Nature News 2016: People can detect flashes of light as feeble as a single photon

Nature News 2016: People can detect flashes of light as feeble as a single photon

People can detect flashes of light as feeble as a single photon, an experiment has demonstrated — a finding that seems to conclude a 70-year quest to test the limits of human vision.

The study, published in Nature Communications on 19 July1, “finally answers a long-standing question about whether humans can see single photons — they can!” says Paul Kwiat, a quantum optics researcher at the University of Illinois at Urbana–Champaign. The techniques used in the study also open up ways of testing how quantum properties — such as the ability of photons to be in two places at the same time — affect biology, he adds.

“The most amazing thing is that it’s not like seeing light. It’s almost a feeling, at the threshold of imagination,” says Alipasha Vaziri, a physicist at the Rockefeller University in New York City, who led the work and tried out the experience himself.

Nature News 2016: People can detect flashes of light as feeble as a single photon

“The most amazing thing is that it’s not like seeing light. It’s almost a feeling, at the threshold of imagination,” says Alipasha Vaziri, a physicist at the Rockefeller University in New York City, who led the work and tried out the experience himself.

It's interesting that the protocol involved the subjects deciding which of two intervals contained the photon and which was a dummy. Overall, they picked the interval with the photon slightly more often than they should have done by chance (51.6%), and that was associated with a (non-significant) p value of 0.0545.

Of course, that would be counted as a failure in a parapsychology experiment, but they obtained a more decisive result by looking at only the 12% of cases where people were most confident that they had picked the right interval. Then the probability of correctness went up to 60%, and the p value fell to 0.001.

I believe their conclusion, But I would expect psi sceptics to be concerned that there's no indication that the authors pre-registered an intention to focus on the trials where the subjects were most confident, and therefore to be suggesting that their p value should be corrected for the possibility of multiple hypotheses. 

It's interesting that the protocol involved the subjects deciding which of two intervals contained the photon and which was a dummy. Overall, they picked the interval with the photon slightly more often than they should have done by chance (51.6%), and that was associated with a (non-significant) p value of 0.0545.

Of course, that would be counted as a failure in a parapsychology experiment, but they obtained a more decisive result by looking at only the 12% of cases where people were most confident that they had picked the right interval. Then the probability of correctness went up to 60%, and the p value fell to 0.001.

I believe their conclusion, But I would expect psi sceptics to be concerned that there's no indication that the authors pre-registered an intention to focus on the trials where the subjects were most confident, and therefore to be suggesting that their p value should be corrected for the possibility of multiple hypotheses. 

Not sure what a psi sceptic is: someone who is sceptical of psi or a psi proponent sceptical of these results? I may be the latter as I wonder whether the choice was made due to sight or psi (psight?).

Not sure what a psi sceptic is: someone who is sceptical of psi or a psi proponent sceptical of these results? I may be the latter as I wonder whether the choice was made due to sight or psi (psight?).

I was thinking of people sceptical of psi. Given similar evidence in a parapsychology experiment, I think at best the sceptics would be saying it was interesting but needed to be replicated.

As for wondering whether it was psi, I agree the overall result is well within the range of the hit rates that parapsychology experiments produce, and in any case the p value would be inconclusive. The 60% hit rate in the highest confidence cases seems more clear-cut, but obviously even when they're most confident they've seen something the subjects are still wrong 40% of the time, so it's not straightforward:
https://www.nature.com/articles/ncomms12172

but obviously even when they're most confident they've seen something the subjects are still wrong 40% of the time, so it's not straightforward

Which is precisely why I'm ever sceptical about stats. They ignore subjectivity. What is confidence? 

Quick diversion: I like to play lazy sports like pool, darts and, recently, lawn bowls. Obviously practice makes any player better but there are moments when, at the moment of delivering the cue/dart/bowl, I have supreme confidence that it is going to hit home and it does. Anyone who plays such sports will tell you the same but they will also tell you - if they are honest - that it is practically impossible to recreate that confidence at will. Some superstars may be better tuned into their psyche and visualisation techniques might help but there are no guaranteed methods otherwise every golfer would be a Jack Nicklaus or better. 

So Psi testing like Radin does and P values might be interesting but, for me, they are missing the point. Psi seems to be something we - in general - have either lost (and some animals seem to be better at it than we are) or that we are not "meant" to have access to. At least for now. So the sceptics can dismiss testimony as anecdote because the experience is not repeatable to scientific standards. The lab test effects are so small that they are inevitably doubted. Even I look at those results and think, so what? It just doesn't happen on demand - it is spontaneous and requires God knows what conditions to be right in that precise moment.

In Vaziri’s experiment, three volunteers sat in total darkness for around 40 minutes, and then stared into an optical system. When they pushed a button they heard two sounds, separated by one second. Sometimes, one of the sounds was accompanied by the emission of a photon. The participants had to say on which occasion they thought they saw a photon, and how confident they were (on a scale of 1 to 3) about their sighting.

I wonder how the results would compare if the apparatus was modified such that an opaque screen was fitted to stop any and all photons from reaching the eye?

That would turn it into a typical psi guessing experiment. Which may be what it is in any case. Until the experimenters eliminate psi from the results, they haven't demonstrated that the photon is literally being seen.

I wonder how the results would compare if the apparatus was modified such that an opaque screen was fitted to stop any and all photons from reaching the eye?

That would turn it into a typical psi guessing experiment. Which may be what it is in any case. Until the experimenters eliminate psi from the results, they haven't demonstrated that the photon is literally being seen.

The trouble is, there's no way of eliminating the possibility of psi from any experiment, particularly if it's considered in the broad sense that includes experimenter psi acting on the experiment as a whole. I don't know whether this lies behind some of the more extreme scepticism, but if the existence of psi were accepted, a large part of experimental science would become questionable. Not just the "softer" sciences like experimental psychology, but also multi-billion dollar physics projects like the Large Hadron Collider. In the worst case, that could turn out to have been the world's most expensive micro-PK experiment.

The trouble is, there's no way of eliminating the possibility of psi from any experiment,

But the converse in this case is not a problem. It is possible to eliminate the possibility of direct visual detection of the photon.

But the converse in this case is not a problem. It is possible to eliminate the possibility of direct visual detection of the photon.

True. And if the effect were still seen that would put the cat among the pigeons. Though I don't think we could be sure, if it's a psi effect, that it would still be seen, given that the expectation of the experimenters would be different.

True. And if the effect were still seen that would put the cat among the pigeons. Though I don't think we could be sure, if it's a psi effect, that it would still be seen, given that the expectation of the experimenters would be different.

Experimenter expectation I agree. I was thinking that the participant should not be told, at least they could be told of the various possibilities, but during a particular test they should not know what the configuration of the apparatus is.