The smallest increment of time a human can percieve?

[Researcher X]

I tested myself with 120fps in Flash, and I could detect a single changed frame (and the next frame is the same as all the other before the change), so that means I can perceive a 120th of a second. Flash doesn't actually go any higher, but I'm wondering at what point I wouldn't be able to see that quick change anymore.

Have there ever been tests done on this?

 

[mgb_phys]

It doesn't quite mean that - your eye/brain can detect short lived large changes by simply integrating the signal. It depends more on how different the frame was rather than just the speed.
You could for instance detect a femtosecond laser pulse if put in front of a sufficiently powerful laser.

 

[Researcher X]

I'm sure your mind would stretch out that laser pulse in terms of how you perceive it though. I can't fathom a femtosecond passing. I think sharp changes in light tend to linger in your eyes. Like an old style camera flash.

I understand that your eyes capture all light, but there must be a limit as to the periods of time your mind can divide and document.

An image appearing that doesn't noticeably change in light intensity would be a more accurate test, right? Like in Flash, I tested it with a mid tone green background and then a large red squiggle of the same luminosity in the center appearing for the single frame. Very quick even though I saw it. Sometimes the image looks broken up a little. Maybe this is based on concentration and my mind is missing part of the image, or maybe my computer can't display it effectively every time.

Let's say you are using google, about to make a search, and then the whole screen is replaced by a picture of a grizzly bear for a 1000th of a second while you're staring at it. Would you notice anything at all?

 

[Ygggdrasil]

IIRC, the integration time of photoreceptor cells in the eyes is on the order of 0.1-0.2 seconds. That is, the signal produced by photoreceptor cells is proportional to the number of photons collected by that cell in 0.1-0.2 s. Perhaps a better test would be to insert a dark frame (as opposed to a frame of a different color) and see whether you can see that dark frame.

 

[Borek]

I am not sure I understand the question - seems to me like you have not defined what you want to check. "Incerement of time a human can perceive" is not the same as "eye inertia" - and basically all things proposed so far revolve about the latter.

 

[Researcher X]

Eye inertia?

Well, I'm thinking from the line that: light enters the eye, and as long as it's intense enough to activate the cells, your eyes capture all the light that goes into them in whatever time period, meaning that the eyes are quicker than the brain. Our brains can't perceive time frames that are too low, so the test would reveal what is too quick for the brain to perceive.

 

[Borek]

I don't think you will be able to test the brain - seems to me like you will be testing eye characteristics most of the time.

 

[Ygggdrasil]

Here's an idea for a test that would involve how the brain processes the signal. It's based on the phenomenon of aliasing: frequencies that appear lower than they actually are due to undersampling of the signal. An example of this is when you look at the wheels of a car that is traveling very quickly (especially in a video) and the wheels appear to be spinning backward.

First create an image that can rotate very quickly and you can adjust the frequency of rotation. Next, increase the frequency of rotation until you see some strange effects (e.g. the rotation seems very slow or the rotation seems to be going in the opposite direction than it should be). From the Nyquist Theorem we know that aliasing will occur when the sampling rate is less than twice the frequency of the signal being observed. Thus, the frequency at which you observe aliasing will allow you to obtain an estimate of the sampling rate of your brain.

 

[Andy Resnick]

I tested myself with 120fps in Flash, and I could detect a single changed frame (and the next frame is the same as all the other before the change), so that means I can perceive a 120th of a second. Flash doesn't actually go any higher, but I'm wondering at what point I wouldn't be able to see that quick change anymore.

Have there ever been tests done on this?

You could be asking several related questions, including the dynamics of a (instantaneously) stimulated rod or cone, or how fast a flashing light can be perceived to be flashing rather than continuous, or how long a stimulus must exist in order to register a response, or others. The same questions can be asked for sound or touch, for example.

There has been a considerable amount of measurements on the details of visual perception, and of the sensitivity to moving objects as a function of intensity, color, size, etc. (in addition to the other senses) Unfortunately, my reference book has been packed up for moving, and I won't be able to retrieve it for some time.

I recommend you look for a physiology text covering sensory perception- a good one will be filled with data. Unfortunately, the interweb search engines turned up stuff that goes though how the senses work, rather than the measured data.

 

[Manchot]

It's difficult to quantify, because it also depends on the speed of whatever you're looking at. I once did a test with a pulsed LED operating at a few kHz, and though I couldn't really perceive the pulsing while the LED was stationary, I could easily see it when I moved it fast enough. By the same token, it was perceivable if I moved my eye fast enough.

Also fun: playing with an optical chopper to produce visual difference frequencies. Tune the chopper to a multiple of 60 Hz and look at an analog oscilloscope screen through it, and you see a number of stationary dark regions on said screen. Detune it a little bit, and that region flickers at a rate given by the difference frequency.

 

[atyy]

I tested myself with 120fps in Flash, and I could detect a single changed frame (and the next frame is the same as all the other before the change), so that means I can perceive a 120th of a second. Flash doesn't actually go any higher, but I'm wondering at what point I wouldn't be able to see that quick change anymore.

Have there ever been tests done on this?

http://www.ncbi.nlm.nih.gov/pubmed/...nel.Pubmed_DefaultReportPanel.Pubmed_RVDocSum

 

[apeiron]

You are asking a loaded question. In principle, eyes can detect light down to a single quanta, which is pretty brief. But then you need to put this in the context of the hierarchy of processing needed to experience this as "an event at a moment".

In rough terms, it takes the retina 50 milliseconds at least to integrate information. Then a minimum of 100 ms for the brain to integrate. And more like 300 to 500 ms to integrate "consciously" as an act of attentive awareness.

So to be aware of event x at time x, the brain is having to suppress quite a lot that happened to either side of it. There is an act of signal enhancement or noise suppression to make for a high contrast, brightly resolved event.

For this reason, the question what is the smallest moment that can be perceived must be explained alongside other questions, like how long before the brain is ready to pick up a second such event. (The attentional blink, etc).

The raw performance figures were the subject of much study by psychophysicists such as Wundt and Donder of course.

If you are really curious, there is also the opposite question of when do discrete events turn into a continuous blur. See cutaneous saltation as an example of such studies.

 

[Researcher X]

If you are really curious, there is also the opposite question of when do discrete events turn into a continuous blur.

I'm actually trying to find a frame rate in which a single frame is imperceptible in order to answer a question about subliminal build-up, where you can have things which are imperceptible on their individual time frames joining together to make a perceivable whole.

Movie frame rate doesn't cut it. 24 frames per second isn't good enough. 120fps is very quick but still perceptible.

If you reach the point where each frame is imperceptible on it's own, what will happen if you play a movie with a completely different image each frame? Let's say that a one frame image of a bear shown at 1000 fps can't be detected, but that a 500th increment can be, and in the next frame it is an image of a crocodile. This suggests that you will detect the sum of the two frames, so that by the time you detect the image change, you see an image of a bear and a crocodile simultaneously. You could make some interesting art or coded messages using this method, if I'm correct.

 

[apeiron]

To "see" each frame shown in succession - I reckon you would be lucky to process 3 to 5 fps.

In a stream of disconnecte images, you might be able to pick out one to focus on and be properly aware of at a rate of about 2 per second. So every now and then, as effectively your brain tightens up and then relaxes to trap an event.

There is no end of actual research into actual human performance here. If you explained your purpose - other than perhaps simple curiosity - I might be able to narrow down your research a bit.

If you just want to make art, there is stuff like pictures projected on to a fast moving wand that rely on phi effects. Much like writing in the air with sparklers.

 

[Researcher X]

To "see" each frame shown in succession - I reckon you would be lucky to process 3 to 5 fps.

Sorry to reiterate, but I don't think I've explained it properly.

I think you are thinking I mean seeing each frame individually without the illusion of motion, whereas I mean seeing the image in that frame full stop. After all, I could still see a change in 120fps. I figured this question would be one of time perception, but apparently in can be divided depending on what you are viewing and effects like intensity, magnitude of change, and so on.

Nonethless, my base question is: how quickly would a movie frame have to pass for me to miss that image? Like my example with the flash frame of the grizzly bear. Or it could be a flash frame of anything. If I was watching a movie and some guy was jogging, and then an image of Cthulu appeared for only a single frame. How high a frame rate would I need to hide that image from the human eye/brain in that single frame? Maybe it's not an exact number, due to color and the magnitude of the frame, but I'm sure there's at least a minimum, and I already know it's above 120fps.

It's just out of curiosity, but it would be interesting to experiment with my subliminal build up idea from this, if I can achieve the frame rate.

Perhaps a better test would be to insert a dark frame (as opposed to a frame of a different color) and see whether you can see that dark frame.

I tried this, and I can still see it, but it creates a very weird effect. I can only see the black flash in a rectangle of where my eyes are focused, not the full square. Possibly it's too quick for all of it to be processed by my brain?

 

[apeiron]

If I was watching a movie and some guy was jogging, and then an image of Cthulu appeared for only a single frame. How high a frame rate would I need to hide that image from the human eye/brain in that single frame?

Can't give you precise answer off the top of my head as it has been 30 years since I was really into the literature. But it this is certainly a much studied issue - all that 1960s subliminal advertising furore.

Your question is still too loose as well. You would notice a discontinuity - a sort of blink - before you had any idea of what the frame contained.

The controversy was over the idea you could take in such images at a subconscious level and then be primed accordingly (flash up a coke bottle before the film interval and crank up sales). Priming exists but does not work out so neatly.

But again, I think you would be suprised at how slow the frame rate could be and you would not notice an oddball frame for what it was. Google change blindness for instance.

 

[Pinu7]

Actually, according to wikipedia, it takes 200 femtoseconds for your eyes to react to light.
http://en.wikipedia.org/wiki/1_E-15_s

 

[Researcher X]

You would notice a discontinuity - a sort of blink - before you had any idea of what the frame contained.

Always? No matter how high the fps is? Or are you just saying that that's you would see it in the subliminal adverts of the 60's? The thing about that is that 24 fps is way too slow, though sudden and confusing, it's actually liminal, even though it was called subliminal.

 

[apeiron]

Actually, according to wikipedia, it takes 200 femtoseconds for your eyes to react to light.
http://en.wikipedia.org/wiki/1_E-15_s

That's just the first chemical reaction of the pigment. Then you have at least four levels of neural processing in the retina before the long trip up the optic nerve even begins.

Missing from this discussion is the role that anticipation plays in real vision. The retinal neurons actually are primed to fire predictively on the basis of developing expectation. So activity 200ms or more ahead of time then causes a cell to fire "as the event happens".

 

[atyy]

If you reach the point where each frame is imperceptible on it's own, what will happen if you play a movie with a completely different image each frame? Let's say that a one frame image of a bear shown at 1000 fps can't be detected, but that a 500th increment can be, and in the next frame it is an image of a crocodile. This suggests that you will detect the sum of the two frames, so that by the time you detect the image change, you see an image of a bear and a crocodile simultaneously. You could make some interesting art or coded messages using this method, if I'm correct.

http://en.wikipedia.org/wiki/Summing_localization