ADRIAN M. WENNER
PATRICK H. WELLS

Library of Congress Cataloging-in-Publication Data

Wenner, Adrian M.
Anatomy of a controversy:
the question of a 'language" among bees
Adrian M. Wenner, Patrick H. Wells.
p. cm.
Includes bibliographical references.
ISBN 0-231-06552-3
1. Honeybee-Behavior.
2. Animal communication.
I. Wells, Pairick H.
II. Title.
QL568.A6W46 1990
595.79'90459-dc2O
90-30539
CIP

Columbia University Press New York

CONTENTS

Preface
Acknowledgments
1. Science, Controversy, and the Question of a Honey Bee Language
2. Philosophers and Paradigms
3. Relativism and Strong Inference
4. Two Hundred Years of Uncertainty
5. The Odor-Search Paradigm: History and Revision
6. The Dance Language Paradigm: Evolution and Acceptance
7. A Parade of Anomalies: Learning
8. A Parade of Anomalies: Odor
9. TransItion in Approaches: Testing the Dance Language Hypothesis
10. Multiple Inference and "Crucial" Experiments
11. The Social Network
12. Reaffirming the Dance Language Hypothesis: Initial Attempts
13. The Realism School and Interpretation of Behavior
14. The Dance Language Controversy: Conflicts Between Paradigms
Excursuses
References Cited
Index

"We have to understand first how many elements can be brought to bear on a controversy; once this is understood, the other problems will be easier to solve.'
-Bruno Latour 1987:62

Interviewing successful scientists is a technique often used by sociologists and philosophers of science when they wish to ascertain the reasons for scientific achievement. On the other hand, many mistakes are made in projects that are begun with the best of intentions; projects started may fail early in the planning stages due to faulty assumptions. Such attempts are rarely the subjects of discussion. One can be certain in any event that the interview technique is unlikely to reveal many truly embarrassing episodes.

What propels a research project forward? Sometimes the desire to know suffices. At other times adversity plays an important role; the desire to prove oneself correct and others wrong can be quite an incentive. One might even say that controversy fuels important scientific progress. However, a great many such incidents in science go unreported and thereby completely escape the attention of sociologists, psychologists, historians, and philosophers of science.

Under the current anonymous peer review grant system, unfortunately, adversity can lead to the early termination of a project. Another social factor prevails, despite claims to the contrary; the scientific community is uncomfortable with controversy, unless the issue is rather unimportant (see chapter 14). All of this means that discussion of only a very small portion of conflict resolution actually reaches print.

Our volume would not be complete without a recounting of one such experience Wenner had at the Salk Institute during the mid-1960s. It was an incident that led to the first real test of the dance language hypothesis.

The first question one might ask is, "Why should a bee researcher be involved with the Salk Institute?" The occasion appears to have been the intended launching of a major research project spawned by Jacob Bronowski. He had apparently convinced the Salk Institute leadership that they could unravel brain function by conducting research on the "dance language" of bees. A "kickoff" seminar was to be given by Harald Esch of Notre Dame University.

The following account of that experience is necessarily written in the first person singular by Wenner. The accuracy of the account was verified by one of the participants, the renowned honey bee geneticist Harry Laidlaw of the University of California at Davis. When he read the account, he remained silent for a long while and then commented, "Yes, that's how it was." After a brief pause he added, "But are you sure you should publish it?" Wenner's account follows.

THE HARALD ESCH SEMINAR

Jacob Bronowski telephoned in 1965 and invited me to serve as a "discussant" immediately following a talk to be given at the Salk Institute by Harald Esch of Notre Dame University. Esch and I, who had never met before then, had independently recorded sounds made by forager bees (see chapter 6) during their dance maneuver within the hive. Both of us had also appreciated the potential significance of those sounds in terms of the purported "dance language" (see Esch 1961; Wenner 1959, 1962, 1964). We both reasoned at that time that bees maybe used sound rather than the dance configuration during communication.

The invitation was quite puzzling for at least two reasons. First of all, the Salk Institute is not known for its interest in natural history studies. Secondly, the formal use of a "discussant" after scientific seminars is a rather rare event. In that arrangement, one person or a group of people give presentations. Then a designated "discussant" provides pro and/or con arguments about the material presented. This procedure seems to be a means by which some notion or other can be "legitimized" in the minds of those present.

Since this invitation came when all of us still worked within the dance language paradigm, it would appear that the occasion was expected to proceed smoothly. Esch would give his presentation, and I would discuss the material and give a "stamp of approval" on the idea that the 'dance language" hypothesis of bees was valid.

By the time that Bronowski extended his invitation, however, my colleagues and I had already succeeded in conditioning honey bees to respond as if to a language upon the presentation of a stimulus (see chapter 7). We had also already recognized the implications of that conditioned-response behavior during the rerecruitment of foragers in nature. That is, we knew that experienced bees did not need to "use" information contained in the dance (see Wenner 1974) as they once again traveled to food sources that they had visited earlier.

Those conditioning experiments had thus revealed to us what von Frisch (1950) and Ribbands (1954) had meant when they reported that experienced bees could be rerecruited to food sources "without the need for a dance." As indicated in chapter 7, at any one time virtually all foraging bees are experienced. That means that foragers would rarely be recruited to food sources by means of the presumed "dance language."

At the time of the invitation from the Salk Institute, we had already perceived that the "dance language" of bees, if it existed (and we were no longer certain of that), would be useful primarily for the recruitment of naive bees. Once those naive recruits learned the location of nectar, however, they could very well spend much of their remaining life visiting that one source. Rerecruitment each day could be by means of conditioned-response behavior.

While inviting me, Bronowski asked if I knew of a good bee geneticist, which I did. Harry Laidlaw of the University of California at Davis had had a long-term association with my beekeeper relatives in Northern California and was perhaps the world's leading bee geneticist at the time. Subsequently he was also invited to the Salk Institute at the time of Esch's seminar.

Another factor entered in. When Bronowski invited me to be a discussant for Esch's talk at the Salk Institute, we had already completed and submitted two manuscripts to journals. Those papers described the results of our experiments on simple conditioning (Wenner and Johnson 1966) and communication by means of conditioned response (Johnson and Wenner 1966). I therefore felt it my professional responsibility to provide the results of these experiments to both Esch and Bronowski before the forthcoming event, even though I was unaware at the time of the reason for Esch's forthcoming talk at the Salk Institute (see below). Bronowski acknowledged receipt of the manuscript but encouraged me to come despite that new development and despite any possible implications of those results to the question of honey bee recruitment.

PRECIPITATION OF A CRISIS

The seminar setting at the Salk Institute totally surprised me. I was expecting a small, relatively informal seminar, as is customary at academic institutions. Instead, perhaps three hundred people were in attendance, as well as television crews and reporters for major news outlets. This was obviously not a routine academic seminar, but no one had informed me of that fact.

Just before Esch began his talk, Bronowski requested that I not mention any of our latest experimental results on honey bee learning (the content of the manuscripts sent earlier) during my "discussion" of Esch's seminar at its conclusion. He said, "That matter can be handled tonight at the dinner."

Bronowski's request caught me unawares, because scientists pride themselves on being open and receptive to new ideas and information. Why then, I wondered, the sudden insistence on even a temporary suppression of new results? Claude Bernard had criticized such action, as follows: "True science suppresses nothing, but goes on searching, and is undisturbed in looking straight at things that it does not yet understand" ([1865] 1957:223).

Esch's talk was, to me, a quite elementary treatment of the presumed evolution of the honey bee "dance language." He outlined a scheme whereby the intricate "recruitment dance" pattern of European bees (which contains information on the distance and direction of food sources; see chapter 6) could have evolved from the behavioral patterns found in related genera of bees living in tropical and subtropical countries.

Esch failed to mention an important point; one could argue equally convincingly that evolution could have proceeded the other way. For example, stingless bees may be considered more "advanced" than European bees, because they have secondarily lost their sting. That is because the sting is a modified ovipositor found in most bees, wasps, and ants.

Taxonomists now recognize that honey bees are not as closely related to stingless bees (Kimsey 1984) as once thought (Michener 1974). Rather, stingless bees are in another (earlier) branch of the family Apidae. Honey bees are thus more closely related to the bumble bees and euglossine bees, which have no dances, than they are to the stingless bees.

During Esch's presentation my thoughts were in turmoil. How could Bronowski, a renowned scholar, mathematician, and philosopher, insist on suppressing results, even if only temporarily? Furthermore, how could I "discuss" Esch's exposition on the "evolution of bee language" in front of three hundred people and reporters, when our experimental results on learning had the potential of relegating the entire "dance language" hypothesis into, at most, a minor facet of honey bee recruitment?

I found myself in a position that was probably incomprehensible to
the rest of the audience. I had already undergone a "paradigm shift" in the Kuhnian sense and had resolved the "crisis" state in my own mind, it had been evident to my colleagues and to me that our experimental results on conditioned responses had matched almost perfectly the earlier experimental results described by von Frisch (1950) as evidence of "dance language" use.

However, von Frisch had not addressed the importance of learning during recruitment and had insisted that the rerecruitment of experienced bees (which we now recognized as conditioned responses) was "proof" that bees had a "language." By contrast, we now realized that two interpretations existed that could both fit that same set of experimental results.

The circumstance we were in was remarkably similar to that encountered by Thomas Kuhn when he was able to perceive that scientists behaved differently from the pattern perceived or advocated by Karl Popper (the "wearing of a new set of spectacles"). How could I then discuss the "evolution of bee language" in front of the audience at the Salk Institute when the very foundation of that hypothesis had been shaken to its roots in our minds? We had undergone a "gestalt switch" and viewed the same results from the vantage point of a new paradigm (see chapter 5 and excursus OS). Once that happens, there is no going back.

Eventually there came the moment of truth. Esch ended his seminar to resounding applause. Then it was my turn to say something to that same audience, an audience that was content with the dance language hypothesis and that was totally unaware of our experimental results and the implications of those results for that hypothesis.

A LACK OF ADEQUATE CONTROLS

As I walked to the front of the room, my dilemma became resolved. For the very first time, I realized that von Frisch's original experiments had lacked adequate controls against forager flight paths and extraneous odor cues. I had "created the image" at that instant, on the way to the front of the room, in the sense meant by Atkinson (1985) and Wenner (1989) (see also figure 3.1).

The idea that von Frisch's original experiments lacked adequate controls was conceded later even by proponents of the dance language hypothesis (e.g., Seymour Benzer in 1966; in a personal communication; Gould 1976). As Gould wrote at that time: "Throughout the dance-language controversy, Wenner has made perceptive and valuable contributions. Von Frisch's controls do not exclude the possibility of olfactory recruitment alone" (1976:241).

However, that open and tolerant attitude did not prevail at the Salk Institute that day. Nevertheless, I had to address the issue as I saw it with my "new spectacles," as Kuhn phrased it. If a "dance language" had never really been established as 'fact," there was little need to address, at that time, the points made by Esch in his seminar. Any discussion of evolution of "the dance language" (in the teleological sense) would be pointless.

Instead, I strode to the blackboard and drew sketches of the experimental design and the results of von Frisch's classic "step" and "fan" experiments. I then highlighted the missing controls and described how those same results fit an odor-search model (e.g., Wenner 1971a, 1974; see also our chapter 5 and excursus OS). The consequence was that I ended up questioning the entire honey bee dance language hypothesis.

The audience reaction immediately turned from what might best be termed one of euphoria to one of intense hostility. It was certainly not the reaction one would have expected from an audience of scientists. The reaction of Theodore Bullock, an eminent physiologist, was fairly typical; he shouted: "What's the matter, don't you believe anything unless you have done it yourself?"

Shortly thereafter the seminar ended. As far as I know, nothing appeared in either the newspapers or on television, despite the presence of all the media and the extensive film footage taken.

THE DINNER CONFERENCE

Bronowski had earlier informed me that a small group would have dinner together and exchange views on honey bee dance language research. That dinner party included several luminaries, including Jonas Salk, Francis Crick, Jacques Monod, Jacob Bronowski, and Theodore Bullock. Harald Esch, Irving Bengelsdorf (the Los Angeles Times science editor), Harry Laidlaw, and a few others were also present. After the dinner itself, the dishes were cleared away and a rather intensive discussion of honey bee communication began.

Eventually (by now unfettered by Bronowski's request that our new experimental results not be presented), I began to describe our conditioned-response experiments. I presented the results of those experiments and indicated the significance of the experimental results in terms of their importance to recruitment efficiency in honey bees.

At one point I said, "One must consider the ecology of the whole system, not just whatever behavior may occur between two individual bees." At that point Bullock came forth with another outburst, "Ecologists be damned, there's not a scientist among the lot of them!"

The discussion became quite heated at that point, and many others joined in. Bengelsdorf, who had remained quiet until that point, finally said, "Wenner's correct; von Frisch's experiments were not well enough controlled." (Bengelsdorf had a doctorate in chemistry.) Shortly after that, Francis Crick leaned over to Jonas Salk and said quietly (I was close enough to hear), "Perhaps we had better not go ahead." The dinner engagement was over shortly thereafter.

HARRY LAIDLAW'S INPUT

After returning to the hotel, I could not relax. Clearly something had been transpiring about which I had not been informed. The last comment directed to Salk by Crick indicated that something big had been in the offing. Since Laidlaw had been a longtime friend, I felt free to telephone him in his room and ask if he knew what had been going on during dinner. He replied, "Yes, didn't they tell you?"

He then came to my room and told me all he knew of what had been planned. As he understood it, the Salk Institute, under Bronowski's prompting, was interested primarily in the functioning of the brain. The "instinctual signaling system" of bees appeared to be an appropriate material for the investigation of brain function. After all, that would be true if, as von Frisch had said: The astoundingly precise adherence to the direction indicated, regarded from the viewpoint of sensory physiology and psychology, stands as a great accomplishment in reception and evaluation of information" (1967a:231).

Apparently, Bronowski's brainchild" was to quick-freeze a foraging bee while it was engaged in its dance maneuver; then neural pathways could be traced. Other races of bees could be used as well, and hybrids of those various races could be bred. More freezing and slicing of brains could then be done at the time of communication, and brain function could be elucidated eventually.

What was the role of Esch in all of this activity? As we later surmised, it appears that he was to have performed the field experiments with the bees during their "bee language" dance. The seminar was apparently what is known as a "recruitment" seminar. Apparently my role as "discussant" was to "legitimize" that activity and the project. In retrospect, I concluded that it probably had eventually dawned on those present at the dinner conference that the honey bee dance language hypothesis may not have been on as solid a foundation as was stated by proponents of that hypothesis.

THE STIMULUS FOR BETTER EXPERIMENTAL DESIGN

The Salk Institute experience provided an additional strong stimulus for our further research on honey bee recruitment. Better experimental designs were obviously needed for investigating honey bee recruitment to food sources. We then gradually developed the more rigorous double-control and "crucial" experimental designs (see chapters 9 and 10). (It is also noteworthy that Gould and co-workers later reverted to single-control experimental designs in their "verification" approach; see chapter 13.)

Esch's seminar at the Salk Institute occurred in March 1966; that very summer we set out to repeat von Frisch's original experiments. Our eyes had been opened wider by now (we suddenly had new "spectacles"). We moved back to the "exploration" approach (Atkinson 1985; see also our figure 3.1) and away from our former "verification" approach of attempting to "prove" that bees used sound signals as part of their "dance language."

We already knew that we could obtain at will results similar to those obtained by von Frisch. We also knew that several controls were missing from von Frisch's experimental designs. The question then became, "Were any essential controls missing from von Frisch's experiments?"

That summer we repeated von Frisch's experiments many times and with many variations. During that process we realized that his single-control experimental design did not exclude the possibility that searching recruit bees could exploit odors and the flight paths of other bees during their search. While watching newly recruited bees approach our feeding stations, we could clearly see that they always approached from far downwind (the use of binoculars helped). They obviously were not flying directly out from the hive.

Through a trial-and-error process, we slowly came to the realization that we could conduct a double-control experiment (a design rarely used in field behavior experiments). That realization was facilitated by the fact that we just happened to have another hive in the area with its different color of bees. The experiments described in chapter 9 were the outcome of all of those deliberations.

The trauma resulting from the Salk Institute encounter cannot be described adequately; it is something that has to be experienced to be believed (but not something one would wish on others). However, sometimes it is an encounter of that sort that enables one to appreciate fully some of the human elements involved in the conduct of science. The Salk Institute affair may account for the surprisingly confrontational attitude we encountered at the Twenty-First International Apicultural Congress a few months later and at subsequent national meetings.

The Salk Institute episode also apparently precipitated events that led to experiments conducted by James Gould and co-workers (see chapter 11).