Soon after our family came to Santa Barbara in 1960, we became friends with the Culler family, partly because we had children of the same age. A professional liason with Glen began in the early 1960s, arising out of our common interest in communication - his with human-computer interactons and mine with honey bee communication. In particular, I recorded and analyzed sounds produced by bees, including those sounds produced during the recruitment of naive bees to food sources already exploited by hivemates. I also dabbled somewhat in the analysis of human speech sounds, recognizing (as others before me had) that spoken words actually had simple components. Glen was interested in the possibility that we could speak commands to a microphone, have the elements of these sounds automatically analyzed, and "instruct" a computer to do specific tasks. We had these ideas, as we now know, decades before such action was technologically feasible.
A more immediately fruitful contribution from Glen occurred when he pressed me to proceed with a multifactorial analysis when that type of analysis by computers was still in its relative infancy. While trying to gain supportive evidence for the notion of a "language" in honey bees, my co-workers and I had encountered anomalies in the recruitment behavior of bees, anomalies that could not be accommodated by the traditional and by-then fabled hypothesis. That is, Karl von Frisch of Germany had invented an appealing hypothesis to explain how newly recruited bees could (presumably) efficiently locate a food source already visited by experienced forager bees (for which he received the Nobel prize in 1973). The anomalies we had unearthed indicated, instead, that such efficiency simply did not exist and that the original hypothesis was ill-founded.
By running a factor analysis of the results of a complex experimental study step-by-step, with Glen's encouragement, we learned that the simple correlation approach -- as used in studies by researchers in that field at the time -- did not necessarily give one a true picture of how Nature functions.
We published the results of our experiments in 1967 as perhaps the first application of factor analysis in an animal behavior study. Much to our chagrin, biologists were mathematically ill-prepared to recognize the superiority of a factor analysis approach to experimental studies, as compared to the prevailing practice of merely studying one correlation at a time. Nearly three decades have now passed, and a significant number of animal behaviorists and bee researchers has finally begun to abandon the bee language hypothesis, a story that has now become firmly entrenched as "fact" in the public mind.
Once again we have just one more example of how far Glen Culler was before his time!
| Previous Tribute |
Tribute Contents |
Next Tribute |
 Back to Glen Culler Honorary Lecture Home Page |
 Back to College of Engineering |