Very cool visual illusion confirmed (updated with new poll)

i-eca0cf2af9fc3ac4445c7dff7d8aab70-research.gifThere's nothing cooler for a perception researcher (or writer) than a new visual illusion. When I learned about this one, I spent half the day Thursday trying to recreate it, but I couldn't get it to work. Finally, in five minutes on Friday morning, I think I figured it out. (Update: Actually, as it turned out, I didn't figure it out at all. In the meantime, Chris from Mixing Memory pointed me to Shams' web page with much better demos. So let's try this again, using one of Shams' demos.) Play the movie and watch for a dot flashing in the middle of the screen. Make sure you've got the sound turned up on your computer -- the flash will be accompanied by beeps. How many times does the dot flash?

Let's make this one a poll:

If this works, most respondents should say the dot flashed two times. Actually it only flashes once. You can convince yourself of this by moving the slider slowly back and forth.

This illusion was discovered by Ladan Shams, Yukiyasu Katmitani, and Shinsuke Shimojo, who claim that the three auditory beats cause the phenomenon to be perceived visually. But is the illusion really a visual phenomenon, or is there some other explanation? In a separate study, Shams' team systematically explores the phenomenon. First, they showed viewers movies with different numbers of auditory stimuli accompanying the flash (they used beeps instead of drumbeats). Here's the result:


As the number of beeps increases, so does the perceived number of flashes. Next, they tried the same test and actually varied the number of flashes, while playing just one or no beeps. Viewers were accurate in reporting how many flashes they saw.

In these tasks, the beeps and flashes were spaced about 60 milliseconds apart. Will the illusion still occur, no matter fast or slowly the beeps are played? In a second experiment, the viewers saw movies with one flash and two beeps. One beep was always synchronized with the flash, but the second beep was played at varying distances from the first beep, up to 250 milliseconds before or after. Here's a chart of how likely viewers were to see multiple flashes


The phenomenon only occurs when the beeps are about 100 milliseconds or less apart. Here's a movie with beats about 200 milliseconds apart:

Do you see the difference from the first movie?

Shams's team also notes that this is a one-way phenomenon: You don't perceive multiple beeps when there are many flashes, but you do perceive multiple flashes with multiple beeps. Fascinating stuff!

Shams, L., Kamitani, Y., & Shimojo, S. (2002). Visual illusion induced by sound. Cognitive Brain Research, 14, 147-152.

More like this

You may have seen this illusion in a post from earlier in the week over at Cognitive Daily, but I thought I'd say a little bit more about it, and talk about a related illusion. First, click play (from Sham's demo site) If the illusion is working -- Dave at Cognitive Daily had a bit of trouble…
Last week, I talked about the sound-induced flash illusion, in which presenting a single flash with two or more auditory beeps caused people to see two or more flashes. This week, a study showing that the same effect can be obtained by replacing the beeps with touches. Here's the setup (…
There was plenty of interest in yesterday's audio-visual illusion. In case you missed it, I'll post it again here: Play the movie with the sound turned up. If the illusion works, then you'll see a dot flash twice, accompanied by two beeps. But actually the dot only flashes once. Unfortunately, we'…
Just about two weeks ago, I posted this visual illusion (if you haven't seen it yet, make sure to watch it with the sound turned ON): How many flashes do you see? In fact the dot only flashes once, but according to the study I report on in the post, the two beeps are supposed to throw you off. If…

I saw only one flash, but it wasn't until I read to the end of your article that I realized you intended for me to watch the movie with sound. Given that the early poll results contradict the expected findings (expected: most respondents will report multiple flashes; actual: 15 report single flashes and seven report multiple flashes), I'm guessing that I'm not only one who did not realize that sound's important for this one.

By Ami M. Echeverri (not verified) on 23 Oct 2006 #permalink

Ami, you may be right. I've changed the instructions. We had 23 votes for 1 flash and 6 votes for 2 flashes when I changed them -- we'll see if that makes any difference.

I find the conclusion is only true with me, when you don't know where the dot will flash. Knowing it will appear in a specific location, its simple to detect it only occurs once.

Uh-oh, still not working, even with the correct instructions. One difference between the study and my replication is that people participating in the study see the illusion over and over. I'm noticing that I seem to see the illusion more frequently now that I've played around with it. Anyone else experiencing this?

I think it's possible your instructions are tipping people off. I know I paid more attention to the sight than the sound, thinking that's where the trick was.

Of course, if it's a 'trick', it's not really an illusion is it? I mean, in an illusion the lines seems to bend and the circles to be different shapes even though I know what's happening.

"Of course, if it's a 'trick', it's not really an illusion is it? I mean, in an illusion the lines seems to bend and the circles to be different shapes even though I know what's happening."

Brian, I think there are two possible explanations for our lack of replication here. The first is that it takes awhile to accommodate to the illusion, so you're not going to see it the first time you watch the movie. The second is that I wasn't able to generate an appropriate "beep" -- the best I could do was a drumbeat that reverberated much longer than in the original study, so it may not stimulate the visual "flash" as readily.

In the original study, both naive and non-naive observers experienced the illusion, so it is a genuine illusion.

>> One difference between the study and my replication is that people participating in the study see the illusion over and over.

I hesitated between one and two flashes when I saw the movie for the first time. As I noticed immediatly that there were very close beeps, I suspected that the goal of the experiment was to make people see two flashes instead of one, what I verified by seing the movie another time. That's why I replied 1. But I may have said 2 during another experiment.

Does this somehow relate to numerical priming, anchoring or hindsight bias?

These audio-video illusions seem to fail again with me! Can someone post the theory as to why our brains should link visual and audio clues in such a fashion? I can imagine hearing a loud bang and imagining all kinds of things as something to run away from!

The first movie, I see very clearly just one flash, the second I am not sure, I think I see two. It's an odd experience anyway, looking at the screen and not knowing if I saw a second flash come with the second beep.

By G. Shelley (not verified) on 27 Oct 2006 #permalink

I saw one flash. When I heard the two beeps I "knew" it was designed to fool me into seeing two flashes so I still saw one. I guess I am not a good test subject.

All this reminds me that people often (sometimes?) report hearing a sound when they see a meteor. It happened once to me, and I couldn't find any physics that would account for the sound--the meteor is to far away for slow moving sound to be heard simultaneously, for example.

Are these related phenomena, somehow, I wonder?

I saw two events, but only saw one dot flashing. The two events are white changing to black, then black changing to white.

If I had not known what to expect, I might have thought I saw two flashes.

In my younger days, I might have saw two flashes. My 67 years of experience told me that the two events were not two dots flashing, but the dot appearing and disappearing, so my reasoning told me it was one dot.

When I went to the link that showed the same thing (with reversed colors) with two beeps and with one beep, It was obvious that the dot flashed only once with one beep. I didn't have to use any reasoning to interpret it. With two beeps, I noticed the two events (color change) much more dramatically, and had to use reasoning to conclude there was only one flash.