The Gap

by Thomas Suddendorf

  Animals are by no means insensitive to temporal matters. There is evidence that some of them detect the time of day (some dogs will get ready for the mail carrier to arrive) and can track short time intervals (if you feed a dog every half an hour, it will start to salivate just before the next feeding). Yet, as the psychologist William Roberts showed, these competencies can be achieved through basic mechanisms, such as simple associations with states of the natural body cycle, rather than anything akin to mental time travel.

  Our most direct evidence for episodic memory in humans comes from people’s verbal reports. As already discussed, some apes have been trained to communicate using human symbol systems, but they have so far not acquired tense and have not told stories about the past or visions of the future. These projects have provided some compelling evidence of semantic knowledge, however. After all, these animals have learned which symbols go with which objects or actions. The chimpanzee Panzee has even used a symbol board to announce what food was hidden outside her enclosure and pointed to get a human to retrieve it for her. This shows she knew where the food was hidden, but it does not necessarily mean she remembered the hiding event itself. One can know things without knowing how one has come to know them. For instance, you may know that Mount Kilimanjaro is the highest mountain in Africa but probably not the occasion on which you learned that fact—unless you learned it just now.

  Although it is safe to conclude that animals have procedural and semantic memory systems, there is no obvious demonstration that they have episodic memory. Rats appear to use their hippocampus to create cognitive maps of their environment. Most species, even insects, demonstrate sophisticated navigational skills. But do they mentally reconstruct the particular events that shaped their knowledge? Do they reminisce about days of yore?

  RECENTLY, RESEARCHERS HAVE MADE THE case for certain animals having something like episodic memory. Psychologists Nicola Clayton, Anthony Dickinson, and their colleagues at the University of Cambridge have produced some intriguing results that they say may reflect mental time travel in animals. Scrub jays hide food for later consumption. Clever experiments on caching and retrieval capacities indicate that these birds know what was cached where and when. For example, they adjust their search differently for cached worms, which rot quickly, and peanuts, which keep fresh longer, depending on how long ago they had stored them. They do not bother searching for the worms if they had been stored a long time ago because they would be rotten. Jays show this search pattern even when there are no cues, such as smell, to guide them. The researchers conclude that the birds have memory for the past occasion on which they stored these foods. They refer to this as “episodic-like” memory, leaving open the question whether the birds are conscious of the past.

  This approach has stimulated headlines, debates, and a profusion of such studies on other animals. Although several species have failed to pass similar tests, some species, such as mice, rats, and chimpanzees, have passed them. These successful species are therefore also said to have a capacity for episodic-like memory. But how like episodic memory is episodic-like memory? Despite the cautious terminology, romantic proponents of the what-where-when approach often imply if it walks like a duck and quacks like a duck, then it probably is a duck. In other words, these species probably can travel mentally into the past. But is this really a duck hunt or a wild goose chase?

  I have argued that evidence for episodic-like memory is neither necessary nor sufficient for mental time travel. You can know that something happened and yet not remember the event at all. For instance, I know what happened on November 24, 1967, in Vreden, Germany. I was born. Yet I have, of course, absolutely no recollection of the event. Conversely, you may recall a particular episode and yet be factually wrong about what precisely happened where and when. Recall (if you can) that episodic memory is notoriously unreliable. Thus evidence that animals can draw on accurate information about the what, where, and when of a particular event does not show that they travel mentally in time.

  How then to explain the scrub jays’ clever retrieval behavior? It is possible, and quite plausible I think, that scrub jays know what food is where and whether it is still good to eat without having to remember the caching episode itself. Here is a simple alternative explanation. As time passes, memories fade. Scrub jays may learn when it is still worthwhile to search for a particular type of food by associating the strength of their memory of the food location with whether it is still good to eat on recovery. They then simply apply a rule along the lines of: worms aren’t worth searching for once the memory of their location has weakened beyond a certain point. The experience with nuts leads to a rule that they are worth searching for even when memory is much weaker. Such rules can effectively provide use-by dates for different types of stored food (without requiring conscious recollection of the caching event).

  If this what-where-when approach does not demonstrate episodic memory, how could one show animals have episodic memory, if indeed they have it? Without language they cannot tell us about their time travel. As we saw, humans can also express past episodes through mime or dance. However, there is no suggestion that animals do this too. If evolution selected for episodic memory in animals, then this capacity must have benefitted survival and reproduction. Evolution could not have selected for it as a private indulgence without a tangible effect. Given the evidence of close links between episodic memory and foresight, and the fact that foresight offers clear fitness benefits, one would expect that animals that have mental time travel capacities should be able to control their future prudently. They should hatch plans and plot their way to future happiness.

  Long-term planning . . . is something utterly new on the planet, even alien. It exists only in human brains. The future is a new invention in evolution.

  —RICHARD DAWKINS

  RICHARD DAWKINS AGREES THAT THERE is something unique about the human capacity to think about the future. Human long-term planning appears to have no obvious rivals in the animal kingdom. Yet many species act in ways that improve their future: animals construct nests for breeding, they migrate at the right time to warmer climates, and they search for food where it is likely to become available. Learning about recurring patterns in the context of mating, food, and predation has obvious evolutionary benefits. Over long periods species have acquired ways of taking advantage of what is regularly recurring. Even bacteria demonstrate future-directed capacities in this sense. Right now, Esherichia coli are moving through your digestive tract from an environment rich in lactose to one rich in maltose—and they have prepared for this by turning on genes necessary for digesting maltose. This does not mean each individual bacterium looks ahead and decides to prepare as it goes down your gut. Evolution has selected for this order of events over many generations of bacteria: E. coli that happened to show this pattern of preparatory gene activation survived and reproduced better than those that did not. Many species have evolved innate mechanisms that take advantage of long-term regularities.

  These innate behaviors look clever, but the lack of foresight involved becomes clear when circumstances change. A classic example is the digger wasp. The wasp always inspects the nest before dragging its prey inside to feed its larvae. If in the meantime a mischievous human moves the food a few centimeters, then the wasp will regather the food, and repeat the sequence again by dropping it at the entrance and inspecting the nest. This can be repeated again and again, without the wasp breaking out of its behavioral program. Although provisioning the young appears to be a complex, future-directed behavior, the wasp executes it without any apparent thoughts about seeing them grow up. If the entrance is destroyed, the wasp will not feed its larvae but trample over them in its frantic search for the entrance. Similarly, various animals hoard food for the winter without necessarily understanding why they do it. Young squirrels, for instance, will hoard nuts even if they have never experienced a winter. These behavioral solutions to recurring seasonal changes may not be all that different from physical a
daptations to the same problem—such as storing food for winter in body fat.

  Innate mechanisms are reliable but not very flexible. Memory, by contrast, enables an individual, rather than a species, to learn about potentially recurring events during its lifetime. All memory, you may recall, is in a sense future-oriented, allowing an individual to adapt to its environment. If a stimulus reliably predicts a situation, like the bell predicted food for Pavlov’s dog, animals can learn to take advantage of this association. A pigeon, for example, may learn to peck on a button for food only when a light indicates food is available. Associative learning can account for some complex-looking behavior, as we saw in the case of Clever Hans. Behaviorists have documented the rules that govern such learning, and one of their main findings has been that two things have to happen together or nearly together if they are to be associated. The link between an action and a consequence is typically learned only if they are separated by no more than a few seconds.

  There are rare exceptions. Perhaps the most extreme is that rats can learn to associate the taste of food with feeling nauseous some hours later. Given that rats keenly explore novel food sources, learning to avoid foods that will make them sick has obvious survival value. So this appears to be a special capacity. The rats learn only that a taste predicts later sickness; they cannot learn that a sound or a sight has the same predictive relationship. Yet instinct and learning may combine to create sophisticated future-directed behaviors, some of which are not yet well understood. For example, how do gray squirrels learn to bite out the seeds of white oak (though not red oak) acorns before storing them, thereby preventing their germination? Curious though they are, in these cases, as in the food-caching scrub jays, the future-direct competences seem limited to a particular type of problem and show none of the open-ended flexibility evident in humans.

  We saw in Chapter 3 that our closest animal relatives have a basic capacity to imagine other possible worlds. Can they use their minds to plot future actions? Maybe they can for some problems that lie in the immediate future. For example, when presented with a treat that is out of reach, great apes can go around the corner, where the treat can no longer be seen, and select a stick of the appropriate length to solve the problem. I had to learn this the hard way, when the chimpanzee Ockie noted that a stick at one side of his enclosure could be used to reach a TV I had put up earlier on the other side—with smashing consequences. They also carry tools over short distances in the wild. One of the most impressive examples of apparent forethought comes from chimpanzees in the Tai forest of the Ivory Coast. These apes use stones to crack open nuts and sometimes carry stones for a hundred yards or so to the site of the nuts. Presumably they pick up the rocks as a result of developing an appetite for nuts and a plan of where to get them.

  In spite of these signs of short-term foresight, animal mental access to the future may be restricted in several important ways. Recall the range of processes involved in a theater production of the future. Shortcomings in any of them may limit capacity. Chimpanzees may have some capacity to imagine alternatives, but they may be fundamentally restricted by other requirements. One possibility is a limited understanding of one’s self as an actor in a future scenario. The psychologists Norbert Bischof and Doris Bischof-Köhler have claimed that animals may not be able to imagine drives and wants they do not currently experience. You can easily imagine being thirsty even when quenched, and hence may want to secure future sources of drink. Recall the comparison between full-bellied lions and humans—we often try to get things we do not (currently) need.

  Such a limitation could explain various curious animal behaviors. Take the case of laboratory monkeys that were fed biscuits only once a day. William Roberts recounts how in the 1970s the cebus monkeys in Michael D’Amato’s laboratory would hungrily eat until they were full and then throw the remaining biscuits out of their cages. To their dismay, they would find themselves hungry again some hours later. You might wonder why they did not learn to guard their food to satisfy their future hunger. If you cannot imagine being hungry again, then perhaps biscuits’ utility may lie in their quality as fine projectiles. There is no point in acting now to secure a future need you cannot conceive of. This Bischof-Köhler hypothesis is much debated. In some sense it cannot be quite right, because any hoarding of food is behavior that secures future needs. What we don’t know is whether hoarding animals are thinking about future hunger and burying their treats in anticipation of satisfying those future cravings. Despite several attempts to disprove the hypothesis,6 it still has currency. Animals do not seem to keep and refine their tools for repeated use, and there is no evidence of the kind of greed present in humans.

  Another essential capacity for foresight is the ability to forego current temptation in pursuit of a more desirable future outcome. When monkeys are given the choice between a small reward now and a larger reward later, they, like most other animals, find it extremely difficult to wait if the delay is more than a few seconds. Great apes can do better and have been shown to delay gratification for several minutes. For a reward forty times larger than the immediate reward option, chimpanzees may wait up to eight minutes. This is impressive, but it is considerably less dramatic than humans delaying gratification for months, years, or even a lifetime. Still, our closest relatives are much better at waiting for something than other animals.

  Perhaps the most prominent case for ape foresight comes from a study in which three bonobos and three orangutans were trained to use a tool to obtain a treat from a feeder apparatus. The animals were then ushered into a waiting room, and the experimenter removed all remaining tools left in the experimental room. After an hour’s wait, the apes were allowed back into the experimental room. On almost half of the trials the apes carried a tool with them from one room to the other and back—so they could use it to get more treats. Some apes did a lot better than others, and two animals even succeeded to hang on to their tools when they returned to the experimental room after an overnight delay. Unfortunately, because it was always the same tool the apes had to select on all trials, it remains unclear here, as in subsequent studies, whether the apes anticipated a specific future situation or had merely learned to associate the tool with rewards. In another study ten chimpanzees were taught that they could exchange a token for a food reward and were then given the opportunity to collect these tokens in anticipation of an exchange session an hour later. Over several experiments the chimpanzees failed to take tokens more often than other useless objects. They just did not think ahead.

  An unusual report made headlines in 2009 when it suggested that a chimpanzee at a Swedish zoo may have spontaneously planned its projectile attacks for hours in advance. Three zookeepers reported that in the late 1990s a male chimpanzee tended to collect rocks and concrete into piles early in the morning so he could excitedly hurl this ammunition at zoo visitors a few hours later. Nothing quite like such planning has yet been reported from the wild or, for that matter, from any other captive chimpanzees. If chimpanzees have this kind of foresight, one would perhaps expect to hear about many more such examples—it’s possible that this is what the future will bring.

  So what are we to conclude? Although some research suggests that our closest relatives may have some, albeit limited, capacity for foresight, we have seen that other studies suggest they are profoundly shortsighted. Animals clearly share with humans some procedural and semantic memory capacities. However, there is little evidence that they have episodic memory. The best evidence for episodic memory should come from signs of episodic foresight, given that mental time travel in both directions is intimately linked. Yet animals do not overtly express any of the obvious manifestations of such a capacity. Only in Orwell and other fiction do they conspire to rebel. Animals can learn that one thing predicts another if the events are separated by only a few seconds. Many species have also evolved instincts that equip them to act in preparation for the future. However, evidence for flexibility in such future-directed behavior is scant.
The cases that exist all deal with the immediate future only. There is no compelling evidence that animals flexibly generate mental scenarios of remote future events the way humans do, or communicate mental scenarios to one another to obtain feedback or to coordinate actions. In the next chapter, we will consider whether animals can even appreciate that others have minds in the first place.

  Mental time travel is essential for explaining a vast array of characteristics of the human mind, ranging from emotions (such as hope and regret) to motivations (such as plans and revenge). It allows us to understand how things got to where they are and to wonder where everything is headed. It has given us unheralded powers to control plant and animal life to our advantage. But it also confronts us with that most unwelcome of all realizations that I remember pondering as I lay in my childhood bed staring at the ceiling: our mortality.

  1In a dramatic documentary Clive is presented with a video recording from earlier in the day in which he can be seen conducting a choir. He acknowledges that it is him in the video but refuses to identify with his image. Instead, and as always, he insists that he has just woken up from unconsciousness. The fact that people can show him his past on tape makes no difference to him. Tulving argued that retrieval of episodic memories is characterized by a particular “self-knowing” consciousness and contrasted it with the “knowing” consciousness involved in semantic memory of facts, and the “not-knowing” consciousness associated with procedural memory.