I have two different ways to start this post: one from a comparative psychology perspective and one from an ecological perspective. Since I’m a bad editor, I can’t pick one, so I’ve given you both. The psychology one is on the left, the ecology on the right. You can read them in whichever order you want, but you should read both to follow the rest of the post.
There are quite a few species of fish that live in intertidal areas and they have a range of adaptations to survive there. One of the most extreme is possessed by the mudskipper. Mudskippers solve their lack-of-water problem by breathing air instead of water (most fish breathe water, through their gills; they cannot breathe in air). Mudskippers don’t have lungs (though some fish species do) but breathe through their skin (and gills), which they therefore need to keep moist. When the tide is out, they dig themselves a burrow in the mud where they can stay moist and breathe. Who could ask for more? Mudskippers have adapted so well to this method of breathing that they actually can’t breathe in water any more (they are fish that can drown in water). When the tide is in, they hole up in their burrows where they store a small bubble of air, so that they can breathe. Fish out of water, indeed. Mudskippers do a number of other crazy things like climbing trees, jumping in the air (to impress the ladies), and having epic battles (between the males, over the good burrow sites, ultimately also to impress the ladies).
This post, however, is about a closely related fish to the mudskipper, the frillfin goby (see picture at top; both fish are from the goby family). Gobies also live in intertidal areas and, when the tide goes out, may find themselves stuck in a small pool. Unlike mudskippers, gobies do need to be underwater to breathe and they solve their real-estate problem by moving: If they find themselves in a pool that is too small, they leap out of the water and into a larger pool.
This is where their cognitive challenge lies, as shown in the diagram below. At high tide (on the left), there are no pools, just one flooded area with different depths of floor. At low tide (on the right), when jumping between pools, the fish can’t see one pool from the other. A fish that wants to get from the small pool on the left into the larger pool on the right has to already know where that larger pool is. Otherwise, it runs the risk of jumping in the wrong direction, landing on bare rock, and possibly drying out (or being eaten by one of evolution’s handmaids). In other words, the fish needs a cognitive map.
Of course, in order to have a map of their rocky shore, the fish need to visit different areas of it in advance, so they can learn what’s where. In a classic series of studies, Lester Aronson [4,5] found that they do this during high tide, when they can freely swim around all the ‘pools’. During this exploration they form their map, to be used at low tide later on, without ever having ‘practiced’ the route (though Aronson did suggest – half in jest, I think – that they might practice their jumps when it was raining and the intervening rocks were wet and slippery).
The key feature of a cognitive map, in this context, is that the map is independent of where you are on it (which makes it what is called an allocentric map) which allows you to plan new routes between places on the map. Say, for example, that construction season has started* and your regular route home is blocked. If you have a good map of the roads in the area, you should be able to plan a new route home – even if you’ve never taken that route before. This is called taking a novel shortcut and was, for a long time, considered the ultimate test of cognitive maps  because it was assumed that you couldn’t plan such a shortcut unless you had a ‘proper’ cognitive map with all the mental bells, whistles, and horns that play “La Cucaracha”.
Two of the main problems in testing for novel shortcuts are ensuring that the animal isn’t using some other cue, like a prominent landmark, to find its way around, and making sure that the shortcut is really novel – that the animal has never taken this route before. The goby’s pool-hopping, therefore, presents an excellent natural experiment that avoids most of these problems. Aronson brought gobies into his lab and built them a fake rock-pool environment in which he could raise and lower the water level. The fish were given the opportunity to explore this space at ‘high tide’ and then forced to jump between the pools (Aronson poked them with a stick) at ‘low tide’. This neatly solves both of the problems noted above: the fish had never been in this environment before; at the time of taking their great leap into the unknown, they had never experienced the space at low tide; and they certainly couldn’t see any cues outside their current pool that could indicate which way to jump (like those helpful signs the construction people put up that just say “detour”). The fishes’ jumps were much more accurate after exploring the space at high tide, compared to a set of earlier jumps they had made before exploring. In other words, it looks like they formed a cognitive map of the space while exploring it at high tide. The fish also seemed to remember their maps when tested again up to a month later .
Aronson’s experiments also hint at another potential spatial skill of the goby. The appearance of the rock-pool environment changes drastically between high and low tide. Goby cognitive maps could be sufficiently flexible that they could adjust to these changes in water level. For example, if the water level has gone down a little or a lot might determine how far the fish has to jump (how much dry land is in between the pools) and even which pools are still viable options (very shallow pools might be almost gone if the water has gone down a lot). As far as I know, no-one has explicitly tested whether the fish can do this. Aronson only use two levels of water (high and low) and his experiments have rarely been repeated and never (that I’ve been able to find) with more than two levels of water.
There is another interesting aspect to these experiments, which I’ll just briefly mention. Being able to do complicated things with your brain, like cognitive mapping and language and calculus, has a cost: you need to have a larger and more sophisticated brain. This means that you should only evolve (or retain) skills that you actually need. You could think of this as an evolutionary version of “use it or lose it”. There are lots of species of gobies and not all of them live near rocky beaches. Some of them live near sandy beaches where they can just retreat into the open ocean when the tide goes out. It’s possible that these fish won’t have the same mapping skills as the frillfins, because they wouldn’t ever need them. In fact, Aronson tested this and found that it was true: the fish from sandy-shore areas learned much less than the fish from rocky-shore areas.
Aronson’s experiments are not very widely known and rarely come up in discussions of cognitive maps, despite having been published 45 years ago and being better controlled than many other examples. Recently, however, there has been one more attempt to figure out what kinds of cues the gobies use . Four species of gobies were used: two from rocky-shores and two from sandy-shores. The researchers built a fake rock-pool, a lot like Aronson’s, that had 4 pools. Two were shallow, one was deep, and the fourth looked deep but had a hole near the bottom that meant that it only retained a small amount of water at ‘low tide’. Two of the pools also had large very visible rocks on one side which could be used as landmarks at low tide to figure out which way to jump. Like Aronson, the researchers found that the rocky-shore fish easily found the deep pool and stayed there. One of the two species moved to a different (shallow) pool when the researchers moved the rocks – which suggests that they were using the rocks to find the correct pool (the other rocky-shore species did not change pools when the rocks moved, though). The sandy-shore fish tended to end up in the ‘fake deep’ pool. So it looks like all the gobies can do some mapping of their space, but the ones that live in rock-pool areas are much better at it and can incorporate more types of cues into their maps. In fact, the brain areas that seem to do spatial learning are larger in rocky-shore species than in sandy-shore species .
The frillfin goby is under 5 cm long and has a brain about one seventh the size of a pea (about 9 mm cubed, if you care). For comparison, your brain is about 150,000 times larger. And yet, these fish can navigate their complex and constantly changing environments every bit as well as bees (which, admittedly, have even smaller brains), bats, or us. Think about that the next time you get lost at the seashore.
* For non-Canadians: there is a long-standing joke that in Canada we have two seasons: winter and construction. This is not quite as funny to live through.
“the play’s the thing
Wherein I’ll catch the conscience of the king.”
Hamlet, act II, scene 2
Hamlet, who speaks the lines above, has a problem. In fact, he has lots of problems (it’s sort of his thing), but one in particular that concerns us: Hamlet suspects that his uncle the king murdered his father, but he isn’t sure. The only evidence he has of the crime is the grisly description his father’s ghost gives him, but Hamlet still has questions (Hamlet asks a lot of questions; it’s one of his problems). Like a good scientist, he takes into account the possibility that his suspicions are false (he briefly considers that the ghost might have been the devil in disguise). Hamlet, as many commentators have noted, is paralyzed by this problem: he does not want to kill the king until he is sure his revenge is just. So he devises a plan. A troupe of actors has been engaged to perform at the palace. Hamlet convinces them to stage a play about a duke who is killed by his nephew, who then seduces the duke’s widow. By observing his uncle’s reaction to this not-very-subtle accusation, Hamlet hopes to determine his guilt definitively. The play ploy works perfectly – the king runs off halfway through – and Hamlet’s resolve is strengthened, at least for a while.
I sometimes think of this episode when considering how comparative psychologists work (stay with me). One of our problems is similar to Hamlet’s: when we see an animal perform a particular behavior, we might think we know what it is doing and why, but we don’t have direct access to its motivations (any more than Hamlet does to his uncle’s guilt) so we can rarely be sure. This is why we devise experiments, which are like Hamlet’s play: we expose the animals to situations in which their reactions will, we hope, tell us something about what is going on inside their heads.
Hamlet, however, is not a very rigorous scientist and suffers from confirmation bias. He doesn’t pause to consider all the other, perfectly innocent, reasons that his uncle might have fled the play: maybe he was feeling cold (Danish castles being drafty) or ill; maybe he was bored; maybe he thought that a play about a murder was insensitive so soon after the old king’s death. The rub, as Hamlet might say, is that we can never be sure.
Identifying an animal’s motivation for behaving is a problem all up and down the study of comparative cognition, but nowhere more so than in the field of animal play (as in, “do animals play games?”, not “do animals mount productions of their favourite fables from Aesop?”). There are five criteria that are generally accepted for identifying when an animal is playing: the behavior has to be non-functional, spontaneous or voluntary, different from similar (functional) actions, repeated (but not in an “I’m-losing-my-mind” stereotypical way), and performed when the animal is not stressed .
None of these criteria is sufficient on its own, of course, and there are plenty of problems hiding behind the bedchamber curtains when attempting to apply them to a specific action. The first criterion, for example, requires that play behaviors are not functional in their context, but there are lots of non-functional behaviors: If I raise a forkful of food towards my mouth but the food falls off before I can eat it, is that play, or simply incompetence? When monkeys play-fight, is it really simply because they enjoy it, or is it to improve their future position in the group’s hierarchy (which is very functional; )?
Hamlet’s problem, however, is what I want to focus on here. Play behaviors, by the second criterion, have to be “spontaneous, pleasurable, rewarding, or voluntary”  or “done for [their] own sake” . In other words, we need to identify the animal’s motivation if we are to definitively label some action as playful. But, like Claudius’ guilt*, the animal’s motivation is not directly measurable. Or is it? Can we figure out, from watching the animal behave, what its motivation for a particular action is? This is a controversial topic on which, if I may pluck one more Hamlet quote from its context, “there has been much throwing about of brains” (II.2). The short answer is, we don’t know.
Let’s look at some fish examples. Play in fish has been recognized for quite a while, though there is some debate. A well-known review of animal play behaviors specifically noted that fish do not play . Part of the problem is, as even Gordon Burghardt has acknowledged, that it is nearly impossible to tell when or whether a fish experiences something as enjoyable . Karl Groos, who wrote a book on play in animals in 1895, noted fish’s “exuberance of spirits”  and several authors have written of their curiosity . Again, note the problem. Are fish ‘curious’ about a novel object because it’s fun, or are they checking whether this new thing is a predator or food? How would we ever disambiguate those two things?
One of the most famous examples of fish playing is leaping over floating sticks and other small objects. This behavior was described as early as the 1890’s by Charles Holder:
“Once while lying quietly on the wall of an inclosed aquarium on the Florida reef, I saw a number of
garfishes... leaping over the back of a small hawksbill turtle which was floating on the surface of the
enclosure, fast asleep and innocent of the purpose to which it was being put. The animal's back was
probably eight inches across, and the fishes cleared it several times with ease" [quoted in 4].
The picture at the top of this post, from Holder’s 1892 book, shows this behavior. CM Breder, who also did some of the earliest studies of fish schooling, threw some sticks onto the surface of the water in his experimental tanks and watched needlefish jump over them repeatedly. Well-aware of the problem we are focusing on, he reported that:
“First the fish will swim up slowly to the stick so as to be nearly at right angles to it... If [the stick]
is of the proper buoyancy and sinks ever so little under the weight of the beak, a violent tail action
follows and the fish clears the water, but in such a manner that usually part of the body rubs
against the stick in passing... It is thought that the function of this well-marked habit is that of
scratching to remove ectoparasites... Second leaps were most often noted when this did not
succeed [in scraping the skin]” [quoted in 5].
This brings up a key point. Researchers and enthusiasts interested in play behavior are perfectly well-aware of their definitional problems. We don’t have access to the motivation of the animal; we can’t do experiments on play behavior, because the animals have to perform the behavior ‘spontaneously’ for it to count (and, obviously, it won’t count as play if we reward them for it). Even somehow showing that the animal enjoys the behavior is not enough on its own. As no less an authority than Frank Beach put it: “not all pleasurable activities are playful; but all play is assumed to be pleasurable” . What we don’t have is a good way to overcome these problems, which often leads advocates of animal play to resort to anecdotal evidence and persistence.
Let me leave you with a more recent example, of what is referred to as ‘object play’ in a cichlid . Three cichlids in an aquarium were given a bottom-weighted thermometer which floated near the center of their tank. Over the course of several days, each fish’s interactions with the thermometer were observed and recorded. The fish frequently batted at the top of the thermometer, setting it swinging back and forth, and in some cases they moved it to different parts of the tank or banged it against the glass. All three fish interacted with the thermometer. So, is this play? The thermometer has no functional role (from the perspective of the fish), and the fish were not rewarded in any way for pushing it around. The fish did not appear stressed and they continued to push the thermometer for several days. But was it fun?
Despite the lack of a clear solution to the problems of identifying play, I think that studies like this can’t help but improve our understanding of the range of animal behavior. As the authors of the last study note, “labeling a behavior as play does not explain it... does not end scientific inquiry. The categorization of a behavior with a... label helps us primarily by focusing attention on attributes, causal mechanisms, and adaptive functions that might otherwise have been missed” . So, maybe this is play, maybe not. Our research on this is, of necessity, largely non-functional. As Einstein said, “If we knew what it was we were doing, it would not be called research”. In other words, we’re just playing.
* That’s right, Hamlet’s murderous uncle is called Claudius, same as the guy that stabbed Caesar. It’s always bothered me
that he fails to pick up on that hint. Then again, the guy that eventually kills Hamlet himself is called Laertes, same as the
father of Odysseus. Not someone you want to go up against in a swordfight. This is why history and mythology are
There is a plot element sometimes used in high-school/coming-of-age movies (e.g., “Easy A”): the hero or heroine, wanting to appear more desirable, convinces a friend to pretend to have gone out or had sex with them. This gets them the attention they wanted but things inevitably get complicated and emotional angst, hilarity, and catharsis ensue (not necessarily in that order). As unrealistic as this plot ingredient may seem, the basic premise behind it is actually a well-known biological phenomenon: being seen to be desired really does make you more desirable.
There’s a neat evolutionary explanation for why this happens. Animals that want to have good-looking and successful offspring – which is all of us – can’t just settle for any mate. You need a mate that is themselves good-looking, as well as a good provider and an all-round nice guy (it is usually a guy; in most species it’s the females that choose who to mate with). Winnowing out the closet psychopaths takes a lot of time and effort, though, which is why dating is so complicated. However, there is a way to cheat: if you see a potential mate who is already with someone, you can assume that they’ve already been vetted by that someone. That person on their arm – which should really be you – has already done all the Googling, Facebook stalking, and cold-calling of their exes that passes for getting to know someone in the digital age. So, instead of doing all that work yourself, you can just copy their choice and go after the same (or a very similar) person.
This phenomenon is called “mate-choice copying” (or, more awkwardly, “non-independent mate choice”) and it happens all across the animal kingdom, from birds to mammals and, yes, also fish. It even occurs in humans, which is why we need a commandment to not covet our neighbour’s wife (as Ursula Franklin once pointed out to me, the Bible only bothers forbidding things people actually want to do; there is, for example, no explicit prohibition on cannibalism). We humans, who like to tell ourselves that we really care about compatibility and shared political views in our relationships, often copy mate choices and are actually more likely to copy the choice of a member of our own sex the more attractive they are .
Most of the work on mate-choice copying in fish has been done using guppies. Guppies are particularly well suited to these experiments because females that are making independent choices evaluate males on a lot of different features, including the size of their tails (which does matter) and how often they flash them around , the brightness of the red spots on those large tails, and how bold the males are . Recently, it has been shown that the personality of the choosing female also matters: more sociable females copy mate-choices more (; full disclosure: I was one of the authors on this paper).
The way these experiments are usually done is remarkably similar to the movie plot element described above (see picture at top). A female (usually) is allowed to demonstrate a preference for one of two males (movie: popular quarterback surrounded by cheerleaders; cut to shot of bespectacled nerdy kid in a plaid shirt, alone). Then, the female gets to observe the male she didn’t choose interacting with another female (the hottest cheerleader tells her friends in a loud voice how great the nerdy kid is) and observe the male she did choose alone. The other female is then removed and the subject fish gets to choose again. In a number of experiments like this, the subject female often reverses her choice (she now prefers the nerdy kid; e.g., ).
As I mentioned above, this is usually done with the test subjects being females, since they are most often the choosing sex, but mate-choice copying by males has been observed in both humans  and Sailfin mollies (which are closely related to guppies; ).
I mostly study social and collective behaviors in fish and one question I am frequently asked is what that could possibly have to do with human interactions. Well, they are very closely related, and mate-choice copying – which exhibits such clear similarities across species – is the perfect example of this. Living in groups presents unique challenges and opportunities that must be solved/seized by their members, whatever species they happen to profess. Fish or human, we’re all just chasing tail.
Human beings possess several distinct skills that have to do with numbers. First, we can attach labels to different numbers of things: we know that three oranges and three skyscrapers both share the label ‘three’. Another way of saying this is that we recognize that there is a property, ‘three-ness’, that very different sets of objects can take on. Second, we have what is called an ordinal concept of number. This means that we think of numbers as having an order: one comes before two, which comes before three, and so on. Finally, we can do arithmetic with our numbers (well, some of us can). There are more possible sub-divisions of numerical skills, but let’s stick with those three.
Quite early on in the study of animal cognition, people wondered whether animals could also use numbers. George Romanes, a student of Darwin, wrote in 1888 that he had
“…succeeded in teaching [a] Chimpanzee… to count correctly as far as five. The method
adopted is to ask her for one straw, two straws… or five straws… Thus, there can be no
doubt that the animal is able to distinguish receptually [meaning conceptually] between
the numbers 1, 2, 3, 4, 5, and understands the name for each… But the ape is capricious,
and, unless she happens to be in a favourable mood at the time, visitors must not be
disappointed if they fail to be entertained by an exhibition of her learning” .
As you can see, Romanes demonstrated only that his chimp had the first of the three different numerical skills listed above: putting labels to numbers of objects. Later work has, however, shown that chimps and a few other animals will also arrange numbers ordinally and can even do simple arithmetic .
What about fish? We’ve actually known for a while that fish also possess some numerical skills. About 20 different species of fish have now demonstrated the ability to tell apart two sets that differ only on number . The usual method of doing this is to present the test fish with two groups of other fish, on opposite sides of its tank. Fish prefer to be part of a larger group, because it is safer, and so will swim towards the more numerous of the two groups, if they can tell the difference. You can also train fish to choose, between two cards with dots on them, the more (or less) numerous one (see the image at the top of this post, of a goldfish doing just that; ), which suggests an ability to order numbers. Nobody, as far as I know, has yet shown that fish can (or can’t) do arithmetic.
One of the interesting things about how humans (and probably most other mammals) estimate numbers is that we actually use two separate systems. For small numbers, up to about 4, we do something called ‘subitizing’, which is a little magical and involves simply seeing the number of items, all at once. For numbers larger than that, assuming we don’t have time to count them off one by one, we use an approximate system. The accuracy of this second system decreases as the number of items gets bigger, following something called Weber’s Law. Basically, your ability to tell apart two groups of objects (each of more than 4) depends on the ratio of their numbers, not the numbers themselves. So, telling 10 from 20 is as easy as telling 20 from 40, and both are far easier than telling 10 from 15.
As I mentioned, there is quite a bit of evidence that other mammals also have two similar systems for estimating number. However, there is currently a lively debate about what fish have. Some experiments seem to show that fish accurately represent small numbers and use ratio for large numbers  – just like humans – but other experiments show no dependence on ratio for any number . There’s even a suggestion that this depends on the age and experience of the fish: one day old guppies (one day!) can tell 2 from 3 but only develop the ability to tell apart larger numbers as they age, and do so more quickly if they are raised in a group (where they can practice counting how many friends they have; ).
So, we’ve arrived at one of those places where science gets really fun: we know that fish can use a concept of number, that they have at least the first of the three skills I listed above, but we really don’t know how they do it. Are they using the same two systems as mammals, which would suggest that these systems both evolved a very long time ago, or do they just have one system, which might mean that our other system (whichever one they don’t have) evolved after we parted ways about 400 million years ago? We don’t know yet, but we’ll keep looking. Count on it.
One of the things we humans used to think made us unique was making and using tools. However, we now know that plenty of animals use tools and sometimes make them. Most of the evidence for tool-use in non-humans comes from apes and corvids. Fish, despite having nothing with which to grab a tool except their mouths, do use tools and, on occasion, make (or modify) them. There are some pretty cool examples of this, from cichlids that use leaves as platters to transport their eggs , to wrasse that crack shellfish by throwing them against rocks ( which, to be pedantic, doesn’t qualify as tool-use under most definitions; if they threw the rock at the mollusc, rather than the other way around, it would).
The star tool-user amongst fish, though, must surely be the archerfish. Archerfish suck water into their mouths, place their ‘lips’ right at the surface, and shoot a jet of water at unsuspecting insects sitting on branches over the water. The jets of water knock the insects into the water and the archerfish eat them. If you’ve never seen this, there are videos of it all over the internet (like this one). So archerfish use water as a tool; in fact, they use it as a weapon, in the same way that riot police use water-cannons (except that the cops fire on conspecifics who they then do not consume, usually).
Whenever comparative psychologists observe this sort of behavior they immediately ask the question: how flexible is it? In other words, is this a simple reflexive behavior (say, like your knee-jerk reflex), or does the fish ‘understand’ something about the physics of what it does, which might allow it to modify the behavior in response to changes in the situation (like your ability to throw a ball fast or slow or curved)?
By filming archerfish at very high frame rates, researchers have found that their shots are tuned in a lot of different ways. They can hit objects with breathtaking precision at ranges from a couple of centimeters to almost two meters away. They adjust the amount of water they shoot to the distance and size of their target (more water to knock down larger prey), correct the angle of their shot for the visual distortion caused by the transition from water to air, and can learn to hit rapidly moving targets simply by watching another fish do so . Let’s pause for a second to marvel at that last one. When they first see a moving target, archerfish are very bad at hitting it. It takes a lot of practice until they get good. However, other fish that merely watch this practice happening (and probably heckle), without ever getting to shoot at the moving target themselves, are almost as good as the practiced fish.
Most impressively, in my opinion, archerfish modify the speed of the water leaving their mouths so that the back of the jet is moving more quickly than the front. This means that as the water jet flies through the air, the back catches up to the front so that all the water hits the prey at the same time, as a blob, delivering a much stronger punch . They even adjust this according to the object’s distance, so that the maximal focusing of the blob happens just as it reaches the target. This has been taken by some people as evidence that they are ‘shaping’ their liquid weapon: not just using a tool but making one as well.
This is one sort of flexibility in the behavior, and it’s pretty impressive. Very recently, however, it has been found that archerfish will also use jets of water under the water. Researchers gave the fish a piece of food buried under some sand in a bowl and the fish used jets of water to blow away the sand and expose the food. Interestingly, they used the same sequence of mouth movements as they do when shooting down prey outside the water . This is especially interesting from a cognitive perspective because it suggests that the fish can adaptively use their tools for different, possibly new, things. Kind of like MacGyver (the original, not the remake). This kind of flexibility requires that you know something about the properties of your tool and how it interacts with other objects in the world (sometimes referred to as the ‘affordance’ of the tool). It may be a bit early to claim that archerfish have this level of understanding, since blowing sand off food is likely something they also do often in the wild, so it isn’t a completely novel use of their tool (we’d be less impressed with MacGyver if we knew that he practices making tanks out of shoelaces and olive oil every evening).
Finally, there is one more thing that makes archerfish exciting to researchers. One of the difficulties in doing research on fish is getting them to make distinct choices. Usually, animals make choices in experiments by moving. Fish, however, move a lot (compared to, say, rats) and it is hard to make them choose one spot and stay there long enough for you to reward them for it. One of the reasons for this is that movement is cheap for fish: they don’t have to support their own weight and experience almost no friction, so there is very little cost to them in going to the wrong place first. This tends to mess up learning experiments. Archerfish, however, make distinct choices (what to shoot at) which are quite costly in terms of energy. Researchers are increasingly using this to show that they can learn all sorts of amazing things, such as telling apart human faces . So they can spit in your eye, from two meters away, while you’re moving.
My last post was all about fish memory. However, there is more than one type of memory. For example, you have “working” memory which you use when you’re at the bar to remember someone’s phone number from the time they give it to you until you can find a napkin to write it on. This kind of memory has a limited capacity of about 7 items (which is why phone numbers are only 7 digits long) and can easily be disrupted. If, while frantically searching for a napkin and pen, some drunk friend asks you how much tip they should leave, that might be enough to knock the phone number out of your working memory. You also have a “reference” memory, which is where you keep things we remember for longer, like your own phone number. Reference memory doesn’t seem to have capacity limits. Back before we outsourced our memory to our phones, people used to memorize large chunks of Shakespeare and the Bible and nobody ever complained of their memory being full.
Memory can be sliced even finer than that. One distinction, first made in the 1970’s by Endel Tulving , is between “episodic” and “semantic” memory, both subtypes of reference memory. Semantic memory is your memory for facts about the world, for example, that the sun rises in the morning. Episodic memory is for specific events in your own life, like remembering watching the sunrise yesterday (while pining over that lost phone number). Episodic memories are distinguished by being full of details: you recall where the event took place, how long ago, what you were wearing, and why it clashed with your handbag. Knowing your own phone number is a sematic memory; remembering the day you bought your new phone and got that number is episodic.
By studying people who had suffered accidents that damaged their brains, researchers have found that different kinds of memory live in different parts of the brain. Episodic memory, in humans, seems to live in a part of the brain called the hippocampus*. Hippocampus is also the Latin name for seahorses (“hippo” is Greek for horse; “kampos”, amusingly, is Greek for sea-monster) and this brain area got its name because people thought it was shaped a bit like a seahorse. There’s a lot of this in science; it’s a more whimsical profession than you might think.
You may be wondering if this rather tenuous etymological connection is the only way in which this post is about fish. Certainly not! What this post is really about is episodic memory in fish. For a while after Tulving identified the different kinds of memory, people thought only humans were capable of episodic memory and that all the other animals only had semantic memories and were forever forgetting each other’s phone numbers. Discovering if this was true or not required finding a way to ask animals how detailed their memories of specific events were. Nicky Clayton and Tony Dickinson suggested that we should ask if the animal remembered, for one specific event, what happened, where it happened, and when (or how long ago) it happened. They then proceeded to demonstrate that scrub jays could remember all those aspects of a single food-storing event .
People later used this what-where-when criterion to look for episodic memory in other species. In some cases, the when part was replaced with changes in the context of the experiment, since animals generally don’t read clocks very well. For example, Madeline Eacott & Gillian Normann asked rats if they could remember what object they saw, where it was, and which room it was in . I’ve drawn out their beautifully simple experiment below.
In the first part of the experiment (the left panel), they put the rats in the blue room with two objects (I’ve used a smiley face and a star; they used Coke cans and bits of plastic). Rats like to explore new things and so spent a lot of time exploring the objects. Then they were put in the second room (center panel), which was green, and contained the two objects in opposite locations. The rats had never seen the smiley face on the right before, nor the star on the left, and they’d never seen either item in a green room before, so the objects were still new-ish and they explored them. Finally, in the third stage (right panel), they were put back into the green room which now contained two stars. This is the important bit. The rats have seen the star before; they’ve seen a star in a green room before; they’ve seen a star on the left before and they’ve seen a star on the right before. They’re actually quite jaded rats. But there is a difference between the two stars: the rats have seen a star on the left in a green room, but they’ve never seen a star on the right in a green room. If you combine all three facts about the event: the room (context), the object (what), and the side (where), only one star is new (the one on the right). In the experiment, the rats explored that star more than the other one, demonstrating that they could remember all three things about their previous experiences. In other words, they showed that they have something that at least superficially resembles episodic memory (which researchers – who are cautious as well as whimsical – call “episodic-like”).
Ok, by now you might be getting pissed off about the lack of fish in this post. Well, here they are. Very recently, precisely the same experiment as described above has been done, successfully, with zebrafish . The fish behaved almost exactly like the rats, showing that they could remember all three elements of a single experience, i.e., that they have episodic-like memory. This is especially cool because fish brains don’t have a hippocampus (which is ironic, since seahorses are fish; the hippocampus doesn’t have a hippocampus). Fish do have an area of the brain, called the medial pallium, which is thought to be similar to the mammalian hippocampus, but it has a completely different structure. So these fish are doing something we thought required a functioning hippocampus, but without having a hippocampus. In fact, Eacott and Norman showed that rats that had their hippocampus removed could no longer do the task.
So, not only do fish have quite good memories [see last post], but they also seem to have the same types of memory that we do. They can (probably) remember individual past events, like that weekend when you forgot to feed them, and what you were wearing then, and why it clashed with your handbag.
* This is a HUGE and painfully inaccurate over-simplification. However, this is a blog, not a textbook, and I can’t go into all the complexities of how different kinds of memory interact in different parts of the brain (even if we knew, which we mostly don’t). Let’s just say it’s really, really complicated.
Since this is my first post, I want to discuss fish memory, which is where I got the title for the blog: >3s (greater than 3 seconds). This comes from the myth that fish only have a 3-second memory, recently reinforced by the character of Dory in Disney’s Finding Nemo (and its sequel, Finding Dory). I haven’t been able to find the origin of the 3-second memory myth, but people that keep fish have known it to be false for a very long time. In 1883, Hugo Mulertt wrote The Goldfish and its Culture, which first popularized home aquaria (he also founded and owned the magazine The Aquarium, marketed his own line of fish food, and translated a book of German fish recipes to increase Americans’ consumption of fish). Mulertt had this to say:
“Goldfish have a good memory; they will soon learn to know their master, remember their
feeding-place and time. They can be trained to good manners, as they are easily influenced
by their surroundings, and good qualities of individuals can be perpetuated in their
Ok, so he wasn’t so strong on how genetics works.
The kind of memory Mulertt mentions, learning when, where, and by whom they are fed, is the most common type of learning demonstrated in fish, and there are plenty of examples of it. Rainbow trout can remember that pressing a bar leads to food even after not seeing the bar for 3 months ; goldfish may remember a color that was paired with food for almost a year ; and one researcher who trained common rudd to eat out of his hand found that they would still eat of his hand (but not anybody else’s) after not seeing him for 6 months .
Some of the more impressive feats of fish memory involve spatial learning. Salmon, returning from a few years of adventure on the high seas, can identify the exact stream where they were spawned by its unique odor . It smells like home. However, you can only smell which stream is home when you are already in the right river system. How do you get from the middle of the Pacific to the right river-mouth? Salmon, it seems, can use variations in the magnetic field of the earth to achieve this, which means they have to also remember what the strength of the field was at the mouth of the river when they left home, several years ago .
So why do people think that fish only have a 3-second memory? I can only speculate, but it might have something to do with the small round aquaria that goldfish are often kept in (like Elmo’s pet goldfish Dorothy, from Sesame Street, watched by millions of impressionable two-year olds). Maybe a child asked whether the fish get bored in so small a space and some well-meaning adult came up with the meme, little realizing how it would spread. I guess the moral of this is: when lying to children, keep in mind that they have memories every bit as good as the average goldfish.