An anxious person finishes a test question. They know the answer. Then they sit there, reconsidering. Thirty seconds later, they’re less sure. A minute after that, they’re convinced they’ve failed. The extra time didn’t help them think more clearly. It gave their brain more runway to catastrophise.
Women doing the same task show a different pattern. Initial doubt, yes. But the longer they sit with the decision, the more that doubt fades. Same hesitation on the surface. Opposite direction underneath.
University College London researchers have spent four experiments working out why, and the answer isn’t about personality or socialisation. It’s computational. The two groups are running different programmes in their heads, both producing underconfidence, neither producing it the same way.
What Happens in the Gap
Most psychology experiments treat the pause between making a choice and reporting confidence as dead time. Dr Sucharit Katyal’s team tracked it obsessively. They ran 1,447 people through simple perceptual tasks—more red dots or blue dots, that kind of thing—then measured not just confidence ratings but how long people sat before giving them.
For high-anxiety participants, longer pauses meant lower confidence. For women, the reverse.
“We identified two different types of underconfidence,” Katyal says. Same output, different engine.
Negative Drift
The researchers used drift-diffusion modelling, borrowed from neuroscience, to simulate how evidence accumulates in the brain. In most people, once you’ve made a decision, the accumulation stops. The verdict’s in.
Not for anxious participants. Their brains kept piling up evidence after the task ended, except the new evidence was contaminated. Biased. Doom-flavoured. The technical term is “negative drift,” which undersells how brutal it is: your mind continuing to argue against you even when there’s nothing left to argue about.
The longer an anxious person reflected, the more their metacognitive efficiency collapsed. That’s the term for how well your gut sense of performance matches reality. Anxiety turned the gut sense into static.
Thresholds, Not Spirals
Women showed no negative drift at all. Doubt spiked immediately after a decision, then declined. No rumination. No invented catastrophe. Just a stricter threshold for what counts as “confident.”
The model suggests women aren’t misjudging their performance. They’re applying a harsher conversion rate when translating “I think I got this” into a number. Extra time gave them space to check the actual evidence rather than defer to the reflex.
Critically, women without anxiety had normal metacognitive efficiency. Cautious, not miscalibrated.
“By revealing the mechanisms behind these biases, we may be able to design targeted interventions,” Professor Steve Fleming, a co-author at UCL, says. “For example, helping anxious individuals interrupt the accumulation of negative self-evaluations.”
The practical conclusion cuts both ways. Anxious people might do better trusting their first answer before the spiral starts. Women—or anyone with threshold-based doubt—could benefit from deliberate slowdown. One visible problem, two invisible machines. In a hiring decision or an operating theatre, knowing which doubt you’re dealing with might matter more than the doubt itself.
DOI: https://doi.org/10.1017/S0033291725102808
If our reporting has informed or inspired you, please consider making a donation. Every contribution, no matter the size, empowers us to continue delivering accurate, engaging, and trustworthy science and medical news. Independent journalism requires time, effort, and resources—your support ensures we can keep uncovering the stories that matter most to you.
Join us in making knowledge accessible and impactful. Thank you for standing with us!
