Do unusual objects attract our attention faster?

It shouldn't take you long to notice what's wrong with this picture:


ResearchBlogging.orgObviously Nora is defying gravity in this shot -- you can't help but notice it. But in your first glance at the photo, how quickly do you notice what's wrong? Do you spot the oddity faster than you'd notice Nora in the original, unaltered picture?


A 1978 study by Geoffrey Loftus and Norman Mackworth found that people respond quicker to unusual or inappropriate objects in line drawings, such as an octopus instead of a tractor in a farm scene. They moved their eyes an average of 7 degrees to fixate on the unusual objects. But in 1999, a team led by John Henderson found the opposite -- people spotted a microscope in a bar scene no sooner than a cocktail glass in the same location, and only moved an average of 3 degrees from where they had been looking previously.

Why the discrepancy? It's hard to know: Maybe, for example, Henderson's team's drawings were just too complicated for viewers to spot differences easily. Or perhaps a microscope in a bar is less unusual than an octopus on a farm. How can we ever measure what's "more unusual" in a given context?

A team led by Mark Becker has developed a different way of inserting oddities into a scene. Instead of replacing one object with another, they altered parts of photos using Photoshop, as I did with the picture of Nora above. To reduce the chance of viewers looking for unusual parts of pictures, they had volunteers look at just two pictures: one altered scene and one unaltered one, from this set (if they saw the altered face, they saw the unaltered truck, and vice versa):


Ninety-four viewers' eye movements were tracked as they fixated on a small cross in the middle of the screen, then viewed each picture for 8 seconds. Here are the results:â¨


Whether people saw the altered truck photo or the group portrait, they fixated on the altered object significantly faster than the unaltered object. They also fixated on significantly fewer objects before moving to look at the altered object compared to the unaltered object. They moved their eyes an average of 8 degrees to the altered objects, suggesting that unusual objects can be located even when they are far from where we're looking.

The team conducted a second experiment where the color of objects was modified -- a man's was turned green in one picture, and a stop sign was green in a second picture. In this case the results were mixed: while the hand-modification matched the results of the first experiment, there wasn't as clear an advantage for the stop sign. As you can see from this picture, another object in each photo was modified to be green in the "unaltered" photos to make sure people weren't simply attracted to the color green.


People not only looked at the green stop sign faster, they also looked at the green license plate faster than the unaltered license plate. But since in California where the study was conducted, plates are normally white, people may have seen the plate as just as "unusual" as a green stop sign.

So, this research suggests that we do notice unusual parts of a scene faster than "normal" parts of a scene. It's also possible, the researchers say, that different types of unusual objects might do more (or less) to attract attention. Threatening things, disgusting things, faces, or other objects might each have a different weighting in the internal competition to attract our attention. But this study does make it clear that some objects definitely do attract attention faster than others.


Becker, M.W., Pashler, H., Lubin, J. (2007). Object-Intrinsic Oddities Draw Early Saccades.. Journal of Experimental Psychology: Human Perception and Performance, 33(1), 20-30. DOI: 10.1037/0096-1523.33.1.20

More like this

The picture looks like it was stitched together from a mountain scene and one from a hiking scene. It's rare to see climate that has fresh green grass and no trees at that sort of altitude, even though there any many tall trees in the mountain scene.

Of course, this kinda get defenestrated when we are looking for something specific. As classic examples show, when we are looking for something specific, such as the number of shots taken in a few minutes of basketball footage, we will often ignore unusual events or items, such as a person walking through the game dressed in a costume and waving.

nice post though, human perception is always fun to read about.

oh, man. My brain must be different from everyone else's because I thought the thing wrong in the picture was the mountains! Nora looked like she and the hill she were on were just normal as can be to me. I don't know how I feel about that.

Also, in response to Freiddie's comment: There is a tree line above which trees will not grow on a mountain. But grass grows above it. I've seen many places looking like this picture in Colorado & Washington.

I also thought the first picture was wrong. The flowers grow in the same direction as 'tilted Nora,' so I was immediately drawn to the moutains as looking tilted relative to the grass. I barely noticed Nora until I compared the 2 pictures.

I wonder if the noticed objects aren't attracting attention not becasue they are unusual, but because they stand out at a low-level of processing. As far as I can tell, they are all breaking some fundamental visual features that will tend to trigger early attention.

Especially for the green-colored things I know my reaction was "that looks fake", not "the hand is green". The color is just slathered on afterwards, and badly so, making it really obvious it wasn't part of the original picture (and thus standing out in early attention processing as well). The only exception is the cup, which is seamless, follows the lighting cues and is pretty well done.

I would want to see a complementary picture test, where you use a skin-colored mug, and color the subjects hand green with food coloring. Then paint the hand pink and the cup green afterwards, with the same lack of care as here.

The truck also breaks all expectations on lighting and perspective in the rest of the scene. Actually flipping a real truck is too expensive (would love to see the equipment grant proposal though) but probably necessary to avoid early attention influence. And faces are enough of a special case for humans that I'd not be surprised if any facial manipulation is picked up almost immediately.

Great - I am red-green colorblind. Couldn't tell one lick of difference in the last set of photos. :( Even when I read what the difference was I had a hard time convincing my brain that the stop sign on the left was green.

Well, if you've gone this far withoout noticing, it can't be all that important, right?

And it's the stop sign in the right picture that is green, but the license plate in the left. (Sorry.)

i like reading studies that tell me what i think i already intrinsically knew. it is like a reaffirmation of my humanity...
humans are curious. it is probably a evolutionarily perpetuating species characteristic. if you are not looking at someone and they say, "do NOT look at me right now," what is the first thing your brain wants to do? right. and there are myriad possible reasons that it would be so. it could be a survival instinct-- if there is something out of the ordinary going on, defensively we should want to know about it, correct?

By bryan simmons (not verified) on 12 Mar 2008 #permalink

nice post though, human perception is always fun to read about.

Posted by: redx | March 11, 2008 3:58 PM

I agree! Our brains and minds and how they are hardwired is fascinating! Also studying the differences between individuals, re: stronger and weaker brains development and chemistry could make for a full fascinating lifetime!
Dave Briggs :~)

What interests me most is that people have posted comments that say, in essence, "Oh, I got it wrong!"

Assuming every commenter is intelligent and credible (and I do): 1. why not question the universality of the study, 2. why not assume you've developed your mind in a way that can override the most ordinary response?

For example, I'm an artist/asst. professor of art (which = lots of training looking at pictures). I almost always get these things "wrong".

So, as far as the mountain image goes, the corner-corner diagonal line of the grassy mountain is a very strong visually dominating element. Having that combined with the angular intersection of the mountain's horizon line, I'm sort of surprised anyone can force their eyes to the hiker within the first 2 or 3 seconds of looking at the image. The unusual strength of the intersection of those two lines (and the knowledge of the existence of Photoshop) could lead a person to assume that that is what's "wrong", rather than the hiker.

I'd like to suggest that any study of this kind is flawed if it does not account for the influence of design/compositional elements.