The Gonzo Scientist

Results From the Science Dance Match-Up Challenge

See allHide authors and affiliations

Science  05 Jun 2009:
Vol. 324, Issue 5932, pp. 1262
DOI: 10.1126/science.324_1262b

The results are in from the Science Dance Match-Up Challenge. Between 17 April and 17 May, hundreds of people around the world tried to guess which of four scientific research articles inspired which of four modern dances commissioned by AAAS (publisher of Science). The results are impressive but also surprising--that is, if you think access to information helps you make complex decisions.

These are the correct matchings between dance and science:

Dance 1: Scientist C
"The Business of Building" was choreographed by Jenn Liang Chaboud, working with Vince LiCata on "Salt Dependence of DNA binding by Thermus aquaticus and Escherichia coli DNA Polymerases."
Dance 2: Scientist D
"Q minor" was choreographed by Christopher McCray, working with Markita Landry on "Mechanism of Force Generation of a Viral DNA Packaging Motor."
Dance 3: Scientist B
"Whoosh" was choreographed by Helena Reynolds, working with Miriam Sach on "Unified inflectional processing of regular and irregular verbs: a PET study."
Dance 4: Scientist A
"Sun Starved" was choreographed by Chloe Jensen, working with Sue Lynn Lau on "Loss of ARNT/HIF1β Mediates Altered Gene Expression and Pancreatic-Islet Dysfunction in Human Type 2 Diabetes."

Question 1: Can modern dance encode science?

If not, then people should have done no better than random at matching the four dances with the four research articles that the choreographers used to create them. But overall, people guessed far more accurately than random. The 341 people who took part made an average of 1.86 out of 4 possible correct matches between the dances and research articles. That's far more accurate than the expected 1 out of 4 correct if everyone guessed randomly. So the rigorously tested answer is yes, dance really can encode science.

Question 2: Which group performs better at decoding science-based dance, online or live audiences?

Credit: Katrien Kolenberg

Over the course of this 1-month experiment, people were exposed to the dances and made their guesses under one of two conditions. A total of 229 of them viewed the dances and made their guesses online here at Science, and 112 people took part in eight live events around the world. I personally hosted two of these events: one at the ScienceGallery in Dublin, Ireland, and another at Harvard University. The others were homegrown experiments in the United States, Canada, Sweden, and Austria. (See full event listing here.) The organizers of those experiments did not know the correct answers.

While it may be more fun to be part of a live audience, I expected the online guesses to be more accurate. After all, the online audience had access to complete information. Not only did they have the summaries and abstracts of the research articles, but they also had the full papers, links to the scientists' and choreographers' homepages, and even a transcript of a research explanation that one of the scientists sent to her choreographer during the creative process. The live audiences only got the titles and summaries of the articles. They watched the very same dances on video, and they were forbidden from seeing each other’s answers, just like the online population. So the only difference seems to be that live audiences made their guesses in the same physical location and had access to far less information.

And yet, the live audiences were more accurate than the online readers of Science. The average live audience member guessed 2.06 out of 4 correctly, compared to the Science online average of 1.76, and this difference is significant (p < 0.05, two-tailed t test).

Question 3: So who won?

Congratulations to the clear winner, the University of Vermont! A live audience led by molecular biology Ph.D. student Ramiro Barrantes-Reynolds got an average of 2.64 out of 4 guesses correct, 50% more accurate than online readers of Science (n = 28, p = 0.001, two-tailed t test). Impressively, nearly half of this group guessed all four dances correctly.

"I explained [to the audience] the importance of independent samples," says Barrantes-Reynolds, a University of Vermont Ph.D. student studying protein evolution. "No cheating, no commenting in between, and people seemed to take this very seriously."

Did his audience have any advantages? Perhaps. At least half of them had "dance experience," says Barrantes-Reynolds, who has more than a decade of formal dance training himself. Also, nearly all of them are in the sciences, "mostly molecularly oriented." That might explain it.

But it doesn't solve the mystery of why live audiences seem to be smarter in general. It's unlikely that the online experiment systematically attracted people with less science or dance expertise. Nor is access to information likely to make the difference. (Probably few online participants took the time to read the full papers.)

I propose a simple explanation. Being part of a live audience focuses your attention in a way that staring at a lonely computer screen never can. It's equally true of art and science.

Full results and analysis:

View Abstract

Stay Connected to Science

Navigate This Article