Wednesday 22 October 2014

Do people really not know what running looks like?

Faster, higher, stronger -
When we run, our arms and legs swing in an alternating rhythm. Your left arm swings back as your left leg swings forward, same with the right. This contralateral rhythm is important for balance; the arms and legs counterbalance each other and help reduce rotation of the torso created by swinging the limbs. 

It turns out, however, that people don't really know this and they draw running incorrectly surprisingly often. Specifically, they often depict people running in a homolateral gait (with arms and legs on the same side swinging in the same direction at the same time; see the Olympics poster). I commented on a piece by Rose Eveleth at the Atlantic about a paper (Meltzoff, 2014) that identifies this surprising confusion in art throughout history and all over the world, and that then reports some simple studies showing that people really don't know what running is supposed to look like.

Rose covered the topic well; I wanted here to critique the paper a little because it's a nice example of some flawed cognitive psychology style thinking. That said, I want to say that I did like this paper. It's that rare thing - a paper by a single author who just happened to notice something and think about it a little then report what he found in case anyone else thought it was cool too. This is a bit old school and I approve entirely.

Meltzoff reports a bunch of examples of artists drawing running incorrectly, with the homolateral rhythm. These include examples from Ancient Egypt, Greece and Rome, as well as from the Renaissance (including a da Vinci!) and more modern examples. It's hard to judge how common the error actually is, though; the specific examples are interesting but it's often couched in a story about how the artist did it right in lots of other works. Some of the examples are also depictions of people posing, which we will get to shortly. The funniest examples come from two drawing 'how-to' books which provide examples of the incorrect pose; people actually got taught this"

Meltzoff then runs a couple of little studies. He showed 30 people (a bunch of his mates, by the description :) an ambiguous stick figure and asked them to label the left and right arms and legs; only 5 got it right. He then asked 43 people waiting in an airport to do the same task with a different picture and a little over half managed it this time. 

Finally, he looked at pictures at A phoon is a pose where people stand 'looking like an athletic runner'; see the example. 
Phoons. Note the top person is doing it 'correctly' while the people on the bottom are doing it 'incorrectly'. Source:
He found only 35/243 pictures with people in the 'correct' contralateral pose; people don't seem to know what running is supposed to look like.

The problem with all this
I really don't know why artists get it wrong sometimes, and I think it's hilarious you can find instruction texts telling you to do it wrong. But the idea that we should be able to get it right because we all walk and run so much is flawed.

The basic cognitive idea lurking here is that what we learn when learning to run is an abstract representation of how to arrange your limbs over time in space. This representation would be a motor programme, a set of instructions for telling your limbs what to do. If we have this instruction set lurking in our minds labelled 'running', one hypothesis is that we should use that information any time we do anything running related, up to and including talking and drawing about it. Why not? The information is encoded in an abstract fashion, meaning that it's not tied to the execution of running, it's just a generally applicable set of information about running. Meltzoff's surprise is premised on the notion that using such a well learned instruction set shouldn't be riddled with this many errors.

His data actually point to the fact that this is not what we learn at all. When I learn to run, I'm actually to perceive the task relevant information and use this information to assemble a running motion that suits the local task demands (Is the terrain flat or hilly? Am I tired or fresh?). There is no abstract knowledge anywhere in the system; it's all grounded in information about the task. 

The phoons case is also interesting. Meltzoff assumes that just because the phoon instructions say 'look like an athletic runner' that this is what people are actually doing. What they are actually doing is trying to maintain a static balance on one leg, and it's my guess that standing in the homolateral pose is better for this kind of balance. I think that with your left leg down, having your left arm forwards and your right arm back out allows you to easily place your centre of mass directly over your base of support (the definition of 'balance'). Throwing your right arm and right leg out (which is what you have to do to be 'running') pulls your CoM to the right of your foot. I will test this when I get my Wii Fit balance board talking to Matlab :)
Me testing my hypotheses with my older brother circa 1978

The point here is that people's behaviour is shaped by the demands of the task they are actually solving, and that might not be the task you asked them to do. Asking people to pose as if running is actually asking them to stand on one leg in place, and from their point of view these are two very different things with, potentially, two different solutions. 

So I think people 'know' what running should look like, because we don't fall over that often while doing it. I just think we also 'know' how balance works and that matters more.

I really did like this paper, I promise! It made me smile and I will now always look at pictures of people running to see if they got it right. In a world of science driven by impact and looking flashy, this was just a bit of good old fashioned curiosity about something weird:
The most exciting phrase to hear in science, the one that heralds new discoveries, is not “Eureka” but “That’s funny...”
Isaac Asimov (1920–1992)
Meltzoff, J. (2014). Errors in the Making and Perception of Art Images of Human Gait: Psychology of Aesthetics, Creativity and the Arts, 8(3), 321-329. Download ($$)


  1. I have the same take as you on the phoon pose. It reminds me of this yoga pose:

  2. I find it interesting that all this concerns artistic depictions of running. An artist, I assume, would want to show as much of the whole body as possible. In the "real" running position, part of the body might be hidden between the crossing arm and leg. in the case of the phoons, the person posing would want to look towards the camera, which is easier (again, I assume) when arms and legs are in the "wrong" position, as the person would have more room to turn the head. It is possible that the artists chose instinctively the position that they feel is easier and/or more effectful.

    1. I actually tend to agree; Meltzoff talks about it in his Discussion and says a) he thinks artists could probably draw it properly without compromising the things you mention and b) there are lots of examples of people doing it wrong while nominally trying to be accurate. A little weak, I agree, but there it is.

  3. This is great. It inspired me (a week later) to write a post in a similar vein about the task of describing language.

    While thinking about it, I also discovered that when athletes are standing in attention preparing to take their first step in a run, they tend to match the forward leg to a forward arm, as well.