"You completed this block in 91.3 seconds. You were 18.3 seconds faster than the average."
That second sentence was a lie. Every participant who saw it was told they were beating the pack, and the margin was calculated from their own time so the flattery would land with roughly equal force for everyone. It is a standard trick in motor learning research, and the theory behind it is straightforward: tell people they are doing well, and their brains release dopamine, and the dopamine helps the new skill stick.
Allison Lewis, Rachel Bohnenkamp, and Jill Campbell Stewart of the University of South Carolina wanted to know whether that trick works equally well for everyone. Their answer, based on 52 adults and a saliva sample from each: probably not.
The task was deliberately dull. Participants sat at a desk, gripped a joystick with their right hand, and steered a cursor into small circles that popped up one at a time in twelve possible spots on a laptop screen. Hold the cursor inside a circle for half a second and it counted as a hit, then the next one appeared. Hidden in the stream was a repeating eight-target sequence, shuffled in among random ones so that nobody would consciously notice it. Almost nobody did. Only 4 of 60 participants could reproduce any part of the pattern afterward.
On day one, participants ran 28 blocks of practice, and after every block a line of text appeared. Half the group got their time and nothing else. The other half got their time plus the news that they had outpaced everyone else. On day two they came back, ran the task again with no feedback at all, and the researchers measured how much faster they had gotten. That overnight change is the standard measure of learning, as opposed to the temporary sharpness that comes from being warmed up.
The genetic split
After the second session, everyone spat into a tube. A third-party lab, blind to the study's aims and to how anyone had performed, genotyped four spots in the DNA linked to dopamine: three dopamine receptors (DRD1, DRD2, DRD3) and COMT, the enzyme that clears dopamine out of the synapse. Each spot earned a score from 0 to 2 depending on whether the participant's version pushes dopamine signaling up or down. Added together, that gave every person a single number from 0 to 8. The team then split the sample into a Low group (1 to 4) and a High group (5 to 8).
Everyone learned. Response times dropped substantially overnight across the board, which is what you would expect from 28 blocks of practice. But the improvement depended on the combination of genotype and feedback, in a three-way statistical interaction that cleared the usual threshold with a little room to spare (p = 0.043).
The pattern sits in the Low dopamine group. Those who got plain timing feedback sped up by about 1.8 seconds on the repeated sequence. Those who got the flattery sped up by about 1.1 seconds. The High dopamine group improved by similar amounts either way; feedback type made no detectable difference to them.
So the praise did not merely fail to help the low-dopamine participants. In this sample, they did worse with it than without it. That runs against the intuition the researchers started with, and against earlier work in which exercise or L-Dopa medication helped exactly this group. Lewis and colleagues offer a possible reason, and they are explicit that it is speculation: drugs and exercise flood the system with dopamine slowly and broadly, while a congratulatory message probably triggers a brief spike. A system with fewer receptors or faster dopamine clearance may simply miss the spike. The study measured no brain chemistry at all, so this remains a story rather than a finding.
Why it matters
Encouragement as a training tool has a shaky evidence base. A 2022 meta-analysis found that only about half of the studies testing positive social comparative feedback saw it improve motor learning, and the rest saw nothing. The usual explanations point at methodology. This work raises another possibility: that the technique genuinely works for some people and not others, and that studies mixing both kinds of participant will average out to noise.
That matters most where motor learning is the actual job, in physical rehabilitation after a stroke or an injury, where therapists routinely use encouragement as part of practice. If the benefit really is genotype-dependent, a technique that is free and harmless might still be the wrong choice for some patients.
The caveats are real and the authors list them plainly. Fifty-two people is a small sample, powered only to catch moderate effects. The participants were young, right-handed, and healthy. The absolute differences were fractions of a second on a laboratory task with no established threshold for what counts as a meaningful change, and the four-gene score is a crude stand-in for a system involving many more genes than that. When the team looked at each gene on its own, nothing reached significance. The authors call the result preliminary and ask for larger samples. That is the right reading.