Henry Molaison, known for years as “H.M.,” was famously unable to form new memories. If someone he had met left the room only to return several minutes later, he would greet that person again as if for the first time. Because of surgery to treat intractable epilepsy, H M. lacked a sea-horse-shaped brain structure called the hippocampus and had amnesia. His case helped establish the hippocampus as an engine of memory.
In recent years scientists have discovered another essential deficit that burdens people with hippocampal amnesia: they can’t envision the range of possibilities that must be considered to make future plans. When researchers asked a group of people with hippocampal damage to describe themselves in a fictitious scene—say, lying on a white sandy beach—they came up largely blank, producing only fragmented images. Brain scans of healthy people, by contrast, showed that their hippocampus was engaged even more when they imagined the future than when they summoned the past.
Studies of neural activity in rats have since come along to support the idea that the hippocampus plays a central role in imagination. “It’s still responsible for creating memories of what is happening right now,” says Loren Frank, a systems neuroscientist at the Howard Hughes Medical Institute and the University of California, San Francisco. “And now it seems it is also responsible for rolling out possibilities.” Frank and his colleagues make their case in a paper entitled “Imagination as a Fundamental Function of the Hippocampus,” which was published in the Philosophical Transactions of the Royal Society B.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
That dual role makes sense, experts say, in part because imagination depends largely, if not exclusively, on memory. “Why do we talk about imagination separately from memory? From the public point of view [talking about them together] is a crazy idea. But you can put it in a simple way: there is absolutely no way you can imagine anything without the past,” says György Buzsáki, a systems neuroscientist at New York University, who was not involved with the paper.
In addition, both skills involve essentially the same process: combining bits and pieces of experience with emotions, inner commentary and things people have read or heard about, says Donna Rose Addis, a cognitive neuroscientist at the Rotman Research Institute in Toronto and the University of Toronto, who was also not involved with the recent review. This process can even distort memories by mixing them with imaginary material. “Memory is a form of imagination,” Addis says.
From Frank’s point of view, imagination gives memory a purpose: helping us make decisions based on what we’ve learned—for instance, deciding to avoid a food that once made us sick. “From an evolutionary perspective, we are reasonably sure that the purpose of memories is actually in the future,” Frank says. “Memories allow you to take experiences that you have and retrieve them to make predictions about what will happen next.” This chain of neural events even loops back on itself. We also need to form memories of our simulations of the future so that when we have an experience, we have something to draw on. “We have found that the encoding of an imagined simulation also involves the hippocampus,” Addis says.
Much of the recent evidence for imagination’s roots in the brain draws on a Nobel Prize–winning discovery in the 1970s of “place cells” in the hippocampus. When a rat runs a maze, the activity of these cells changes in a predictable way based on the animal’s position in the maze. These hippocampal cells tell an animal where it is in the world. This function seemed distinct from imagination until Frank and his colleagues showed that the activity of these cells does not always represent an animal’s actual location.
The firing patterns of place cells repeat about eight times per second in rats, forming what is called the theta rhythm. And within each cycle, the researchers found, the patterns progressively change to represent three different locations for the animal that are separated in time: the place it just was, its current position and, late in the cycle, a possible upcoming location. “The neural activity has this unmistakable structure where, at certain time points, it looks like what the animal is experiencing in the present. At these other time points, it looks like an imaginary experience,” says Kenneth Kay, a postdoctoral researcher at the Mortimer B. Zuckerman Mind Brain Behavior Institute at Columbia University and a co-author of the paper. (The paper’s first author is Frank’s graduate student, Alison Comrie.)
What a rat seems to be imagining in any given cycle varies. When the rat is approaching a T-junction in a maze, the late theta activity alternates between two possible immediate futures: a turn to the left in one cycle and a turn to the right in the next. It’s as if the animal is planning its next move, akin to a soccer player who is running toward a ball and flipping through various scenarios before deciding on a play.
In other instances, that late theta activity denotes a more distant place in the maze, as if the animals’ mind wandered to some other scene or scenario, perhaps some place it would rather be. The researchers also found instances in which this imagination portion of the cycle reflected a hypothetical direction of travel that differed from the animal’s actual directional heading. “They are representing things that can roughly be thought of as possibilities or hypotheticals, things that could be but aren’t necessarily the case in terms of a possible future or just an alternative reality,” Frank says.
The mere existence of spontaneous activity within the hippocampus that is not necessarily tied to a specific place, some experts say, hints at an internal thought process that is separated from reality. “That rhythmicity [of the theta wave] is not coming from the environment,” Kay says. “That’s highly reminiscent of the notion that our imaginings are coming from ourselves, and they are not from this external reality.”
Another form of imagination seems to occur when an animal isn’t traveling through space but is eating, grooming or zoning out. At these times, scientists have detected bursts of activity in the hippocampus called “sharp wave ripples,” which also occur during sleep, that seem to represent mental replays of past events. The replays occur about 10 times faster than the original event, a reenactment that is reminiscent of human experience. “One huge advantage of using our minds to think about things sometimes is: we can quickly play through things, we can quickly simulate them,” Kay says.
While these mental replays are a form of recollection, they can also represent events the animal has not experienced, Frank says. Some sharp wave ripples appear to connect two trajectories that an animal had experienced separately but not together, he says. The ripple activity may, in essence, build a mental map so that the animal can mentally traverse new paths, such as shortcuts and detours. In this context, the hippocampus seems to be acting to combine past events in new ways, something that “is more like imagination than it is just replaying the past or predicting the future,” says Lynn Nadel, an emeritus professor of cognitive science and psychology at the University of Arizona, who did not contribute to the recent paper.
The experiments of neuronal activity in rodents are important, experts say, because they place the idea of imagination into a physical reality: that of the brain itself. “This gives us an opportunity to take a fuzzy cognitive concept like imagination” and link it to brain activity, says Daphna Shohamy, a cognitive neuroscientist at Columbia University, and director and CEO of Columbia's Zuckerman Institute, who was not involved in those studies or the review paper.
Humans’ internal worlds are rich, however, and the studies of place cells in rats may not represent all types of human imagination. The animal results connect most directly with imagination that is based in experience and action, as in planning out a strategy for moving through the world, Nadel says. But other experts believe the hippocampus has a much broader repertoire: it may also forge ties between ideas and information. “I don’t think the hippocampus cares, really, about what you’re connecting,” Addis says.
Some of Shohamy’s work supports the idea that the hippocampus might be important for mental simulations that are not rooted in time or place. She has found that people with damage to the hippocampus are much slower than those without brain damage to choose between food items—say, a Kit Kat versus M&Ms—that they like about equally well. The problem seems to be that they have trouble imagining what the options are like. “It looks as if they are spending more time trying to conjure up the evidence,” Shohamy says. In the end, they make a choice at random.
Although the hippocampus may play a central role in imagination, it is by no means performing a solo act. It needs the cooperation of other brain areas. Frank likens the hippocampus to an orchestra conductor that cues up neurons in other regions that represent the sights, sounds and smells that either are part of a recollection or “fit together in some imagined thing.”
One mystery is how people separate a real symphony from music playing in their head. “It’s amazing that we’re not all psychotic all the time, that we’re not all delusional, because our brains are clearly making stuff up a lot of the time about things that could be,” Frank says. New data from Frank’s group suggest the brain may use sensory input—say, the feeling of a foot hitting the ground while walking—to flag what is real versus what is just in the mind’s eye and so ground this hive of neural activity in the physical world. The brain, he says, separates fact from fiction by reconciling the information it receives from the outside world with its own internal models.