insync

January 5, 2021 10:05 AM

Synchronized violin players reveal uniqueness of human network "An unusual experiment involving 16 violinists trying to synchronize their playing while wearing noise-canceling headphones yielded some intriguing results, according to an August 2020 paper published in Nature Communications. The study concluded that human networks are fundamentally different from other networks in terms of synchronized behavior because of our decision-making ability."
posted by dhruva (12 comments total)

This feels very Steven Reich.
posted by robotmachine at 12:54 PM on January 5, 2021 [2 favorites]

I'm having trouble understanding what's happening. Can the musicians hear each other on the headphones? Can they hear each other live? What is this "delay" they mention, is it just unavoidable lag in transmitting the signals or is there a delay imposed by whoever is controlling this experiment? Can the musicians see each other? Are they trying to synchronize with each other by visual cues (looking at the other musicians) as well as by sound piped into their headphones?
posted by MiraK at 1:57 PM on January 5, 2021 [1 favorite]

From the article:

>>>The participating violinists donned noise-canceling headphones and began playing the same musical phrase on repeat, without looking at or listening to the other players.

It seems like they are saying the violinists cannot see or hear any other players, but ---

>>>They could only rely on what they heard through the headphones, which were connected to a computer system.

What COULD they hear through the headphones?

>>>The researchers then introduced intermittent delays in signals between coupled violinists, varying the delays and the combinations of violinists.

What are coupled violinists? Are these violinists that who are selectively allowed to hear each other through the headphones? So... the violinists CAN hear each other through the headphones and they are sometimes artificially delayed.


>>>most network models assume that in such a frustrated state, each node will attempt to find a middle ground between all the various inputs. Instead, Fridman et al. found that the players reacted by adjusting their playing, quickening or slowing their tempo to better synchronize with their fellow violinists. ... Fridman told The Jerusalem Post, "In a state of frustration, [humans] don't look for a 'middle,' but ignore one of the inputs.

Okay.

So this experiment is about seeing if musicians can modify their own playing on the spot to keep themselves synced up with.... multiple and conflictingly delayed playing piped into their ears? If a machine had been given this task, it would have averaged out all the delays in the musical cues and picked that average for itself to sync up with. However, human players did not try to find the average delay from all the musical cues they were given; rather, human musicians picked one cue randomly, ignored the rest, and synced themselves with their chosen cue.

I'm really confused about the insight or discovery here. It's cool as a piece of art, but what exactly is the scientific significance of the "discovery" that humans can't spontaneously calculate the average delay out of all the delayed musical cues they are being given? I feel quite lost.
posted by MiraK at 2:24 PM on January 5, 2021 [3 favorites]

What are coupled violinists

I mean, I assume they're not fucking
posted by axiom at 2:31 PM on January 5, 2021 [6 favorites]

I’d also like to hear more, as I don’t totally grasp the finding here. I’m reminded of Terry Riley’s composition In C. If you don’t know it, I will let Wikipedia give you the rundown:

In C consists of 53 short, numbered musical phrases, lasting from half a beat to 32 beats; each phrase may be repeated an arbitrary number of times at the discretion of each musician in the ensemble. Each musician thus has control over which phrase they play, and players are encouraged to play the phrases starting at different times, even if they are playing the same phrase. In this way, although the melodic content of each part is predetermined, In C has elements of aleatoric music to it.[5] The performance directions state that the musical ensemble should try to stay within two to three phrases of each other. The phrases must be played in order, although some may be skipped. As detailed in some editions of the score, it is customary for one musician ("traditionally... a beautiful girl," Riley notes in the score)[6] to play the note C in repeated eighth notes, typically on a piano or pitched-percussion instrument (e.g. marimba). This functions as a metronome and is referred to as "The Pulse". Steve Reich introduced the idea of a rhythmic pulse to Riley, who accepted it, thus radically altering the original composition by Riley which had no pre-determined rhythm.[7]

It’s a marvellous thing to see or even just hear performed. There’s a remarkable pulse of change within stasis in the piece. I’ve heard versions recorded without the steady eighth-note pulse, and the players just naturally all gravitate to the same andante tempo. (Note that the sheet music itself doesn’t even suggest a tempo.)
posted by ricochet biscuit at 2:32 PM on January 5, 2021 [5 favorites]

I *think* this is what's happening:
-The "noise canceling" headphones prevent each violinist from hearing anyone else "live".
-Instead, they hear what the computer is feeding them.
-These players are all at a high-enough level that they can play "by ear"; i.e., they hear a note and don't have to consciously think 'oh that's an E flat'.
-At any time, the computer may be feeding what other players are doing to any one player's headphones; only one ear; a mix of multiple players; etc.

So this is an interesting sense of a network - a bunch of nodes (stereo nodes, even!) in a bunch of non-one-to-one connections.
posted by notsnot at 7:08 PM on January 5, 2021

Also- ricochet biscuit - kinda reminds me of how Japancakes got it start. Get some players together, jam in D until you're done jamming in D, then pick a different key and jam on that, maybe in a different tempo...
posted by notsnot at 7:11 PM on January 5, 2021 [1 favorite]

It is a bit confusingly written but it sounds like “decision-making” is supposed to be the key. The human players didn’t get hopelessly lost trying to average out multiple conflicting inputs, they made a decision (maybe not consciously) about what to synchronize to and ignored the rest. This apparently doesn’t happen in similar situations where humans aren’t the ones doing the synchronizing, so it’s an interesting finding and can maybe be applied to situations where machines or algorithms are trying to synchronize. Is what I got out of it.
posted by No-sword at 7:29 PM on January 5, 2021

What are coupled violinists

I mean, I assume they're not fucking

You do gain two additional f-holes when you become a violinist.
posted by Kabanos at 9:06 PM on January 5, 2021 [7 favorites]

I wonder how human networks would behave for simple aural cues as opposed to complex sets of music notes that only make sense in that sequence (I think humans would pick the average of C if fed C sharp and C flat).

I suspect it's only when there's a series of sounds which only makes sense in sequence that human networks do not attempt averages. Because, being not-machines, we know the music makes no sense when notes are averaged out.
posted by MiraK at 4:05 AM on January 6, 2021

I've been struggling to understand this also. As a musician myself, it just sounds like the musicians are doing what we're trained to do - play with the people we can hear. Musicians can't average out and try to play in between two things which are wrong, we have to pick one and decide it is right. Obviously this is not the ideal situation - we'd rather everyone never made a mistake - but it does happen and that's how you deal with it. Obviously in this experiment there was no proper source of leadership as a real ensemble would have (even if it didn't have a conductor), but this is what musicians do.

So... did they discover musical training?
posted by mathw at 5:43 AM on January 6, 2021 [2 favorites]

It does seem like kind of a strange way to describe musical training. I've never really learned an instrument but I've sung in choirs and I definitely was picking one voice to follow out of many even without any real training on the subject (outside of music classes as a child where we learned to follow our teacher's voice so maybe that counts).

I get that the bridge was averaging out the motions and the humans were picking one of many inputs instead of averaging them and that's the difference. It would have helped me if there was more done to tie together how a group of humans performing music is a network and how that is conceptually similar to a bridge swaying.
posted by Nec_variat_lux_fracta_colorem at 8:31 AM on January 6, 2021 [1 favorite]