Imagine you're walking to work when the unspeakable occurs: Your favorite coffee shop — where you stop every day — is closed. You groggily navigate to a newly opened coffee shop a couple blocks away, which, you're pleased to discover, actually makes quite a good morning brew. Soon, you find yourself looking forward to stopping at the new location instead of the old one.
That switch probably alters more than just your morning routine. Each time you visit that new coffee shop, the experience likely strengthens a neural map marking the positions of rewarding experiences — a map that can guide you back to those experiences even from miles away.
While the existence of a reward map is familiar from previous work, Wu Tsai Neuro researchers working with mice were surprised to find that the map persists even when mice move many meters away from a treat, and that it updates almost immediately when the location of the treat changes.
"No matter where we moved the reward, the reward map adapted almost immediately" said Lisa Giocomo, a professor of neurobiology at Stanford Medicine and one of Wu Tsai Neuro's deputy directors. "I wasn't expecting the change to be so quick."
The results could shed light on a variety of human conditions, including dementia and addiction, as well as illuminating an important aspect of how animals survive in the wild.
To study how rewarding experiences are encoded in the brains of mice, Giocomo and her colleagues first replaced a bit of a mouse's skull with a transparent window. That allowed the researchers to use a sensitive technique called two-photon microscopy to observe neural activity in real time in a region of the brain called the hippocampus. The mice had to stay beneath a stationary microscope for the technique to work, so instead of allowing the animals to search for rewards in real life, the researchers used a virtual reality environment.
"People picture mice wearing miniature virtual reality goggles, but actually it's more like an IMAX Theatre situation," said Marielena Sosa, a postdoctoral scientist who performed the work along with former Giocomo Lab graduate student Mark Plitt, a member of Wu Tsai Neuro's Mind, Brain, Computation and Technology program.
The mice ran on wheels while surrounded by three large monitors that played a video of a hallway that advanced as they ran, kind of like a first-person (first-rodent?) video game. The researchers programmed the apparatus so that it dispensed a drop of sugar water to the mouse when it approached a particular location in the virtual environment. The team then moved the virtual location of the sugar water and watched as neural activity in the brain adapted.
As they expected, one population of neurons in the hippocampus maintained a stable map of the environment. Less expected was the observation that a separate population of neurons turned on and off fluidly as the reward's location changed, adding a layer of adaptability that let the mouse easily adjust when the location of the food changed — an important tactic in the wild.
The neurons that make up the two maps are also not completely distinct. Rather, the neurons that make up the spatial map can sometime switch to being involved in the reward map and vice versa. However, the total number of neurons in the spatial map remained relatively constant, whereas the number of neurons making up the reward map increased as the mouse revisited the reward. Eventually the reward map could keep track of an animal's position relative to the sugar water from meters away. On a human scale, that could be equivalent to many city blocks.
When the researchers changed the position of the treat, "the switch at the neural level was obvious even before the switch at the behavioral level," Sosa said. That could mean that the neural reward map shapes a mouse's future behavior as much as its past behavior shapes the map.
There's evidence that humans also store locations they intend to return to in a separate set of neurons than those used to map their general surroundings . Understanding the connections between these two maps could bear relevance to dementia and addiction disorders, the researchers believe.
People with dementia tend to forget the sequences in which events occurred, which could be explained, at least in part, by a loss of the ability to visualize where they were when they completed certain tasks. For example, someone with dementia may not remember whether they drank their morning coffee in the kitchen or brought it with them in the car, whereas a healthy person would distinctly remember sitting at their kitchen table with a cup in hand. That difference could be explained by an uncoupling between the reward map and the spatial map in people with dementia.
In the case of addiction, memories of rewarding experiences, like using drugs, are often tightly tied to spatial locations. "Someone who first uses drugs at a concert might always be triggered to seek out drugs when they're at a concert, for instance. And that can be a big problem because it causes people in recovery to relapse when they encounter those triggering environments," Giocomo said. That could be because the link between the reward map and the spatial map created by the powerful drug is unhealthily strong.
Understanding the neural links between spatial information and rewards could ultimately lead to therapies that weaken these links and help people overcome cravings brought on by specific locations. Conversely, therapies that strengthen the links could help people living with dementia.
The research team has more questions, one of which is how reward maps guide animals as they explore their natural environments. "If you're a mouse, how do you decide if you want to explore the possibility of finding food in new locations or return to a place where you've found food before?" Giocomo asks. Studying such questions could also reveal how humans decide when to seek out novel experiences and when to return to the comfort of old habits.
The researchers also wonder whether enjoyable experiences besides food are mapped using this type of neural structure. For example, are the locations of social interactions marked on reward maps? Mice and humans are both highly social animals, but do we use the same type of mental maps to find our friends as we use to find food?
Giocomo thinks her lab has the tools to find out. "The technology is finally at the stage where we can start to tackle some of these more complex behaviors," she said.
Funding Sources
The research in the Giocomo lab was supported by the National Institutes of Health (NIH grants 1R01MH126904-01A1, R01MH130452, BRAIN Initiative U19NS118284, P50 DA042012); The Vallee Foundation; The James S. McDonnell Foundation; The Simons Foundation (grant 542987SPI), the Champalimaud Vision Award, and the Howard Hughes Medical Institute. Sosa's work was supported by a Helen Hay Whitney Foundation fellowship and an NIH BRAIN Initiative Pathway to Independence award (K99MH135993).
