Re-identification Risk

The risk that an individual can be singled out from a supposedly anonymized neural dataset by matching the recording against a reference. Both functional signals (EEG event-related potentials, resting-state connectivity) and structural features have been shown to act as fingerprints: functional-connectome fingerprinting demonstrated that a person's connectivity pattern can identify them across scan sessions with high accuracy in modest cohorts.

Because the discriminative information lives in the signal, removing metadata does not anonymize a neural recording the way it might a spreadsheet. Aggregation, coarsening, and formal privacy mechanisms are needed, and each trades identifiability against scientific or decoding utility. Re-identification risk also grows with database linkage: a neural fingerprint that is harmless alone becomes sensitive once it can be joined to a named record elsewhere.