Understanding Human Decision-Making

A team of MIT researchers has successfully modeled how people approach solving a complex problem by adopting distinct decision-making strategies. This study focuses on understanding how individuals predict the path of a ball traveling through a maze – a task made difficult by the inability to track the ball’s trajectory directly.

Breaking down complex problems

The human brain excels in problem-solving by dividing complicated tasks into simpler steps. For instance, everyday tasks like obtaining coffee involve sequential actions: leaving the office, navigating to the coffee shop and making the purchase. This breakdown helps people deal with challenges, like a broken elevator, without significantly altering the overall process.

Despite the ability to manage such tasks effectively, it has been challenging to experimentally capture the computational strategies the brain uses to solve problems. In their study, MIT researchers developed a model to understand how humans navigate such complex decision-making tasks.

Modeling decision-making strategies

The researchers designed an experiment where participants had to predict the path of a ball hidden in a maze. While the ball’s trajectory is invisible to the participants, auditory cues at key points in the maze provide partial information. The task, inherently impossible to complete with perfect accuracy, requires participants to adopt different strategies to make the best prediction possible.

The study identified two prominent decision-making strategies: hierarchical reasoning and counterfactual reasoning. Hierarchical reasoning involves breaking down a problem into smaller sections, while counterfactual reasoning involves considering what would have happened if a different choice had been made. The researchers sought to understand under what circumstances people switch between these strategies.

Human adaptability in complex scenarios

Humans tend to excel in simple tasks with clear solutions. However, in more complicated scenarios, where no single solution is optimal, they use heuristics – mental shortcuts that help in decision-making. The hierarchical and counterfactual reasoning strategies are examples of such heuristics, commonly used to navigate complex tasks.

“What humans are capable of doing is to break down the maze into subsections, and then solve each step using relatively simple algorithms. Effectively, when we don’t have the means to solve a complex problem, we manage by using simpler heuristics that get the job done.”

Dr. Mehrdad Jazayeri.

The MIT researchers crafted an experiment in which volunteers predicted the movement of a ball through four possible paths in a maze. Though it is not possible for humans to track the ball’s trajectory perfectly, participants were able to predict its path based on auditory cues. The results revealed that most participants relied primarily on hierarchical reasoning but would switch to counterfactual reasoning when necessary, especially when the information gathered did not align with their initial assumptions.

The role of memory in decision-making

The study also demonstrated how memory influences decision-making. When participants chose a path and then heard a cue that did not match their prediction, they sometimes revised their decision. This revision was influenced by the strength of their memory of previous cues, showing that people use counterfactual reasoning based on how reliable their memories are.

Computational models and human behavior

The researchers further validated their findings by training a machine-learning model to perform the same task. The model, initially capable of making accurate predictions, began to mimic human-like decision-making when cognitive limitations, similar to those faced by humans, were introduced. When the model’s ability to track all possible paths was impaired, it adopted strategies similar to those used by human participants.

“What we found is that networks mimic human behavior when we impose on them those computational constraints that we found in human behavior. This is really saying that humans are acting rationally under the constraints that they have to function under.”

Dr. Mehrdad Jazayeri.

By manipulating the model’s memory capabilities, the researchers observed that the strategy-switching process happens gradually, not at a clear-cut point, reflecting the flexible nature of human decision-making.

Conclusion and next steps

The study provides new insights into how humans solve complex problems by using simplified decision-making strategies under cognitive constraints. The research suggests that the brain deploys rational strategies based on available resources, providing a model for understanding human decision-making in uncertain situations. The researchers are continuing to investigate what happens in the brain during these shifts in strategy.

Reference: Ramadan M, Tang C, Watters N, Jazayeri M. Computational basis of hierarchical and counterfactual information processing. Nat Hum Behav. 2025. doi: 10.1038/s41562-025-02232-3

This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source. Our press release publishing policy can be accessed here.

This content includes text that has been generated with the assistance of AI. Technology Networks' AI policy can be found here.