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


A computer can figure you out as well as your spouse, study finds

Jan. 14, 2015
Courtesy of the University of Cambridge
and World Science staff

A com­put­er can read your per­son­al­ity as well as your spouse if you give it 300 Face­book “Likes” to an­a­lyze, a study has found.

The re­search­ers be­hind the work de­vel­oped a com­put­er mod­el that they said could pre­dict a per­son’s per­son­al­ity more ac­cu­rately than most of their friends and fam­i­ly, just from Face­book Likes. Such cheap, au­to­mat­ed and ac­cu­rate per­son­al­ity as­sess­ments could be use­ful for ap­plica­t­ions rang­ing from job re­cruit­ment to dat­ing ser­vic­es, they added.

The find­ings were pub­lished Jan. 12 in the re­search jour­nal Pro­ceed­ings of the Na­tio­n­al Aca­de­my of Sci­en­ces

“In the fu­ture, com­put­ers could be able to in­fer our psy­cho­log­i­cal traits and re­act ac­cord­ing­ly, lead­ing to the emer­gence of emotionally-intelligent and so­cially skilled machi­nes,” said lead au­thor Wu Youyou, from Uni­vers­ity of Cam­bridge in the U.K. But he and his col­leagues added that the re­sults raise pri­va­cy con­cerns, so they sup­port poli­cies giv­ing us­ers full con­trol of their dig­it­al foot­print.

The study found that a com­put­er could more ac­cu­rately pre­dict the sub­jec­t’s per­son­al­ity than a work col­league by an­a­lyz­ing just ten Likes; more than a friend or a room­mate with 70, a par­ent or sib­ling with 150, and a spouse with 300 Likes.

An av­er­age Face­book us­er has about 227 Likes, the study au­thors said.

The find­ings build on pre­vi­ous work from Cam­bridge, pub­lished in March 2013, which showed that Face­book Likes could serve to pre­dict a va­ri­e­ty of psy­cho­log­i­cal and de­mo­graph­ic char­ac­ter­is­tics with startling ac­cu­ra­cy.

In the new stu­dy, re­search­ers relied on 86,220 vol­un­teers on Face­book who com­plet­ed a 100-item per­son­al­ity ques­tion­naire through the ‘myPer­sonal­ity’ app.

The find­ings pro­vid­ed self-reported scores for what psy­chol­o­gists call the “big five” per­son­al­ity traits: open­ness, con­sci­en­tiousness, ex­tra­ver­sion, agreea­bleness, and neu­rot­i­cism (or “O­CEAN.”) Through this, re­search­ers could es­tab­lish which Likes equat­ed with high­er lev­els of par­tic­u­lar traits. For ex­am­ple, lik­ing Sal­va­dor Da­li or medita­t­ion showed high “open­ness.”

Users of the “myPer­sonal­ity” app were then giv­en the op­tion of in­vit­ing friends and family to judge their psy­cho­log­i­cal traits through a shorter ver­sion of the per­son­al­ity test. Re­search­ers were able to get more than 22,000 par­ti­ci­pants judged by one or two friends or family mem­bers.

To gauge the ac­cu­ra­cy of these mea­sure­ments, the on­line per­son­al­ity judg­ments were cor­rob­o­rat­ed with an anal­y­sis of pre­vi­ous stud­ies on how peo­ple judge per­son­al­ities. Re­search­ers found their on­line results si­m­i­lar to the av­er­ages from years of per­son-to-per­son re­search.

Mi­chal Kosin­ski, co-au­thor and re­searcher at Stan­ford Uni­vers­ity in Cal­i­for­nia, said machines have a cou­ple of key ad­van­tages over peo­ple: the abil­ity to re­tain vast amounts of in­forma­t­ion, and to an­a­lyze it with al­go­rithms, or for­mu­las.

“Big Da­ta and machine-learning pro­vide ac­cu­ra­cy that the hu­man mind has a hard time achiev­ing, as hu­mans tend to give too much weight to one or two ex­am­ples, or lapse in­to non-ra­t­ional ways of think­ing,” he said. Nev­er­the­less, the au­thors con­cede that de­tec­tion of some traits might be best left to hu­man abil­i­ties, those with­out dig­it­al foot­prints or de­pend­ent on sub­tle cog­ni­tion.

* * *

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

A computer can read your personality as well as your spouse if you give it 300 Facebook “Likes” to analyze, a study has found. The researchers behind the work developed a computer model that they said could predict a person’s personality more accurately than most of their friends and family, just from Facebook Likes. Such cheap, automated and accurate personality assessments could be useful for applications ranging from job recruitment to dating services, they added. The findings were published Jan. 12 in the research journal PNAS. “In the future, computers could be able to infer our psychological traits and react accordingly, leading to the emergence of emotionally-intelligent and socially skilled machines,” said lead author Wu Youyou, from University of Cambridge in the U.K. But he and his colleagues added that the results raise privacy concerns, so they support policies giving users full control of their digital footprint. The study found that a computer could more accurately predict the subject’s personality than a work colleague by analyzing just ten Likes; more than a friend or a roommate with 70, a parent or sibling with 150, and a spouse with 300 Likes. An average Facebook user has about 227 Likes, the study authors said. The findings build on previous work from Cambridge, published in March 2013, which showed that Facebook Likes could serve to predict a variety of psychological and demographic characteristics with startling accuracy. In the new study, researchers used a sample of 86,220 volunteers on Facebook who completed a 100-item personality questionnaire through the ‘myPersonality’ app. The findings provided self-reported scores for what psychologists call the “big five” personality traits: openness, conscientiousness, extraversion, agreeableness, and neuroticism (or “OCEAN.”) Through this, researchers could establish which Likes equated with higher levels of particular traits. For example, liking Salvador Dali or meditation showed high “openness.” Users of the ‘myPersonality’ app were then given the option of inviting friends and family to judge their psychological traits through a shorter version of the personality test. Researchers were able to get more than 22,000 participants judged by one or two friends or family members. To gauge the accuracy of these measurements, the online personality judgments were corroborated with an analysis of previous psychological studies over decades which looked at how people’s colleagues, family and so on judge their personality. Researchers found their online values similar to the averages from years of person-to-person research. Michal Kosinski, co-author and researcher at Stanford University in California, said machines have a couple of key advantages over people: the ability to retain vast quantities of information, and to analyze it with algorithms, or formulas. “Big Data and machine-learning provide accuracy that the human mind has a hard time achieving, as humans tend to give too much weight to one or two examples, or lapse into non-rational ways of thinking,” he said. Nevertheless, the authors concede that detection of some traits might be best left to human abilities, those without digital footprints or dependent on subtle cognition.