The brain's 'body map' doesn't shift: Longitudinal fMRI shows stability of hand representations even after amputation

The classic idea is that if an arm is amputated, the orphaned body map region in the primary somatosensory cortex (S1) is quickly taken over by its neighbors, primarily the lips and face. A new paper in Nature Neuroscience breaks that mold. The researchers followed three adult patients longitudinally, before and for up to five years after amputation, and compared them with controls. The hand map in S1 and the motor cortex (M1) remained remarkably similar to the original, and there was no “expansion” of the lip region into the “hand.” In other words, amputation itself does not trigger large-scale cortical “rewiring”—adults retain a stable internal body model even without peripheral input.

Background of the study

The classic picture of somatotopy (the very same Penfield's "homunculus") was long supplemented by the thesis of "remapping" of the cortex after amputation: supposedly, the hand zone in the primary somatosensory cortex (S1) quickly loses input and is "captured" by the neighboring projection of the face/lips, and the degree of such remapping is associated with phantom pain. This idea was supported by cross-sectional fMRI/MEG studies and reviews, as well as individual clinical observations of the "transfer" of sensations from the face to the phantom hand. But the evidence base relied mainly on comparisons of different people and "winner-takes-all" methods, sensitive to noise and threshold selection.

In recent years, more accurate maps have emerged that show the complex and often stable organization of the face and hand in S1 in amputees: some of the signals taken for lip "invasion" may be an artifact of the analysis, and the relationship with phantom pain is inconsistent. Critics have pointed specifically to the "winner takes all" methodology, small ROIs, and lack of consideration of phantom movements and top-down influences. Multivoxel approaches and RSA provide a more nuanced picture, where obvious "capture" by the face is often not visible.

A new longitudinal study in Nature Neuroscience closes the main gap - a comparison "with oneself" before amputation and months/years after. In three patients, the authors compared activations during finger movements of the hand (before) and a "phantom" hand (after), as well as lips; there were also control groups and an external amputation cohort. Result: hand and lip maps remained remarkably stable, and no signs of "expansion" of the face into the hand were found; a decoder trained on "before" data successfully recognized "after". Conclusion - in adults, somatosensory representations are supported not only by peripheral input, but also by internal models/intentions.

Hence the practical and theoretical implications: brain-computer interfaces and prosthetics can rely on surprisingly stable “maps” of the amputated limb, and the “pain = remapping” hypothesis requires revision in favor of other mechanisms of phantom pain. More generally, the work shifts the balance in the long-standing debate about plasticity: mature somatotopy in humans turns out to be much more stable than neuroscience courses assumed.

How did they check it?

The authors used a longitudinal design: fMRI was recorded from the same people before surgery and then at 3 months, 6 months, and later points (1.5 or 5 years). In the scanner, participants were instructed to move their fingers (before amputation) and “phantom” fingers (after), purse their lips, and bend their toes.

• Sample and controls: 3 patients with elective upper limb amputation; 16 healthy controls (with repeat scans); additional comparison with a cohort of 26 chronic amputees (mean 23.5 years post-amputation).
• Map metrics: centers of gravity (COG) of activity in S1, pre/post pattern-to-correlate correlations for each finger, linear SVM motion decoding (training before amputation → test after and vice versa), assessment of lip penetration into the hand area.
• Key numerical results: longitudinal correlations of finger-to-finger patterns were high (r≈0.68-0.91; p<0.001), the accuracy of the decoder trained “before” remained above chance when tested “after” (≈67-90%), and the boundaries of the “lip map” did not expand into the “hand zone” even by 1.5-5 years.

Why is this important for neuroscience and clinical practice?

The work shows that “body” representations in S1 in adults are supported not only by peripheral sensory signals, but also by top-down influences from motor intentions and internal models. This explains why attempting to move a “phantom” hand elicits activity similar to that of a normal hand, and why previous cross-sectional studies may have overestimated face “intrusion” due to a “winner-takes-all” approach that does not account for phantom activity. This is good news for brain-computer interfaces: a detailed and stable “map” of an amputated limb is suitable for long-term applications. For phantom pain therapy, the implication is more subtle: current surgeries and neural interfaces do not “restore” the map because it is already there; therefore, other pain mechanisms need to be targeted.

What to check next

The authors conclude carefully but directly: there is no evidence of deficit-driven "remodeling" of S1 somatotopy after amputation in adults; preservation and reorganization are not conceptually mutually exclusive, but large "capture" by the lips is not visible in longitudinal measurements. It is important to expand the sample and standardize the tasks:

• Expand N and age ranges, test speed/limits of card preservation for different amputation causes and preoperative motor control levels.
• Add objective peripheral markers, including stump electromyography and neurostimulation, to separate the contributions of descending and peripheral signals.
• Rethink remapping protocols from winner-take-all to longitudinal, multi-voxel, and classification analyses that explicitly account for phantom motion.

Briefly - the main points

• Stability instead of 'grab': Hand and lip maps in S1/M1 in adults remain stably positioned for up to 5 years after amputation.
• Phantom is not imagination: attempts to move "phantom" fingers produce patterns statistically similar to preoperative hand movements.
• Implications: a robust basis for BCI prosthetics; reconsidering the concept of deficit-driven plasticity; new targets for phantom pain therapy.

Source: Schone HR et al. “Stable cortical body maps before and after arm amputation,” Nature Neuroscience, August 21, 2025 (Brief Communication). DOI: https://doi.org/10.1038/s41593-025-02037-7