Neuroplasticity has become the word people reach for when they want to say that change is possible. It is doing an enormous amount of rhetorical work, most of it unearned, and the honest way to test it is to look at the cleanest available case: adults deliberately learning something demanding over a sustained period, with their brains measured before and after.

What Is Not in Dispute

That adult brains change with experience. Structural and functional changes have been reported in association with skill acquisition, and the underlying capacity for synaptic change does not switch off after childhood.

That is the floor, and it is a real one. The claims worth examining are the ones built on top of it.

Those claims usually go further in two directions at once: that the changes are large, and that they generalise — that practising one thing makes the system better at things in general. Neither follows from the floor.

A Good Test Case

Adult second-language learning is close to ideal for this question. It is demanding, sustained over months or years, unambiguously effortful, and it happens well after the developmental windows that made plasticity famous.

A 2026 systematic scoping review took that literature seriously — PRISMA 2020, searching PubMed, Scopus, Web of Science and PsycINFO to June 2025, covering longitudinal and cross-sectional neuroimaging studies of adult second-language plasticity.

It found fourteen studies across structural MRI, diffusion imaging and functional MRI (Frontiers in Human Neuroscience, 2026).

Fourteen is a small literature for a question this widely discussed, and that discrepancy is itself worth noticing.

It is worth being clear why fourteen studies is a problem rather than a starting point. Neuroimaging studies of training are expensive, so samples are small; small samples produce unstable effect estimates; and a literature assembled from unstable estimates supports confident conclusions in neither direction. The review’s own emphasis on heterogeneity is the consequence of that arithmetic.

What the Review Actually Concluded

Two statements, both from the authors, both more restrained than the popular version.

First, that findings remain heterogeneous and are influenced by differences in study design, learner age, training intensity and imaging modality. In other words, what you find depends substantially on how you looked and who you looked at.

Second — and this is the more consequential one — that the relationship between structural remodelling and functional reorganisation, and their relevance to aging and cognitive reserve, remains unclear.

That second sentence dismantles the most common inference in this area. The popular argument runs: the brain changes structurally with training, therefore it is functioning better, therefore that protects cognitive capacity later. The review is saying that the link between the first and second steps is not established, let alone the third.

Flexibility Is a Different Claim From Change

It is worth separating two things that get merged.

Change means the system is different after training than before. That is well supported — it is what learning is.

Flexibility means the system has become better at changing — quicker to adapt, better at switching between modes, less locked into whatever it has practised most. That is a claim about a capacity rather than about a state, and it requires evidence that improvement transfers beyond the trained task.

The reviewed literature is about the first. Reading it as evidence for the second is the step that popular accounts take without noticing.

Transfer is the specific thing that would have to be shown, and it has a poor history across cognitive training research generally. People reliably get better at what they practise; whether that improvement reaches anything else is the question that has repeatedly returned disappointing answers, and it is not one the second-language literature was designed to settle.

Where Flexibility Is Genuinely Visible

The clearest evidence that regulatory capacity is not fixed comes from developmental data rather than training studies.

Effortful control follows a U-shaped trajectory across ages 10 to 19 — declining into the mid-teens, then rising. Co-regulation between parent and child measurably reduces as a child develops their own regulatory skills. Early adversity co-patterns with biological aging markers across adolescence in ways that vary by kind of adversity rather than simply accumulating.

None of those are interventions. They show that these systems move — over years, in patterned ways — which is a weaker and more defensible claim than that they can be deliberately made more flexible.

The broader version of this, and how badly the popular framing overstates it, is covered in what the adaptability evidence supports.

The Honest Answer

Can the brain become more flexible? The defensible answer is: it changes with sustained experience, demonstrably; whether that change constitutes increased flexibility, and whether it transfers or protects anything, is not established by the literature that gets cited for it.

That is an unsatisfying answer and it is the accurate one. The alternative — asserting that any sufficiently determined effort reshapes the brain toward general improvement — is not supported by a fourteen-study review whose own conclusion is that key relationships remain unclear.

What follows practically is modest. Sustained, demanding, genuinely novel activity is associated with measurable change, which is a reasonable thing to know. It is not a guarantee about cognition later, and anyone selling it as one is ahead of the evidence.