Study shows brain keeps internal body map intact after amputation

New fMRI evidence shows that the brain’s hand and lip maps remain stable years after amputation, reshaping our understanding of neuroplasticity and guiding future prosthetic and rehabilitation strategies.

Study: Stable cortical body maps before and after arm amputation. Image Credit: SeventyFour  / Shutterstock

In a recent study published in Nature Neuroscience, a group of researchers tested whether adult arm amputation triggers large-scale cortical reorganization by tracking hand and lip representations with functional magnetic resonance imaging (fMRI) before and up to 5 years after surgery.

Background

For decades, students learned that after a hand is lost, the face invades its territory in the brain. Yet amputees feel vivid phantom fingers and can try to move them, suggesting preserved maps. Classic animal studies and early human imaging supported extensive reorganization in the primary somatosensory cortex (S1). Newer human work points to stability, with phantom movements engaging handlike patterns.

The authors also highlight that phantom activity can be confirmed by residual limb muscle contractions, indicating genuine motor attempts rather than imagined actions. The field lacks longitudinal data following the same people across surgery. 

Comparisons of affected and unaffected hemispheres, and replication in primary motor cortex (M1), are essential to reconcile models, and further research is needed to identify who reorganizes, when, and why.

About the study

Three adults scheduled for unilateral arm amputation (P1, P2, P3) were studied twice before surgery and at 3 months, 6 months, and at follow-up (1.5 years for P1; 5 years for P2). Sixteen able-bodied controls (Ctrl) were scanned at four sessions over 6 months; a separate group of 22 younger controls was additionally analyzed for P1.

A chronic amputee group (n = 26) provided context. fMRI was acquired on a 3 Tesla scanner. T1 weighted magnetization prepared rapid acquisition gradient echo anatomy and blood oxygenation level dependent (BOLD) echo planar imaging were collected (repetition time (TR) 1.5 s, echo time (TE) 35 ms, field of view (FOV) 212 mm, 2 mm isotropic voxels).

During scanning, participants performed movements of each finger, lips, and feet; after surgery, they attempted phantom hand movements and completed imagery control scans. Analyses focused on S1, Brodmann area 3b (BA3b), with replication in M1. A general linear model (GLM) estimated condition beta weights.

Multivoxel analyses included support vector machine (SVM) decoding across sessions and representational similarity analysis (RSA) using cross-validated Mahalanobis distances. Region of interest (ROI) definitions, center of gravity (COG) shifts, and smoothing parameters (full width at half maximum (FWHM) 3 mm) followed procedures.

a, Experimental timeline. Scans before and after amputation were conducted across 4–5 time points: twice before, and at 3 months, 6 months and 1.5 (P1)/5 years (P2) after amputation. b, Illustration depicting the three participants 6 months after amputation, including their subjective description of their phantom limb position. c, Phantom movements are not imaginary. Univariate activity (z-scored) contrast map displaying a participant’s attempts to open and close the phantom hand versus imagining movement, 6 months after amputation. d, Participant’s hand (red) and lip (blue) cortical activation maps (contrasted against feet movements) in the affected hand hemisphere across 4–5 sessions. All maps were minimally thresholded at 33% the maximum z-statistic and used a common color scale (the participant’s maximum z-statistic > 4.5). Participants agreed to have their image reproduced. Brain illustrations in a were created in BioRender.

Study results

Across sessions spanning before and after surgery, all three case participants generated phantom hand sensations and could volitionally attempt phantom finger movements; phantom movement produced stronger activity than imagining movement, and was accompanied by residual muscle contractions in the stump, confirming genuine motor control.

Hand and lip activity in the hemisphere contralateral to the missing hand were tracked across time using matched tasks. COG shifts for the hand and individual fingers fell within the able-bodied Ctrl distribution at 6 months, and voxelwise finger patterns before amputation correlated strongly with those at the final scan.

Decoders trained on pre-amputation finger pairs classified post-amputation patterns above chance, and RSA with cross-validated Mahalanobis distances confirmed significant consistency across sessions, including long-term follow-ups. Together, these multivoxel metrics indicate stability of the hand map in S1, with similar evidence in M1.

The study also tested a central prediction of remapping: increased lip activity in the deprived S1 hand region and expansion of lip map boundaries toward hand territory. Neither pattern emerged. Across timepoints, lip univariate activity in the hand region stayed within the Ctrl range, the lip COG did not drift toward the hand, and lip map boundaries showed no expansion.

One participant (P2) did show a temporary increase in lip–thumb representational distance at 6 months, which returned to the typical range by 5 years. When compared with a chronic amputee cohort (n = 26; approximately 23.5 years since amputation) and a secondary Ctrl cohort (n = 18), the case participants’ phantom hand and lip COGs fell within group distributions, and lip activity in the S1 hand region matched the chronic amputee pattern. 

The authors note that some case participants showed slightly, though not significantly, higher lip activity than chronic amputees.

Transient deviations were detected but were idiosyncratic rather than systematic. At 6 months, P1 showed a temporary reduction in average finger selectivity relative to Ctrls that returned to the typical range by 1.5 years; P3 showed a similar transient reduction in decoding at 6 months. P2 already had reduced classification accuracy before surgery due to impaired motor control, which complicates the interpretation of post-amputation changes.

The authors note that some transient reductions did not survive statistical correction and may reflect classifier sampling variability across long intervals. In the unaffected hemisphere, intact hand finger selectivity and lip metrics showed typical session-to-session variability relative to Ctrls.

A separate analysis found that for the missing hand, only P3 showed a brief reduction in average inter-finger correlation at 3 months, which normalized by 6 months.

Overall, the longitudinal evidence demonstrates stable hand and lip cortical maps across amputation, with only brief and participant-specific reductions near the early post-amputation window.

Conclusions

In adults, arm amputation did not trigger large-scale remapping of S1 or M1. Hand representations and lip maps remained topographically stable, while multivoxel structure could be decoded across surgery, with brief dips early after amputation.

The findings support a deeper conceptual shift: S1 is not a passive relay of peripheral input but maintains an internal, resilient body model that persists even after sensory loss. This stability matters clinically, as it supports training strategies that leverage intact hand representations for prosthetic control and for targeted neurostimulation or sensory feedback. The longitudinal design, small sample, and adult cohort limit generalization, and child plasticity may differ. 

Future work should test diverse etiologies, ages, and rehabilitation paths to determine when cortical maps reorganize and how interventions shape outcomes.