Flies That Do Calculus With Their Wings 107
DudeTheMath (522264) writes "Cornell University scientists studied how fruit flies respond to flight disturbances (instead of wind gusts, they used carefully controlled magnetic pulses) and found that the flies recover in as little as three wing beats (at 250 per second) by doing some kind of calculus in a little 'integrated circuit' of neurons that control the wings directly. The pitch and yaw results are already published, and the roll study is forthcoming."
It's hardwired (Score:3, Informative)
how calculus? (Score:2)
Can you or anyone explain how what the neurons are doing is "calculus"?
Do they mean that were humans to mimic the neurons in a simulation, we'd have to do calculus in an algorythm to achieve the desired result?
Re:how calculus? (Score:5, Informative)
Can you or anyone explain how what the neurons are doing is "calculus"?
Calculus is how we scientifically communicate nature to each other, not away for nature to implement mathematics. Flies are not doing calculus any more than you catching a thrown ball is doing calculus. This headline, and perhaps the grant proposal, is written for stupid people. I hope this explains it for you.
on a different note... (Score:1)
aren't we part of nature?
so Calculus is "not a way for nature to implement mathematics" but humans use it to communicate nature
does that mean you consider the act of abstraction "non-natural"?
Re: (Score:2)
aren't we part of nature?
Yes. And so are flies.
Flies are made out of atoms.
So are we.
Atoms don't touch eachother.
There's no single sharp separating surface you can draw between the atoms of a fly and the atoms of a person.
Therefore, we ARE the flies.
And WE are doing the calculus.
Re: (Score:2)
another different note (Score:2)
yo hawkinspeter, can you or StripedCow explain why someone would mod my "on a different note..." post as "Troll"?
i genuinely don't understand the mod, and am honestly a bit more confused about Calculus than when I started
Re: (Score:2)
Re: (Score:2)
word I was being serious about the "non-natural" thing...it's a whole other discussion, but it relates to cybernetics
if the act of "abstraction" really is a human-only trait of all life we observe in the universe that could mean something
i think it's probably just a big conversation leading to nowhere but maybe there will be a chance to discuss it in the future
Re: (Score:2)
Off topic, but you and I seem to have a mutual fan manifest by slashdot's disdain for serif fonts. Note these are two different URLs:
http://slashdot.org/~gIobaljus... [slashdot.org]
http://slashdot.org/~gIobaIjus... [slashdot.org]
This one will probably be taken soon:
http://slashdot.org/~globaIjus... [slashdot.org]
I mostly ignore him, but had to modify my signature to make the distinction clear when a few accusations were thrown my way, and I just happened to notice that he's up to it again a minute ago.
Re: (Score:2)
thnx AlphaWolf_HK
I wonder who's up to this? And does /. know/care?
could this be APK in Beta???
Re: (Score:2)
hey just wanted to add that I've been aware of my doppleganger for about 2 months now
it's weird...
I haven't changed my sig b/c i'm almost afraid that the dupe "globaljustin" will change theirs...
one of mine makes quasi-constructive comments and gets upmodded
i recently addressed one who **replied to one of my comments** and this is what happened: http://slashdot.org/comments.p... [slashdot.org]
he says, "please ignore the troll pretending to be me"
which i didn't reply to
Re: (Score:2)
Yep, he probably will change his to match yours. This is mainly because on slashdot, most people recognize other people by their signature rather than their name. If he doesn't copy your signature, most people won't ever mistake him for you, and he knows that. So, he did the same thing to me.
At first I just made a joke of it and kept changing my signature to say different goofy things, and just watched him copy it. After a while I stopped and created my current signature, which he never copied. (Though give
Re: (Score:1)
Re: (Score:2)
you catching a ball is way, way more calculus.
this is more akin to analog circuit fly-straight autopilot doing calculus.
Re: (Score:3)
you catching a ball is way, way more calculus.
It really isn't. Catching a thrown ball requires practice, during which you learn more or less how a ball moves after being thrown without any actual understanding of the math behind it.
If you were doing calculus, you'd know exactly where the ball was going as soon as you saw it moving and you could simply put your hand in the right place and wait for the ball to arrive.
In reality, you get yourself into what looks like it might be the path the ball is going to take, then you constantly correct your positio
Re: (Score:1)
It's more like a series of guesses where you're constantly told whether you're getting closer or further away from the correct answer and that is not calculus.
Perhaps you don't understand calculus?
One thing is for certain, you're not very good at catching balls. Some of us can tell exactly where it's going to land.
Re: (Score:2)
Fom the thrower's release, your brain has likely already determined 1) approximately how much time before the ball reaches you
Calculus will tell you exactly how long it will take for a ball to get to you, not an approximate guess.
You've confused "being familiar with thrown balls so that you can make a good guess" with "actually doing math."
Re: (Score:2)
Some of us can tell exactly where it's going to land.
No, you can't. Calculus can tell you exactly how far the ball will travel down to the last millimeter if you account for all of the variables.
You can make an educated guess, but you'll completely fail to catch the ball if you then close your eyes and assume that it will be exactly where you predicted it to be. If you can tell exactly where is going to land, why would you need to see it and make corrections?
Only an idiot would believe that they can tell exactly where a ball is going to land from the instan
Re: (Score:2)
You are aptly named, is it because you lack skills that you are so grumpy?
Re: (Score:2)
If you were doing calculus, you'd know exactly where the ball was going as soon as you saw it moving and you could simply put your hand in the right place and wait for the ball to arrive.
Really good outfielders can do that. Willie Mays would turn around and run straight to the spot where the ball would come down without watching it - as soon as it left the bat.
Re: (Score:2)
If they could do that, they wouldn't have to keep watching the ball as they ran and caught it. There's a reason they keep their eyes on the ball. Constant corrections.
Re: (Score:1)
Re: (Score:2)
What's not calculus about that? You approximate the result and further refine it with additional data/time. With even more practice, you certainly CAN see where it's going to land (barring interference by another force, such as wind).
We are quite good at this. How do you think things like hitting moving targets with arrows happens?
Re: (Score:2)
Calculus is not a series of guesses. If you think it is, you should probably go back to school.
Re: (Score:2)
Approximation != guessing.
Re: (Score:2)
So, the fly has a biological PID controller [wikipedia.org]?
Re: (Score:2)
Calculus is how we scientifically communicate nature to each other, not away for nature to implement mathematics. Flies are not doing calculus any more than you catching a thrown ball is doing calculus. This headline, and perhaps the grant proposal, is written for stupid people. I hope this explains it for you.
When somebody throws a ball to you, how do you figure out how to catch it?
Mechanical/electrical systems can do calculus, stupid.
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
It almost always happens with reporting of pretty much anything scientific.
It should always sensitize your bullshit detector.
Err, not always (Score:2)
I'd like to see someone try and implement chess or a 3D game purely in hardwired TTL. It might be theoretically possible but I doubt it would be more efficient (ie faster , uses less energy) than software running on a processor.
Re: (Score:1)
Re: (Score:2)
IMO I think something like that would be better off implemented in analog hardware - you could virtually mimic the neurons 1 to 1 and it doesn't matter if a bit of noise gets in the system.
Re: (Score:2)
TTL chess might be faster, since no game would last longer than 255 moves.
Re: (Score:1)
Interesting (Score:5, Funny)
Any way of installing these wings on college students?
Re: (Score:2)
Bad weather was not a factor in that incident
So... (Score:4, Funny)
Re: (Score:2)
In other words.... (Score:4, Insightful)
What the fly does naturally requires the use of calculus to mimick artificially. Seems pretty natural to me. The laws of physics and mathematics are inseperable.
Re: (Score:2)
Calculus is not mimicry.
Calculus? (Score:5, Insightful)
A human-built bug might have to do the calculus, but the natural bugs don't.
Re:Calculus? (Score:5, Insightful)
Actually, it is doing calculus with a highly optimized analog computer. Amazing what Mother Nature can do given enough time.
Re:Calculus? (Score:5, Interesting)
Indeed. This is more like the smart bomb sights or artillery computers from WWII. Analogue all the way, and because of that, incredibly fast.
A dog catching a hubcap-like plastic object is a more complex operation, and the brain is involved, running an evolved trajectory program that isn't very fast, nor very accurate, and tends to freeze when run in parallel. But it's fast enough and accurate enough that the dog catches the thing most of the time.
Presumably, some far distant ancestor caught falling fruit or jumping fish, or catching tidbits flying from your parents ferocious eating, and the ability to just do slightly better than your peers meant greater chance of survival for you and your offspring.
These days, the genes might be favored again, because we like dogs to play with us.
Well, that's how it works outside Oklahoma and Alabama, anyhow.
Re: (Score:2)
Don't forget the emphasis on feedback - visual and otherwise. The dog catching the Frisbee isn't strictly obeying classical physics, so you can't say they're doing physics. Instead, the dog is reacting
Re: (Score:2)
I've found very few dogs that will run to the expected location without checking the object trajectory, signifying feedback is quite important in the process.
No, but I had one that would look at the thrown ball, then start running under it without looking again, passing it, and turning around to catch it where she felt it should land. She seldom was off by more than a dog could twist, even for pretty long throws.
I'm not sure many humans could do the same, except baseball and cricket pros, but they don't run nearly as long distances and don't overtake the ball.
So in some respects, I think the dog's ability to estimate a thrown object's landing position is at lea
Re: (Score:2)
It is both more and less amazing than that. It's still not doing calculus. It's just doing whatever feels right. What's amazing is that nature has produced such an efficient nerve cluster which can provide the appropriate reaction. What's hilarious is that because it fits a mathematical model, we think the insect is doing the math. No, it's just moving the only way it can move, and the results are best described by calculus, a math we devised to understand things we saw in nature.
Re: (Score:2)
To be fair, each neuron is probably doing some mathematical function (in the form of a series of chemical reactions) and then passing the results to the next neuron. The end result is the same as an analog computer performing some hard-wired calculus function. In the end we are arguing semantics, not what is happening in actual fact. I'm much less interested in how each Slashdotter defines "calculus" and much more interested in how fruit flies do so much with so few neurons.
Re: (Score:2)
Yes, No. We are arguing semantics and semantics are language features and math is a language so we are talking about what's happening in actual fact.
Math is a language of precision. English is, for the most part, not. When we need precision, we have to use some subset like "legalese". Complaining about imprecise language in a scientific journal is one thing - complaining about the precision of a NY Times headline is quite another. Expecting the NY Times to write headlines that sound like scientific paper titles is not realistic IMHO.
FWIW, in the article they clarify with:
"Humans use calculus to solve these kinds of problem involving multiple changes in
Re: (Score:2)
Re: (Score:2)
A tuned feed-forward loop is pretty interesting, and it will be very enlightening to see how it is done by a natural neural network. If you did this with a man-built system, you would definitely use calculus to tune the system.
Re: (Score:2)
Re: (Score:2)
You might very well end up implementing your routine with simpler math, but I'd bet a dollar that you'd first design the system in something like MATLAB before simplifying. And depending on the application, you might stop there because controller chips are so cheap now. PID with feedforward is possible on almost anything.
Re: (Score:2)
Calculus is OUR way of DESCRIBING the motion or action that is happening. The fact that you can use calculus to determine the area under a curve doesn't mean that the area under a curve is 'calculus'. Or that the volume of an arbitrary vessel is 'calculus', just because you USE calculus to determine the volume.
The fly has evolved a set of behaviours that corrects for certain environmental conditions while it's flying. That's all. There's no real computation going on.
Re: (Score:2)
Tere's no evidence calculus is being done beyond an apple doing addition when it falls from a tree onto other apples. Just because addition happens, and that's how we'd describe it, doesn't mean that's what the apple is doing.
Re: (Score:2, Insightful)
That's like saying that a dog catching a ball or frisbee is doing calculus. Nope, it's experience.
It's not "experience" either. Basically, it's calculus. But the fly isn't doing the calculus. Evolution did the calculus using a massively parallel Monte-Carlo optimization akin to "stupidsort" (stupidsort randomly permutes a sequence and then checks whether it is sorted, iterating when it isn't).
Re: (Score:1)
Re: (Score:2)
A better way of phrasing it would have been that it's engaging in an analog computation that provides an approximation of calculus. That is, using analog systems, it does the calculations necessary to come up with a result that is remarkably close to that which calculus would tell us is the correct answer, yet it does so without any of the higher-order reasoning and in a fraction of the time that it would take us to do so exactly.
As you said, dogs do the same thing when they run to catch a disc, just as we
Re: (Score:2)
Re: (Score:2)
I believe I did a poor job explaining myself, since it seems you misunderstood what I wrote. I'm in full agreement with you, and was merely trying to provide an alternative and more accurate way of describing what was happening, yet your reply suggests that you were trying to contradict a point I didn't make.
As you said, an apple falling from a tree might be capable of having its descent described in terms of math, but that in no way suggests that the apple itself (nor the tree it fell from, obviously) is e
Re: (Score:2)
your reply suggests that you were trying to contradict a point I didn't make.
Right back at you.
while a fly, dog, or person may not be engaged in calculus, they are engaged in some primitive computation, else they wouldn't be able to move to the right place or respond to a change in their environment,
I said nothing to contradict that. Humans have a well-studied "muscle memory", and this sounds a lot more like that than any conscious computation.
Again, this does not mean that they are "doing calculus", and to say that they are would indeed be a misstatement, as you and I both agree
Yup. I think they intentionally used inflamatory language to make it sound more interesting. My point is that they were "lying".
It's likely a feedback loop. There are many that have been examined and proven in nature (dogs catching things, for example). This doesn't look like anything new, just (maybe) faster or more accurate than seen bef
Re: (Score:2)
Fair points, all around, and it looks like we're in vehement agreement with one another. ;)
Re: (Score:2)
Some kind of calculus? (Score:1)
When I learn they can analytically solve arbitrary partial differential equations I'll take even more pleasure in swatting the little buggers.
Time, distance, motion, acceleration (Score:4, Informative)
Re: (Score:2)
Agreed. Calculus is a specific mathematical, symbolic approach to computation. Solving the same kinds of problems by a different means is not calculus.
Re: (Score:2)
If fruit-flies use calculus then so do amoebas.
Rocks do calculus when they roll down hills, since they always make sure to only move a distance which exactly matches the integral over time of their velocity.
Circles do calculus too, since they always choose their area such that it corresponds to the integral over their radius of a circumference.
Re:Time, distance, motion, acceleration (Score:4, Funny)
This is nothing. My desk lamp does REAL-TIME ray tracing!
Especially when I put a couple reflective spheres on a checkerboard underneath it.
Pffft NewBees (Score:2)
Call me back when they do my taxes
What a retarded headline (Score:2, Insightful)
So when a dog catches a frisbee, is it doing calculus with its teeth?
Next they'll discover... (Score:2)
that humans do signal processing with their brains, and that such processing involves complex analysis. One day they'll learn that those squiggly symbols in maths books actually mean something. It's an embarrassment to science that these insect chasers are called scientists rather than sciensecoolhuhwowists. End rant of an old school fundamentalist.
Which kind? (Score:2)
doing some kind of calculus in a little 'integrated circuit' of neurons
If it's using a dedicated integrated circuit the calculus they perform must be integral.
no more than (Score:3)
Re: (Score:2)
Re: (Score:3)
Little known calculus wizardry of Babe Ruth (Score:2)
Boring: A neural net can approximate diff-eq (Score:2)
The fly’s brain is not doing calculus (or rather, differential equations). It’s a neural net that has evolved to respond to stimulus in a way that appears like what we’d use diff-eq for. Within certain bounds of range and accuracy, we can make artificial neural nets do this. So why is it surprising that meatware that evolved over millions of years can do the same thing?
Re: (Score:1)
The story is not "the fly's brain does calculus". The story is "a region in the fly's nervous system outside the brain has an autonomic role in fast recovery from perturbations of the flight (statespace) trajectory".
You know that kneejerk reaction (the real one, not the figurative one)? That's the same deal - the signal doesn't get to the brain before the knee jerks, but the jerk is mediated by neuron groups in the spinal cord. So the spinal cord is not just a communication channel, it also embeds contro
Re: (Score:2)
Someone should mod up the AC’s comment.
Nonsense (Score:2)
This is no more 'fly doing calculus' any more than people do calculus when they throw a snowball at a target.
Not to say that I'm not occasionally amazed at the staggering mathematical underpinnings of some of what we internalize as fairly simple things:
- seeing - and interpreting - an image with light and color.
- hitting a moving target with something
- the signal complexity involved in muscle coordination to do just about anything.
Hmm... (Score:1)
RC (Score:2)
You what simple circuit implements calculus (i.e. can integrate and differentiate)?
An R-C circuit.
Move along, happy Friday.
KCDoodle (Score:2)
The nerve cells immediately after the rods and cones in your eyes (and most other animal kingdom eyes) also perform calculus. Edge detection is done BEFORE visual stimuli makes it to your brain. The image and the edges reach your CPU at the same time. This lets you know where things start and where they end. It is a great asset when hunting chasing and running away.
However, it can get confused. This is the reason zebras have stripes and run in herds. With a large number of
What about ants? (Score:2)
I didn't do calculus well. :(
(IRO(CASM)) (Score:1)