Nice Robots Finish First

Seen something interesting in the news or on the intertubes? Discuss it here.

Moderators: Zamfir, Hawknc, Moderators General, Prelates

User avatar
podbaydoor
Posts: 7548
Joined: Sun Sep 02, 2007 4:16 am UTC
Location: spaceship somewhere out there

Nice Robots Finish First

Postby podbaydoor » Tue May 10, 2011 3:46 pm UTC

Discover Magazine
http://blogs.discovermagazine.com/80bea ... can-evolve

Nice Robots Finish First: Simulation Shows How Altruism Can Evolve
What’s the News: The diminutive, unassuming Alice robot has helped a Swiss research team test a core tenet about the evolution of altruism, called Hamilton’s rule. The researchers’ new study shows that even simple robots operating with simple evolutionary rules can recreate evolution’s complex interplay of selfishness and selflessness.

How the Heck:
Each Alice bot is a trundling cube equipped with two wheels and 33 “genes” that reflect the make-up of its artificial nervous system. These genes control how well they can move around and push small disks that represent food into their nests. Because researchers wanted to do the study on a big scale (500 generations of 1,600 robots each), they actually carried out most of the action in a computer simulation of the robots, where it could be done faster and cheaper. (Previous work showed that the software version of Alice evolution was a good model of the hardware version.)

At the beginning of the simulation, each robot was randomly assigned a value for each of its genes. Then survival of the fittest took over: The robots that couldn’t gather food effectively didn’t have their genes passed on to the next generation, which the researchers created by recombining successful genomes, just as would happen in an animal population.

The researchers wanted to see how the robots would behave when given a chance to collaborate with robots closely related to them. So they gave the robots the option of sharing their food equally among all the others in their group. They found that when robots were around robots with similar genes, they gradually evolved altruistic behavior, choosing more frequently to share the bounty than hoard it.

The question at the heart of the simulation was whether robotic evolution of altruism followed Hamilton’s rule, an inequality that connects the fitness cost to the altruist, the fitness benefit to the beneficiary, and their genetic relatedness. They tested out 25 different combinations of these values, and found that the transition to altruistic behavior always occurred at the point predicted by the rule.

What’s the Context:
Kin selection is a theory that attempts to explain why, if we evolved through a system of survival of the fittest, we take care of others instead of ruthlessly brushing them off. It holds that because we share genetic material with our family, helping them out is really just a way of helping ourselves (or, more accurately, our genomes) and thus doesn’t conflict with goal of passing our genes on.

The idea is not without controversy: E.O. Wilson, the scientist who popularized it, announced last year that it failed to hold up to close mathematical scrutiny. But it has consistently offered one of the few convincing possible explanations for altruism and still a subject of much study.

The group has previously published studies using the Alice bots as model organisms. A 2010 paper has a detailed description of their earlier work with evolving altruistic Alices, which involved getting them to cooperate on collecting food disks too large for a single bot to push (as in the top image).

Reference: Waibel M, Floreano D, Keller L, 2011 A Quantitative Test of Hamilton’s Rule for the Evolution of Altruism. PLoS Biol 9(5): e1000615. doi:10.1371/journal.pbio.1000615


I'd suggest clicking through to the article, since it provide links, including one to Alice bots and Kin Selection, and the E.O Wilson controversy. No Ten Commandments necessary.
tenet |ˈtenit|
noun
a principle or belief, esp. one of the main principles of a religion or philosophy : the tenets of classical liberalism.
tenant |ˈtenənt|
noun
a person who occupies land or property rented from a landlord.

User avatar
Jahoclave
sourmilk's moderator
Posts: 4790
Joined: Thu Oct 25, 2007 8:34 pm UTC
Contact:

Re: Nice Robots Finish First

Postby Jahoclave » Tue May 10, 2011 7:21 pm UTC

Is it sad that I totally just want to run this simulation with cute little robots?

User avatar
podbaydoor
Posts: 7548
Joined: Sun Sep 02, 2007 4:16 am UTC
Location: spaceship somewhere out there

Re: Nice Robots Finish First

Postby podbaydoor » Tue May 10, 2011 9:22 pm UTC

Perhaps outfit them with brainwave-controlled cat ears.
tenet |ˈtenit|
noun
a principle or belief, esp. one of the main principles of a religion or philosophy : the tenets of classical liberalism.
tenant |ˈtenənt|
noun
a person who occupies land or property rented from a landlord.

Enokh
Posts: 473
Joined: Mon Jan 17, 2011 1:55 pm UTC

Re: Nice Robots Finish First

Postby Enokh » Tue May 10, 2011 9:55 pm UTC

Altruism always made sense to me in the self-preservation, survival-of-the-fittest sense, as it's sort of like buying insurance: when you help other people out, it makes them more inclined to help you out when you need it. So long as you do not help someone out in a way that is incredibly detrimental to yourself or your chances of survival, it's a great system.

fr00t
Posts: 113
Joined: Wed Jul 15, 2009 11:06 am UTC

Re: Nice Robots Finish First

Postby fr00t » Wed May 11, 2011 12:23 am UTC

Enokh wrote:Altruism always made sense to me in the self-preservation, survival-of-the-fittest sense, as it's sort of like buying insurance: when you help other people out, it makes them more inclined to help you out when you need it. So long as you do not help someone out in a way that is incredibly detrimental to yourself or your chances of survival, it's a great system.


The basic idea behind Hamiltons rule is that genes "try" to propagate themselves not individuals, which means that altruism doesn't necessarily make sense for self-preservation unless you happen to be a gene. The intersection between morality and evolution is interesting because there is a clear causal link, yet it is generally wrong to derive values from extrapolating what evolution 'should' want you to do.

Can't memetics also provide an answer for altruistic behavior? Societies that emphasize such values may be more dominant than others, for instance, as we do generally value human life that is unrelated to us, or even unknown to us.

It's an interesting article but I'm not convinced it necessarily shows anything conclusive, as we already know altruism did evolve, and this is way too reductive to give any more insight than "sometimes working together can be advantageous"

Tirian
Posts: 1891
Joined: Fri Feb 15, 2008 6:03 pm UTC

Re: Nice Robots Finish First

Postby Tirian » Wed May 11, 2011 1:16 am UTC

fr00t wrote:It's an interesting article but I'm not convinced it necessarily shows anything conclusive, as we already know altruism did evolve, and this is way too reductive to give any more insight than "sometimes working together can be advantageous"


Based only on my layman's understanding, it would be interesting if the experiment validated the kin hypothesis as the authors seem to claim. I mean, if we imagined that if you deliberately planned a group that was made up of half robots that were optimized for finding food in the field and communicating that and half of robots that were optimized for hauling that food back to the nest, our intuition might suspect that would be an advantageous synergy that would lead to a dominant society. The kin hypothesis rejects this notion, saying that a Gatherer will only "want" to share food with other Gatherers because a Seeker is too genetically dissimilar and that our planned community will therefore not thrive. That's at the core of the Wilson controversy, because the guy who championed the kin hypothesis is now suggesting that it is too naive and doesn't explain cross-species altruism which is also observable in nature.

User avatar
the_bandersnatch
Actually not so frumious.
Posts: 939
Joined: Wed Jan 09, 2008 11:46 am UTC
Location: on a bed in a room inside a TV in a hotel room in a hotel on a Monopoly board

Re: Nice Robots Finish First

Postby the_bandersnatch » Wed May 11, 2011 9:09 am UTC

Tirian wrote:That's at the core of the Wilson controversy, because the guy who championed the kin hypothesis is now suggesting that it is too naive and doesn't explain cross-species altruism which is also observable in nature.


The Selfish Gene theory, as championed by Dawkins et al, and which can be seen as an extension of the ideas of Wilson and Hamilton, fully explains cross-species altruism the same as it explains intra-species altruism. This article recaps some of the ideas: http://www.prehensile.co.uk/thingstothinkabout/?p=17, but in short, if you apply the rules of game-theory and work through the mathematics, you can see how altruism arises from a selfish-gene perspective.
In girum imus nocte, et consumimur igni


Return to “News & Articles”

Who is online

Users browsing this forum: No registered users and 10 guests