fNIRS Observation-Operator Notes#

fNIRS is not a NeuroTwin pivot. It is useful because it validates the observation-operator worldview: measured signals are lossy observations of latent neural and physiological fields, not the field itself.

Observation Story#

Latent cortical activity drives hemodynamic activity. Hemodynamics interact with scalp/skull physiology, systemic signals, optode geometry, and motion artifacts. The fNIRS measurement is therefore:

latent neural field
-> hemodynamic field
-> optical propagation and systemic physiology
-> artifact processes
-> observed optical density

Rytov Approximation#

\[\log\left(\frac{\phi_1(t;\theta)}{\phi_0}\right) \approx J\delta\mu_a(t;\theta)\]
\[\phi_1(t;\theta)\approx \phi_0\exp(J\delta\mu_a(t;\theta))\]
\[\Delta OD(t;\theta)=\log\bar\phi_{\mathrm{base}}-\log\phi_1(t;\theta)\]
\[Y_{\mathrm{fNIRS}}(t)=\Delta OD(\phi_0\exp(J\delta\mu_a(t)))+\epsilon\]

Here J is a sensitivity matrix and delta mu_a is absorption change.

HRF and Physiology#

Future synthetic realism should separate:

  • neural field

  • hemodynamic field

  • physiology nuisance field

  • artifact field

  • modality observation operator

The hemodynamic bridge can be written:

\[F_{\mathrm{hemo}}(t)=(h_{\mathrm{HRF}}*g(F_{\mathrm{neural}}))(t)\]

Why This Helps NFC#

fNIRS makes it obvious that a modality is not the primitive. The primitive is the latent field plus a measurement operator. That supports the NFC framing for fMRI, EEG, behavior, and future modalities.

Claim Boundaries#

  • No fNIRS implementation claim is made.

  • No clinical, MDD, diagnostic, or treatment claim is made.

  • No SIMR/private work or personal contact information belongs in this repo unless public, licensed, and explicitly approved.

  • fNIRS notes must not block the A100 synthetic diagnostic.