Pharyngula::Smarter than Feynman!

Pharyngula has moved to http://scienceblogs.com/pharyngula/

Thursday, July 28, 2005

Smarter than Feynman!

PZ Myers • 25 Comments (last page)

A short essay in Nature lays out what seems to be a simple problem, but it's one that stumped Richard Feynman. Ants have to forage, and they lay down chemical trails from the nest to food sources. These trails may form fairly complex branching trees and represent a kind of maze—how do ants know which branch to take?

Even simple trees introduce problems. Imagine an ant that has walked from its nest to food source A, and is heading home with a tasty bit of caterpillar. It arrives at the intersection of the Y-shaped trail…how does it know to turn right to go home, rather than left to end up at food source B?

There are a number of possible solutions. The one Feynman came up with is that the trails have a pattern of chemicals within them that impose directionality; that there actually is an arrow in the chemistry of the trail that tells the ant which way the nest lies. This solution, however, is too complex and impractical to set up, and has not been demonstrated. Another possibility is that the ants can use global cues, like the position of the sun or local landmarks, to have a general sense of the right direction towards the nest.

I'm proud to say that when I first read of the problem I immediately thought of an even simpler solution, and one that the article then goes on to describe as having been demonstrated in one species of ant, Monomorium pharaonis: just make the ant reluctant to make sharp turns. If all branches from the trail home diverge at shallow angles, simply taking the path that is closest to your current trajectory will put you on the right track home. Clearly, that swift insight means I'm smarter than Richard Feynman. Of course, it also means that ants are simpler than Feynman expected, and my superiority is based on my ability to think more like an ant than Feynman, which does sort of diminish the accomplishment.

I also cheated. The ant problem is much like a problem I did some work on a while back, on how branching networks of neuronal growth cones arborize over a sheet. For a while, I thought there must be some kind of gradient present to impose a general directionality on their growth, but after making a great many observations and seeing how growth cones turned and branched, a simpler solution presented itself: growth cones just don't make sharply angled turns, probably due to cytoskeletal constraints, and running a few computer simulations showed that just putting limits on the degree of branching produced distributions similar to those we were seeing in real animals. And now that it has been brought to my attention, I see that there are some interesting parallels between ant trails and nerve pathways that I'm going to have to think about some more.

Ratnieks F (2005) Outsmarted by ants. Nature 436:465.

Trackback url: http://pharyngula.org/index/trackback/2642/7SdATx66/

Comments:

#33045: — 07/28 at 07:51 AM
Do ants return to a given food source preferentially once multiple sources are found?

#33046: bill — 07/28 at 08:02 AM
I think I'd be happy being somewhere between an ant and Feynman.

#33048: — 07/28 at 08:12 AM

when I first read of the problem I immediately thought of an even simpler solution

Right there with you making the same guess/deduction before the exposition. :-D It's all too obvious for someone sensitive to geometry rather than mere topology.

#33049: — 07/28 at 08:13 AM
The way I remember it, Feynman said that the ants don't use directional trails, but he claimed that one species of south american ant was known to use directional signals.

Anyway, I wonder if there might be physics applications of this. Places where theory says something goes in any direction, where really it's biased toward continuing at close to the starting direction.

#33050: — 07/28 at 08:16 AM
If I remember correctly, Feynman's ants were able to tell which direction to go even if he placed them on the path with a random orientation (he also played with some ants that were unable to do this). I suppose it's possible that the ants noticed that while the path they were travelling was smooth there were an inordinate amount of backtracks branching off, but that would seem to require them to sniff out the direction of the branches they were passing. Feynman doesn't describe any of this behavior (in fact, I got the impression that their were no branches on the part on which he was experimenting), but he has been known to alter stories for the retelling...

#33052: Alon Levy — 07/28 at 08:40 AM
Doesn't these ants' inability to make sharp turns hurt them when they need to take sharp turns to get to food sources? For example, what do they do when the intersection is the shape of a capital T, with the vertical stroke leading home and the horizontal one leading to two food sources?

#33054: Raven — 07/28 at 09:11 AM

It's all too obvious for someone sensitive to geometry rather than mere topology.

Mere topology, SEF--are you trying to start yet another religious war here, albeit one much more interesting than Windows vs. Linux or Mac?

If you want to argue geometry vs. topology, bring it on!

#33057: — 07/28 at 09:22 AM
Fascinating ant stuff PZ, but I have questions:

Now here's an "interesting thing", I did the same experiments with ant-ferrying as Feynman did and came to the same conclusions, importantly, BEFORE reading about Feynman etc. Does this therefore mean that I am dumber than PZ (likely) and as smart as Feynman (unlikely)?

I also thought of retrosynthetic analysis (although I didn't call it that) when I first started thinking about organic chemistry at school, which was BEFORE I'd heard of EJ Corey (another Nobel Laureate) and his work. Does this mean that I am as smart as Corey (unlikely)?

I'm guessing that good ideas sometimes land in brains even as tiny as mine! (Pratchett's model of inspiration appeals here!)

But seriously:

On the ant thing, isn't one of the ways that ants know which direction the chemical trail is going in, down to the age of the trail? I.e. earlier bits of the trail are more "aged" by degredation/dispersion of the scent chemicals than later bits. You don't need chemicals A and B in order as the article mentions, you just need chemical A which disperses/degrades with time, so older scents can be distinguished from newer ones.

Also don't ants produce different scent chemicals for different activities, i.e. there is a specific "food found" scent emitted when found is found on the return journey to the nest, and a "foraging" scent emitted when leaving the nest seeking food. Also isn't the behaviour of the ant seeking food very different from the behaviour of the ant that has found food?

Didn't Feynman also note that the trails got ever more efficient as several ants moved along them. So the first ant's trail was very wiggly, and the second's slightly less so etc. Isn't it even simpler to think that some ants imperfectly follow the strongest trail of scent chemical back from the food to the nest. Of course different ants have different solutions to this problem.

Maybe I am missing something, but from a chemical perspective one aspect of this problem doesn't seem so difficult.

#33059: Arun — 07/28 at 09:38 AM

"If all branches from the trail home diverge at shallow angles, simply taking the path that is closest to your current trajectory will put you on the right track home."

How does ant exploration work?

An ant wanders away from home, after going some distance, finds nothing, and backtracks. It backtracks for a bit, and then, meanders along a different direction. Now, are sharp turns allowed during the backtrack - which would be required to preserve the property that all branches from the trail home diverge at shallow angles?

Or does the ant backtrack, turn around again, and then move off a shallow angle, in order to preserve the property that all branches from the trail home diverge at shallow angles?

Unless ants have global positions systems, I think also that ants, in their meandering, may cross the trail home at a sharp angle, it is not clear to me why ant trails will be trees.

#33060: — 07/28 at 09:54 AM
My initial question was trying to get to whether ants could follow a path back to a preferred food source. That would indicate some mechanism at work other than not making sharp turns.

#33061: — 07/28 at 09:58 AM
No war, Raven. I like topology too. I'm just pointing out that in this particular instance concentrating on the topology loses an observer some rather valuable information.

Arun, why would the ant not know whether it is going out or coming home? Even if it has a poor memory, the carrying of food or not would be better than a knotted handkerchief. Ants have also been observed to use a fibonacci angle when starting a forage en masse. So they don't overlap an area they've previously explored until later enough for those areas to be repopulated with prey and their chemical trails to have worn out. It's like real trees and their branches (though those form a spiral with the additional vertical dimension).

#33068: — 07/28 at 10:46 AM
It's all very interesting, but will it help me keep ants out of my house?

#33069: — 07/28 at 11:04 AM

It's all very interesting, but will it help me keep ants out of my house?

Yes. You obviously just have to build your house at sharp angles from the ants' paths.

#33078: — 07/28 at 12:06 PM
I vaguely recall a book reviewed decades ago in The Whole Earth Catalog, mostly dealing with tree limbs, honeycombs, etc, which offered the conclusion that nature shows a strong proclivity towards thirty-degree angles, with 45s & 90s being extremely rare.

Assuming the neurological patterns PZ is pondering can also be found in ants, this looks like a perfect case for fractal analysis!

#33089: — 07/28 at 02:35 PM
My physics adviser, who was a student of Feynmann's, happened to be talking about this last week (not ants, physical insights), and he told a story about Feynmann and Gell-Mann when they were discussing whether to grant tenure to a researcher at Cal Tech.

Gell-Mann said something like, 'Whoever we pick, he must have a profound feeling of physical insight.'

And Feynmann shot back, 'That lets you out, Murray.'

There's no one way; and, as for ants, my observation of the ones who used to march across my patio in Iowa was that a) they encountered ants coming in the opposite direction and dickered long enough to (perhaps) acquire information in addition to whatever was on the scent trail; and b) a small but not insignificant proportion of ants heading down the trail toward food got distracted and wandered off on their own.

#33114: — 07/28 at 07:30 PM
My hypothesis was that each ant has a "handedness", always traversing the tree in the same direction, and reversing it to return to the root.

I also considered the ant choosing the weakest scent, but that seemed more difficult practically.

#33117: — 07/28 at 08:48 PM
I'm guessing that it is some sort of quantum neurodynamical and/or bifurcationary and/or symmetry-breaking type of deal, the ants (or, all of their brains) being seen as one coherent entity.

Deep down inside, I'm hoping that Dr. Feynman (for at least one fleating moment) considered the very same thing.

Once in a while you get shown the light, in the strangest of places if you look at it right.

-Jerry Garcia

#33125: — 07/28 at 10:12 PM
Ouch! I thought the scent mechanism was verified. Now I have to unregister old stuff and register new stuff. Wait ... wait ... done! And faster than Windows, too.

J Thomas:
"Anyway, I wonder if there might be physics applications of this. Places where theory says something goes in any direction, where really it's biased toward continuing at close to the starting direction."

Crack formation in surfaces and solids. In fact I think there may be some stuff there for PZ to look into.

#33130: — 07/28 at 11:15 PM
Uh, no, this is probably nothing for PZ (but there are models of why small angels develop).

And to be clear; it's stressinduced crack propagation. New cracks split off in shallow angels.

#33132: — 07/29 at 02:50 AM

there are models of why small angels develop

Do give them their proper title: cherubs. :-D

PS Newton spent a long time looking into this area of classification and cladistics. So it would be nice if there now was a working hypothesis.

#33134: — 07/29 at 04:29 AM
When I first read your problem, I immediately thought of an even simpler solution.

Imagine that each ant constantly lays down a trail of scent. As each ant moves along a trail, the scent on that trail grows stronger. When they come to a fork on their outbound journey, some of the ants will take the left fork and some will take the right fork, but all have travelled down the path from home so the scent on it will always be stronger. When an ant is returning home, it simply has to take the path with the strongest scent to lead it home.

I thought this mechanism was already well understood. Am I missing something here?

#33139: jmorrison — 07/29 at 08:11 AM
i was going to say the exact same thing as cjb. first thing that popped into my head. the road home would always be the most heavily tagged and thus would stand out. could be wrong of course but seems the simplest solution.

#33157: — 07/29 at 03:05 PM
When Samuel Eliot Morison retraced Magellan's voyage through the Strait of Magellan, one of the puzzles he wanted to answer was how Magellan could have confidently taken one arm of a Y channel and not the other.

On a map, there seemed nothing to choose between the two, to a naive voyager.

When Morison got there, it was obvious. One channel was ocean colored, one was not.

#33167: — 07/29 at 07:31 PM
"Do give them their proper title: cherubs. :-D"

That was a new angle.

#33289: — 07/31 at 03:29 PM
I knew a guy who did his master's thesis on fire ant tracks.

He did time-lapse photography and measured the paths, and then tried to approximate the result with computer simulations.

His conclusion was that when they forage they first travel in a broad arc that could be part of a large ellipse or circle. But after some time they're likely to pick a smaller arc. They follow the smaller arc until they run into their own trail or somebody else's, and they take that trail home.

When they follow somebody else's trail they follow it awhile and occasionally wander off of it. They'll go a little ways and then veer back to join the main trail. But if the main trail turned and they meet up with it without veering back, then they just keep going. I don't understand why others then are more likely to follow them on it, until the main trail over time gets shorter and straighter.

He said sometimes ants got turned around and just went home without getting where they were going. And sometimes heading home they'd get turned around and go back to the site. He thought they were pretty stupid that way.

He claimed his research could show how to spread bait so that fire ants would be more likely to get it than other ants which had different two-stage foraging paths, but I didn't actually see any recommendations about that.

My memories about this seem clear except the part I knew I forgot, but I might be misremembering part of it. I hate it when that happens.

Page 1 of 1 pages

Next entry: Riding the pseudoscientific wave

Previous entry: Coincidence, or sigils of an evil plan?

<< Back to main

Pharyngula has moved to http://scienceblogs.com/pharyngula/

Thursday, July 28, 2005

Smarter than Feynman!

Info