From insects to humans, cooperation is a powerful survival strategy. Working together allows more efficient food gathering, defense against threats and the spread of useful behaviors through observation^1–3, but it’s also cognitively demanding. Cooperation requires individuals to anticipate a partner’s next move, adjust their own behavior and stay focused on a shared goal, all in real time. People with autism spectrum disorder, especially those with fragile X syndrome, often struggle to connect with others and to adjust their behavior as a social situation unfolds. Fragile X syndrome has a single genetic cause, mutations in the gene FMR1, making it especially useful for studying how one specific biological change can shape social behavior in animal models^4,5.

To see how cooperation might be affected in Fragile X syndrome, a team led by Shantanu Jadhav of Brandeis University, a SFARI grantee, along with co-first authors Ashutosh Shukla and Edward Rivera, designed a new task for pairs of rat. Each rat was free to navigate its own maze, which consisted of a central corridor branching into three arms each ending in a reward well. Paired rats navigated two mazes mirroring one another, with arms pointing inward toward a shared transparent divider, letting each rat see, hear and smell its partner without making physical contact. To earn a reward, both had to choose matching arms at the same time.

The researchers compared Fmr1-knockout rats, in which the Fmr1 gene is disabled to model the loss of function in fragile X syndrome, to wild-type rats using this paradigm. Both groups learned to cooperate at rates above chance, but wild-type rats consistently outperformed Fmr1-knockout rats throughout training.

Two follow-up experiments helped rule out simpler explanations for this gap. First, the researchers blocked the rats’ view of each other. Cooperation dropped in both groups, confirming that the task depends on watching a partner. Second, they replaced the partner with a simple light cue that indicated the correct well, removing the social element entirely. Both groups of rats learned to associate the cue with reward equally well, although the Fmr1-knockout animals took longer to respond to the cue, reflecting a difference in response time rather than in learning or sensory processing. Together, these results suggest that Fmr1-knockout rats do not have a general problem learning this type of task, but rather have trouble specifically in cooperating with another animal.

A closer look revealed why wild-type rats came out ahead. Both groups relied on a reactive strategy, watching where a partner went and then moving to match it. Wild-type rats, however, also learned to anticipate a partner’s next move, allowing the near-simultaneous arrival of the pair. Fmr1-knockout rats did not develop this more efficient strategy.

To confirm this, the researchers tracked how often each rat looked toward its partner and whether a rat’s choices tracked its partner’s recent behavior. As training went on, wild-type rats relied less on watching their partner and, as the task got harder, grew more likely to cooperate if their partner had recently cooperated with them. Fmr1-knockout rats showed neither change. They kept watching their partner just as closely throughout training, and they matched their partner’s recent cooperation only in the easier, fully rewarded version of the task.

These results suggest that Fmr1-knockout rats can still engage socially, but they struggle to build an internal model of a partner’s likely behavior and to use that model to predict rather than simply react. This echoes broader reports of difficulty with flexible thinking in people with fragile X syndrome⁶. The task’s design lends extra confidence to this interpretation. Because each rat ran on its own maze, the researchers could be sure that successful coordination reflected genuine monitoring of a partner’s actions, not avoidance, physical jostling or competition for space. This same separation also makes it possible to study brain activity during cooperation without that physical complication. It offers researchers a more direct way to explore the neural basis of predicting a partner’s behavior and to pinpoint where it is disrupted in autism.

1. Axelrod R. and Hamilton W.D. Science 4489, 1390–1396 (1981) PubMed
2. Nowak M.A. Science 5805, 1560–1563 (2006) PubMed
3. Giraldeau L.A. and Caraco T. Social Foraging Theory (Princeton University Press, 2018)
4. Salcedo-Arellano M.J. et al. Neurobiol. Dis. 136, 104740 (2020) PubMed
5. Treccarichi S. et al. Int. J. Mol. Sci. 27, 3278 (2026) PubMed
6. Schmitt L.M. et al. Brain Sci. 9, 15 (2019) PubMed