Summary: A new study shows the brain can rapidly reconfigure itself to encode rewarding experiences—like finding food—even when the reward’s location changes. Using virtual reality and two-photon imaging in mice, researchers observed flexible “reward maps” in the hippocampus that update within moments after the reward moves.
These reward maps remain detectable over long distances within the virtual environment and can influence future choices before behavior visibly changes. The findings clarify how spatial memory and reward information interact and could have implications for understanding dementia, addiction, and other conditions that involve disrupted spatial or reward-related memory.
Key facts
- Dual mapping system: The hippocampus supports a stable spatial map of the environment alongside a separate, flexible reward map.
- Rapid adaptation: Reward-related neural representations update almost immediately when the reward location changes.
- Clinical relevance: Altered coupling between spatial and reward maps may contribute to symptoms in dementia and to context-triggered relapse in addiction.
Source: Stanford
Imagine this: your regular coffee shop is unexpectedly closed, so you walk a few blocks to a new café that turns out to be great. After a few visits, you begin to prefer the new place. That behavioral switch likely reflects more than habit: each rewarding visit strengthens a neural representation that marks where good experiences occurred—a reward map that helps guide future choices, even from far away.
Researchers at the Wu Tsai Neuro program at Stanford used head-mounted windows and two-photon microscopy to watch hippocampal activity in real time while mice navigated controlled virtual environments. Because two-photon imaging requires the animals to remain stationary under the microscope, the team placed mice on a running wheel surrounded by large screens that displayed a first-person hallway view. As the mice ran, the virtual scene moved, and the setup delivered a small reward—sugar water—at designated virtual locations.
When the experimenters changed the virtual position of the sugar water, the hippocampal network adapted. One neuronal subpopulation maintained a consistent spatial map of the virtual environment. A distinct, more flexible subpopulation shifted its activity so that neurons fired at positions defined relative to the new reward location—effectively forming a reward-centered map that spanned the entire task. Over repeated visits, more neurons joined the reward map, making the representation more robust.
Importantly, the two maps are not entirely separate: some neurons can contribute to both spatial and reward representations at different times. Even so, the overall size of the spatial map stayed relatively stable while the reward map grew with experience. Because the reward-relative activity often changed before the animal’s overt behavior did, the study suggests that neural reward maps can shape future behavior as much as behavior shapes the maps.
The rapid, flexible encoding of reward locations helps explain how animals adapt when food sources move and may shed light on human memory and motivation. Other studies have hinted that humans also use distinct neural signals to mark places they intend to return to versus the general layout of their surroundings. If the link between those representations weakens, it could contribute to memory problems seen in dementia—such as forgetting where or when an event occurred. Conversely, an overly strong coupling between reward cues and places may underlie context-driven cravings in addiction, where environments tied to prior drug use trigger relapse.
Understanding how the hippocampus links spatial information with rewards could lead to new therapeutic approaches. For people with addiction, interventions that weaken the pathological association between a place and a drug reward might reduce relapse risk. For those with dementia, strengthening relevant associations could help preserve memory for where meaningful events occurred.
The research team plans to investigate how reward maps influence natural exploratory decisions: how an animal chooses between revisiting a known food source and searching for new opportunities. They also want to know whether non-food rewards—such as social interactions—use the same hippocampal mechanisms. As methods for imaging and manipulating neural activity improve, these questions become increasingly tractable.
Funding sources
This work 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. Marielena Sosa received support from a Helen Hay Whitney Foundation fellowship and an NIH BRAIN Initiative Pathway to Independence award (K99MH135993).
About this neuroscience and brain mapping research news
Author: Nicholas Weiler
Source: Stanford
Contact: Nicholas Weiler, Stanford
Image credit: Neuroscience News
Original research (open access): “A flexible hippocampal population code for experience relative to reward” by Lisa Giocomo et al., published in Nature Neuroscience. DOI: 10.1038/s41593-025-01985-4
Abstract
A flexible hippocampal population code for experience relative to reward
Remembering events that lead up to and follow rewards helps reinforce beneficial behaviors. Hippocampal place cell activity covers spatial and non-spatial episodes, but it has been unclear whether hippocampal ensembles encode full sequences of experience relative to a reward. Using two-photon imaging of CA1 while mice navigated virtual tracks with changing reward locations, investigators found a subset of neurons that shifted their firing fields to maintain the same relative position to reward, creating behavioral-timescale sequences across the entire task. With learning, this reward-relative code grew stronger as additional neurons were recruited, and neural changes often anticipated behavioral adjustments after reward relocation. At the same time, spatial coding was preserved by a parallel, dynamic neuronal subpopulation rather than fixed cell classes. These results show how hippocampal networks can flexibly represent both spatial context and reward-related information while highlighting behaviorally relevant signals.