1. An Experiment Where Bees Picked Empty Cards as the Smallest Number
In an experiment, bees were first trained to choose the smaller number of shapes on cards. They received a reward for picking the card with fewer shapes (1 to 6). Then, researchers showed them a white card with no shapes at all. The bees treated the blank card as the smallest number [1]. The study involved researchers from RMIT University in Australia and the CNRS Animal Cognition Research Center in Toulouse, France [1].
According to Japanese commentary on the study, when shown the zero card for the first time, bees spent 60–70% more time on the white card. They were also more accurate when the difference in numbers was large, such as 6 versus 0 [2]. Note that this 60–70% figure refers to the proportion of time spent, not the accuracy rate of their choices. The paper’s abstract says that this ability to order "zero" is comparable to the understanding seen in non-human vertebrates such as African Grey parrots and monkeys, and in preschool children [1]. However, this means bees can place zero as the smallest number, not that they perform calculations with zero.
Before believing bees truly understand numbers, we must check for other clues. Animals often use cues humans do not expect.
2. Clever Hans the Horse: Why His Math Answers Came from Reading Human Faces
In the early 20th century, a horse named "Clever Hans" became famous for tapping his hoof on the ground to answer math questions. Observers believed the horse was calculating [3]. However, psychologist Pfungst investigated and found that the horse was watching the posture and facial expressions of the person asking the question. As the horse’s taps got closer to the correct answer, the person’s posture and expression changed. The horse used these changes as a signal to stop tapping [3].
This episode shows why animal number studies need control experiments that rule out unconscious human cues and clues other than number, such as size or distance. In the bee landmark-counting study discussed next, researchers changed the spacing and types of markers and tested objects the bees had never seen [4]. So, how do animals actually count when these tricks are removed?
3. How Honeybees Count Up to Four Landmarks on the Way to Food
In 2008, Dacke and Srinivasan reported that honeybees can count up to four objects they encounter in sequence during flight. Bees used the number of landmarks they passed to remember locations on their way to food [4]. The results did not change even when the spacing or type of landmarks was altered. Because bees could count unfamiliar objects, researchers described this as a counting method that does not rely on object type [4]. This type of task also works in experiments where distance is changed to prevent answering based on travel distance alone [5].
The question of whether bees can count landmarks was addressed as early as 1995 by Chittka and Geiger in a paper titled "Can honeybees count landmarks?" [5]. The specific numerical details of that older paper are not covered here. The limit of four is specific to the task of passing landmarks in sequence; it is not a general limit on bee number sense. In this landmark task, bees appeared able to count up to four. Is the ability to distinguish small numbers innate?
4. How Newborn Guppies and Chicks Notice Differences in Number
Fish have a natural tendency to join larger groups of their own kind. Using this trait, Bisazza and colleagues tested guppies one day after hatching. The fish were asked to choose between a group of 2 and a group of 3. The guppies could distinguish these numbers. However, they could not distinguish between 4 and 8, or 4 and 12. By 40 days old, they could distinguish 4 from 8 [6]. The study concluded that the ability to distinguish small numbers is innate, while the ability to distinguish larger numbers appears later through growth and social experience [6].
This experiment used the fish’s group-joining instinct. It does not prove fish "count" like humans. Another review suggests that fish number abilities are very similar to other vertebrates. It posits that separate mechanisms for precise small-number distinction and ratio-based large-number distinction are shared with monkeys and human infants [7].
Chicks show a different setup. Chicks trained to associate number differences with food could distinguish 5 vs. 10 and 10 vs. 20 at three days old, even when non-numerical cues like size were controlled [8]. Because they were trained with food, we cannot say that chicks are simply born able to count. Chicks a few days old could also distinguish 2 vs. 3 and 4 vs. 8, suggesting the same mechanism might be used for both small and large numbers [9]. The ability to distinguish number differences seems to be present in animals earlier than expected. But do any animals "produce" numbers themselves, like humans counting aloud?
5. How Crows Plan Their Calls and Use Number-Responsive Brain Cells
A 2024 study in Science reported experiments with three carrion crows by Nieder’s lab at the University of Tübingen in Germany. The crows produced 1 to 4 calls in response to colored numbers (visual cues) or specific sounds (auditory cues), finishing by pecking [10]. Researchers noted that the crows seemed to plan the number of calls before starting. They suggested this is similar to how young children count before understanding number words, possibly an evolutionary precursor to true counting [11]. Reaction time increased as the number of calls increased, but the time did not change whether the cue was visual or auditory [11]. This result comes from only three birds, so it cannot be applied to all crows, and the study does not show that crows understand number words.
Number-responsive cells have also been found in crow brains. In 2015, Ditz and Nieder reported neurons that responded only to specific quantities, regardless of dot size or shape. These neurons functioned remarkably similarly to those found in the association areas of primate brains [12]. Number-related behaviors have been reported in animals with very different kinds of brains. However, the tasks and conditions differ, so it is hard to bundle them together as evidence of one "counting" ability.
6. A Tapping Activity: Can You Match Sounds Without Counting?
Ask someone else to tap a table a few times, then tap back the same number of times. Do not say "one, two, three" out loud or count in your head. Does it get harder as the number increases? This is not a reproduction of the crow experiment; it is only a way to feel what it is like to handle a number of taps without counting. It does not show that the same mechanism works in crows.
Another observation is to watch fish in a tank. See if they move toward the larger group. The guppy experiment used this instinct [6]. However, results may vary by fish species and situation, so observe without assuming a specific outcome.
In the table-tapping game, what clues could allow you to match the number without counting? You might use rhythm or the interval between taps. How could you remove those clues?