The Headline That Changed Its Mind
The honeybee did not cheat. That is the real headline behind a reanalysis by Mirko Zanon, a neuroscientist at the University of Trento in Italy, and Scarlett Howard, a zoologist at Monash University in Australia, published in Proceedings of the Royal Society B: Biological Sciences. Their target was a 2019 study claiming that honeybees (Apis mellifera) could match invented symbols to specific quantities, a result a 2020 critique dismissed as a trick of the eye rather than a feat of the mind. Zanon and Howard went back to the original test cards and checked something nobody had checked before: how a bee actually sees them. The accusation did not survive the check. The bees come out looking sharper than the critics gave them credit for.
What the Bees Actually Did
The experiment under review dates to 2019. Researchers trained free-flying honeybee foragers to associate an invented symbol with a specific number of shapes printed on a card. During training, the bees learned the association quickly and reached high accuracy. On brand-new test cards they had never seen before, accuracy held at 60 to 65 percent. Chance alone would have produced 50 percent, so that figure sat well above guessing. Nobody ever disputed it. The argument was always about what was producing it.
Seeing Like a Bee, Not Like a Grad Student
A honeybee’s compound eye resolves far less fine detail than a human eye does, especially at the distances used in these experiments. Zanon and Howard built a mathematical model of honeybee spatial acuity, the fineness of detail an eye can actually pick out, and ran every original test card through it. Instead of judging the cards the way a researcher glancing at a printout would, they produced a version of each card the way a foraging bee’s visual system would actually register it. Any shortcut the bees might have used had to survive that translation. As Howard put it, “we must put the animal’s perspective first when assessing their cognition or we may under or overestimate their abilities.”
The Accusation
A 2020 critique offered a deflating alternative. Maybe the bees were not responding to quantity at all. Maybe they were simply picking whichever card looked visually busier, treating clutter as a stand-in for “more,” since cards with more objects in the training set often did look more crowded to a human eye. The paper argued that the original 2019 claim, that bees could acquire something like a symbolic sign for numerosity, a basic sense of quantity, had never been properly tested against this simpler explanation. It was a serious objection. It swapped a surprising cognitive feat for a boring visual reflex, and for a while, it held.
The Shortcut That Wasn’t There
When Zanon and Howard reran the busyness analysis, not on the raw images but on the bee-vision versions, the correlation between visual clutter and object count mostly disappeared. Cards with more shapes did not reliably look busier once filtered through real bee acuity. Some of the cards that looked most cluttered to a human eye turned out to be some of the plainest to a bee’s. The shortcut the 2020 critique proposed only works if you assume a bee reads a printed card the way a person does. It does not. Take that assumption away, and picking the busy card stops being a strategy a bee could even run.
The Second Why
None of this matters to a bee because of a laboratory card. It matters because foraging is a running tally. A honeybee has to judge how many flowers are worth working in a patch, how many trips it has already made, how many landmarks it has passed on the way back to a reliable food source. Research on bee navigation has already shown that bees count landmarks in sequence to judge how far they have traveled. A rough sense of more, less, and none, what researchers call numerosity rather than counting in the human sense, is cheap, general-purpose currency for an animal whose entire economy runs on whether a patch is worth the trip. The lab card is not a party trick. It is a stripped-down version of a problem the bee already solves every time it leaves the hive.
What Nobody Can Say Yet
The reanalysis worked entirely from the original 2019 images and a vision model. No new bees flew through a corrected version of the experiment. So this does not prove the original bees were tracking number instead of clutter back in that room; it proves the clutter explanation no longer has visual evidence behind it. It also says nothing about how high a honeybee’s sense of quantity can climb, or whether the same process scales past the small numbers used on these cards. Howard has pointed out that a human brain runs on roughly 86 billion neurons while a honeybee’s holds fewer than a million, and whether that much smaller brain is running a tiny version of what ours does, or something built on entirely different lines, is still unresolved. The alibi for cheating is gone. Nobody has yet run the live trial that would close the case for good.