Kindness enriches rather than depletes. It is one of those claims that almost nobody disputes and almost nobody measures, which is an odd combination for a sentence repeated as often as this one.
It has been measured, twice, by two unrelated groups asking different questions. One asked how much the person performing a kind act gains from it. The other asked how accurately that person judges what the act is worth to the person receiving it. Put side by side, the answers suggest the claim is true, that the return is smaller than the phrasing implies, and that the part we get most wrong is not on our side of the exchange at all.
Putting a number on the return
In 2018 a team led by Curry published a systematic review and meta-analysis in the Journal of Experimental Social Psychology, pooling 27 studies covering 4,045 participants that had tested whether performing acts of kindness improves the wellbeing of the person performing them.
The Artful Age
A weekly letter on aging well, family across generations, and the creative life after the kids leave home.
It does. Their pooled estimate was a delta of 0.28, which they describe as small to medium, and which they translate into something a reader can actually picture: roughly a 0.6-point increase on a 0 to 10 happiness scale.
So the direction everyone assumes is the direction the evidence shows. The magnitude is the part that never makes it into the slogan. Half a point, give or take, on a ten-point scale. Worth having. Not a transformation.
Reasons to trust it, and reasons to hold it loosely
The authors are unusually candid about their own result, in both directions.
Against it: many of the pooled studies were under-powered, and the samples lean heavily on Western students, which is a recurring problem in this corner of psychology rather than a flaw unique to this paper. The review also did not systematically vary who the kindness was aimed at, so whether it matters more to help a friend, a colleague or a stranger is an open question here, not a settled one.
In its favor: the result does not look like an artifact of selective publication. A file-drawer analysis indicated that it would take 1,919 unpublished or unretrieved null studies to drag the average effect down to non-significance, and a formal test for funnel asymmetry did not flag a problem. That is worth stating plainly, because it would be easy to quote only the limitations.
Why it rarely feels like half a point
If performing kindness reliably lifts the performer, why is it so easy to talk yourself out of? One answer is that the arithmetic most people run is not about their own return at all. It is about whether the act will matter to the other person, and on that question we appear to be reliably wrong.
Kumar and Epley ran eight experiments, published in 2022 in the Journal of Experimental Psychology: General, in which one person performed an ordinary kind act for another and both sides then rated it. The acts were deliberately unremarkable: a hot chocolate handed to someone at a skating rink, a coffee, a greeting card, flowers, baked goods, a lift, a handwritten note.
Performers rated how big a deal their act had been. Recipients rated the same act. Across the first four experiments, which paired 66, 55, 75 and 49 sets of people, the performers gave 3.96, 4.23, 3.76 and 2.67. The recipients gave 6.03, 6.01, 7.00 and 7.27.
Mood showed the same shape. Performers predicted how good the recipient would feel, on a scale running from minus five to plus five, and estimated 2.15, 1.96, 2.73 and 2.08. The recipients reported 2.98, 2.24, 3.52 and 3.47. Every one of those differences was statistically significant.
It is worth being precise about what this does and does not show. The performers were not blind to the effect. They knew the coffee would land well, and they predicted a positive mood in every case. They were wrong about how positive, every time, in the same direction.
What a systematic underestimate costs
A benefit you underestimate is a benefit you under-supply. If you think the note is worth a four and it is in fact worth a seven, the decision about whether to send it is being made with the wrong figure in the equation. That is a more useful thing to know than the reassurance that kindness is good for you, because it names a specific error rather than a general virtue.
Kumar and Epley are equally careful about the edges of their own work. Response rates in their field studies were imperfect, raising the possibility of selection effects, which they addressed by running laboratory versions where everybody responded. The mechanism behind the miscalibration is unresolved, and they say as much. Acts that are obligatory rather than chosen may behave differently. And their samples were drawn largely from United States university and community populations, so how far any of it travels is unknown.
Where the claim stops
All of this concerns discrete acts. A coffee, a card, an afternoon of help. The meta-analysis pooled studies of acts, and the experiments staged single gestures between two people.
Sustained caregiving is a different subject, and nothing above says it cannot exhaust the person doing it. “Kindness enriches, not depletes” is a defensible statement about the small, optional, repeatable kind. Stretched past that, it turns into an instruction to feel guilty about being tired, which is not a finding and should not be presented as one.
Within its proper limits, then, the claim holds, with the numbers attached. Performing a kind act tends to lift the person performing it, by something on the order of half a point out of ten. And the thing you are most likely to be mistaken about is not what you get back. It is what they got.