[vhfpmr (OP)]

The title says it, what is the smallest increment in brightness that the eye can detect? I've tried consulting Professor Google, but I can't get him to spill the beans. From my student days I recall that the resolution of the ear is about 1dB for changes when the listener is present, and 3dB otherwise, so I'm thinking that the eye is likely to be of a similar order. Does anyone know?

[Colin2B]

Try this as a starting point http://math.ucr.edu/home/baez/physics/Quantum/see_a_photon.html

[vhfpmr (OP)]

Thanks Colin, but it was the resolution (minimum change) that I was looking for, rather than the sensitivity (minimum intensity).

[RD]

... it was the resolution (minimum change) that I was looking for ...

Maybe "contrast threshold" ?

The contrast threshold can be defined as the minimum contrast that can be resolved by the patient.

https://en.wikipedia.org/wiki/Contrast_threshold#Contrast_sensitivity_and_visual_acuity

[evan_au]

Normal JPEG images use 8-bit encoding, allowing 256 shades between zero and full intensity, for each of the red, green and blue colors.

If you take a photo of a uniform blue daylight sky, and examine it on a computer screen, you can see oval areas of equal color, surrounded by ovals of slightly different brightness. So the human eye can easily detect brightness differences of 0.4%.

Some cameras are now promoting high dynamic range, which allows smaller intervals between black and full intensity, which will make these oval artifacts in the sky less obvious.

[vhfpmr (OP)]

Thanks Evan/RD, I wasn't thinking of contrast at the time I posted, so I hadn't noticed that the question is a bit ambiguous.

Imagine you are in an artificially lit room, you then leave the room to make a cup of tea and whilst you're gone I adjust the brightness of the light. What is the smallest change in light level you would notice?

The contrast chart on the Wiki page was quite interesting, I hadn't seen that before.

[RD]

Imagine you are in an artificially lit room, you then leave the room to make a cup of tea and whilst you're gone I adjust the brightness of the light. What is the smallest change in light level you would notice?

Feedback-control via changes in pupil-size, and inhibitory neurones , will interfere with that tea-break experiment. I think a side-by-side comparison is your only option to find the smallest detectable change in brightness. 

[vhfpmr (OP)]

They won't interfere with the experiment, because the experiment is testing the question I'm trying to answer. Put it another way: if someone was trying to sell you a brighter light, how much brighter would it have to be in order to notice any difference?

[Colin2B]

What RD was pointing out is that the brightness of the room you make the tea in will affect the result. Eyes adapt eg night vision, colour of the light will also affect what you are measuring eg make tea in a room with red light you will get a different result with white light of the same intensity.

[chiralSPO]

The smallest difference that registers consciously is probably much larger than the smallest difference that is registered unconsciously.

Perhaps you could test how quickly people are able to read a reasonably long passage depending on the brightness of the room (the longer the passage, the more resolution you're likely to get...)

[RD]

... Put it another way: if someone was trying to sell you a brighter light, how much brighter would it have to be in order to notice any difference?

Your vision is constantly adapting to the brightness of the scene you are presented with , like a camera with auto-exposure.  The amount of Depth-of-Field would be a clue your pupils had changed size, ( even though subjectively you perceived the brightness as unchanged ).  As mentioned by Colin2B, if lighting is incandescent, a change in the colour-temperature could be a give-away sign that the brightness had changed.

[Colin2B]

Well, you live and learn (sometimes!).
I think this is what you are looking for http://www.telescope-optics.net/eye_intensity_response.htm
Quite unlike the ear's sensitivity which varies with frequency and spectral content

[vhfpmr (OP)]

Thanks Colin, that's just what I was looking for. It's an excellent website, I could read it for hours, what keyword did you have to feed Google? I see from Fig.238 that the minimum detectable change varies at low light level, but from moonlight to daylight it's fairly constant at between 2-5%. I'm assuming that those measurements were made with the subjects present at the time the level was changed.

Yes, the ear's sensitivity to change varies a bit with level and frequency, but figures from Reisz, and Dimmick & Olson are comparable with what I quoted above from my student days.

[Colin2B]

what keyword did you have to feed Google?

Weber's law
(I cheated and asked my wife)

Yes, the ear's sensitivity to change varies a bit with level and frequency, but figures from Reisz, and Dimmick & Olson are comparable with what I quoted above from my student days.

Yes, we were taught Reisz but I was doing some background research for a presentation and came across some more recent work '80s?. All very similar ballpark but different methodologies, this used pink noise rather than beat frequencies. I must revisit this as I have a project where it might be relevant.

[AndroidNeox]

For a healthy eye, about 1 in 7 photons get absorbed by rods or cones. Under optimal conditions, a person can detect a single photon (that one of seven).