Bystander Apathy

The by­stan­der effect, also known as by­stan­der ap­a­thy, is that larger groups are less likely to act in emer­gen­cies—not just in­di­vi­d­u­ally, but col­lec­tively. Put an ex­per­i­men­tal sub­ject alone in a room and let smoke start com­ing up from un­der the door. 75% of the sub­jects will leave to re­port it. Now put three sub­jects in the room—real sub­jects, none of whom know what’s go­ing on. On only 38% of the oc­ca­sions will any­one re­port the smoke. Put the sub­ject with two con­fed­er­ates who ig­nore the smoke, and they’ll only re­port it 10% on the time—even stay­ing in the room un­til it be­comes hazy. (Latane and Dar­ley 1969.)

On the stan­dard model, the two pri­mary drivers of by­stan­der ap­a­thy are:

  • Diffu­sion of re­spon­si­bil­ity—ev­ery­one hopes that some­one else will be first to step up and in­cur any costs of act­ing. When no one does act, be­ing part of a crowd pro­vides an ex­cuse and re­duces the chance of be­ing held per­son­ally re­spon­si­ble for the re­sults.

  • Plu­ral­is­tic ig­no­rance—peo­ple try to ap­pear calm while look­ing for cues, and see… that the oth­ers ap­pear calm.

Cial­dini (2001):

Very of­ten an emer­gency is not ob­vi­ously an emer­gency. Is the man ly­ing in the alley a heart-at­tack vic­tim or a drunk sleep­ing one off? … In times of such un­cer­tainty, the nat­u­ral ten­dency is to look around at the ac­tions of oth­ers for clues. We can learn from the way the other wit­nesses are re­act­ing whether the event is or is not an emer­gency. What is easy to for­get, though, is that ev­ery­body else ob­serv­ing the event is likely to be look­ing for so­cial ev­i­dence, too. Be­cause we all pre­fer to ap­pear poised and un­flus­tered among oth­ers, we are likely to search for that ev­i­dence placidly, with brief, cam­ou­flaged glances at those around us. There­fore ev­ery­one is likely to see ev­ery­one else look­ing un­ruffled and failing to act.

Cial­dini sug­gests that if you’re ever in emer­gency need of help, you point to one sin­gle by­stan­der and ask them for help—mak­ing it very clear to whom you’re refer­ring. Re­mem­ber that the to­tal group, com­bined, may have less chance of helping than one in­di­vi­d­ual.

I’ve mused a bit on the evolu­tion­ary psy­chol­ogy of the by­stan­der effect. Sup­pose that in the an­ces­tral en­vi­ron­ment, most peo­ple in your band were likely to be at least a lit­tle re­lated to you—enough to be worth sav­ing, if you were the only one who could do it. But if there are two oth­ers pre­sent, then the first per­son to act in­curs a cost, while the other two both reap the ge­netic benefit of a par­tial rel­a­tive be­ing saved. Could there have been an arms race for who waited the longest?

As far as I’ve fol­lowed this line of spec­u­la­tion, it doesn’t seem to be a good ex­pla­na­tion—at the point where the whole group is failing to act, a gene that helps im­me­di­ately ought to be able to in­vade, I would think. The ex­per­i­men­tal re­sult is not a long wait be­fore helping, but sim­ply failure to help: if it’s a ge­netic benefit to help when you’re the only per­son who can do it (as does hap­pen in the ex­per­i­ments) then the group equil­ibrium should not be no one helping (as hap­pens in the ex­per­i­ments).

So I don’t think an arms race of de­lay is a plau­si­ble evolu­tion­ary ex­pla­na­tion. More likely, I think, is that we’re look­ing at a nonances­tral prob­lem. If the ex­per­i­men­tal sub­jects ac­tu­ally know the ap­par­ent vic­tim, the chances of helping go way up (i.e., we’re not look­ing at the cor­re­late of helping an ac­tual fel­low band mem­ber). If I re­call cor­rectly, if the ex­per­i­men­tal sub­jects know each other, the chances of ac­tion also go up.

Ner­vous­ness about pub­lic ac­tion may also play a role. If Robin Han­son is right about the evolu­tion­ary role of “chok­ing”, then be­ing first to act in an emer­gency might also be taken as a dan­ger­ous bid for high sta­tus. (Come to think, I can’t ac­tu­ally re­call see­ing shy­ness dis­cussed in analy­ses of the by­stan­der effect, but that’s prob­a­bly just my poor mem­ory.)

Can the by­stan­der effect be ex­plained pri­mar­ily by diffu­sion of moral re­spon­si­bil­ity? We could be cyn­i­cal and sug­gest that peo­ple are mostly in­ter­ested in not be­ing blamed for not helping, rather than hav­ing any pos­i­tive de­sire to help—that they mainly wish to es­cape an­tihero­ism and pos­si­ble re­tri­bu­tion. Some­thing like this may well be a con­trib­u­tor, but two ob­ser­va­tions that miti­gate against it are (a) the ex­per­i­men­tal sub­jects did not re­port smoke com­ing in from un­der the door, even though it could well have rep­re­sented a strictly self­ish threat and (b) tel­ling peo­ple about the by­stan­der effect re­duces the by­stan­der effect, even though they’re no more likely to be held pub­li­cly re­spon­si­ble thereby.

In fact, the by­stan­der effect is one of the main cases I re­call off­hand where tel­ling peo­ple about a bias ac­tu­ally seems able to strongly re­duce it—maybe be­cause the ap­pro­pri­ate way to com­pen­sate is so ob­vi­ous, and it’s not easy to over­com­pen­sate (as when you’re try­ing to e.g. ad­just your cal­ibra­tion). So we should be care­ful not to be too cyn­i­cal about the im­pli­ca­tions of the by­stan­der effect and diffu­sion of re­spon­si­bil­ity, if we in­ter­pret in­di­vi­d­ual ac­tion in terms of a cold, calcu­lated at­tempt to avoid pub­lic cen­sure. Peo­ple seem at least to some­times hold them­selves re­spon­si­ble, once they re­al­ize they’re the only ones who know enough about the by­stan­der effect to be likely to act.

Though I won­der what hap­pens if you know that you’re part of a crowd where ev­ery­one has been told about the by­stan­der effect...


Cial­dini, R. (2001.) In­fluence: Science and Prac­tice. Bos­ton, MA: Allyn and Ba­con.

Latane, B. and Dar­ley, J. (1969.) By­s­tan­der “Apa­thy”, Amer­i­can Scien­tist, 57: 244-268.