A child completes a set of problems and does well. An adult says one of two things: "You're so clever" or "You found a good way to work through that."
Those two sentences have been compared experimentally many times, and the results are consistent enough to have entered popular consciousness — usually in a flattened form. It is worth recovering the actual finding, because the flattened version has produced a lot of confused practice.
The original result
In the classic studies, children praised for ability after a success were more likely, when subsequently given a harder task, to choose an easier option when offered the choice, to show reduced persistence, and to report lower enjoyment. Children praised for effort or strategy showed the opposite pattern on each measure.
The interpretation is that ability praise sets up an implicit contract. If success is evidence of being clever, then difficulty is evidence of not being clever. Once that inference is available, avoiding difficulty becomes a way of protecting the identity.
Process praise does not create that contract. It attaches success to something the child did, which remains available when the next task is harder.
Where the popular version goes wrong
The advice that emerged was "praise effort, not ability," and it has been applied in ways the research does not support.
Praising effort on work that was not effortful is transparent to children and reads as consolation. Studies looking at effort praise following failure find it can be interpreted as a signal that the adult has low expectations — "she thinks I'm working as hard as I can and this is the result."
The more accurate version of the finding is that praise should attach to something specific and repeatable: a strategy, a decision, a revision, a choice to try a different approach. Generic effort praise is only marginally better than generic ability praise, and in some contexts is worse.
Replication work on the broader mindset literature has also moderated the claims. Large-scale trials of brief mindset interventions have found effects that are real but small, and concentrated among lower-achieving students rather than distributed evenly. The mechanism appears to be genuine; the magnitude in popular retellings was inflated.
Why this lands differently on high-achieving girls
Two patterns intersect here, and the intersection is the interesting part.
The first is that ability praise is more damaging to students for whom it has been abundant and unbroken. A child who has been told she is clever since she was five, and who has not yet encountered work she could not do on first attempt, has an unusually large amount invested in the ability account and unusually little evidence that difficulty is survivable.
The second is that early school environments reward compliance, and girls are on average more compliant earlier. This produces a group of students who succeed early, are praised heavily, and are rarely stretched to the point of visible struggle — the exact profile that ability praise damages most.
The consequence tends to appear late: at the transition to a selective institution, or in the first genuinely hard course, or at the point where the answer stops being findable in the material provided. Teachers of first-year undergraduates describe this pattern regularly, though the systematic evidence is thinner than the anecdote.
The specific failure mode
It is not that these students give up. Typically they work harder — but they work harder at the strategy that previously succeeded, which is thoroughness. Rereading. Recopying. More hours on the same approach.
What they are less likely to do is the thing that actually resolves genuine difficulty: attempt something, be wrong publicly, ask a question that reveals confusion, or abandon a method that is not working. Each of those requires being visibly not-clever for a period, which is precisely what the ability contract forbids.
This is worth naming because it is often misread as a motivation problem. It is not. It is an extremely motivated student applying the wrong tool with great diligence.
What helps
Deliberately introducing tractable failure is the most-supported response. Work that is not solvable on first attempt, presented as normal rather than as a test, gives a student evidence that difficulty precedes competence rather than disproving it.
Making the struggle of competent people visible helps too. Students very rarely see what a working mathematician's page actually looks like, or how many drafts a published piece went through. In the absence of that, the reasonable inference from observation is that competent people simply know things.
And there is value in naming the pattern directly to a student who is in it. "You are good at this and you have never had to be bad at it first" is a sentence that a lot of high-achieving seventeen-year-olds find clarifying, because it reframes their current distress as a stage rather than a verdict.
A caveat worth keeping
None of this makes praise dangerous. Children who are told nothing do not thrive. The finding is about which of several available sentences to use, not about withholding approval.
And the effects here are modest at the level of populations. They are worth acting on because the cost of acting is close to zero, not because a change in phrasing transforms outcomes. Anyone promising the latter is selling something.