[1993 Wenner, A.M. Science as a process: The question of bee "language." Bios. 64:78-83.]
By ADRIAN M. WENNER
Professor of Natural History, Emeritus,
Department of Biological Sciences
University of California, Santa Barbara,
Santa Barbara, California 93106
Abstract. Scientists normally receive very little formal training in scientific method or in the philosophy, sociology and psychology of science. Consequently, individual scientists tend to become committed to hypotheses as end products rather than as entities that will be replaced. Competing hypotheses espoused by others then may lead to confrontation (controversy), and application of the scientific method becomes a collective and inefficient process. The on-going 25-year controversy about the honey bee “dance language” hypothesis serves as a good example of how scientific progress is a process instead of a series of accomplishments.
Science as Practiced vs. Science as Envisioned
Only rarely are academic biologists interested in the philosophy, sociology and/or psychology of science-most often biologists not only ignore deliberations of those scholars but scoff at their efforts. It is no wonder that these subjects are not graduation requirements for biology majors in most colleges and universities throughout the country. Graduate students in biology fare little better; graduate programs are built upon the premise that the undergraduate curriculum has provided an adequate amount of intellectual breadth. Graduate education largely consists of an apprenticeship in the “objective” search for knowledge.
Many scientists often scoff when they hear the phrase “scientific method.” Instead they rely upon narrow methodology such as ”verification” or ”null hypothesis” approaches in their research. That same attitude is reflected in textbook treatment of “scientific method” (See Excursus SCI in Wenner and Wells 1990). This unfortunate set of circumstances led Theocaris and Psimopoulos to comment (p. 345 in Wenner and Wells 1990):
“The hapless student is inevitably left to his or her own devices to pick up casually and randomly, from here and there, unorganized bits of the scientific method, as well as bits of unscientific methods.”
One can argue that there must be a scientific method if for no other reason than the fact of a remarkable amount of scientific progress these past three centuries. No other facet of our culture has exhibited a similar progression in accumulation of knowledge. Furthermore, application of the scientific method has become more effective through time and has obviously been more efficient in some fields than in others – witness the different rates of progress in the different areas of science.
Despite the fact that individual scientists are not taught the scientific method, scientific progress occurs – collectively and inefficiently – by an unconscious group application of a definable method. That method and its gradual evolution these past few hundred years can be illustrated by a diagram, the details of which can be found in Wenner (1989) and in Wenner and Wells (1990; chapter 3).
The numbers attached to the names of people in Figure 1 represent a sequence through time from Aristotle (~330 B.C.) to Atkinson (in 1985). A short summary of the collective scientific process follows (generally clockwise around the diagram).
Atkinson provided a thought that completed the picture for us. An individualistic (creative) scientist generates some new perception about nature when in the exploration mode (lower right corner), often after having observed an anomaly in a set of results during the progress of “normal” science. Recognition of that anomaly permits a transition from an attitude of “Realism” to “Relativism” – an interpretation considered ”fact ” before is recognized as perhaps no longer adequate to explain known facts. That scientist then “creates an image” (in Atkinson’s words) – forms an alternative explanation – and attempts to convert others to the same point of view.
If others can be convinced, that scientist (perhaps others as well) may try to verify the find (lower left corner of the diagram) – a very necessary part of the process. For many, adequate verification may suffice to establish “truth,” but others might insist on a test of the hypothesis and move either to the upper left corner of the diagram (falsification approach) or to the upper right corner (inference approach) and test the hypothesis.
In some fields of science, a philosophy termed “logical positivism” or “logical empiricism” (verification and/or falsification approaches) dominated thinking about how science should be done – textbook authors and workers in some fields are still locked into those approaches. Thomas Kuhn sent shock waves through the philosophical community in 1960 by suggesting that science was not practiced solely by verification and/or falsification. It was he who brought out most vividly the importance of anomalies. Attention to an anomaly can propel one out of the comfortable notion that we have a good perception of reality and back into the exploration mode.
As far back as 1890, Thomas Chrowder Chamberlin recognized the weakness of an over reliance on verification (“ruling theory”) and/or falsification (”working hypothesis”] methods. He advocated instead an application of what he termed ”The Method of Multiple Working Hypotheses,” a process by which scientists continually pit hypotheses against one another and attempt to falsify all of them during experimentation. After results are in, new alternative hypotheses are generated that might explain known facts. Experiments are then designed to eliminate as many hypotheses as possible. In that manner scientists continue recycling ideas with no attempt to reach ultimate ”truth.”
Unfortunately, Chamberlin’s thoughts disappeared from view until the early 1960s when John Platt began to wonder why progress was especially rapid in fields such as nuclear physics, genetics and moulecular biology. He concluded that those fast moving fields employed a ”strong inference” approach – the equivalent of moving up and down on the right half of the diagram – repeatedly exploring and inferring without irrevocable commitment to given hypotheses (paradigm hold). Platt persuaded Science to re-publish Chamberlin’s 1890 paper, a paper that is having ever more impact with time.
The sum total of ideas contained in the theme diagram constitutes the collective scientific process for scientists. Although fast moving fields of science routinely include the strong inference approach in their apprenticeship programs, scientists usually conduct their research under quite different philosophies from one another – each has a unique background, certain values and set notions of procedure. Only rarely does one find a scientist who can move from one approach to another with ease (as Pasteur did). Duclaux, biographer of Pasteur, recognized the root problem a hundred years ago: “However broad-minded one may be, he is always to some extent the slave of his education and of his past” (p. 31 in Wenner and Wells 1990).
Controversy as the “Fuel of Progress” in Science
One may try to “prove” (verify) a notion as an irrefutable hypothesis and refuse to face the simple question, ”Is it conceivable that this hypothesis is not true?” Alternatively, we may go to a great amount of trouble phrasing a null hypothesis and going through the motions of trying to prove that null hypothesis true in an effort to defeat the hypothesis we have labored so long to verify. Unfortunately, deep down inside we may really like the hypothesis about to be tested (a paradigm hold) and not want to see it negated by a critical test.
Another scientist may view a given set of results from another perspective and controversy may erupt. In that sequence, one person has a creative insight during exploration, generates a hypothesis, and provides some verification for that hypothesis. If the hypothesis is “attractive,” others may accept it. Given enough time, a subset of the scientific community may treat that hypothesis (rather than the data) us ”fact” and therefore ”not open to question.” When anomalies continue to arise, either those anomalies are dismissed as unimportant, the hypothesis is altered (and weakened), accessory hypotheses are proposed that can explain away discrepancies, and/or supportive evidence gets preferential treatment. All too often, new experiments that provide additional verification may be considered ”definitive” or ”elegant” even when the hypothesis in question has failed tests repeatedly. Science then slows to a near halt.
Science moves ahead again only when someone recognizes that an anomaly simply cannot be dismissed, explores those implications, and generates a new hypothesis that can explain all available evidence better. However, the scientific community by then may be too entrenched to recognize that the old hypothesis is no longer adequate. The controversy that erupts eventually leads to totally new types of research after the usual rush to confirm (re-verify) the old hypothesis has subsided. In fast moving fields of research, the time span can be measured in months or years; in slow moving fields, decades may pass.
The Honey Bee Dance Language Controversy
The question of “language” among bees dates back centuries with odor-search and ”language” advocates firmly convinced about their “final” answer each time. Karl von Frisch proposed his waggle dance ”language” hypothesis in 1946. Earlier, though, in a 1937 article (in English) he strongly advocated an odor-search hypothesis (Science Progress, reprinted a year later in the Smithsonian Institution’s 1938 Annual Report). That 1937 article disappeared from all citation lists after that time. (For example, he did not cite that contribution in his classic 1946 paper nor in his massive 1967 review volume; neither was it listed in the official list of publications at the time of his death.)
The temporary loss of that early von Frisch article and his accumulated knowledge to that time was very unfortunate in that it interfered with science as process during the later bee language controversy. Under the strong inference approach, both odor-search and “language” hypotheses should have been kept as viable options during experimentation (strong inference approach). Just how the entire episode relates to our theme diagram is worth elaborating upon here because of its implications for science teaching. (For a more complete account, See Wenner and Wells, 1990.)
In the more than two decades before his 1937 article, von Frisch had patiently studied how naive bees are recruited to food sources visited by experienced bees (foragers). In 1937 he wrote with confidence:
“. . . [recruited] bees first seek in the neighborhood, and then go farther away, and finally search the whole flying district. . . . I succeeded with all kinds of flowers with the exception of flowers without any scent . . . the scent of the flower is taken up by [the forager's] body-surface and hairs, and when it dances after homing [then] the interested bees . . . perceive the specific scent on its body and know what kind of scent must be sought to find the good feeding-place announced by the dancing bee. That this view is correct can be proved easily . . .”
Later, von Frisch found that he could tell from elements in the dance maneuver of successful bees just where they had been in the field. On the basis of new experimental results (exploration mode), he concluded that not only we but potential recruits could “read” the dance maneuver and fly straight out to the same site. With that verification, he published his classic 1946 paper on the “dance language” of honey bees. However, he (and many others of us) failed to realize that results in one section of his paper contradicted interpretation in another section and vice versa.
In later experiments, “improved” experimental designs actually funneled searching bees into the center of an array (See Excursus PN in Wenner and Wells 1990), leading von Frisch and others to the notion that searching recruits “used” dance information with great precision. Others using his experimental design could easily get the same results (verification approach). This exotic hypothesis then swept the world, even without ever having been tested. (Exotic hypotheses are usually easier to “sell.”)
Fifteen years after generation of the language hypothesis, I noticed (awareness of anomaly) that information in the dance maneuver was not sufficiently accurate to account for the “precision” of recruit bee performance. Rather, random concentration of searchers as a consequence of experimental design (as an odor field) could explain his results. Later it became apparent (more anomaly) that some evidence in support of “language” use (e.g. in von Frisch’s classic 1950 Cornell University Press book) had actually been obtained by experimenting with bees that had been re-recruited to familiar sources-bees that already knew the location of the food source (recruitment by conditioned response, not “language”).
Eventually, my colleagues and I felt compelled to test, for the first time, the dance language hypothesis. We used a double controlled experimental design (falsification approach) and a “crucial” experimental design (strong inference approach). The results obtained clearly refuted the dance language hypothesis. However, that evidence had little influence on the scientific community – the hypothesis had become too deeply entrenched as “fact” (the “ruling theory” of Chamberlin) and was “not open to question.” Instead of attending to the results of our test, subsequent verification experiments by others provided more supportive evidence for the language hypothesis.
I left bee research in the early 1970s but recently resumed research in that area. There is now a willing audience for our newly formalized odor-search hypothesis (based in part on von Frisch’s 1937 contribution), particularly among those engaged in applied research. The language hypothesis has been of essentially no use to them in its 45 years of existence, whereas the odor-search hypothesis has immediate practical value – especially in countering the threat of Africanized bees. There is also a possibility that the odor-search hypothesis (e.g. chapter 5 and Excursus OS in Wenner and Wells 1990) will be of immense practical value for those interested in honey production and pollination of crops.
It is ironic that 56 years have now passed since von Frisch had been so adamant that searching bees relied solely on odor cues. We only hope that this episode can be used in teaching at all levels to illustrate how science is a process rather than a series of accomplishments (as it is so often taught).
Chamberlin, T. C. The method of multiple working hypotheses. Science 148:754-759;  1965.
Duclaux, E. Pasteur: the history of a mind. Trans. E. F. Smith and F. Hedges. Philadelphia: W. B. Saunders;  1920.
Wenner. A. M. Concept-centered vs. organism-centered research. American Zoologist 29:1177-1197; 1989.
Wenner. A. M.; Wells, P. H. Anatomy of a controversy: the question of a “language” among bees. New York: Columbia University Press; 1990. 400 pp.