 nonreinforced trials, yet the presentation of the red color alone failed to evoke salivation. The dog Kikimora received 108 reinforced and 121 nonreinforced stimuli. Interestingly, after the 100 nonreinforced trials to the presentation of the green stimuli, the dog began to salivate to the green and to the red stimuli. Orbeli thought that the conditional reflex was formed to the lowered luminosity of the stimuli, but not to specific colors. After realizing that the reaction to the red and green stimuli was similar, Orbeli substituted blue for the green stimuli. Red and blue stimuli were projected alternatively, but only the former were reinforced. There were 28 presentations of the red, and 46 presentations of the blue stimuli. No difference in reaction was found to the red stimulus on the one hand, and to the blue and green stimuli on the other hand. Orbeli favored the interpretation that it is the degree of brightness rather than color that evokes a conditional reaction. So, using the dog, Kikimora, the luminosity was lowered and the dog was

135

reinforced after the presentation of the color red. After about 10 to 15 presentations, the conditional reflex was firmly established and then blue, violet, green, black, and smoky colors were presented. The salivary responses to these colors were quantitatively similar. However, the diminished luminosity produced a salivary response. This was seen most dramatically in response to the color black, which evoked the most powerful salivary reaction. In a final analysis Orbeli (1909) concluded that changes in brightness produce a reaction, but color differences do not: in regard to the controversial issue whether clogs are able to react differently to various wavelengths, we have to express ourselves negatively; in our investigation we have found no evidence on the discriminatory action of rays of different wavelengths (p. 81). 9

.

.

This conclusion, however, was not Orbeli's last statement on color vision in dogs. Contemporary researchers obtained different results, challenging Orbeli's findings. Thus, to determine whether dogs react to brightness or color, Samojloff and Pheophilaktowa (1907) trained a dog to approach two boxes, the first marked by a green circle, and the second by 1 of 50 shades of gray to control for brightness. In 1173 trials, with an error rate of about 20%, the dog did discriminate green from grays. The researchers concluded that dogs discriminated colors provided they were subjected to prolonged training in familiar conditions. However, faced with complex conditions, dogs ignored colors and relied on other cues. Nagel (1907) reported that Himstedt continued the 1902 work on color discrimination and found that a dog could distinguish between red, blue, and green objects. However, Pavlov's German student Nicolai (1907) was unable to obtain color discrimination in dogs using the conditional reflex method. Far more extensive experiments were performed by Kalischer (1909), who developed a training procedure known as Dressur--essentially a variant of the motor discrimination method--that enabled him to study sensory processes in dogs. A dog was placed in a wooden container that had a small window. Light bulbs of different colors were placed near the window in such a way that the rays could be visible to the dog. A translucent glass plate was placed between the window and the bulb. Near the window was a small opening through which the experimenter placed a piece of meat in the box. The training procedure consisted of making the dog eat, or not eat, the meat in response to light stimuli. When the red light was on, the experimenter allowed the dog to eat the meat. When the light was extinguished, thus making the box dark on the

136

Pav. J. Biol. Sci. October December 1989

WINDHOLZ

inside, the experimenter held the meat in his hand preventing the dog from eating. Soon the dog learned to grab the meat when the red light was on, and not to touch it when it was out. Subsequently, Kalischer substituted blue light for darkness and prevented the dog from eating. The dog learned to discriminate between red and blue light. Next, Kalischer lighted bulbs of different colors. The frequency of mistakes by the dog declined progressively in response to blue, green, and orange color stimuli. To control for brightness, Kalischer shifted the distance of the translucent glass between the light source and the window or used bulbs of different power output. Kalischer concluded that dogs discriminate among colors. Orbeli (1913) attempted to determine the reason for diverse experimental findings and resolve the issue of dogs' color discrimination. Orbeli used, with some modification, Kalischer's training procedure. In a semidark room, the dog stood on a table facing a lantern. In front of the lantern stood a small shelf, on which the experimenter placed meat powder. A rope over a horizontal crossbeam located over the table could be used to restrain the dog from eating the meat. Five or six times a day, with a few minutes intervals, the lantern was lighted for the duration of 1 minute and the dog allowed to eat the meat powder. The dog was restrained from eating the meat powder when the lantern was not lighted. Soon the dog learned to discriminate between the lighted and the unlighted lantern, eating when the lantern was lighted and not eating when it was extinguished. When this training was completed, the dog was successfully trained to eat when the lantern beamed red light, and not to eat when the lantern beamed blue-violet light. Discrimination between red and green was also accomplished. The controls used by Orbeli in establishing chromatic discrimination were important. The light in the lantern was produced by two light bulbs that obviously generated heat. Because different colors conduct heat more or less efficiently, the dog could have discriminated temperatures rather than colors. To control for temperature, a narrow container of water was placed between the dog and the lantern. The different color sensations were obtained by placing colored glass plates in the lantern. The mechanics of changing the glass plates could have influenced the dog. To control for this, the presentation was in r a n d o m order and the experimenter's movements were standardized. The most intractable problem was the control for brightness. To achieve some certainty that the discrimination was made on the basis of chromaticity rather than brightness, the intensity of light was varied, as well as small variations in the basic

color. Orbeli determined that color discrimination occurred because the dog consumed the meat powder in response to the hues of redness, but not to blueness or greeness. Following these experimental findings, Orbeli commented on color discrimination in dogs. Only some dogs were able to discriminate colors and then only after a prolonged and difficult training process. The colors must be bright because it is possible that in dogs the absolute thresholds for different wavelengths are higher than in humans. In the final analysis, color vision in dogs is at best rudimentary and probably plays a minor role in dogs' interaction with the environment.

Orbeli's Conclusions in Light of Subsequent Research Three quarters of a century have passed since Orbeli's conclusions about color discrimination in dogs. Let us now consider his findings in relation to subsequent research. In a 1969 review of relevant studies, Rosengren concluded that 9 out of 16 studies indicate that dogs have color vision9 Her own experiment on color discrimination also supported this conclusion. Orbeli's experiments, as well as those of many others, were conducted with only the most rudimentary measurements of chromaticity and optics. Orbeli was convinced that conditioning was superior to methods that relied on motor movements--the motor discrimination method--of the dog insofar as the salivary response could be described with greater precision than could a dog's erratic movements. This assumption was apparently unjustified. Using conditioning, both Orbeli and Nicolai were unable to obtain color discrimination and so Orbeli switched to Kalischer's training method. However, Orbeli and Nicolai were not alone in placing too much hope on Pavlov's method. As late as 1916, John B. Watson wrote that conditioning was suited to the investigation of sensory modalities in animals. However, as history has shown, Pavlov's methods were most fruitful in the study of learning.

Acknowledgments The author thanks J. R. Kuppers, P. A. Lamal, J. G. Storey and R. E. Vogel for their suggestions and editorial assistance and librarians B. J. Lisenby and A. A. Cobb for their help in providing me with the source material.

References Boldyrev, V. N. (1906) Obrazovanie iskusstvennykh uslovnykh (t. e. psikhicheskikh) refleksov i svol*stva Lkh (Prevrashcheme mestnago okhlazhdema kozhl

Volume 24 Number 4

COLOR DISCRIMINATION

v vozbuditel~ otdelenil'~ sl~ny) (The formation of artificial conditional (i.e. psychic) reflexes and their characteristics (The transformation of the localized cooling of the skin as evocator of salivary flow [in Russian]). Trudy Obshchestva Russkikh Vrache7 v S.-Peterburge s Prilozheniem Protokolov Zasedan~ Obshchestva za 1905-1906 God, 73, 198-228. Graber, V. (1884) Grundlinien zur Erforschung des Helligkeits-und Farbensinnes der Tiere (Guidelines for Investigating Brightness and Color Sense in Animals [in German])9 Prag: F. Tempsky. Himstedt, F., and Nagel, W. A. (1902) Versuche fiber die Reizwirkung verschiedener Strahlenarten auf Menschen- und Thieraugen (Experiments on the effect of diverse rays on human and animal sensory receptors [in German]). Festschrift der AlbrechtLudwigs- Universitii t in Freiburg zum Fiinfzigjiihrigen Regierungs-Jubiliium Seiner K6niglichen Hoheit des Grossherzogs Friedrich. Freiburg: C. A. Wagner's Universit~/ts-Buchdruckerei. Kalischer, O. (1909) Weitere Mitteilung fiber die Ergebnisse der Dressur als physiologischer Untersuchungsmethode auf den Gebieten des Geh6r-, Geruchs- and Farbensinns. (Additional reports about the results of training as a physiological research method in the auditory, olfactory, and color perception senses [in German])9 Archly fiir Physiologie, 303-322. Kashereninova, N. A. (1908) Matertaly " k Izuchenuu "" Uslovnykh Sl~nnykh Refleksov na Mekhanicheskoe Razdrazhenie Kozhi u Sobaki (Contributions Toward the Study of Conditional Salivary Reflexes in Response to the Tactile Stimulation of a Dog.~ Skin [in Russian]). St. Petersburg: Tipografila Shtaba Otdelnago Korpusa Zhandarmov. Kvasov D. G., and Fedorova-Grot, A. K. (1967) Fizmlogwheskala Shkola I. P. Pavlova (The Physiological School of I. P. Pavlov [in Russian]). Leningrad: Izdatel'stvo Nauka. Lubbock, J. (1888) On the Senses, Instincts, and

137

Intelligence of Animals. New York: D. Appleton. Nagel, W. A. (1907) Der Farbensinn des Hundes (The color sense of the dog [in German]). Zentralblatt fdr Physiologie, 21, 205-206. Nicolai, G. F. (1907) Die Physiologische Methodik zur Erforschung der Tierpsyche; ihre M6glichkeit und ihre Anwendung (The physiological method for investigating the animal psyche; Its possibility and application [in German]). Journalfttr Psychologie und Neurologie, 10, 1/2, 1-27. Orbeli, L. A. (1909) Uslovnye refleksy s glaza u sobaki (The visual conditional reflexes in a dog [in Russian]). Arkhiv Biologicheskikh Nauk, 14, 31 136. Orbeli, L. A. (1913) K voprosu o razlichenii tsvetov sobakami (On the issue of color discrimination in dogs [in Russian]). Voprosy NauchnotMeditsmy, *" ''" 1, 513-524. Oberli, L. A. (1966) Vospommanna" x~ (Reminiscences [in Russian]). Moscow: Izdatel'stvo "Nauka." Rosengren, A. (1969) Experiments in colour discrimination in dogs. Acta Zoologica Fennica, 121, 319. Samojloff, A., and Pheophilaktowa, A. (1907) Uber die Farbenwahrnehmung beim Hunde (About color perception in dogs [in German]). Zentralblatt far Physiologie, 21, 133-1399 Tolochinov, I. (1903) Contribution ~ l'etude de la physiologic et de la psychologic des glandes salivaires (Contribution to the study of the physiology and psychology of salivary glands [in French]). Fgrhandlingar rid Nordiska Naturforskare- och Liikarem6tet i Helsingfors den 7 till 12 Juli 1902. Helsingfors: Helsingfors Centraltryckeri & Bokbinderaktiebolag. Vul'fson, S. G. (1898) Rabota Sl~nnykh Zhelez (The Function of Salivary Glands [in Russian]). St. Petersburg: Trenke i F~sno. Watson, J. B. (19t6) The place of the conditionedreflex in psychology9 Psychological Review, 23, 89116.