The Surprising Complexity of Animal Memories

Chimpanzees, birds, and even rats have shown signs of reviewing their own past to prepare for the future.

A bird holding a peanut
Rebecca Richardson / Getty

Chrysippus, a Greek philosopher from the third century B.C., is said to have recounted how a hunting dog arrived at a spot where three roads met. The dog smelled the two roads by which the quarry had not passed, then without hesitation or any further sniffing set off on the third. According to the philosopher, the dog had drawn a logical conclusion, reasoning that if the quarry had not taken two of the roads, it must have taken the third.

Facing a fork in a maze, mice often hesitate for a few seconds before continuing. Recent studies suggest that in order to decide which way to go, a mouse has to project itself into the future. We know that rodents replay previous action sequences in their hippocampus, so the wavering mouse in the maze probably compares the memory of old routes with imagined future ones. In order to do so, it will have to be able to tell the difference between experienced and projected actions, which requires a primal sense of self.

This post is adapted from De Waal’s new book.

At least this is what the scientists doing these experiments assume. I find this fascinating, because in this thought experiment, we postulate that humans would need a sense of self to make the same decision, which we then take as evidence for a sense of self in another organism. This extrapolation is generally satisfying, but not risk-free, because it hinges on the assumption that there is only one way to solve a problem.

Chrysippus’s dog is a great example of apparent inferential reasoning. Fortunately, we now have tests of inferential reasoning. In the 1990s, the psychologists David and Ann Premack presented their chimpanzee, Sarah, with two boxes, putting an apple in one and a banana in the other. After a few minutes, Sarah would watch one of the experimenters munch on either an apple or a banana. This experimenter then left the room, and Sarah was given a chance to inspect the boxes.

She faced an interesting dilemma, since she had not seen how the experimenter had obtained his fruit. Invariably, however, she would go to the box with the fruit that the experimenter had not eaten. She must have concluded that the experimenter had taken his fruit from the corresponding box and that the second box would still contain its original fruit. Most animals don’t make any such assumptions, the Premacks note; they just see an experimenter consume fruit. Chimpanzees, by contrast, always try to figure out the order of events, looking for logic, filling in the blanks.

In another test, apes were presented with two covered cups after they had learned that only one would be baited with grapes. Both cups were covered and shaken. As expected, the apes preferred the cup in which they could hear the grapes make noise. But then the experimenter shook just the empty cup, which obviously made no sound. The apes picked the other cup.

I once watched another such causal inference unfold at the Burgers’ Zoo when the chimpanzees in the indoor colony watched us carry a crate full of grapefruit, which they find delicious, through a door that went outside onto their island. They seemed interested enough. But when we returned to the building with an empty crate, pandemonium broke out. As soon as they saw that the fruit was gone, 25 apes burst out hooting and hollering in a most festive mood.

Animal consciousness is hard to investigate, but we are getting close by exploring examples of reasoning, such as those given above, that we humans cannot perform unconsciously. We cannot plan a party without consciously thinking about all the things we need; the same must apply when animals plan for the future. The latest neuroscience suggests that consciousness is an adaptive capacity that allows us both to imagine the future and to connect the dots between past events. We are said to have a “workspace” in the brain where we consciously store one event until another one comes along.

Take, for example, taste aversion in rats. It is well known that rats avoid certain toxic foods, even if they don’t become nauseous until hours later. Simple association fails to explain this. Could it be that rats consciously go over the recent past in their minds, thinking back to every food encounter to determine which one was most likely to have made them sick? We certainly do so ourselves after food poisoning and gag at the mere thought of the particular food or restaurant that we believe caused a shock to our digestive system.

The possibility that rats have a mental workspace where they review their own memories is not so farfetched, given the growing evidence that they can “replay” memories of past events in their brain. This kind of memory, known as episodic memory, is different from associative learning, as when a dog learns that by responding to the command “sit,” he will be rewarded with a cookie. To create the association, the trainer has to give the dog the reward right away—an interval of even just a few minutes is not going to be helpful. In contrast to this kind of learning, episodic memory is the capacity to think back to a specific event, sometimes long ago, the way we do when we think of, say, our wedding day. We remember our clothes, the weather, the tears, who danced with whom, and which uncle ended up under the table.

Episodic memory must be at work for the wild chimpanzees that forage among the fruit-bearing trees of Taï National Park, in the Ivory Coast. The Dutch primatologist Karline Janmaat has described how the chimps build their night nests en route to such trees and get up before dawn, something they normally hate to do because of the danger of meeting a leopard. Despite their deep-seated fear, the apes would set out on a long trek to a specific fig tree where they had recently eaten. Their goal was to beat the early fig rush by other animals, from squirrels to hornbills. Remarkably, the chimps would get up earlier for trees far from their nests than for those nearby, arriving at about the same time at both. This suggests calculation of travel time based on distance. All this makes Janmaat believe that Taï chimpanzees actively recall previous experiences in order to plan for a plentiful breakfast.

In a classical experiment, Nicky Clayton at Cambridge University studied western scrub jays (a corvid) to see what they remembered about foods they’d cached. The birds were given different items to hide, some perishable (wax worms), others durable (peanuts). Four hours later, the jays looked for the worms—their favorite food—before they looked for the nuts, but five days later their response was reversed. They didn’t even bother to look for the worms, which by that time would have spoiled and become disgusting. But they did remember the peanut locations after this long time.

We also have studies of metacognition, which refers to the knowing of knowing. Let’s say someone asks me whether I’d rather answer a question about 1970s pop stars or about science-fiction movies. I’d right away pick the first category, because that’s what I’m better at. I know what I know. These kinds of experiments have been conducted with animals (including monkeys, apes, birds, dolphins, and rats), showing that they, too, have different levels of confidence about what they know. They perform some tasks without hesitation, but at other times they can’t make up their mind, exhibiting doubt.

In one early study, a dolphin named Natua was asked to discriminate between a high tone and a low tone. His level of confidence was quite manifest. He swam at different speeds toward the response, depending on how easy or hard it was to tell the tones apart. When they were quite distinct, Natua swam at full speed with a bow wave that threatened to soak the electronic apparatus. The scientists had to cover it with plastic sheets. If the tones were similar, though, Natua slowed down, waggled his head, and wavered. Instead of touching one of the paddles to make his choice, he selected the opt-out paddle (asking for a new trial), which meant he knew that he’d probably flunk the task.

Even if this and other studies fail to directly tell us how aware animals are of their own memories, it is hard to deny the possibility that animals consciously travel along the time dimension and rack their brains for knowledge and experiences. We now have the beginnings of an idea of what consciousness is good for and why it evolved. I suspect that animals capable of consciously probing their experiences and memories also have the capacity to explicitly recognize the bodily upheavals that we call emotions. If they’re going to make decisions based on memories of the past, it probably helps to realize how those experiences made them feel.

This post is adapted from De Waal’s new book, Mama’s Last Hug: Animal Emotions and What They Tell Us About Ourselves.