"Long before it's in the papers"
January 27, 2015


Robots learn to share, and hint that relatedness is the key

May 3, 2011
Courtesy of Public Library of Science
and World Science staff

Sci­en­tists working with groups of ro­bots have found that the mach­ines can “evolve” a will­ingness to share. And the con­di­tions un­der which they do, re­search­ers say, support a the­o­ry that re­lat­ed­ness is be­hind the ev­o­lu­tion of al­tru­ism.

It’s been one of bi­ol­o­gy’s most en­dur­ing puz­zles: Why do most so­cial an­i­mals, in­clud­ing hu­mans, ac­tively help each oth­er? 

Al­tru­ism, the sac­ri­fice of in­di­vid­ual gains for the great­er good, seems at first glance to flout the “sur­vival of the fittest” prin­ci­ple crit­i­cal to ev­o­lu­tion­ary the­o­ry. That the­o­ry states that spe­cies grad­u­ally change be­cause each spe­cies’ best-functioning mem­bers con­tin­u­ally spread their genes at the ex­pense of weaker mem­bers’ genes. Since al­tru­ism of­fers no ap­par­ent ben­e­fit to the al­tru­ist, it would seem few or no “al­tru­ism genes” should ex­ist.

But they do, sug­gest­ing that per­haps there is a sort of strength in al­tru­ism. Work­er ants, for in­stance, make the ul­ti­mate “sac­ri­fice” by trans­mit­ting no genes at all—they are ster­ile—to help en­sure the sur­viv­al of the queen’s ge­net­ic make­up.

The sac­ri­fice of the in­di­vid­ual to en­sure the sur­viv­al of a rel­a­tive’s ge­net­ic code is known as kin se­lec­tion. In 1964, bi­ol­o­gist W.D. Ham­il­ton pro­posed a set of con­di­tions un­der which al­tru­istic be­hav­ior may evolve, now known as Ham­il­ton’s rule of kin se­lec­tion. Here’s the gist: If an in­di­vid­ual family mem­ber shares food with the rest of the fam­i­ly, it re­duces his or her per­son­al like­li­hood of sur­viv­al but in­creases the chances of family mem­bers pas­s­ing on their genes, many of which are com­mon to the en­tire fam­i­ly. Ham­il­ton’s rule simply states that wheth­er or not an or­gan­ism shares its food with anoth­er de­pends on its ge­net­ic close­ness (how many genes it shares) with the oth­er or­gan­ism.

Test­ing the ev­o­lu­tion of al­tru­ism us­ing stud­ies in live or­gan­isms has been largely im­pos­si­ble be­cause ex­pe­ri­ments need to span hun­dreds of genera­t­ions and there are too many vari­a­bles. But ro­bots cre­at­ed by robotics pro­fes­sor Da­ri­o Flo­re­ano of the Fed­er­al Poly­tech­ni­cal School in Lau­sanne, Switz­er­land, evolve rap­idly us­ing sim­u­lat­ed ge­net­ic func­tions, so sci­en­tists can meas­ure the costs and ben­e­fits as­so­ci­at­ed with the trait. 

The robots are fitted with simple “brains” whose de­cis­ions are controlled by “gen­omes” whose make­up, in turn, is de­term­ined by a com­puter pro­gram designed to simulate evo­lu­tion.

Ham­il­ton’s rule has al­so long been ques­tioned be­cause its equa­t­ion seems too sim­ple to be true. But “this study mir­rors Ham­il­ton’s rule re­markably well to ex­plain when an al­tru­istic gene is pas­sed on from one genera­t­ion to the next, and when it is not,” said Un­ivers­ity of Lau­sanne bi­ol­o­gist Lau­rent Kel­ler, who works with Flo­re­ano.

Pre­vi­ous ex­pe­ri­ments by Flo­re­ano and Kel­ler showed that for­ag­ing ro­bots do­ing sim­ple tasks, such as push­ing seed-like ob­jects across the floor to a des­tina­t­ion, evolve over mul­ti­ple genera­t­ions. Ro­bots unable to push the seeds to the right place are struck from the “gene pool” and can­not pass on their code, while ro­bots that do bet­ter see their code re­pro­duced, mu­tat­ed, and re­com­bined with that of oth­er ro­bots in­to the next genera­t­ion, mim­ick­ing the pro­cess of sex and ev­o­lu­tion.

The new study by re­search­ers with the two in­sti­tu­tions adds a twist: once a for­ag­ing ro­bot pushes a seed to the prop­er des­tina­t­ion, it can de­cide wheth­er it wants to share it or not. Ev­o­lu­tion­ary ex­pe­ri­ments last­ing 500 genera­t­ions were re­peat­ed for sev­er­al sce­nar­i­os of al­tru­istic in­ter­ac­tion—how much is shared and at what cost—and of ge­net­ic re­lat­ed­ness in the popula­t­ion. The re­search­ers cre­at­ed groups of re­lat­ed­ness that, in the ro­bot world, would be the equiv­a­lent of com­plete clones, sib­lings, cousins and non-relatives. The groups that shared along the lines of Ham­il­ton’s rule were found to for­age bet­ter and pass their code on­to the next genera­t­ion.

In the tests, “Ham­il­ton’s rule al­ways ac­cu­rately pre­dicts the min­i­mum re­lat­ed­ness nec­es­sary for al­tru­ism to evolve,” the re­search­ers wrote, re­port­ing their find­ings in the May is­sue of the re­search jour­nal PLoS Bi­ol­o­gy.

The re­sults matched sur­pris­ingly well the pre­dic­tions of Ham­il­ton’s rule even in the pres­ence of mul­ti­ple in­ter­ac­tions be­tween dif­fer­ent genes, a com­plica­t­ion Ham­il­ton did­n’t orig­i­nally take in­to ac­count, they said. “We have been able to take this ex­pe­ri­ment and ex­tract an al­go­rithm that we can use to evolve coop­era­t­ion in any type of ro­bot,” added Flo­re­ano. “We are us­ing this al­tru­ism al­go­rithm to im­prove the con­trol sys­tem of our fly­ing ro­bots and we see that it al­lows them to ef­fec­tively col­la­bo­rate and fly in swarm forma­t­ion more suc­cess­ful­ly.”

* * *

Send us a comment on this story, or send it to a friend


Sign up for

On Home Page         


  • St­ar found to have lit­tle plan­ets over twice as old as our own

  • “Kind­ness curricu­lum” may bo­ost suc­cess in pre­schoolers


  • Smart­er mice with a “hum­anized” gene?

  • Was black­mail essen­tial for marr­iage to evolve?

  • Plu­to has even cold­er “twin” of sim­ilar size, studies find

  • Could simple an­ger have taught people to coop­erate?


  • F­rog said to de­scribe its home through song

  • Even r­ats will lend a help­ing paw: study

  • D­rug may undo aging-assoc­iated brain changes in ani­mals

Using groups of computer-simulated robots, scientists have found that artificial creatures are willing to share. And the conditions under which they do so back up a theory that relatedness is behind the evolution of altruism. It’s been one of biology’s most enduring puzzles: Why do most social animals, including humans, actively help each other? Altruism, the sacrifice of individual gains for the greater good, seems at first glance to flout the “survival of the fittest” principle critical to evolutionary theory. That theory states that species gradually change because each species’ best-functioning members continually spread their genes at the expense of weaker members’ genes. Since altruism offers no apparent benefit to the altruist, it would seem few or no “altruism genes” should exist. But they do, suggesting that perhaps there is a sort of strength in altruism. Worker ants, for instance, make the ultimate “sacrifice” by transmitting no genes at all—they are sterile—to help ensure the survival of the queen’s genetic makeup. The sacrifice of the individual in order to insure the survival of a relative’s genetic code is known as kin selection. In 1964, biologist W.D. Hamilton proposed a set of conditions under which altruistic behavior may evolve, now known as Hamilton’s rule of kin selection. Here’s the gist: If an individual family member shares food with the rest of the family, it reduces his or her personal likelihood of survival but increases the chances of family members passing on their genes, many of which are common to the entire family. Hamilton’s rule simply states that whether or not an organism shares its food with another depends on its genetic closeness (how many genes it shares) with the other organism. Testing the evolution of altruism using quantitative studies in live organisms has been largely impossible because experiments need to span hundreds of generations and there are too many variables. But simulated robots created by Robotics professor Dario Floreano of the Federal Polytechnical School in Lausanne, Switzerland, robots evolve rapidly using simulated genetic functions and let scientists measure the costs and benefits associated with the trait. Hamilton’s rule has also long been questioned because its equation seems too simple to be true. But “this study mirrors Hamilton’s rule remarkably well to explain when an altruistic gene is passed on from one generation to the next, and when it is not,” said University of Lausanne biologist Laurent Keller, who works with Floreano. Previous experiments by Floreano and Keller showed that foraging robots doing simple tasks, such as pushing seed-like objects across the floor to a destination, evolve over multiple generations. Those robots not able to push the seeds to the correct place are removed from the gene pool and cannot pass on their code, while robots that do better see their code reproduced, mutated, and recombined with that of other robots into the next generation, mimicking the process of sex and evolution. The new study researchers with the two institutions adds a twist: once a foraging robot pushes a seed to the proper destination, it can decide whether it wants to share it or not. Evolutionary experiments lasting 500 generations were repeated for several scenarios of altruistic interaction—how much is shared and at what cost—and of genetic relatedness in the population. The researchers created groups of relatedness that, in the robot world, would be the equivalent of complete clones, siblings, cousins and non-relatives. The groups that shared along the lines of Hamilton’s rule were found to forage better and pass their code onto the next generation. In the tests, “Hamilton’s rule always accurately predicts the minimum relatedness necessary for altruism to evolve,” the researchers wrote, reporting their findings in the May issue of the research journal PLoS Biology. The results matched surprisingly well the predictions of Hamilton’s rule even in the presence of multiple interactions between different genes, a complication Hamilton didn’t originally take into account, they said. “We have been able to take this experiment and extract an algorithm that we can use to evolve cooperation in any type of robot,” added Floreano. “We are using this altruism algorithm to improve the control system of our flying robots and we see that it allows them to effectively collaborate and fly in swarm formation more successfully.”