Something is always said before a cognitive test begins. A clinician explains the task, a daughter in the next chair says nothing and hopes, and somewhere in the room the phrase "this one is hard" is in the air. A review published online on September 18, 2026, in Neuroscience and Biobehavioral Reviews argues that those words are doing measurable work, and proposes a reason: belief and the task may be drinking from the same cup.
The paper, by Miriam Braga, Bernardo Villa-Sanchez, Sara Assecondi, Veronica Mazza and Mirta Fiorio, is a narrative review. It gathers what placebo and nocebo research has found across cognitive domains, and the picture it reports is uneven. Reward-based learning and emotional processing show some susceptibility to expectation. For memory, attention and working memory, the authors describe the evidence as limited and inconclusive.
That unevenness is usually read as a sign that cognition is stubborn, immune to suggestion in a way that pain is not. Braga and colleagues read it differently. They propose that producing a placebo or nocebo response is itself effortful, leaning on frontal control systems, and that a demanding task competes for the very resources the expectation effect needs. On a hard working-memory task, there may be nothing left over for belief to work with.
Think of it as one lane of traffic. A gentle task leaves room for expectation to ride alongside. Pile on the load, and the lane fills.
The authors are explicit that this is a hypothesis awaiting controlled, directly comparable experiments. It is a proposal, published in a review, not a result.
It does rest on something. In March 2025, a team led by Aliz Mostauli and Stefanie Brassen reported in Scientific Reports on 49 healthy volunteers, mean age about 27, scanned while they believed they had received oxytocin. The placebo instruction lifted mood and reduced how much fearful faces pulled attention away, and it did so specifically when attentional resources were free for the job. Participants with greater cognitive control capacity showed the larger placebo effects. Forty-nine young adults in a scanner is a small window, and it is a window onto emotion rather than memory.
The work that will matter more to a family happened earlier this year. In a randomized trial published in Psychological Research on May 15, 2026, Alessandra Barbon and colleagues tested healthy adults aged 65 to 80 entirely online, with no researcher in the room. Everyone heard the same inert sound. One group was told it was neutral, one was told it would help their performance, one was told it would hurt it. From 204 people recruited, 101 reached the experimental session.
Then they did an oddball task, pressing to the odd tone or image among the standard ones, before and after the sound.
The control and placebo groups improved on the auditory version, the ordinary practice gain you get from doing a thing twice. The group told the sound might impair them did not improve. Their performance simply stayed flat. The positive suggestion moved what people said they expected and left their scores alone.
The detail worth sitting with: fewer than half of the people in the negative-suggestion group reported, when asked, that they expected to do worse. The effect showed up in the scores of people who did not say they believed it.
None of this is new in outline. A 2017 meta-analysis in The Gerontologist by Bonnie Armstrong and colleagues found that exposing older adults to negative stereotypes about aging reliably shifted their memory scores, with effect sizes of 0.253 for episodic memory and 0.373 for working memory. Framing has been known to move the numbers for years. What the 2026 review adds is a candidate mechanism with a testable edge to it, and a prediction that runs against intuition: the harder the task, the less room expectation has to act.
For a clinician, the practical content is modest and real. A score obtained in one session carries the conditions of that session, including how the session was introduced. Research teams already control for this with standardized instructions. The review suggests the standardization matters most on the tasks where people have the most spare capacity.
For the daughter in the next chair, the translation is narrower than she might want, and more useful than nothing. The way a test is framed can shift a measured score in healthy older adults by a small amount. That is a finding about test conditions. It says nothing about what is happening in a brain over years, and it will not move a biomarker.
Everyone in these studies was healthy. Nobody had a diagnosis. The oddball task runs for minutes.
Still, there is a reason to care. The clinic is where fear and measurement meet, and the fear arrives first. Whatever it costs, it is being paid in the same currency the test is counting.