Careful planning feels most valuable when the future can be anticipated. When circumstances become unreliable, volatile or difficult to control, people appear to respond in a surprisingly consistent way: they plan less.
A new series of experiments published in Nature Communications found that adults made their first decisions faster as uncertainty increased across three different forms of environmental randomness. Computational modelling suggested that participants also became less sensitive to differences in the value of competing options, effectively simplifying how they made decisions when detailed planning became less useful.
The results offer a more nuanced view of decision-making under uncertainty. People did not appear to respond to unpredictable conditions by calculating every possible future more carefully. Instead, they reduced the cognitive effort devoted to planning and relied on simpler decision policies.
A treasure hunt designed to measure planning
Jordan Lei and colleagues at New York University recruited 300 unique adults in the United States for three online experiments, with 100 participants assigned to each experiment. Mean ages were approximately 37 to 38 years across the samples.
Participants completed a branching decision task built around treasure chests carrying point values. At each branch they chose a route through the tree with the aim of collecting as many points as possible. The structure allowed the researchers to alter one aspect of uncertainty at a time while observing both choices and the time participants spent before making their first move.
The three experiments focused on reliability, volatility and controllability. These represent distinct problems people encounter outside the laboratory. Information can be unreliable, future conditions can change after a plan has been made, and an intended action does not always produce the expected outcome.
In the reliability experiment, some mystery chests displayed values that did not reliably predict the reward participants would ultimately receive. In the volatility experiment, rewards further along the decision tree could be resampled while a participant moved through it, meaning that a distant reward considered during planning might change before it was reached.
The controllability experiment changed the relationship between choice and outcome. Participants could choose to move left or right, but the selected direction could be reversed with probabilities ranging from zero to 50%. They knew the level of stochasticity operating in each block, so the uncertainty was part of the decision environment rather than a hidden trick.
More randomness produced faster first decisions
The same broad pattern appeared in all three experiments. As the environment became more stochastic, participants took less time before making their first choice. The researchers treated this first-choice response time as an indirect measure of planning effort because the opening decision is the point at which participants have the greatest opportunity to evaluate the routes ahead.
People therefore did not simply deliberate longer when the task became less predictable. They shortened their initial decision time when visible rewards became less trustworthy, when future rewards were more likely to change, and when their chosen movement was less likely to produce the intended transition.
Performance moved with the environment as well. Participants earned more points when volatility was low than when it was high, and they performed better when reward information was more reliable or when they had greater control over where their choices took them.
This distinction matters. Faster decisions under uncertainty need not mean that participants became careless in a simple sense. If the future is genuinely unstable, the return from investing substantial cognitive effort in a detailed plan can shrink. A sophisticated plan based on information that may soon become obsolete has less value than the same plan in a stable environment.
People simplified rather than perfectly calculating uncertainty
Response times showed how effort changed, but they could not by themselves reveal the strategy behind participants’ choices. The researchers therefore compared computational cognitive models designed to capture different ways of evaluating the decision tree.
The modelling suggested that participants generally did not calculate the mathematically optimal expected value of every uncertain outcome. Instead, their behaviour was better described by a strategy the researchers call determinizing. People effectively planned as though the world were deterministic, rather than explicitly integrating all of the stochastic possibilities into each prospective calculation.
At the same time, participants became less sensitive to value differences as stochasticity increased. If two routes appeared to offer different rewards, that difference exerted less influence on behaviour in a more unpredictable environment. The authors interpret this as evidence of policy compression: a less precise, simpler decision rule that requires less cognitive effort.
The best-performing models also favoured depth filtering, in which planning depth remained relatively stable while sensitivity to value changed. In other words, the evidence was stronger for people reducing the precision or intensity of planning than for them simply cutting a fixed number of future steps from consideration.
Why planning less can make sense
Planning has a cost. It consumes time and mental resources, and those costs are worthwhile only when looking ahead meaningfully improves the decision. The experiments show how people may adapt that investment when the informational value of planning deteriorates.
Consider a highly stable journey in which travel times, connections and routes are dependable. Detailed advance planning can substantially improve the outcome. If cancellations, delays and route changes become extremely frequent, some of that planning can expire before it becomes useful. The rational response is not necessarily to stop thinking, but the optimal amount of effort may decline.
The same principle can apply to financial decisions, project planning and organisational strategy. When assumptions are stable enough to support forecasting, deeper analysis can carry considerable value. Under genuine and irreducible uncertainty, decision-makers may place greater weight on flexibility, immediate information and the ability to revise a course of action.
Importantly, the study does not show that people make optimal adjustments. The determinizing strategy is explicitly simpler than fully incorporating uncertainty into expected-value calculations. The findings instead reveal how human cognition appears to manage the trade-off between the cost of thinking and the expected benefit of thinking further ahead.
Uncertainty is not all the same
One strength of the research is that the pattern survived three conceptually different manipulations. Reliability concerns whether information about an outcome can be trusted. Volatility concerns whether the environment changes over time. Controllability concerns whether an intended action reliably produces the expected transition.
Those problems can feel similar because all make the future harder to predict, but they imply different real-world responses. Better information can address unreliability. Faster updating can help with volatility. Contingency planning may help when control over outcomes is weak. Finding a common reduction in planning effort across all three suggests that people may have a broader mechanism for adjusting cognitive investment when foresight loses value.
For South African households and businesses, the finding has an intuitive relevance in environments where electricity availability, transport conditions, prices or market assumptions can change. It should not be interpreted as evidence that uncertainty makes planning pointless. Rather, it highlights why adaptable plans and frequent revision can become more useful than ever-more-detailed forecasts when key inputs are unstable.
The laboratory cannot capture every kind of uncertainty
The experiments were deliberately controlled. All 300 participants were US adults completing online tasks, and earning points from treasure chests is far removed from consequential decisions about careers, investments, health or family life.
First-choice response time is also an indirect measure of planning effort. Faster responses are consistent with less deliberation, particularly when supported by the computational modelling, but response time cannot reveal every mental process occurring before a choice.
The authors further distinguish the irreducible randomness tested here from uncertainty that can be reduced through learning. When uncertainty reflects missing knowledge rather than inherently unpredictable outcomes, people may benefit from investing more effort to learn the environment instead of compressing their decision policy.
The study therefore does not imply that humans universally retreat from planning whenever they feel uncertain. Its narrower contribution is to show that when randomness is built into the environment itself, people consistently reduce planning effort across several different forms of unpredictability.
Planning appears to have an internal cost-benefit calculation
The findings add an important behavioural layer to the study of uncertainty. Much decision research focuses on which option people eventually choose. This work asks a prior question: how much cognitive effort do people decide to invest before choosing at all?
Across reliability, volatility and controllability, the answer was strikingly consistent. As the future became less dependable, participants made their opening decisions faster and used less value-sensitive policies. Human planning appears to be flexible not only in what it predicts, but in how much effort it spends trying to predict.
That may be an efficient response to a world in which some uncertainty cannot be eliminated. The challenge is knowing when reduced planning is sensible adaptation and when uncertainty is instead a signal to gather better information.
Source Information
Study Title: Environmental stochasticity reduces human planning effort
Authors: Jordan Lei, Jeroen Olieslagers, Nastaran Arfaei, Daisy Xinlei Lin and Wei Ji Ma
Journal: Nature Communications
Published: 28 September 2026
DOI: 10.1038/s41467-026-78023-9
Study design: Three online behavioural experiments manipulating reliability, volatility and controllability, supported by computational cognitive modelling
Sample: 300 unique adult participants in the United States, 100 per experiment
