As an animal navigates the world, cells in the hippocampus and entorhinal cortex produce rapid, repeating bursts of activity called theta sweeps: Grid and place cells fire in a specific sequence, first plotting the location the animal has just passed, then where it is currently and lastly what lies ahead.
Whether these theta sweeps simply scan the surrounding environment or instead represent the deliberation and planning needed for goal-directed movement is “something that people have been arguing about for 30 years,” says David Redish, professor of neuroscience at the University of Minnesota.
That debate may now be over: Theta sweeps serve both functions, depending on the situation, according to three new studies by independent teams. The brain produces systematic sweeps by default to passively sample an environment, but it switches to active, targeted sweeps whenever an animal is pursuing a goal or focused on something specific, the studies show.
“It changes our conception of what theta sweeps do,” says Edvard Moser, professor of neuroscience at the Norwegian University of Science and Technology and an investigator on one of the new studies, published today in Science. The other two studies appeared last month in Nature Neuroscience.
Theta sweeps occur within individual theta wave cycles, which are around 125-250 milliseconds long. The teams were able to detect the sweeps’ trajectories by recording hundreds of individual neurons at once in 10-millisecond blocks, a time resolution fine enough to see individual theta cycles, Moser says, adding that they are “invisible if you only look at the average.”
Together, the studies show how theta sweeps begin to plan an animal’s next steps rather than sampling the world, says Redish, who was not involved in the work. “It’s really exciting to see how the field is really converging on mechanisms and circuits that can drive [theta sweeps].”
D
irection-selective cells in the parasubiculum coordinate theta sweeps, previous studies show. At the population level, these cells generate an “internal direction” signal that swings roughly 30 degrees to either side of a rat’s actual heading on alternating theta cycles, and grid cells align their sweeps to that internal signal rather than to the head axis itself. When rats passively navigate an environment, this produces theta sweeps that sample space in front of the animal in a coordinated left-right pattern, according to previous work by Moser and his colleagues.“We interpreted this as a search mechanism,” Moser says. “But the question then was, could these sweeps actually be more flexible?”
To find out, Moser and his colleagues made electrophysiological measurements of hundreds of neurons in the medial entorhinal cortex and parasubiculum as rats either searched for food crumbs in an open area or chased a piece of food suspended from a string. During the latter “fishing rod” experiment, the left-right theta sweeps narrowed the area they sampled to point in the direction of the dangling food, even if the animal couldn’t reach it. Rather than tracking the animal’s head direction, the sweeps zoomed in on wherever the rat was paying attention.
When the rat lost track of the moving food target, theta sweeps turned toward the food source before the animal did, decoupling the neural signals from the head axis. This finding suggests that the sweeps do more than just encode a future trajectory, and instead encode an attention-like mechanism.
“It’s very tempting to call this attention, and I think it actually is,” Moser says. “But there are extra requirements for attention, mainly that the animal needs to not attend to something else.”
The work from Moser indeed shows that “theta sweeps do just go wherever [a rat] is thinking about [going],” says Neil Burgess, professor of cognitive and computational neuroscience at University College London, who is an investigator on one of the Nature Neuroscience papers.
And theta sweeps continued during REM sleep, switching between passive left-right sampling and more narrow, attention-like sweeps, the study found. “Since REM sleep is when at least humans have their dreams, you can speculate that this is related to dreaming. But that’s just pure speculation,” Moser says.
I
n Burgess’s study, he and his colleagues recorded from roughly a hundred hippocampal place cells as rats navigated a “honeycomb” maze, made from small, moving hexagonal platforms. Initially, all the platforms were level, and a rat could navigate the maze freely to learn locations where it would receive a bit of food.During subsequent trials, however, individual platforms were raised such that the rat could still move through the maze—but not always in the direction that it wanted to go. Regardless, “the sweeps seem to be going toward the goal, even though the rat can never go there,” Burgess says.
In the other Nature Neuroscience study, rats either foraged randomly for water or learned the locations of three other water sources, whose positions changed from session to session. While the rats were foraging randomly, the theta waves alternated between left and right, neural population recordings from hippocampal CA1 revealed. The sweeps also alternated during the goal-directed task, but were interspersed with sweeps toward goal locations.
The goal-directed sweeps, the team also found, are preferentially replayed during sharp-wave ripples, which are linked to memory consolidation and occur during sleep and periods of rest.
“[Theta sweeps are a] way of anticipating behavior by simulating a potential trajectory,” says study investigator Antonio Fernandez-Ruiz, assistant professor of neurobiology and behavior at Cornell University. “If you close your eyes and you’re thinking about what you want to do after this, you’re mentally traveling. You’re exploring your internal map and simulating—what if I go to the grocery store, then the bar. That type of planning, I think, is the primary function of the hippocampus. These theta sweeps are a mechanism to explore internal representations, irrespective of behavior.”
W
hether goal- and attention-related theta sweeps reflect genuinely different mechanisms, or the same circuit-level process triggered by different kinds of targets, is unclear. “Whether you call it goal orientation or attention, it’s an internal representation,” Redish says.The three studies are “entirely consistent,” Moser says. “But it’s not only a long-distance goal, it can be anything that momentarily demands an animal’s attention.”
Goal-directed sweeps may “not necessarily [be] tied to the external world” as attention-like sweeps are, at least in the hippocampus, Fernandez-Ruiz says. Instead, they may be “primarily a way to explore internal representations independently of the external world,” not just an immediate sensory experience. The exact purpose of the sweeps is “up for debate,” he adds.
Theta sweeps may explore or evaluate future trajectories, Redish says. “The current theory is that these hippocampal sweeps, at least when they’re goal-oriented, are some sort of planning” signal, he says.
Planning in general involves more than just the information a theta sweep provides, Moser says. “This is a way to scan the immediate environment within a very short time cycle. I think this is a short-distance planning mechanism for navigation, and maybe for other kinds of navigation in abstract thought.” How theta sweeps connect to that longer-range planning is still an open question, he adds, and the overall mechanism through which the sweeps arise is also still unclear.
Theta sweeps, like many oscillations in the brain, may convey information to other brain areas, says György Buzsáki, professor of neuroscience and neurology at the New York University Grossman School of Medicine. “The rhythms in the brain are the most important things for neural communication, because, like in every single communication system, you have to have a beginning and end of messages,” Buzsáki says. Theta waves “are giving you punctuation marks. They’re giving you grammar and syntax of how information could be packaged.”
