A recent study published in the journal Psychological Science suggests that a single dose of testosterone does not cause men to abandon deliberate reasoning in favor of gut instincts. In fact, contrary to previous assumptions, the hormone might actually lower a person’s confidence in their own cognitive performance. These findings provide evidence that short-term spikes in testosterone do not significantly alter this specific type of human decision-making.
The human brain can solve problems using fast, intuitive thinking or slow, deliberate reasoning. Psychologists often measure this tendency using the Cognitive Reflection Test, an assessment that presents questions with an obvious but mathematically incorrect intuitive answer. Arriving at the correct solution requires pausing and overriding that initial gut reaction.
Testosterone is a sex steroid hormone that responds dynamically to social and environmental changes. In both animals and humans, testosterone levels can rise rapidly in competitive or mating contexts. This environmental reactivity forms the basis of the challenge hypothesis, a biological theory proposing that testosterone modulates behavior in a context-sensitive way to facilitate fast-acting, instinctive actions under social threat.
Translating this biological theory to human cognition led to a specific hypothesis regarding decision-making. Past researchers proposed that transient testosterone elevations might shift judgments toward intuitive responses. An influential 2017 paper reported exactly this effect, finding that testosterone administration significantly reduced scores on the Cognitive Reflection Test.
Subsequent follow-up studies produced inconsistent or completely null results, creating confusion in the scientific community. To settle the debate, scientists who authored the conflicting papers teamed up to design an adversarial collaboration. This is a research project where scientists with opposing views work together to design a definitive experiment.
Gideon Nave, the Carlos and Rosa de la Cruz associate professor of marketing at the Wharton School of the University of Pennsylvania, helped organize this joint effort. “The original idea came from evolutionary biology and neuroscience,” Nave said. “Testosterone rises in situations such as competition and mating, where acting quickly and instinctively can be more adaptive than careful deliberation, so we hypothesized that a temporary increase in testosterone might reduce cognitive reflection by making people more likely to rely on intuitive responses.”
“Our original 2017 experiment supported that idea, but later studies produced different results,” Nave added. “Rather than arguing over which study was right, the different research teams decided to work together in an adversarial collaboration. We have always had a collegial relationship, despite reaching different conclusions, and jointly designed the largest and most rigorous study we could conduct to settle the question once and for all.”
The team preregistered their methods, meaning they published their exact plan and hypotheses before collecting any data. They recruited 1,000 adult men between the ages of 18 and 45. The study used a double-blind, randomized, placebo-controlled design. This means neither the participants nor the research assistants handing out the doses knew who received the actual hormone and who received an inactive substitute.
Half of the participants self-administered an intranasal testosterone gel, which delivers 11 milligrams of the hormone directly into the nose. The other 500 men received a placebo gel that looked and felt identical. Following a 30-minute waiting period, which allows the hormone to enter the bloodstream and reach the high-normal physiological range, the men completed a variety of behavioral tasks.
The main task of interest was a modified seven-item version of the Cognitive Reflection Test. The authors used a reworded version of the classic test to ensure participants had not seen the specific puzzles before on the internet. To keep participants motivated, the scientists offered a one-dollar cash bonus for every correct answer. Following the test, the men guessed how many questions they answered correctly out of seven.
The test scores failed to support the original 2017 hypothesis. The researchers found no statistically significant difference in test performance between the testosterone group and the placebo group. The men who received testosterone answered an average of 3.01 questions correctly, compared to 2.86 in the placebo group.
When analyzing the self-reported confidence ratings, the researchers found another unexpected result. The team had originally predicted that testosterone would make the men more overconfident in their performance. Instead, the data showed a significant decrease in self-confidence among those who received the hormone. The men in the testosterone group estimated they got fewer questions right compared to the estimates made by the placebo group.
“Not only did we fail to replicate our original finding, but the estimated effect actually pointed in the opposite direction,” Nave told PsyPost. “We also found evidence that testosterone reduced confidence rather than increased it, which runs counter to many existing theories. Because this finding was unexpected, we interpret it cautiously and believe it should be independently replicated before drawing strong conclusions.”
The researchers also explored whether the treatment changed how participants viewed their abilities relative to others. They asked the participants to guess how many questions the average participant in the study got right. They then calculated a relative confidence score by subtracting the participant’s guess about the average person’s score from their own self-estimated score. The data showed that both groups overestimated their performance compared to others, and the testosterone treatment did not significantly alter this gap.
In addition to the main experiment, the authors ran several exploratory analyses based on the participants’ natural hormone levels. They defined baseline testosterone as the naturally occurring hormone level found in the very first saliva sample. The researchers found a negative association between natural, baseline testosterone and test performance. Men with naturally higher baseline testosterone tended to score slightly lower on the cognitive reflection test.
The scientists also looked at cortisol, a hormone closely associated with stress. They found suggestive evidence of a positive interaction between baseline cortisol levels and the testosterone treatment on test performance. Additionally, they tested whether a person’s natural impulsivity changed how they reacted to the hormone dose. An earlier study suggested that testosterone might impair performance in low-impulsivity men, but the current massive sample found no evidence of any interaction between personality traits and the treatment.
Readers should be aware of a few limitations when interpreting these findings. The study only included adult men, which restricts the ability to generalize the results to the broader public. Men and women can process and react to sex steroid hormones differently. Future research needs to include female participants to see if these null results hold true across different populations.
The exact timing and dosage of the hormone administration also present limitations. The researchers waited 30 minutes after the nasal gel administration before starting the behavioral tasks. Although previous evidence favors a rapid timescale for testosterone’s influence on behavior, it remains possible that the hormone influences cognitive reflection over a much longer time frame.
The study only addresses short-term, sudden spikes in hormone levels. Administering testosterone over a longer period of time might influence cognitive reflection differently. Chronic exposure can lead to structural neural changes that differ entirely from those observed in a single-dose experiment.
“Our study examined the effects of a single, short term increase in testosterone in healthy men,” Nave told PsyPost. “It does not address the effects of long term testosterone exposure, hormone replacement therapy, or testosterone during critical developmental periods such as puberty, all of which could have different consequences but are much more difficult, and often impossible, to study experimentally.”
Looking forward, the researchers plan to continue exploring the boundaries of behavioral endocrinology. “One important direction is understanding when hormones do influence cognition and behavior and when they do not,” Nave said. “Our findings suggest that a temporary increase in testosterone does not meaningfully change people’s ability to override intuitive responses and engage in deliberate reasoning.”
The team hopes their methodological approach will inspire other scientists facing contradictory data. Nave hopes this project encourages more adversarial collaborations in behavioral science. “They require researchers to put aside attachment to their previous findings and focus on designing the strongest possible test of a scientific question, which ultimately benefits the entire field,” he said.
“For me, one of the most important aspects of this project is not just the result but the process,” Nave said. “This study brought together researchers who had previously published conflicting findings, yet we shared a common goal of finding the correct answer rather than defending our earlier work.”
“We agreed on the design and analysis plan before looking at the data, and I think this kind of collaboration provides a model for how science can efficiently resolve important disagreements,” Nave added. “More broadly, I hope people see this study as an example of science working as it should. Different laboratories sometimes obtain different results, and the best way to resolve those disagreements is not through debate, but by working together on larger and more rigorous studies that everyone agrees on in advance.”
The study, “Does Testosterone Affect Cognitive Reflection? Evidence From a Double-Blind, Randomized Controlled Study of 1,000 Participants,” was authored by Erik L. Knight, Gideon Nave, Steven D. Shaw, Coren Apicella, Pierre L. Bonin, Anna Dreber, Shawn N. Geniole, Magnus Johannesson, Dylan Manfredi, Pranjal Mehta, Valentina Proietti, Steven J. Stanton, Francesca R. Luberti, Triana Ortiz, and Justin M. Carré.