Scarcity is often described as doing more than limiting what people can afford. One influential psychological account argues that having too little of a resource changes attention itself, drawing people into a narrow “tunnel” around immediate needs while pushing other useful information out of view. The idea has become an important way of explaining why people facing financial pressure might focus intensely on urgent problems yet make choices that carry costs later.
A new peer-reviewed study in Nature Communications challenges a central piece of evidence for that explanation. By reanalysing a landmark scarcity experiment, examining a much larger replication dataset, and running a new conceptual replication, Fiona tho Pesch, Antonia F. Langenhoff and Mahesh Srinivasan found that key behavioural differences previously attributed to an attentional shift could instead be explained by the structure of the experimental game.
The distinction matters. The researchers are not arguing that scarcity has no psychological consequences, nor that poverty does not affect cognition or decision-making. Their narrower claim is methodological: the particular behavioural pattern that helped establish the “tunneling” hypothesis does not, on closer inspection, demonstrate that scarcity caused people to neglect information outside an attentional tunnel.
Revisiting a highly influential experiment
The new paper revisits research published in 2012 in which participants played resource-management games under experimentally created conditions of scarcity or abundance. One key experiment, known as “Angry Blueberries”, gave participants a limited number of shots to hit targets. Participants assigned to a resource-poor condition received three shots per level, while resource-rich participants received 15.
The original results were striking. Resource-poor participants spent longer aiming and earned more points per shot. When borrowing from future rounds was permitted at a high cost, they also borrowed relatively more and ultimately performed worse. These findings were interpreted as two sides of attentional tunneling: scarcity supposedly increased focus on the scarce resource, improving efficiency, while simultaneously encouraging neglect of future borrowing costs.
The new authors first reproduced those behavioural patterns. Their disagreement concerns what produced them.
The apparent performance advantage changed when shots were compared fairly
In the original dataset, 40 resource-poor participants and 28 resource-rich participants in the no-borrowing condition differed in how many shots they were allowed to take. Resource-poor participants did spend longer aiming and accumulated more points per shot across the game. But the new analysis asked whether their longer aiming actually made the comparable shots more productive.
It did not provide convincing evidence that it did. The original study was too small to establish statistical equivalence on the first three shots, but a later large-scale replication was much more informative. That replication originally included 1,010 participants, with 974 retained after exclusions. In those data, performance on the first three shots was meaningfully equivalent between resource-rich and resource-poor participants, with equivalence tests reaching p values below 0.005.
The game itself provides an alternative explanation for the overall points-per-shot difference. Early shots were simply more profitable. Across conditions, participants averaged 2.81 points on the first shot of a level and 1.80 by the third. Resource-rich participants, who could continue shooting, averaged only 0.69 points on shot 15.
Resource-poor participants therefore had a structural advantage on the points-per-shot measure: the three-shot restriction prevented them from taking the later, lower-return shots that pulled down the resource-rich group’s average. In the no-borrowing condition, resource-poor participants used an average of 2.51 shots per level, compared with 5.67 among resource-rich participants.
That does not prove the longer aiming time had nothing to do with attention. It does mean that longer aiming cannot be treated as evidence of beneficial focus merely because the resource-poor group ultimately recorded a higher points-per-shot average. The apparent efficiency advantage can arise from the rules of the task.
Borrowing opportunities were radically unequal
The borrowing result produced an even clearer structural problem. Participants could borrow only after exhausting their available shots while targets remained. Because resource-poor participants started with just three shots, they reached that decision far more often.
They encountered an opportunity to borrow on about 45% of levels. Resource-rich participants, with 15 shots available, encountered one on only about 2%.
That difference changes the interpretation of total borrowing. If one group is repeatedly offered a costly choice and another group almost never sees it, greater aggregate use of that option does not by itself demonstrate that the first group paid less attention to its cost. The reanalysis found that when opportunities to borrow were taken into account, the evidence did not support the claim that resource-poor participants were more prone to borrowing because they neglected the future cost.
This is important because the original behavioural result has often been discussed as an illustration of how scarcity itself can promote counterproductive borrowing. The new study argues that the experimental design confounded scarcity with differences in the decision environment.
A new experiment tested attention with equal opportunities
The researchers did not rely only on reanalysis. They also ran a preregistered conceptual replication designed to give resource-rich and resource-poor participants equal exposure to information about future consequences. The experiment recruited 174 participants, short of the preregistered target of 200 because of budget constraints but above the authors’ stated requirement of 90 for detecting the central effect at 80% power.
Instead of using unequal borrowing opportunities as the crucial test, participants repeatedly faced insurance decisions. Some insurance options had positive expected value, while deliberately poor “scam” options cost more than their expected return. If scarcity caused people to neglect information about future benefits, resource-poor participants should have been less likely to recognise and purchase beneficial insurance.
That prediction was not supported. The mean likelihood of purchasing positive-value insurance was 4.74 among resource-poor participants and 4.92 among resource-rich participants. The difference was not significant, F(1,172) = 0.17, p = 0.682, η² = 0.001. Equivalence testing also supported the conclusion that the groups were statistically equivalent on this outcome.
The result for bad insurance moved in the opposite direction to a simple neglect account. Resource-poor participants were significantly less likely to purchase negative-value scam insurance than resource-rich participants, F(1,172) = 8.24, p = 0.005, η² = 0.046.
The experiment nevertheless reproduced the broader behavioural pattern seen in the earlier work. Resource-poor participants took significantly longer to aim, F(1,172) = 10.93, p = 0.001, and achieved more points per shot across the game, F(1,172) = 21.95, p < 0.001. Yet once again, there was no meaningful difference between groups on the first three comparable shots, with all conventional tests yielding p values above 0.439 and equivalence tests supporting equivalence on each of those shots.
Later shots also became progressively less profitable for everyone. The estimated shot-position coefficient was β = -0.29, t = -34.28, p < 0.001. The resource-poor participants’ higher overall efficiency therefore emerged even though their performance on directly comparable early shots did not improve.
What the findings do and do not overturn
The study challenges a specific evidentiary foundation, not the broader observation that scarcity can influence behaviour. Research has reported effects consistent with increased attention to scarce resources, including memory for prices, resistance to framing effects and faster detection of scarcity-related words. Other work has examined cognitive load under financial pressure.
What remains less firmly demonstrated, according to the authors, is the full tunneling mechanism: scarcity should both intensify attention to immediate resource management and cause neglect of other potentially beneficial information. Showing only greater focus does not establish the second half of that process.
The new findings also illustrate a broader challenge in behavioural research. Experimental manipulations can alter not only participants’ internal psychological states but also the choices physically available to them. When those two changes occur together, behaviour attributed to a mindset may partly reflect the structure of the decision environment.
That point has practical relevance. Policies based on behavioural explanations of poverty can look very different from policies based on structural explanations. If apparently short-sighted choices arise partly because people with fewer resources encounter costly decisions more frequently, changing the environment in which those decisions occur may matter alongside interventions aimed at attention or cognition.
Important limits remain
The authors emphasise several limitations. The Angry Blueberries task is an artificial laboratory game, so neither the original experiments nor the new replication can reproduce the complexity of real poverty. The original study and subsequent replication also used non-representative samples, while the conceptual replication relied on participants recruited through Amazon Mechanical Turk.
The original dataset contained 68 participants, making some comparisons underpowered. The much larger replication helps address that problem, but the new conceptual replication itself stopped at 174 rather than the preregistered 200 participants because of budget constraints.
The researchers also deviated from their preregistered analysis plan. They had planned multilevel analyses but concluded that this would give unequal observational weight to participants depending on how many shots they used, so they instead adopted the analytic approach used in the original work. The change is transparently reported, but it remains relevant when evaluating the study.
Most importantly, the results should not be translated into the claim that scarcity never narrows attention. The evidence shows that the behavioural effects in this influential paradigm can be generated without demonstrating the proposed attentional mechanism. Establishing whether real-world scarcity reliably creates attentional neglect will require designs that manipulate scarcity without simultaneously giving resource-poor and resource-rich groups fundamentally different decision opportunities.
Why the reassessment matters
Scientific ideas often become influential because a memorable experiment appears to make an invisible psychological process visible. That strength can become a weakness if the behavioural measure has more than one plausible explanation.
Here, higher efficiency could be produced by restricting one group to the most profitable shots, while greater total borrowing could be produced by exposing that same group to borrowing decisions far more often. Once those structural features are separated from the proposed psychological mechanism, the case for attentional tunneling becomes substantially less certain.
The study therefore offers a useful correction rather than a simple reversal. Scarcity remains capable of shaping choices, and structural scarcity itself remains consequential. What researchers cannot safely infer from these experiments is that the observed choices necessarily reveal a scarcity-induced attentional tunnel. The next generation of studies will need to measure attention more directly and hold the surrounding decision structure constant if that mechanism is to be established.
Source Information
Study: Revisiting evidence for an attentional tunneling effect caused by scarcity
Authors: Fiona tho Pesch, Antonia F. Langenhoff and Mahesh Srinivasan
Journal: Nature Communications, volume 17, article 10233
Published: 26 September 2026
DOI: 10.1038/s41467-026-78074-y
