"Long before it's in the papers"
April 29, 2009

RETURN TO THE WORLD SCIENCE HOME PAGE

Computers learn “regret”

Feb. 21, 2008
Courtesy Science
and World Science staff

Sci­en­tists in Italy have de­vel­oped com­put­er pro­grams that mim­ic hu­man decision-making by us­ing sim­u­lat­ed “re­gret” to im­prove per­for­mance. The mod­els do a bet­ter job than oth­ers in pre­dict­ing some as­pects of hu­man decision-making, the re­search­ers report.

Sci­en­tists have de­vel­oped com­put­er pro­grams that mim­ic hu­man de­ci­sion-mak­ing by us­ing sim­u­lat­ed “re­gret” to im­p­rove per­for­mance. (Im­age ©MSXO) 
The stu­dy’s bas­ic as­sump­tion was that peo­ple mod­i­fy their be­hav­ior dur­ing stra­te­gic games by look­ing back­ward to what might have been their best move, once they know what the oth­er play­ers’ move was. 

Da­vide Mar­chiori of the Uni­ver­s­ity of Tren­to and Mas­si­mo War­glien of Ca’ Fos­cari Uni­ver­s­ity in Ven­ice built math­e­mat­i­cal mod­els based on bi­o­log­i­cal neu­ral net­works. These use sim­u­lat­ed net­works of “brain cells” to ar­rive at de­ci­sions and learn by tri­al and er­ror.

In­tro­duc­ing an ap­proxima­t­ion of re­gret al­lowed the mod­els to pre­dict hu­man be­hav­ior more pre­cisely than con­ven­tion­al eco­nom­ic learn­ing the­o­ries, the re­search­ers said. Their find­ings ap­pear in the Feb. 22 is­sue of the re­search jour­nal Sci­ence.

“Re­gret refers to the dif­fer­ence be­tween out­comes at­tained and the best out­comes that might have been at­tained,” wrote Mi­chael D. Co­hen of the Uni­ver­s­ity of Mich­i­gan, Ann Ar­bor, in a com­men­tary in the jour­nal. “This is an im­por­tant step in the de­vel­op­ment of a work­a­ble new syn­the­sis,” added Co­hen, who was­n’t in­volved in the stu­dy. The work has ap­plica­t­ions in de­vel­op­ment of eco­nom­ic the­o­ries that pre­dict hu­man be­hav­ior, he added.

The mod­el’s pre­dictions, he con­tin­ued, aren’t based on “con­ven­tion­al, forward-look­ing ex­pecta­t­ions of gain, the no­tion so long at the heart of eco­nom­ic the­o­riz­ing.” Rath­er, its pre­dictions rely on “propens­i­ties that de­vel­op through a back­ward-look­ing learn­ing pro­cess that is driv­en by re­gret.”

* * *

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

 

Sign up for
e-newsletter
   
 
subscribe
 cancel

On Home Page         

LATEST

  • Discov­ery of “furth­est ob­ject” said to pave way for prob­ing early cos­mos

  • A w­arm T­V may drive away feel­ings of loneli­ness, re­jection

EXCLUSIVES

  • Re­port: cells “from space” have un­usual make­up

  • Dol­phins and the evo­lution of teach­ing

  • D­rug may tri­ck body into “think­ing” you ex­ercised

  • Tit-for-tat: bi­rds found to re­pay war­­time help

  • Musi­cal genes may be com­ing to light

MORE NEWS

  • R­ock-hurl­ing zoo chimp stock­ed am­mo in ad­vance: study

  • Fai­th found to re­duce er­rors on psycho­logical test

  • Dood­ling gets its due: ti­ny art­works may aid mem­ory

  • From or­al to mor­al? Dirty deeds may prompt “bad taste” reac­tion

Scientists have developed computer programs that mimic human decision-making by using simulated “regret” to improve performance. The models do a better job than others in predicting some aspects of human decision-making, the researchers found. The study’s basic assumption was that people modify their behavior during strategic games by looking backward to what might have been their best move, once they know what the other players’ move was. Davide Marchiori of the at University of Trento in Trento in Italy and Massimo Warglien of Ca’ Foscari University in Venice built mathematical models based on biological neural networks. These use simulated networks of “brain cells” to arrive at decisions and learn by trial and error. Introducing an approximation of regret allowed the models to predict human behavior more precisely than conventional economic learning theories, the researchers said. Their findings appear in the Feb. 22 issue of the research journal Science. “Regret refers to the difference between outcomes attained and the best outcomes that might have been attained,” wrote Michael D. Cohen of the University of Michigan, Ann Arbor, in a commentary in the journal. “This is an important step in the development of a workable new synthesis,” added Cohen, who wasn’t involved in the study. The work has applications in development of economic theories that predict human behavior, he added. The model’s predictions, he continued, aren’t based on “conventional, forward-looking expectations of gain, the notion so long at the heart of economic theorizing.” Rather, its predictions rely on “propensities that develop through a backward-looking learning process that is driven by regret.”