Well, there are of course interaction-free measurements in the sense that the system probed never interacts in the usual sense with the probe itself, with the classical example being the Elitzur-Vaidman ‘bomb tester’ experiment: imagine a bomb with a trigger so sensitive that the bomb explodes upon even a single photon impinging upon it. If you want to find out whether a bomb is ‘still good’, i.e. has a working trigger, you can then introduce it into an interferometric setup designed such that when both arms are open, you will only ever detect photons in one of the detectors; thus, detecting a photon in the other means that one path was blocked (by the bomb’s trigger), however, the photon can’t have taken that way, because then, it would have been absorbed (and the bomb exploded). That way, you can gauge the viability of a bomb with a 50% chance without exploding it (which can be increased to arbitrary values by ‘chaining’ the setup). (Of course, there’s some semantic qubbling on the meaning of ‘interaction’ here: that the photon hasn’t taken the path with blocked by the trigger is essentially classical reasoning, and its validity may depend on your view on the proper interpretation of quantum mechanics, the reality of the wave function, and similar subtleties.)
This has also led to techniques such as ‘interaction-free imaging’, where instead of a bomb’s trigger, you scan some arbitrary object in a rastered way, and infer when the object blocked the path from your detections in the ‘dark’ detector, thus building up an image with arbitrarily little coupling of photons to the object. In principle, I suppose this could be used to detect a beam of light without interacting with it, perhaps using something like the optical Kerr effect where the refractive index of a medium changes with an incident light beam, or some other QND scheme.
Detection is a form of interference, so I don’t see how you could detect light without interfering with it.
You could deduce it though like, say, a blindfolded man holding a flashlight that’s slowly warming up in his hands. Probably generalizable to many different phenomena. Light has momentum, so when a particle emits light you could detect it’s change in momentum, and deduce it has emitted light that way. Etc.
I’m certainly no expert, but it seems to me the quantum experiments mentioned in previous posts reflect the same sort of distinction. You’re not detecting light so much as deducing that it’s there based on how other particles are affected by it.
It seems to me that this particular splitting of hairs could apply to any method of “detecting” light. Couldn’t you say that detection is always, ultimately, based on other particles being affected by the light?
True, but the difference is whether your device absorbs (or deflects) the photon or not. In the case of a flashlight, it gets warm when the photons are emitted, and thus detecting the associated warmth doesn’t actually block, interfere with or absorb any photons. Whereas detecting them with your eyeballs or a photocell does.
From the photon’s perspective, those are very different scenarios. In the former case, they’re free to continue on their trajectories unmolested, in the latter, they aren’t. I read the OP as asking whether you can detect photons without stopping them or changing their paths, as most traditional “light detectors” would do.
(It occurs to me however that I don’t know how much of a flashlight’s heat rise is due to resistive heating from the electric current, versus absorbing the photons that travel backwards and sideways into the flashlight itself, rather than forward where you are aiming the beam. I’m sure it’s both, but I’d still say this mostly counts as “deducing” rather than “detecting” the light, by my own idiosyncratic definitions at least.)
Virtually all of it is resistive heating. The interior of the half-cavity at the business end of a flashlight is highly reflective, so very little of the light is absorbed rather than reflected towards the target.