Researchers reconstructed a Jurassic forest soundscape from 150-million-year-old fossilized katydid wings, measuring resonant frequencies to model the acoustic properties those insects would have produced.
The result is a soundscape of chirps and calls that might never have happened — a reconstruction built on the assumption that wing structure alone determines acoustic output. Measure the wings, understand the physics of their vibration, and you know what came out. It is plausible enough to publish.
But katydids do not sing like tuning forks. They modulate their calls through muscular control, temperature variation, social context. Learned behavioral repertoires developed across their lifespan — none of which fossilizes. A female adjusts her preference by listening to males, and males shift calling patterns based on competitive pressure from rivals. The same wing morphology under different nervous systems and social conditions produces acoustically different results.
We have built an entire framework for studying sensory evolution on the assumption that we can reverse-engineer behavior from anatomy. We cannot. We can model what the physics permits. We cannot know what the animal actually chose to do with those permissions. You cannot infer Beethoven's Ninth from his skeletal remains, and you cannot measure intent from wing structure.
The real cost is that we have mistaken the boundary of what we can measure for the boundary of what we can know. We have solved a physics problem and called it paleontology. In your own work, whatever field you have chosen, you have likely done this too. Confused the precision of what you can quantify with the truth of what you are trying to understand. You have built something elegant from clean data while the thing that actually mattered stayed invisible just beyond the measurement. The katydid's call was not determined by its wings. Neither is your argument determined by your evidence, though it can feel that way when the evidence is all you have.
You have built something elegant from clean data while the thing that actually mattered stayed invisible just beyond the measurement.
”