Do you recognize the faces in this picture?
Sure you do -- you could recognize the authors of this blog anywhere, even upside-down. It might take you just a bit longer to realize that something isn't quite right with the picture. I'll show you what the problem is at the end of this post.
We've known for decades that the human perceptual system is especially good at recognizing faces, but that ability breaks down in predictable ways when the faces are upside-down. While it takes us a bit longer to recognize objects when they are inverted, faces take even longer compared to other things.
For example, you might be able to tell whether two faces are identical or slightly different when they are upside down, but you'll be quicker to note a similar difference in, say, two houses. When people try to recognize inverted faces, different brain regions are activated compared to recognizing upright faces, but nonface objects activate the same regions whether upside-down or right side up.
But maybe faces aren't the only objects that are special in this way. A team led by Catherine Reed showed observers pairs of pictures of human figures like the one below.
The viewers simply had to indicate, as quickly as possible, whether the figures were in the same position or a different position. Some of the figures they saw were right-side up and others were upside-down. The viewers also saw pictures of houses that were the same or different to a similar degree. Here are the results.
Viewers were significantly more accurate when bodies were upright compared to when they were inverted. However, there was no such significant difference in accuracy with the houses. The same results held for reaction time. In a second experiment, viewers compared bodies and faces. Here are those results:
The results for bodies were remarkably similar to those for faces -- just like faces, it seems, we aren't as good at recognizing upside-down bodies. In a final experiment, half the bodies the observers saw were contorted to impossible positions, like this:
For error rates, the results were very similar to the body - face experiment. But measuring reaction times, a different result was found:
As before, for possible body positions, reaction times were significantly slower when the bodies were inverted. But for impossible positions, there was no significant difference in reaction time, whether bodies were inverted or upright.
Reed's team argues that the reason we are better at recognizing upright faces and bodies compared to upside-down faces and bodies is the same: we're experts at recognizing both bodies and faces (when they're right-side-up). It's something we do all the time. We're not so good at house recognition, so we're equally bad at recognizing both right-side-up and upside-down houses. The impossible bodies represent a sort of in-between case. We're better at these than houses, but still not as expert as we are with body configurations that we actually find in the real world.
Indeed other research has found that when people are trained as experts in a field, such as bird watchers, they have the same difference in recognizing upright versus inverted figures. We're all experts in recognizing bodies and faces, as long as they are right-side up.
So, did you spot the problem in the picture of me and Greta I showed you at the start of this post? Here's the same picture presented right-side up:
Yikes! Either I've been playing with Photoshop again or Greta is in serious need of more coffee!
Reed, C.L., Stone, V.E., Bozova, S., Tanaka, J. (2003). The body-inversion effect. Psychological Science, 14(4), 302-308. DOI: 10.1111/1467-9280.14431
- Log in to post comments
Dave, That's spouse abuse!
"That's just frikkin' weird," I muttered, returning my laptop to its customary, not-upside-down position.
+1 for more coffee!
Very interesting.
I'm wondering though whether part of the difficulty in spotting your photoshopping lies in the fact that Greta's cheeks are still showing telltale "smile lines." Would it be easier to spot an inverted mouth in an expression that did not involve telltale cheek contours?
Karl, I know -- I was going to do myself, but the beard and mustache were too much for my meager Photoshopping skills.
This reminds me of that old book, "Drawing on the Right Side of the Brain"-- don't know about the right-brain stuff, but shaking things up, confoundiing expectations, makes us see or not see afresh. It's a thinking outside the box thing; if you can get rid of the familiar tracks of thought, which can help us to eliminate distractions and get to the point most of the time, you can see the rest of the mostly unsalient world. Kids I think develop this around 6 or 7 and at the same time their artwork becomes sterile and conventional. I bet this applies to the recent research about improvising Jazz musicians too-- damn those frontal lobes, anyway! Um, this is incoherent, but I think there are some ideas in here.
You got me!
I actually looked at the first one thinking that Greta's smile looks so much vivid than yours :)
Nice trickery! The explanation could be somewhat like this.
We 'understand' the face better in normal 'gravity-wise' orientation, because plotting of the graphical co-ordinates of the face by the brain, is a task of the left hemisphere. When the picture is inverted, its the right hemisphere that is involved now. Drawing (NOT interpreting) the inverted figure, in comparison to the upright one, paradoxically, will be more faithful, since our right 'artistic' brain will be involved.
Curiously, I noticed immediately what was wrong with the photo. I'm prosopagnosic. I wonder whether these facts are related?
Interesting. Actually, the Exploratorium has an exhibit that involves the same principle:
http://www.exploratorium.edu/cmp/exnet/exhibits/group5/vanna/index.html
Like Mathematician, I'm also prosopagnosic -- and I also noticed the upside-down smile right away. Do prosopagnosics process right-side-up faces the way normal folks process upside-down ones? I know many of us cope by focusing on individual features and tend to see faces as a collection of parts, not as a whole.