The question asked of a symptom is almost always the same one: what has gone wrong. It is the right question much of the time, and it has a specific failure mode — it produces nothing useful when the investigations come back unremarkable. At that point the available conclusions are that something was missed, or that nothing is happening. A different question is worth putting alongside it: not what has broken, but what is this system doing.

The Assumption

A 2026 paper states the default model plainly: traditional biomedical models assume that symptoms passively reflect structural pathology, and that normal test results imply the absence of illness.

Those two assumptions are a package, and the second follows from the first. If a symptom is a readout of damage, then no damage found means no readout to explain.

It works well where it works. It fails in a specific and common situation — persistent symptoms with normal investigations, or with pathology insufficient to explain the severity — which the same paper describes as among the most common and challenging presentations in general practice, and one that patients and clinicians alike often experience as a dead end.

Where the Reframe Has Firm Ground

The strongest version of this is not philosophical at all. In persistent pain, central sensitisation describes a system whose processing has been amplified — manifesting as hyperalgesia, allodynia, expanded receptive fields and impaired endogenous pain inhibition (Journal of Pain Research, 2026).

Read that as an output rather than a fault and it becomes intelligible. A pain system that has increased its gain is doing something a protective system might reasonably do after sustained demand: lowering the threshold at which it reports, widening the area it monitors, reducing its own damping.

It is measurable, it is assessed with quantitative sensory testing and algometry, and it explains the mismatch that the damage model cannot. This is not a metaphor.

The whole account is in what central sensitisation describes.

Where It Is a Proposal

The broader version — that symptoms in general are constructed rather than reported — comes from a 2026 paper applying predictive processing to persistent physical symptoms in general practice.

Its claim is that symptoms are active perceptual constructions, generated as the brain infers the state of the body from incoming signals and prior beliefs.

One label, and it matters: this is an opinion paper. It develops a clinically applicable account from existing literature and illustrates it with clinical scenarios. It reports no new data, and it should be read as a proposed framework rather than a finding.

That distinction is exactly the sort of thing that gets lost when an idea is appealing, and this one is appealing. The framework and the evidence for it are different objects.

The Frame, Stated Carefully

If a symptom is an output rather than a readout, then the useful question about it is what the system was doing when it produced it.

Hypervigilance is what a threat-detection system does after it has learned that threats appear without warning. Avoidance is what a system does once something has been established as dangerous — and learning that something is not dangerous is a separate, slower process, which is why avoidance persists past the point of usefulness.

Amplified pain is a protective system with its gain turned up. Withdrawal and flatness are what a system does when it is conserving.

Every one of those descriptions is about function rather than fault. None of them says the experience is desirable, proportionate or something to leave alone.

Four Things This Frame Is Not

It is not a claim that symptoms are useful. A response that made sense in one context can be actively harmful in another, and often is. Adaptive in origin is not adaptive now.

It is not a reason to skip investigation. Persistent physical symptoms deserve proper assessment, and a framework about how symptoms are generated is no substitute for finding out whether something specific and treatable is producing them.

It is not a claim that symptoms are chosen or imagined. On this account every perception is actively constructed, including ones with unambiguous physical causes. Constructed describes a process, not a verdict on whether something is real.

And it is not evidence for any treatment. Understanding a symptom as an output does not establish what changes it. The framework paper proposes a way to explain and think about symptoms; it does not demonstrate an intervention, and it does not claim to.

The Question Worth Ending On

If a nervous system has learned one pattern, can it learn another? That is the question the reframe naturally leads to, and it deserves an honest answer rather than an encouraging one.

The evidence is mixed and mostly modest. Regulatory capacity demonstrably changes across development — effortful control follows a U-shape through adolescence, co-regulation reduces as children develop their own skills. Safety learning is a real, separate process, and it is slower than threat learning by necessity. Adult brains change with sustained experience, though whether that constitutes increased flexibility is less established than it is usually presented.

So: these systems move, they move in patterned ways, and the timescales are long. That is a more modest claim than ‘the brain can rewire itself’, and it is the one the evidence carries.

It is also enough to make the question worth asking. What a system is doing, and what it learned to do it for, is a better starting point than what is wrong with it — provided nobody mistakes the reframe for an answer.