Feedback changes what people say about their performance more than what they do next
The clearest pattern across these experiments is a split between stated self-knowledge and actual behavior. In the object-comparison tasks, participants solved problems with one of two strategies — an internal one (rotating an object in the mind) or an extended one (rotating it by hand) — and then reported how well they thought each strategy had worked for them. Explicit feedback about speed and accuracy made those self-estimates more accurate [2][3]. But when participants were then free to choose which strategy to use, the better-informed estimates did not produce better choices [2][3].
That gap matters for anyone hoping that a moment of lowered confidence will steer a learner toward a smarter approach. Knowing more precisely how you performed is not the same as acting on that knowledge. The authors describe the result as increased awareness failing to guarantee improved strategy choice, and they suggest feedback may even tamper with more adaptive implicit mechanisms that ordinarily guide the choice [2][3].
Why explicit feedback can backfire on strategy choice
People do not always consciously track how well each approach is working; some of that monitoring appears to run implicitly, outside deliberate calculation [2][3]. The proposed explanation is that handing someone explicit performance numbers pulls attention toward the stated feedback and away from those quieter, possibly better-tuned cues. In that reading, the feedback does not simply add information — it can displace a process that was already doing useful work [2][3].
This is an interpretation rather than a demonstrated mechanism, and the authors themselves frame it as a possibility worth further study of how explicit and implicit performance monitoring interact [2][3]. What the experiments do show is the outcome: in this setup, explicit feedback did not support adaptive strategy choice, which is why the authors advise against rolling out such feedback prematurely [2][3].
In medical training, feedback did improve calibration — with a caveat
A different picture comes from diagnostic training. Medical interns diagnosing chest X-rays were randomly assigned to receive no feedback, feedback on whether their diagnosis was accurate, or feedback plus an explanation of why the correct diagnosis was right. Both feedback groups showed better calibration between confidence and actual accuracy on a later set of X-rays, with the explained-feedback group improving most [1].
The scope of that result is important. The feedback was about accuracy and reasoning, not a manipulation designed to make interns feel less confident, and the improvement was measured immediately afterward rather than after a delay [1]. The authors also note it is unclear whether the gain reflects better confidence estimates or simply better diagnostic accuracy, since accuracy and confidence both improved [1]. So this study supports feedback as a calibration tool in visual diagnosis, especially for cases that are not too difficult, but it does not show that deliberately lowering someone's confidence would carry those benefits forward [1].
What this means if you are designing feedback for learners
The practical lesson is to separate two goals that are easy to conflate: making people's self-estimates more accurate, and helping them choose better ways to work. Feedback appears reasonably good at the first and unreliable at the second [2][3]. If the aim is better strategy selection, the evidence here gives no assurance that a confidence-damping feedback message will deliver it, and raises the possibility that explicit feedback could get in the way [2][3].
For calibration itself, the medical-intern experiment offers more encouragement, but only for feedback about accuracy and reasoning in a diagnostic task, measured right after training [1]. Whether a temporary dip in confidence would produce lasting calibration gains, or change how people study later, is simply not tested by these studies — the question remains open rather than answered in either direction.
Sources used in this answer
Impact of performance and information feedback on medical interns' confidence–accuracy calibration
In 125 medical interns diagnosing chest X-rays, both accuracy feedback and accuracy-plus-explanation feedback improved confidence–accuracy calibration on a subsequent no-feedback test, though it is unclear whether this reflects better confidence estimates or better diagnostic accuracy.
When feedback backfires – Explicit performance feedback can impair cognitive strategy choice
Explicit trialwise or blockwise performance feedback improved participants' explicit estimates of how well two object-comparison strategies worked but did not improve, and may have harmed, their subsequent strategy choices.
When feedback backfires: Knowledge of results can impair cognitive strategy choice
Written feedback on speed and accuracy improved participants' explicit performance estimates for two object-comparison strategies without guaranteeing better strategy choice, and attending to it may have interfered with more adaptive ways of informing the choice.