Three groups of healthy older adults sat at home computers and listened to a sound that did nothing.
One group had been told the sound might sharpen their attention. One had been told it might blunt it. The third was told nothing either way. Then all of them repeated an attention task they had done minutes earlier.
The group told nothing improved. The group told the sound would help improved too. People usually do better the second time, the way you find the light switch faster on your second night in a rented house. The group told the sound would hurt stayed flat.
That experiment, run through the Prolific platform with adults aged 65 to 80 and published in Psychological Research on 15 May 2026, is a randomized trial of a single afternoon, with fewer than thirty people in each group in each analysis. Of 204 people recruited, 177 finished the first session, 76 were screened out, and 101 were invited back for the experimental phase. The intervention was an inert acoustic frequency. The only active ingredient was the sentence attached to it.
For twenty-odd years this kind of finding has been turning up one study at a time, in journals that do not read each other. In September a team at the University G. d'Annunzio of Chieti-Pescara tried to put the pieces on one table.
Their narrative review, by Fabio Lombardi, Matteo Frisoni, Giorgia Committeri and Carlo Sestieri, appeared online in the Journal of Cognitive Enhancement on 12 September 2026, drawing mainly on the Journal of Interdisciplinary Placebo Studies database. It walks through placebo and nocebo effects across different kinds of memory task and reports where the evidence is thickest: placebo effects on working memory and on resistance to the misinformation effect, nocebo effects on episodic memory.
The review also records the studies that found nothing in the scores, only in the self-ratings, which has led some researchers to argue these effects live entirely in how people judge themselves. The authors' position is that the effects deserve study as a possible tool for cognitive aging in their own right, alongside the training and lifestyle work already underway.
A narrative review is a map, not a measurement. There is no pooled effect size here, no systematic search protocol, no number you can carry to a clinic. What it offers is the shape of a field, and the shape is lopsided.
That lopsidedness is the part worth sitting with. In the older-adult trial, the discouraging message changed performance and the encouraging one did not. The placebo group expected to improve and said afterwards that they felt the sound had helped; their scores matched the control group's. The nocebo group did not report lower expectations at all, and still scored worse. Belief and behaviour came apart in both directions.
The trial authors are careful about why that asymmetry might show up in this age range. They point to decades of work on age-related stereotype threat, where activating negative ideas about aging measurably impairs how older adults perform, and they note that internalized views of one's own aging may make a discouraging suggestion land with more force than an encouraging one. Most placebo research in cognition has been done in young adults in laboratories. This one was done at kitchen tables, in people who have already started wondering about their own memory.
If you have ever watched a parent go quiet after being told, kindly and in passing, that a test is hard for people their age, you already know the scene this research is circling.
What changes, practically, is small and real. For clinicians, the framing sentence before a cognitive test is part of the test conditions, and these two papers together suggest it is the negative framing that needs watching most. For families, the finding is about the people who were studied, healthy volunteers doing an attention task online, and it is not a statement about anyone's relative or anyone's diagnosis. Nobody in either study had dementia.
But the direction of travel is clear enough to name. Expectation is not decoration on top of cognition. In these experiments it was a variable with a measurable effect on what people actually scored, and the effect ran downhill more easily than up.