Fish are far more intelligent than most people assume. They can learn new information after a single exposure, remember it for months or even years, recognize individual companions, use tools, cooperate across species, and in some tasks outperform non-human primates. The old idea that fish operate on pure instinct with a three-second memory is thoroughly debunked by decades of research spanning hundreds of species.
What makes fish cognition especially remarkable is that their brains look nothing like ours. Instead of the layered outer brain structure that mammals use for higher thinking, fish evolved clusters of neurons that appear to perform equivalent functions through simpler but effective circuits. The hardware is different, but the cognitive output is surprisingly comparable.
The Three-Second Memory Is a Myth
Goldfish and other fish species have memories that last months, not seconds. Fish can learn novel information extremely quickly, often after a single trial, and retain it for extended periods. In controlled studies, various species have demonstrated memory lasting months and, in some cases, years for learned tasks and locations.
Tide pool gobies offer a vivid example of rapid spatial learning. These small fish memorize the layout of their rocky habitat so they can leap between pools at low tide without landing on dry rock. In lab studies simulating tidal conditions, 100% of rock pool gobies learned to locate the deepest, safest pool within just ten tidal cycles. They accomplished this over five days, building a mental map of their surroundings that guided precise navigation under pressure.
Tool Use and Problem Solving
Several fish species use tools, which was once considered a hallmark of primate intelligence. The tuskfish is the most well-documented example. When it catches a clam or snail, it carries the prey to a hard surface and smashes it open by striking it repeatedly. These fish use rocks, dead coral, shells, and even concrete moorings as anvils. On average, a tuskfish strikes its prey six times per visit and spends about 84 seconds working at an anvil. Nearly half the time, the fish will try multiple anvils during a single feeding event, suggesting a deliberate process of selecting the best tool for the job.
Cooperative Hunting Across Species
One of the most striking examples of fish intelligence comes from the Red Sea, where groupers actively recruit moray eels as hunting partners. A grouper will swim to a resting moray and shake its head rapidly, three to six times per second, just centimeters from the moray’s face. This signal invites the moray to join a hunt. The moray slithers into crevices where prey hide, while the grouper patrols the open water above. Prey that flee the moray swim into the grouper’s path, and prey that stay hidden are caught by the moray.
The results are dramatic. Groupers hunting with a moray catch nearly five times as many prey per hour compared to hunting alone. Morays benefit too, catching prey at a rate of 0.36 per hour during cooperative hunts, while they were never observed catching anything when solitary. What makes this especially impressive is the grouper’s flexibility. If a hunt fails and prey escape into a crevice, the grouper will hover directly above the hiding spot and perform a distinct headstand-and-shake signal that attracts other predators to the exact location. This is referential signaling, pointing to a specific place to communicate information, a behavior previously thought to require a much larger brain.
Recognizing Human Faces
Archerfish, known for spitting jets of water to knock insects off overhanging branches, can learn to distinguish between individual human faces. In one study, archerfish were trained to spit at a specific person’s face displayed on a screen. They then had to pick that face out of a lineup of 44 different faces, and they did so reliably. Even when researchers removed color, standardized brightness, and made all the head shapes identical (eliminating every shortcut except the facial features themselves), the fish still succeeded with 18 different faces. Their accuracy improved with practice, reaching over 80% correct in later testing blocks. This is notable because fish have no evolutionary reason to tell human faces apart, meaning they’re applying a general visual discrimination ability rather than relying on a specialized brain circuit.
Counting and Ordinal Numbers
Fish don’t just perceive “more” versus “less.” Guppies can learn ordinal positions, understanding concepts like “third in a row.” When researchers set up a line of 12 identical feeders with food hidden in only one position, guppies learned to go to the correct spot. They could reliably find the 3rd position and, with more difficulty, the 5th. Their accuracy dropped off around five, suggesting that’s close to their upper limit for tracking ordinal information, which is comparable to what many mammals and birds can do. Broader research confirms that fish numerical abilities, including judging group sizes and making quantity-based decisions, are on par with those of many warm-blooded vertebrates.
Self-Recognition in the Mirror
The mirror self-recognition test has long been considered a benchmark for self-awareness. An animal gets a mark placed on its body in a spot it can only see in a mirror. If it uses the mirror to inspect or remove the mark, that’s taken as evidence it understands the reflection is itself. Great apes, elephants, and dolphins have passed this test. So has a small tropical fish called the cleaner wrasse.
In a series of experiments, 14 out of 14 cleaner wrasse scraped their throats against surfaces when a brown mark was placed there, but only when a mirror was present. Without a mirror, they ignored the mark. Critically, blue and green marks (colors that don’t resemble natural parasites on their skin) did not trigger the scraping behavior, suggesting the fish weren’t just responding reflexively. They were looking in the mirror, seeing something that looked like it shouldn’t be there, and taking action to remove it. When researchers injected the mark deeper beneath the skin so it was physically irritating, the fish scraped even without a mirror, confirming they could distinguish between seeing the mark and feeling it.
Cultural Knowledge Passed Between Generations
Some fish pass learned behaviors from older individuals to younger ones, creating traditions that persist far longer than any single fish’s lifespan. French grunts in the Virgin Islands follow specific migration routes between daytime resting spots on coral heads and nighttime feeding areas in seagrass beds. Individual grunts rarely live past two years, yet the same resting spots and routes have been documented for at least three years, meaning the knowledge outlives the fish that carry it.
Brown surgeonfish take this further. Not only do they pass along migration routes socially, but seemingly arbitrary movements along the way (specific body pitches and rolls at certain points) also get transmitted to new group members. In bluehead wrasse, researchers tracked 22 breeding locations over 12 years and found that not a single one emerged or disappeared, despite complete generational turnover in the population. Herring and cod similarly learn migration routes by following older cohorts rather than relying purely on instinct. Young cod, for instance, follow experienced adults along a migration “highway” to spawning grounds, learning the route for future trips.
Pain, Feeling, and Awareness
Fish possess the same basic categories of pain receptors found in mammals. Rainbow trout have been shown to have multiple classes of nerve fibers dedicated to detecting harmful stimuli, including fibers that respond to pressure, heat, and chemicals. Zebrafish have at least six types of acid-sensing channels and nine types of heat-sensing channels, many of them functionally similar to those in mammals.
More telling than the receptors is how fish behave when injured. They stop eating. They rub the affected area against surfaces. They rock on the substrate or adopt abnormal postures. They breathe faster. These responses go well beyond simple reflexes because they persist over time, change the fish’s future decisions, and disappear when pain relief is administered. A fish that touches a hot surface doesn’t just flinch and forget. It avoids that area afterward, shifts its priorities, and behaves in ways consistent with an ongoing unpleasant experience rather than a momentary reaction.
How Fish Brains Achieve This
Fish brains are small and structured very differently from mammalian brains. Mammals process higher cognition in the neocortex, a layered sheet of neurons covering the outer brain. Fish have no such structure. Instead, their equivalent region (called the pallium) is organized into dense clusters of neurons rather than layers. Researchers have found that specific neuron types within these clusters closely resemble neurons in specific layers of the mammalian cortex and send similar kinds of connections to similar targets. The emerging view is that fish and mammals share a common blueprint for higher brain circuits, but mammals elaborated theirs into layers while fish packed theirs into compact clusters.
This helps explain why fish can match or even exceed mammals on certain cognitive tasks despite having brains a fraction of the size. Cleaner wrasse, for example, outperform chimpanzees and other non-human primates on a specific decision-making task modeled on biological market theory, where the challenge is to serve a visiting client before a resident one to maximize long-term payoff. The fish solve this faster and more consistently than primates do. A massive collaborative project called ManyFishes recently completed data collection on 444 individual fish across 22 species, representing the largest comparative study of fish cognition to date, with analysis currently underway to map the full landscape of cognitive abilities across the group.