Two things said to a child who has finished a drawing.
The first: you are so good at drawing. The second: you have made the shadow darker on the side away from the window, so it looks like it is really sitting there.
Both are warm. Both are meant kindly. Only the second contains information, and the difference between them turns out to matter more than it looks.
What each one does
Praise evaluates. It delivers a verdict, usually about the person. It tells the child how they are seen. It feels good, it is socially important, and it is a large part of how affection is expressed.
Feedback informs. It describes what was done, how it relates to the goal, and what might be done next. It is about the work rather than the worker.
These are not opposites, and this is not an argument for withholding warmth. It is an argument for noticing that they do different jobs and that only one of them is much use for learning.
The problem shows up at failure
While things go well, evaluative praise costs little. The difficulty appears at the first real setback.
A child whose success has been explained by what they are has an explanation available for failure: they are not, after all, that thing. That is a conclusion about identity, and there is nothing to be done about it.
A child whose success has been described in terms of what they did has a different account available: something in the approach did not work here. That is a conclusion about method, and methods can change.
This is the intuition behind the enormous literature on mindset, and the intuition survives even though the specific research programme has had a difficult time, which is worth being clear about.
What happened to growth mindset
The idea that beliefs about the fixedness of ability shape response to difficulty was compelling, well presented and rapidly adopted. Schools bought programmes. Posters went up.
Then came the larger trials. Effects turned out to be much smaller than the early work suggested, sometimes not distinguishable from zero, and where anything was found it tended to be concentrated in specific groups rather than general. Meta analyses have been contested in both directions.
The lesson is not that the underlying observation is worthless. It is that a modest and conditional effect was converted into a school improvement strategy, and the strategy could not carry the weight. This is a recurring pattern, described in replication in plain words.
Meanwhile the finding that has held up better is the plainer one: feedback that tells a learner where they are relative to a goal and what to try next is among the more consistently valuable things a teacher does. It appears near the top of most evidence syntheses, including the Education Endowment Foundation toolkit, with the caveat that feedback can also be harmful when done badly.
Feedback done badly
Worth stating, because the assumption that more feedback is better is wrong.
Feedback directed at the person rather than the work turns into evaluation regardless of wording. Feedback that arrives with no opportunity to act on it is merely a verdict delivered late. Feedback that overwhelms, listing everything at once, exceeds the workspace described in working memory in plain words. And feedback that removes the difficulty by supplying the answer ends the learning it was meant to support.
Marking that reaches a child after they have moved on is the commonest institutional version, and it is a large expenditure of teacher time for a small return.
The trouble with constant praise
Praise that arrives regardless of what happened stops carrying information, and children detect this earlier than adults assume. By school age, many children read effusive praise for an easy task as a signal that low expectations are held of them, which is the opposite of the intended effect.
There is also a subtler issue. A child praised heavily for outcomes tends to protect the outcome, which means choosing tasks they will certainly succeed at. Since learning happens at the edge of ability, a strategy of guaranteed success is a strategy of not learning much. This does not require a mindset theory. It follows from what is being rewarded.
What this is not
It is not an instruction to stop telling children you are pleased with them, or to convert every domestic exchange into an assessment. A household is not a classroom and should not be run as one.
It is a distinction worth having available. When it matters, and it matters most when something has gone wrong, describing what happened gives a child more to work with than delivering a verdict does. That is not a technique. It is a description of the difference between two things we usually call by the same name.
For the mechanism that makes overloaded feedback fail, see working memory in plain words, and for the general problem of programmes outrunning their evidence, why the parenting press gets it wrong.