The cover story was about pupils. Participants at the University of Padua were told the experiment was testing how a special pair of glasses affected pupil size, which is the sort of dull, plausible detail that stops people from guessing what is really being measured. Then they were told what the lenses would do to their reading: improve it, or spoil it. The lenses were sham. They did nothing at all. What happened next is the finding, published in Acta Psychologica (volume 266, article 106776) and posted online on 7 April 2026 by Sandro Franceschini, Andrea Facoetti and colleagues from Padua, Bergamo, Insubria and Florence.
This was a within-subject, cross-over experiment in 102 young adults, meaning each person passed through the positive and the negative condition and served as their own comparison. Expectations were planted two ways: explicitly, by the experimenter saying out loud what the glasses would do, and implicitly, by rigging the feedback a computer gave participants about how well they were doing. Alongside reading, the team recorded what people said they felt, their Big Five personality scores, their heart rate and their pupil size. Participants duly reported that the lenses had influenced their performance, which is the ordinary, expected part of placebo research.
The less ordinary part is that the performance moved too. Explicit placebo expectations increased reading speed for real words and for pseudowords, the invented letter strings that force a reader to decode rather than recognise. Negative expectations did not significantly impair performance, though the authors note a nocebo effect might emerge in longer tasks. The conditioning phase, the rigged feedback, did not shift expectations; it was the spoken instruction that did the work. Personality did not predict who responded. Heart rate and pupil size showed no significant difference between the placebo and nocebo conditions, so whatever changed was not visible in those two bodily signals.
The same group had already found something similar with a different sham. In Psychological Research (volume 89, article 120, published 8 July 2025), a double blind within-subject study of 49 children with developmental dyslexia and 48 unselected young adults reported that positive expectancy improved word reading accuracy in the children, with an effect size the authors describe as larger than those reported for gold standard training programmes, and replicated the improvement in adult poor readers; typical readers improved only in pseudoword decoding speed. Read the two studies together and the uncomfortable implication is the authors' own: reading training trials that do not control for expectation may be measuring, in part, the words spoken before the task.
That is where this lands for families living with a dementia diagnosis, and it lands sideways rather than straight on. Nobody in either study had dementia. Nobody was given a pill. But cognitive testing is a performance delivered in a room, by a person, after a sentence of framing, and these experiments are a clean demonstration that the framing can register in the output, not just in how the person says they felt. The same machinery sits inside every trial with a placebo arm and every clinic appointment where someone is asked to read a list aloud while a relative watches from the next chair. It does not mean a test score is fiction. It means the score is a measurement of a person in a situation, and the situation includes what they were told.
What it is not is a technique to try at home, and the authors do not present it as one. The honest reading of this work is narrower and more useful: expectation is an active ingredient that researchers should control for and clinicians should be conscious of, which is close to what the Padua team says in its own conclusion. For a family, the takeaway is not a thing to do but a thing to understand, including on the bad days when a score dips and everyone in the room is looking for a reason.