A small human study suggests that age-related differences in preparing to step may partly reflect how people update expectations about their next movement. The findings offer a possible cognitive explanation, but do not establish a treatment or a way to predict falls.
Taking a step feels simple. Before a foot leaves the ground, however, the body has already begun preparing for the change in support. The nervous system must organize that preparation while deciding whether movement is needed at all.
New research in npj Aging suggests that this process becomes less responsive to recent experience in older adults. In a laboratory stepping task, older participants began their preparatory weight shifts later and gained less speed from a sequence of cues telling them to move.
The researchers interpret this pattern as a difference in how expectations are updated. It is one possible contributor to mobility changes, within a much larger picture involving muscles, sensation and the nervous system.
Balance starts before the step
Maintaining balance depends on coordinated information processing and physical responses. Aging can affect muscle strength, force generation, sensory processing and the attention required to move safely. No single mechanism explains every person’s difficulties.
The new study focused on anticipatory postural adjustments, or APAs. These are the preparatory changes in muscle activity and weight distribution that help the body manage an upcoming movement. Researchers can detect their timing using force plates beneath the feet.
Preparation also depends on expectations. If several signals in a row tell you to step, you may begin to expect another. When the pattern changes, that expectation needs revising. Here, “belief updating” means this ongoing adjustment of expectations, rather than a conscious opinion about walking.
A stepping task with changing odds
Alan Bince Jacob and colleagues at Maastricht University compared 24 younger adults with 25 older adults. Their average ages were approximately 23 and 72 years. The older participants had good functional balance and mobility on screening tests, making this a study of relatively healthy adults rather than people selected for recurrent falls.
Participants stood on a computerized stepping mat above force plates. A screen indicated which foot to use, followed by a signal telling them either to step forward or remain standing.
In one block, 75% of signals required a step. In the other, only 25% did. The order of the blocks was counterbalanced across participants. This allowed the team to examine how stepping preparation changed with both the overall likelihood of moving and the recent sequence of signals.
Younger adults completed 240 trials. Older adults completed 160 to reduce the burden of the experiment.
Older adults gained less from a run of “go” signals
On average, preparatory weight shifts began about 249 milliseconds after the signal in younger adults and 302 milliseconds afterward in older adults. That difference is roughly five hundredths of a second.
Both groups prepared faster when stepping was more likely. Both also tended to speed up after consecutive signals requiring a step. However, the influence of recent trial history was weaker in the older group. The age difference was particularly evident when repeated “go” signals gave participants a reason to anticipate another step.
The pattern points toward differences early in movement preparation. The time between lifting the foot and placing it down did not show the same statistically significant interaction between age group and task condition.
That does not prove muscle strength was irrelevant. The study did not establish that its findings were independent of strength or other sensory and motor differences. It suggests that looking only at movement execution could miss part of the explanation.
A possible mechanism, inferred from behaviour
The team fitted the responses to a Hierarchical Gaussian Filter, a mathematical model that estimates how expectations change as new information arrives. This Bayesian approach weighs existing expectations against incoming evidence.
The fitted model indicated lower learning rates in older adults. Simulations were consistent with older participants placing less weight on new evidence that contradicted expectations and relying more on established predictions.
The model also suggested a stronger slowing response to unexpected uncertainty. That result deserves particular caution: its estimate was more sensitive to the different trial counts between groups than the learning-rate finding.
These are explanations inferred from behaviour and modelling. The experiment did not directly record brain activity, identify a particular damaged brain region or show that changing the proposed mechanism restores balance.
What this means for falls and independence
Being able to prepare and adapt movements matters for navigating everyday environments. Understanding how expectations shape the start of a step could therefore help researchers investigate mobility changes that threaten independence.
But a controlled stepping task is only one part of everyday balance. This study did not follow participants to determine who later fell, validate a clinical fall-risk test or test an intervention.
The sample was small and relatively healthy, limiting how widely the findings can be applied. There was only one change between the two probability conditions, limiting conclusions about repeatedly changing environments. Unequal trial counts complicated some model estimates. More comprehensive cognitive, sensory and motor testing would help separate the possible explanations.
The authors propose future studies combining stepping tasks with direct measures of brain activity. Whether training expectation updating can improve real-world mobility or prevent falls remains an unanswered question.
Why it matters for healthy aging
Healthy mobility involves preparation as well as physical execution. This study adds evidence that the way people learn from recent events may help shape age-related differences in stepping. Its value is a more specific hypothesis to test, with larger studies needed before it can guide new interventions.
Research classification: Early Research. Small human laboratory study with computational modelling.
Original research: Jacob AB, Kotz SA, Auksztulewicz R, Köpcke M, Meijer K and Boonstra TW. Reduced belief updating impairs adaptive step initiation in older adults. npj Aging. Published October 14, 2026. Open-access, peer-reviewed article in press; the accepted manuscript may receive further editorial changes.