Case Study: FND After Spine Surgery

In an effort to better understand how FND presentations are described and diagnosed, I often review FND case literature. One pattern I’ve noticed is what is described as the sudden onset of FND immediately after cervical spine surgery. In these cases, the surgery itself is often framed as a psychological trauma or stressor which, within the conversion model, is used to account for the emergence of FND symptoms.

In this post, we examine one such teaching case. The key question is this. Does this really show new onset FND, as interpreted within the predictive processing framework commonly used in FND, or can that very framework also explain it as a lag in updating protective priors (delay in the brain letting go of its old danger warnings) after real preoperative spinal cord compression? Because predictive processing errors do not automatically imply FND.

Clinical case

A 56-year-old man who reported numbness in both hands and an unsteady tandem gait, but had normal, symmetric reflexes on both sides of his body, with no clonus (indicating no nerve pathway damage causing involuntary muscle jerks) and no Hoffman’s sign (indicating no nerve pathway damage affecting hand and arm movements).

A subsequent cervical spine MRI showed severe central stenosis at C3-C4 with T2 cord signal changes, prompting the neurosurgeons to perform cervical decompression surgery.

Right after surgery, he had no movement or feeling in his legs, but his arms worked normally.

• Day 1: Intermittent leg movements observed.
• Day 2: Neurosurgery consulted neurology; consistent toe movement noted, positive Hoover’s sign, and other muscles moved with distraction (diagnosed based on positive signs consistent with FND). Though highly specific for functional weakness, this might reflect maladaptive priors from months of genuine C3–C4 myelopathy-related gait instability rather than pure conversion.
• Day 3: Stood unassisted.
• Day 5: Discharged.

The case becomes more complex when considering the patient’s history: head trauma, intermittent homelessness, drug abuse (a previous surgery was cancelled after he tested positive for cocaine), schizoaffective disorder, a major depressive episode with psychotic features, and suicidality.

Throughout his hospital stay, the patient was verbally abusive to nursing and medical staff, frequently ordering interns/residents out of his room as he stated that he did not want to be taken care of by “child doctors.”

His hostile and demanding behaviour may have coloured the team’s perception, making it difficult to distinguish between possible malingering and genuine functional deficits. This highlights how challenging the diagnosis can be when objective signs overlap with behavioural factors and secondary gain is uncertain. In such cases, labelling the presentation as FND can sometimes serve as a polite or pragmatic compromise when the underlying cause remains unclear.

Alternatives

Which brings us back to the original lesson this paper is trying to teach namely postoperative recognition of FND in neurosurgery.

This lesson feels somewhat premature, as no preoperative assessment for possible functional elements appears to have been done.

Let me propose an alternative hypothesis based on a neuroscientific theory behind FND, the Bayesian brain hypothesis, that the supposed functional overlay may already have been in place prior to surgery.

A personal anecdote may help illustrate this:

In 2014 I had a major accident that shattered my femur. I spent seven weeks in hospital while doctors worked to save my leg. Rehabilitation also took a long time.

During recovery, I developed a fear-avoidance reaction. Stepping on the healing leg caused pain, so I avoided using it. Over time, this avoidance disrupted predictive processing, as described in the Bayesian brain hypothesis.

Even after the leg had healed, I continued to avoid using it, a behaviour I had to consciously unlearn.

How does all of this fit into this case at hand?

This functions as an analogy to explain how the brain can carry forward a learned protective pattern that influences movement even when the original physical trigger is gone.

Months of real gait instability from C3-C4 myelopathy could create a strong, pre-conscious expectation that “my legs are unreliable, better to hold back and not fully commit to using them.”

This expectation can act as a subconscious safety program or inhibitory pattern that selectively blocks voluntary, attended leg movement, such as when the patient is directly asked to lift the leg and focuses on it.

As my personal example illustrates this, after my femur was shattered in 2014, I avoided fully using the leg during recovery because stepping on it caused pain. Even after the leg healed, I would still hold back unless my attention was distracted, at which point I could step on it without pain.

Similarly, in Hoover’s sign as used in the case lesson, when attention is shifted such as during a distraction task or contralateral movement, the observed pattern may reflect the influence of automatic motor pathways that remain intact despite impaired voluntary initiation. This can present as apparent normalisation of strength during distraction, which is then revealed as a positive Hoover’s sign when formally tested.

In predictive processing, the brain builds strong expectations based on past experiences. In this case, years of unsteady walking have created a strong expectation of gait problems. This expectation can interfere with a person’s ability to voluntarily start walking, while automatic movements (like reflexes) remain mostly unaffected.

Distraction may help by temporarily reducing the influence of this unhelpful expectation, allowing normal voluntary movement to come through more easily.

This raises a simple concern: in some cases, clinical signs can become reified into the condition itself, rather than being treated as indicators of an underlying process. When this happens, investigation of possible underlying causes may be reduced once the FND label is applied. This makes me wonder whether “functional overlay” is always a distinct condition, or sometimes a convenient label used when the drivers of the observed clinical signs are not fully explored.

But all of this said, if predictive processing is used to interpret a functional presentation, its value may remain largely theoretical rather than directly clinically actionable if that link cannot be operationalised or demonstrated at the individual level. It is like observing a scar without knowing what caused it or whether the underlying process is still active. This may be useful for conceptual understanding, but its clinical application at the individual level remains limited.

This risks focusing on identifying and classifying “scars”, giving them theoretical labels and groupings, without establishing why they are there in the first place, like focusing on the scar left by a stab wound while the assailant that caused it remains unidentified (and potentially still hiding behind a bush in your garden).