Advertisement

Dance researchers floored as study reveals computers can trace our steps

Disco divas rejoice! Your daggy (or divine) dance moves are unique. And what’s more, computers could use them to recognise you.

Feb 05, 2020, updated Feb 12, 2020
Your dance moves are unique, even if they are as terrible. (Photo: Getty Images: Jessie Casson)

Your dance moves are unique, even if they are as terrible. (Photo: Getty Images: Jessie Casson)

Researchers in Finland used motion capture to record 73 study participants dancing freely to eight different music genres: blues, country, dance/electronica, jazz, metal, pop, reggae and rap.

Then they ran the recorded data through a machine-learning algorithm to see if it could correctly identify what music genre the participants were dancing to each time, based on their dance moves alone.

While the algorithm didn’t perform very well at picking the genre of music — its best results were identifying metal and jazz, with 53 per cent and 35 per cent accuracy respectively — it could correctly identify which dancer was shaking their tail feather 94 per cent of the time.

“We were really sort of floored when we saw the results,” said Emily Carlson, a music psychologist at the University of Jyvaskyla and lead author of the study, which was published in the Journal of New Music Research last month.

After a few weeks of checking that they hadn’t made a mistake in their analysis, Carlson and her colleagues realised they had stumbled across what appeared to be like a fingerprint — a combination of moves — that was unique to each individual regardless of the type of music they were dancing to.

Music makes us move

There has been a lot of work done in the past looking at how we move to a groove.

Previous research has shown that our dance moves can reveal how much we empathise with different dance partners and even aspects of our personality, like how neurotic or extraverted we might be.

“There’s evidence that … the way that we perceive the emotion in music can affect the way we move,” Carlson said.

So the researchers had good reason to expect that there would be some kind of link between people’s dance moves and the genre of music they were dancing to.

“I think there is, I don’t think that we move exactly the same way to every genre,” Carlson said.

“But I think that particular movement piece that we ended up working with, somehow captured an individual aspect that was there across genres.”

To record their participants’ dance moves the research team put them in suits that had markers on them that look like little balls.

They used optical motion capture cameras that shoot out infrared light that bounces off the markers and is then recorded.

This allowed the team to capture a stick figure of the person moving in three dimensions, like you see in the video above, stripping away other physical characteristics like the person’s shape, size and the clothes they were wearing.

“We’re really only recording pure movement,” Carlson said.

When she delved back into the literature to find out why the individual dancers might be so easily recognised by the algorithm, she discovered that previous research had shown humans can identify other people based on their movement, even from a distance, although little is know about how we do this.

InQueensland in your inbox. The best local news every workday at lunch time.
By signing up, you agree to our User Agreement andPrivacy Policy & Cookie Statement. This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

“There really is something kind of biological about it, that was really interesting,” Carlson said.

The machine-learning algorithm was able to replicate that process, although in a more transparent way by telling us which combinations of markers it’s using to distinguish between different people, she added.

Everything we do has a fingerprint

Kim Dunphy, a lecturer in dance movement therapy at the University of Melbourne, wasn’t surprised that there might be a unique individual fingerprint underlying everyone’s dance moves.

“People have particular movement preferences and styles that are recognisable,” Dunphy said.

For example, someone might move very suddenly and show a particular directness in their movements, and this would be recognisable across different dance genres they’re moving to.

It shows there are so many ways in which one can synchronise — or dance — to music, said music psychologist Professor Bill Thompson of Macquarie University.

“People will naturally kind of gravitate to a way that suits their own kind of personal body style and embodied knowledge in a sense because people … have their own way of interacting with the world,” Thompson said.

And that could have implications for how we learn or how we teach people to dance, he said.

“When you’re trying to learn how to dance, you’ve really got to draw on your own personal way of responding to music and build on it, and not try to be like somebody else because you’re not going to move like that other person.”

– ABC / Suzannah Lyons

Local News Matters
Advertisement

We strive to deliver the best local independent coverage of the issues that matter to Queenslanders.

Copyright © 2024 InQueensland.
All rights reserved.
Privacy Policy