Two colours, one tissue
All-optical interrogation means imaging and photostimulating the same tissue at the same time. The enabling trick is spectral separation: pair a green calcium indicator with a red-shifted opsin (or a red indicator with a blue opsin) so the read light and the write light occupy nearly non-overlapping bands. Get this wrong and you have optical crosstalk — the read and write channels leaking into each other.
Crosstalk as the spectral overlap between the imaging light and the opsin's action spectrum. All-optical demands that the indicator's excitation band and the opsin's absorption be nearly orthogonal.
Closing the loop
Once read and write coexist, you can close the loop optically: image activity, infer a state, and write a targeted pattern in response — all in real time. This is what makes optics a genuine closed-loop and bidirectional instrument rather than a mere stimulator. In a research setting it lets you ask counterfactual, causal questions — 'if these cells had fired, what would the circuit do?' — with cell-type and single-cell precision no electrical loop can match.
The scattering wall
Everything optical eventually hits the same wall: light scattering in tissue. Living brain scatters far more than it absorbs; the ballistic photons that form a diffraction-limited focus decay exponentially with depth. Usable two-photon depth is only a few scattering lengths — roughly the thickness of cortex — before the focus dissolves into a background glow.
Ballistic photons — the ones that stay focused — decay exponentially with a scattering length of tens to a couple hundred microns. Depth in the living brain is limited by scattering, not by diffraction.
Pushing deeper: adaptive optics and three photons
Two tactics push the wall back. Adaptive optics measures and pre-corrects the wavefront distortions the tissue imposes, restoring a tight focus deeper than an uncorrected beam allows. Three-photon excitation uses an even higher nonlinear order and a longer wavelength that scatters less, dramatically improving the signal-to-background ratio and reaching subcortical structures through the cortex. Both extend the accessible depth; neither removes the exponential wall.