What causes what? Depends on where you're looking.

Take a look at this video (click on the image to play). It's pretty clear what's going on -- the green dot bumps into the red dot, causing it to move:


But what about this one?


With this movie, it's harder to say: some people would say the green dot passes through the red dot, turning red and then moving on. Others would say the green dot launches the red dot, as before. It's an ambiguous figure.

But now look at this new movie, with two sets of dots. What's happening with the top pair of dots: launching, or passing through?


Most people would now say that the green dot on the top launches the red dot, even though the actual animation is identical to the ambiguous second animation above. The presence of the non-ambiguous launching event below it causes viewers to perceive the ambiguous event as a launching event. Brian Scholl and Ken Nakayama made these observations in two studies conducted in 2002 and 2004.

But Hoon Choi, working with Scholl, wondered if other types of grouping can also affect how we perceive causal relationships. They found the same effect even when the events were proceeding in opposite directions, as long as the two dots being "launched" were connected with a line:


It turned out that what mattered in this display was the presence of the connecting line after the apparent collisions. If the line disappeared at that point, then the ambiguous event at the top was again more likely to be perceived as a pass-through rather than a launch.

Next Choi and Scholl tried grouping the event differently. Here, the red dot ambiguously meets the green dot. But when the three green dots above move along with the red dot, most people perceive a launching event:


If the three dots above the event don't move, then most people see it as a pass-through:


This effect holds whether there are 1, 2, or 3 dots over the event. Even if a single dot is placed in the top position farthest from the event, it is still perceived as a launching event significantly more often than a pass-through, as long as it moves along with the "launched" dot.


Now, consider a slightly different event, when one dot only partially overlaps the other dot:


Most people will see this event as a launching event. But now take a look at this animation. Your job is to decide whether the blue dot launches or passes through the green dot:


Now, you're likely to perceive the blue dot as passing through, even though the animation of that sequence is identical to the previous movie. The only difference is the addition of an ambiguous event above it. So both pass-through and launching perceptions can be induced, depending on the context of the animation.

Next, take a look at this animation. Your job is again to decide if the blue dot launches or passes through the green dot. In addition, focus your attention on the row of the screen where the yellow arrow appears.


Looks like a launching event, right? Now do the same task, with the arrow pointing to a different row:


When you focus on the ambiguous event above, you're more likely to see the same animation below as a pass-through instead of a launch.

Choi and Scholl documented all these effects in a laboratory setting, with 10 to 12 participants repeating each task 20 times. Their conclusion? Nearly 100 years after Gestalt psychologists developed principles of grouping, suggesting that much of our perception of causal relationships is due to how we group objects, this research suggests that grouping does not explain all of how we perceive causal relationships. Instead, the critical factor appears to be where we focus our attention. In some cases, Gestalt grouping principles lead us to direct our attention in a way that affects the results as expected, but in other cases, by simply directing our attention to a focal point, the researchers were able to manipulate our perception of causality in ways that are not predicted by Gestalt principles.

Choi, H., & Scholl, B.J. (2004). Effects of grouping and attention on the perception of causality. Perception & Psychophysics, 66(6), 926-942.

More like this

This was a sort of failure for me, because I see all events as "launching" -I would have never thought by myself of describing any of them as "passing through", and even trying I found it difficult to percieve them in that way. Is that very uncommon?

Even this one? About 90 percent of viewers see this as a pass-through. But that still means that almost 10 percent see it as a launch. None of these tasks will work for absolutely everyone....

I'm with you, Alejandro...although I can sort of distinguish between a typical pool ball type launch and a circle that appears to launch with its back side, almost as if it had a hook on the other end.As a side note, Brian Scholl was one of the best professors I had in college. I'm definitely hot for visiual perception studies now.

By ThePolynomial (not verified) on 09 Feb 2006 #permalink

It's possible that the lower-res online versions are tripping you up -- that the actual animations used in the experiment are smoother. One thing I noticed is that if I play the video twice, the animation is smoother. Anyone else see the "pass through" effect, especially on the animation I link in my earlier comment?

One other point -- in the original experiment, participants were trained to see each type of event, then asked to report on which event they saw in subsequent animations.

The first "passing through" event was quite clear to me, and at least one of the other early ones, but all of the latter events seemed like "launch" events to me. I may have started to anticipate that after watching enough of them.

My perception matches the most common interpretation very strongly for all cases except the last of the close-together group (this one). That one looks like top and bottom are both launch events, no matter where in the frame I look, or how many times I view it.

Okay, now in watching it again whilst checking my link, it sometimes looks like a pass-through. Weird!

By Johnny Vector (not verified) on 09 Feb 2006 #permalink

Except for the first one I saw every one as a pass through event. I do not know why

I'm with Alejandro on this one as well... all the events look like launches to me and I couldn't even force myself to view them as pass-through after multiple reviewings.

I'm with Alejandro too! They all seem like launches. It wasn't until I read the part about "green changes colors as it passes through" that I even thought of the pass-through perception. The overlap events just look like "springy" launches to me.

By James Gambrell (not verified) on 09 Feb 2006 #permalink


This is fascinating stuff. I saw all the animations as you predicted except one of them.

Last night I listened to world famous autism researcher, Uta Frith give two presentations. At one point she was showing the Ebbinghaus illusion.

She said that autistic people are less likely to be fooled by it.

I have Asperger's syndrome and so was disappointed to hear that, since I am fooled by the Ebbinghaus illusion.... but as I sat there looking at her slide from the audience, for some reason I looked at it with my peripheral vision and the illusion disappeared.

I could see plainly that the central circles were the same size.

I wondered if that is how the autistics are avoiding the illusion.

Mottron has a paper out about autistic perception it's in the newest Journal of Autism and Developmental Disabilities. He talks about the way autistics use peripheral vision...

anyway. Thanks for sharing the animations and the research. Very cool.

They all seem like launches to me also. I wonder if watching one after another doesn't have a similar influencing effect as watching two together.

Excellent page. Thanks for posting. I will share this with my K-12 colleagues and expect it will spawn some excellent discussions and reflections.