A cup of hot chocolate is not a life-changing gift. A cupcake is not a grand sacrifice. That is precisely why a set of experiments on kindness caught my attention.
Across ordinary acts such as giving food to a stranger, writing a note to someone and helping a colleague, the people doing the kind thing understood that recipients would feel good. They just did not expect them to feel as good as they actually did.
The finding is less sentimental than it sounds. It is about a measurable error in social prediction: givers tended to evaluate the object or effort, while recipients also registered the human warmth behind it.
That gap could help explain why a person with a kind impulse sometimes talks themselves out of acting on it.
The research compared predictions with recipients’ reports
In the paper, published in the Journal of Experimental Psychology: General, Amit Kumar and Nicholas Epley reported eight experiments conducted in field and laboratory settings.
The key design was simple. A participant performed an act of kindness and predicted the recipient’s reaction. The recipient then reported their actual reaction. That separation matters because it allowed the researchers to compare one person’s forecast with another person’s experience rather than asking givers whether they believed they had done something nice.
In one field experiment at Chicago’s Maggie Daley Park, 84 people were offered hot chocolate. Seventy-five chose to give it to a stranger. On a scale running from minus five to plus five, givers predicted that recipients’ moods would average 2.7. The recipients’ average report was 3.5.
The same general mismatch appeared among MBA students who chose their own acts. These included paying for someone’s lunch and going beyond what was required to help a colleague. Performers estimated recipients’ moods at an average of 2.32, while recipients reported 3.55.
Those numbers should not be treated as universal constants. Some studies depended on recipients answering follow-up messages, and the experiments were conducted within a limited set of American contexts. What they show is a recurring prediction error within this research program, not that every gesture will be welcomed by every person.
The giver saw the cupcake; the recipient also saw the intention
A later experiment helps explain the mismatch. In a Chicago park, some people received a cupcake as compensation from the researchers. Others received the same cupcake because a stranger had chosen to give it to them.
Recipients were happier when the cupcake came through a stranger’s kindness. Yet the people making predictions did not fully account for that difference. They focused more on the thing being transferred, even though the recipients appeared to gain additional value from what the transfer communicated.
That distinction rings true to me. Before offering something small, I can get caught up in whether it is useful enough, generous enough or phrased well enough. The other person may be noticing a more basic message: somebody thought of me.
The researchers describe this as a difference between competence and warmth. Givers worry about the quality of their social performance. Recipients are comparatively attentive to the care behind it. The experiments do not show that recipients ignore quality, only that givers can underweight warmth when imagining how the act will land.
One act changed a later decision involving a stranger
The paper also tested whether the effect stopped at a pleasant self-report. In a laboratory experiment, participants received an item either from the researchers or as a gift chosen by another participant. They then decided how to divide $100 between themselves and an unknown person.
People who had received the item through another person’s kindness gave the stranger an average of $48.02. Those who received it from the lab gave $41.20. The people predicting recipients’ behavior did not anticipate the full difference.
This is evidence of indirect reciprocity within one controlled task. It does not prove that a coffee bought today will set off an endless chain of generosity tomorrow. Still, it strengthens the result because the effect appeared in behavior involving a third person, not only in a mood rating supplied to the researchers.
The undervaluation appeared early, but this is an extension rather than an independent replication
A 2023 follow-up asked whether adults learn this prediction error or whether it is visible earlier in life. Across two experiments involving children ages four to seven, children ages eight to 17, and adults, people performed the same random act of kindness for somebody else.
Across ages, givers underestimated both how big the act would seem and how positive it would make the recipient feel. That suggests the mismatch can emerge quite early.
There is an important qualification. Nicholas Epley was an author of both papers, so the child research is a useful extension of the original program, not a fully independent replication. Different research teams, cultures and kinds of relationship would give us a better sense of how broad the pattern really is.
Kindness is not a treatment plan or a duty to ignore boundaries
Research on kindness often gets flattened into the promise that being helpful will make the giver happier. An independent 2018 systematic review and meta-analysis provides a more measured picture. It combined 27 experimental studies with 4,045 participants and found a small-to-medium average benefit to the giver’s wellbeing, with an effect size of 0.28.
That average does not make kindness a mental-health treatment, and it does not tell any one person how they will feel after giving. The review’s authors called for more specific theories about which kinds of kindness help, for whom and under what conditions.
Context matters on the receiving side too. Unrequested help can feel intrusive. A gift can create obligation. A workplace request dressed up as kindness can conceal unequal power. Warm intention does not erase another person’s boundaries.
The safest interpretation is not “always do more for everyone.” It is that our hesitation may sometimes be based on an overly narrow estimate of what a sincere, low-pressure gesture communicates.
A modest correction to our social forecasts
Earlier today, I wrote about evidence that people underestimate kindness in their communities. This experimental work points to a similar error at the scale of a single encounter.
We can see the cost of a coffee, the size of a cupcake or the awkwardness of sending a message. It is harder to see, in advance, the warmth another person may take from the fact that we noticed them at all.
I do not think the research asks us to become relentlessly cheerful or generous beyond our means. It offers a smaller and more humane correction: when a respectful kind act feels too minor to matter, our first estimate may be too low.
