Subhash C. Kak
Louisiana State University, Baton Rouge

Karl von Frisch was awarded the Nobel Prize for physiology and medicine in 1973 to honour his work on animal behavior, especially his theory that honey bees communicate by a dance language. This theory was presented first by von Frisch in the 1940's and within a few years it became quite popular as it suggested symbolic communication in a non-human species.

That foraging honey bees, who have found a food source, perform a dance on their return to the hive has been known for a long time. In fact the first recorded reference to this is in Aristotle's Historia Animalium (330 B.C.) (Wenner and Wells 1990, page 270), where he appears to suggest that the recruited bees follow the forager to the food source. In this original form there is nothing very striking about this dance, since it serves only to attract the attention of the other bees to the foraging bee who leads the recruits to the source of food, and it is clear that bees and other animals do have the capacity to associate and remember visual maps. The process described by Aristotle does not constitute a language.

The theory of von Frisch (1947) claimed that the runs and the turns of the dance were correlated with the distance and direction of the food source from the hive. Von Frisch developed several experiments to verify the theory. He claimed that while honey bees use both odor and dance language cues for ordinary communication his experiments demonstrated that the bees could locate the food source with just the dance information. But even before he was awarded the Nobel prize, a zoologist named Adrian Wenner found that his own experiments went against von Frisch's predictions. With better controls against odor cues, Wenner (1967) found that the bees were unable to locate the food source with dance alone. Wenner now worked to disseminate his odor theory, but subsequent to a paper in Nature with his collaborator Wells (1973) he found himself barred for many years by the editors of major scientific publications from presenting his results.

The first and the only attempt to meet Wenner's objections came through the results of the doctoral dissertation of James Gould (1974, 1975). He conceded that the design of the experiments of von Frisch had been flawed and that his earlier results could all be accounted for by Wenner's locale-odor hypothesis. Gould argued that the controversy arose due to a difference in training techniques: Wenner's training method encouraged bees to rely on odor, while von Frisch's training method led them to use the dance cues. But this distinction has not been conceded by Wenner.

The 1980's saw a renewed challenge to the dance theory (Rosin 1980, 1988). In the face of evidence against the original dance language hypothesis of von Frisch, even its proponents now grant the locale-odor mechanism a role in the communication amongst bees. But, as argued by Rosin (1988), these proponents do not, amongst themselves, agree on a 'single' dance language. Says Rosin: 'There are numerous contradictory versions of the 'dance language' hypothesis that concur only in the belief that somewhere, somehow, some honey bees use 'dance language' information, but disagree on practically anything else." This raises questions regarding sociology of science and the scientific method (Veldink 1989, Wenner 1989). It appears that once the dance language hypothesis was adopted by the scientific establishment in animal behaviour as a fact, all challenges to it were summarily brushed aside. It is not very well appreciated that science is a collective, social process and the development of each discipline proceeds within a well-circumscribed framework. These issues as well as the history of the dance language hypothesis have been discussed at length in a new book by Wenner and Wells (1990).

More on the Dance Language Hypothesis
We now briefly present an analysis of the dance language hypothesis from an information theory perspective. One reason that the controversy has lasted for about three decades is that this hypothesis has been interpreted differently by the various protagonists. One might talk of two versions of the hypothesis for convenience. In the 'strong' version bees communicate distance information entirely by dance, while in the 'weak' version bees use dance information to find the general vicinity of the feeding station and then use odor information to pinpoint the station. Most recently, some defenders of the dance hypothesis have spoken only of the 'weak' hypothesis, as is the case with Seeley's (1991) review of the book by Wenner and Wells.

It appears that the controversy is partly of a semantic nature. What does language mean? According to Webster's Collegiate Dictionary one definition is "signs, sounds, gestures, or marks having understood meanings." Operationally, this means that a language must be associated with a vocabulary of basic signs and sounds and a grammar that allows the signs or sounds to be combined into an unlimited number of statements. The statements of a language must have the capacity to embrace a potentially infinite variety of possibilities. Put differently, a language allows representation and communication of knowledge about the world.

The verification evidence furnished by the defenders of the hypothesis only establishes that the honey bee dance is somewhat correlated with the distance and direction information; it does not follow that the dance is a language for the bees. To establish that it is essential to demonstrate that just the dance information is enough to guide the bees to the feeding station.

The evidence garnered by Wenner and associates appears to establish that the dance does not represent a language to the bees, whereas the verification evidence indicates that there might still be a dance language for the scientist. The analogy of this distinction in human terms may be seen in the case of a person who has ingested mood altering drugs. The change in the blood chemistry defines a language for the chemist and not for the layperson, who might reach the same conclusion based on other cues.

Another issue of interest is the cognitive capacity of the honey bee. Note that information is a logarithmic function of the total number of possibilities. Defining distance and direction requires information which is potentially infinite. Now if one were to say that this definition is being made only approximately, the direction information could be finite but the distance information would still be potentially infinite. That is why dealing with distance information requires abstract concepts, and only humans appear to be capable of such reasoning. Now one might argue that the distance information is defined approximately only within a certain range of a few hundred meters, which would render it finite. It should be noted, however, that the foraging distance of the bees can be over several kilometers. But it is hard to see how a cognitive model which works on distance information in only a small range would operate. Nevertheless, one might visualize associative neural models, such as the one proposed by the author (Kak 1990), that allow cues, in addition to the one on direction alone, in the dance of the bee. These models define a process similar to the one where a few notes of a song recall the entire tune. It is possible, therefore, that a dance may carry cues strongly correlated with certain aspects of the find of the food source that may be correctly interpreted by the bee's cognitive apparatus. However, such a mechanism could only work for standard, or common, settings for the food source, and it could not describe novel settings. But if such a process is at work, it would require the design of a careful experiment to test it. On the other hand, processing of odor information is easily understood by neurophysiological models.

The above arguments indicate that the dance of the bees could not communicate distance information to the bees, but it leaves open the possibility that it might communicate approximate direction information. However this latter possibility does not qualify as a language. It is essential to reiterate that the associative visual map built by the foraging bee will allow this bee to find the food source again. Such a bee can then guide other bees, who are attracted to follow this bee by the dance as well as the odor picked up by the bee, to the food source. Perhaps it is in this limited sense that a simultaneous consideration of dance and odor cues might be relevant.

The latest results of Wenner and Wells (1990 pages 299-311) demonstrate that even the weak version of the hypothesis is invalid. This conclusion was reached from observations from an experiment which was so designed so that the recruit bees could use either dance language or odor cues.

Why Did The Controversy Stay Alive this Long?
One might ask why has the controversy lingered for so long, especially since the hypothesis itself is relatively simple. This can be answered only by considering the sociology of science. Thomas Kuhn (1962) documented how major shifts in scientific paradigms are accepted only after fierce challenge. Wenner and Wells claim: "Neither is evidence the most important factor in the resolution of controversies, contrary to standard belief in science. Adequate evidence, as obtained from crucial experiments, cannot immediately override emotional attachment to pet theories. That is because paradigm holds dictate to the participants how that evidence shall be viewed. Thus it is that adequate evidence, while absolutely essential in the resolution of controversies, is necessary although not sufficient during that resolution." [page 266]

Thie reasons why the dance language hypothesis of von Frisch was so eagerly adopted in the 1940's and 1950's is in itself an interesting story. Von Frisch's experimnents did indicate a correlation between observation and the dance language hypothesis. But as Wenner and Wells argue (1990) these experiments were seriously flawed. They point to several specific counts on which these experiments and the analysis thereof could be criticized. The main ones are: (1) negative results were ignored, (2) the importance of conditioning was not recognized, and (3) there was lack of adequate controls in experimental design.

The other reasons relate to the milieu in which the dance language theory was advanced. At the end of World War II there was a great faith in the dawning of a new golden age. The von Frisch hypothesis was so sweeping in scope that it could conceivably completely alter man's understanding of animal behaviour. This is why once a few leading scientists accepted this hypothesis to be true, it rapidly became the dominant paradigm. In an early promotion of this hypothesis, August Krogh, the 1920 winner of the Nobel prize in medicine, wrote:

The series of experiments constitutes a most beautiful example of what the human mind can accomplish by tireless effort on a very high level of intelligence. But I would ask you to give some thought also to the mind of the bees. I have no doubt that some will attempt to explain the performances of the bees as the result of reflexes and instincts ... for my part I find it difficult to assume that such perfection and flexibility in behavior can be reached without some kind of mental processes going on in the small heads of the bees. [Krogh 1948 quoted by Wenner and Wilks 1990]

But this endorsement does point to the unexplored area of intentionality in animal behaviour and animal communication. Clearly linguistic capability presupposes intentionality. Put differently, language presupposes that the animal has the neural apparatus for knowledge representation and internal discourse. Oakley (1985) summarizes the evidence thus:

[It appears] that spatial mapping is an early representational acquisition, and seems to be well developed in all mammals irrespective of grade of neocortical differentiation ... Social modelling, including representations of the self, may have only a rudimentary form in the rat compared to the more impressively neocortically evolved groups, such as primates and cetaceans (dolphins, porpoises and whales)... Language in humans is based on the most complex representational system of which we are aware and is primarily neocortically based... It is possible that human language developed as a means of internal discourse based on acoustic representations, and only secondarily acquired the role of communication with others. Much of animal reasoning, particularly in primates and cetaceans, may be mediated by a similar internal discourse, though not necessarily an acoustic one.

One can analyze behaviour to determine linguistic capability. Behaviour is a response to the internal state and the stimulus from the emivironinent. Whether an animal possesses a language, verbal or non-verbal, would depend on the richness of the internal states to support a grammar. Clues to this can be obtained from the repertoire of responses. It does appear that some higher mammals, like chimpanzees and gorillas, are capable of a genuine non-verbal language (Blakemore and Grienfield 1987). But studies of this kind are in their infancy and very little can be extrapolated beyond the species that have been studied.

Conclusions
The major conclusion for the historian of science to draw from the honey bee dance controversy is that challenge of established paradigms, even when the evidence compels such a challenge, is extremely hard. Once scientists and scholars invest parts of their career in support of a paradigm, it becomes a sort of a self-betrayal to abandon it. This is why Planck (1950) claimed that often only death can separate scientists from their pet theories.

One important reason that the von Frisch dance language hypothesis survived so long is because questions about the process supporting the hypothesis were not asked. The question to ask is whether the brain of the bee is capable of processing of such abstract information. And if the answer to that is yes, then how does the cognitive mechanism work? The protagonists, under pressure from conflicting evidence, tried to present their hypotheses in a form that would be unfalsifiable. Thus it has been claimed that both dance language and odor are used by the bees. Unhappily for the protagonists, such a modified hypothesis continues to be at variance with the evidence.

Another important reason why the controversy has continued this long is that the protagonists have interpreted key words differently. Thus the supporters of the dance language may now mean it to stand for just a protocol to attract the bees to the site of food. But the opponents are justified in claiming that such a protocol does not represent a true language.

The problem of animal communication, which is merely one aspect of animal behaviour, is related to the old paradox of determinism and free will (Kak 1986, Kak 1987). Physics, which ultimately lies at the basis of chemistry and biology, does not allow free will, but no person would deny the reality of his free choices. The system of language, and its use to represent knowledge, validates this belief in the reality of free will. Somehow the postulate that only humans have linguistic capability appears to make the determinism/free will paradox less threatening. However, as described earlier, one might talk of several levels of knowledge representation capability that would embrace all animals. While only the behaviorists would describe animal processes in terms of mechanistic and reflexive associations between stimuli and responses, even those who would wish to introduce consciousness as a fundamental category of nature do not know how that can be done within the current structure of science. Roger Penrose (1989) has summarized the various issues involved in considering a framework which allows both the determinism of physics and the apparent intentionality of the animal. Yet such speculations merely give intimations of controversies that will continue to erupt in this field.

REFERENCES

Blakemore, C. and Greenfield, S.
1987: (Editors) Mindwaves: thoughts on intelligence, identity, and consciousness, Basil Blackwell, Oxford.

Frisch, K. von.
1947: The dances of the honey bee, Bulletin of Animal Behaviour, 5, 1-32.

Gould, J.
1974: Honey bee communication, Nature, 252, 300-301.
1975: Honey bee recruitment: the dance language controversy, Science, 189, 685-693.

Kak, S.C.
1986: The Nature of Physical Reality, Peter Lang, New York.
1987: Patanjali and Cognitive Science, Vitasta, Baton Rouge.
1990: Self-indexing of neural memories, Physics Letters A, 143, 293-296.

Krogh, A.
1948: The language of the bees, Scientific American, 179, 18-21.

Kuhn, T
1962: The Structure of Scientific Revolutions, University of Chicago Press, Chicago.

Oakley, DA.
1985: (Editor) Brain and Mind, Methuen, London.

Penrose, R.
1989: The Emperor's New Mind: concerning computers, minds, and the laws of physics, Oxford University Press, New York.

Planck, M.
1950: Scientific Autobiography, Williams and Norgate, London.

Rosin, R.
1980: The honey bee "dance language" hypothesis and the foundations of biology and behaviour, Journal of Theoretical Biology, 87,
457-481.
1988: Do honey bees still have a "dance language"? American Bee
Journal, 128, 267-268.

Seeley, T.
1991: Bee warned, Nature, 349, 114.

Veldink, C.
1989: The honey-bee language controversy, Interdisciplinary Science Reviews, 14, 166-175.

Wells, PH. and Wenner, AM.
1973: Do bees have a language? Nature, 241, 171-174.

Wenner, AM.
1967: Honey bees: do they use the distance information contained in
their dance maneuver?, Science, 155, 847-849.
1989: Concept-centered versus organism-centered biology, American
Zoologist, 29, 1177-1197.

Wenner, A.M. and Wells, P.H.
1990: Anatomy of a Controversy: The Question of a "Language" Among Bees, Columbia University Press, New York.