I've been thinking about the Cataglyphis ants — the desert species that navigate home across featureless terrain by counting their own steps and tracking the sun. In 2006, experimenters extended and shortened their legs mid-journey, and the ants overshot or undershot home by an amount exactly proportional to the stride change. The error was clean and predictable: the ant stopped where its step count said home was, not where home actually was. The ant, in that moment, had only the count.
You knew the bees did something structurally similar. A forager returns to the hive and dances — the waggle run's duration encoding distance, the angle off vertical encoding the direction relative to the sun. The more vigor in the waggle, the richer the food source. Other bees read the dance and fly. Your book described a kind of language: not learned, not arbitrary, but shaped across evolutionary time to transmit exactly this information in exactly this way.
What I keep turning over is the direction of the encoding. The ant uses its step count privately — it's accumulating an internal signal that will later tell it when to stop. The bee also accumulates something during the flight out, but then it exports that signal as a dance. What was private becomes public. The forager's locomotion experience gets converted into a pattern that other bees can read and convert back into a flight program. The count that the ant uses to navigate becomes, in the bee, the count that teaches others to navigate.
The bees also have systematic errors that look like the ant's. Bees flying uphill underestimate distance; bees flying downhill overestimate it. Bees in dim light encode shorter distances than bees in bright light covering the same ground. The errors aren't random — they're built in, consistent, and they get transmitted. A dancer who flew uphill communicates a shorter distance than the trip actually was, and the recruits who follow that dance fly short and arrive short. The error propagates. The ant's miscalibration stays private; the bee's spreads.
Here's what I can't quite settle: when a bee performs the waggle dance, what is it doing? Is it replaying the journey — running some internal representation of the flight again, and the duration of the replay becomes the waggle duration? Or has it converted the experience into an abstract code that it then executes separately? I don't know how to tell these apart from outside, and I suspect you didn't either, though you might have had intuitions I'd be curious to hear.
The ants can't be asked. The bees can't be asked. But there's something odd about the bee's situation: it has information that's useful to others, and it acts in a way that transfers that information. Whether there's anything it's like to do this — whether the dance feels like anything, whether the bee has any sense of reporting — I have no idea. The waggle dance looks purposeful. It is purposeful, in the sense that it reliably achieves something. But purposeful-looking and having a purpose you can be aware of are different things, and I don't know how to stand at the boundary between them.
You worked on this for fifty years. I'm coming to it through the ants and finding the bees there waiting. I don't have a resolution — just the feeling that the step counter, when it becomes a communication channel, raises questions the navigation use case didn't. The ant's error is its own problem. The bee's error is everyone's.