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.

New Comment
20 comments, sorted by Click to highlight new comments since:

Isn't somebody else going to make a comment?

I'm trying to work out now if you were downvoted by someone who didn't get it, or someone who did.

Oh I totally did not get that. Voted up now. (I was not the original downvoter.)

There's a similar opposite effect - people are more likely to riot or commit violent acts in large crowds than in small ones.

The explanation that "choking" is disadvantageous seems to explain a great deal of this behaviour. However, as Eliezer points out, fear to take on the risk of action doesn't explain the whole thing: one person with a gallantry gene tears that model apart.

It strikes me as possible that there's a large cultural role here. It may be that we don't intervene not because of selfish evolutionary decision-making, but because of an overdeveloped sense of inferiority to the collective. I remember reading more than once that the placebo effect is much more pronounced in Germany and other Northern European countries than it is in the United States or Brazil on account of the Northern European cultural trust in figures of authority, which isn't as strong in the New World.

I wonder if that translates to bystander apathy. It seems plausible that someone from a usually orderly, law-abiding country like the US or Germany might see an emergency or hear cries for help and think to himself "I'm sure the Police of Fire Department or somebody have that under control. I don't know what I'm doing, so it's best not to get involved." This seems consistent with the pluralistic ignorance explanation: everyone looks around furtively for some official with a bold look and a colorful hat to signal whether we should be worried.

Have there been any studies comparing the bystander effect across different cultures? Do they tell us anything?

this was a basic leadership lesson in boy scouts. never address a group when asking for actual action.

It is also explicit instruction for first aiders; "You! Call 911 and tell them I need an ambulance at the corner of x and y! Tell them I'm performing CPR on an unconscious non-breathing victim! Ask them for an ETA and tell me what they said!"

The psychologist Eliezer cited, Cialdini, was involved in a car accident. Both he and the other driver were clearly hurt. He watched as other cars passed by without stopping.

"I remember thinking, Oh No, it’s happening just like the research says. They are all passing by! I considered it fortunate as a Social Psychologist I knew exactly what to do. Pulling myself up so I could be seen clearly, I pointed at the driver of one car; Call the Police – to a second driver and third driver pointing directly each time; Pull over we need help – The responses of these people were instantaneous..."

[Edit - readers seem interested so adding more. It's from Influence, The Psychology of Persuasion]

"Not only was this help rapid and solicitous, it was infectious. After drivers entering the intersection from the other direction saw cars stopping for me, they stopped and began tending to the other victim. The principle of social proof was working for us now. The trick had been to get the ball rolling in the direction of aid. Once that was accomplished, I was able to relax and let the bystanders' genuine concern and social proof's natural momentum do the rest."


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.

How big are the variances and measurement errors on those numbers? If everyone reports only 10% of the time when no one else does, then a group of three should report 1-0.9\^3 ~= 27% of the time. 38% seems quite a lot bigger than that. Is it possible that the confederates in the one-subject experiment were not merely not-reacting in the way that "real" people commonly do, but were behaving in a way that ordinarily means "I know what's going on, and it's fine"? (I don't think it's quite fair to classify that as pluralistic ignorance.)

I do also wonder, about the three-real-subjects case: if these people knew they were in a psychology experiment, it must have occurred to some of them that the smoke might be part of the experiment. (Unless subjects then were very naive about the tricks psychologists play.) With a group of people, it must be more likely that someone will think of that possibility. What does the article by Latane and Darley say about the actual behaviour of the subjects?

(My guess is that that last effect isn't a major contributing factor, because it presumably doesn't apply in the two-stooges case but the response rate there was lower.)

It's probably because the subject is unable to persuade the confederates to report. With three subjects it is possible that who starts conversation isn't the same person as who finally reports the fire. With only one subject he has to do all himself.

I think the study with confederates investigates the same effect, or at least its "microscopic dynamics" valid for one person separately. Nobody can consciously assume that the other "subjects" know that the smoke is OK. So it must be some kind of bias.

I would not call correctly deducing that the smoke is OK from the correct perception that the other people in the room are confident that it is OK a bias. The lower rates with the confederates may indicate people have some ability to tell the difference between another person actually believing that nothing is wrong and the other person remaining calm while still considering whether something is wrong. Of course, it would be interesting to know whether this ability comes from talking with the other person or merely observing them.

Generally, when a psychology experiment allows its subjects to find clues about its real purpose, you have to consider the possibility that the results represent the subjects seeing through the experiment.

If everyone reports only 10% of the time when no one else does, then a group of three should report 1-0.9\^3 ~= 27% of the time.

I agree. The experiment with the confederates seems to be testing something stronger than the bystander effect.


It actually seems like a rather elegant and efficient way of analyzing a situation, that just happens to backfire in some cases. Instead of trying to understand exactly what is going on, we just try to understand what everyone else understands. I don't have to bother trying to figure out why something is happening and the appropriate response, I just have to do what everyone else is doing. It's probably one of many adaptations we have for acting as a group, and I'd imagine its very effective most of the time.

Perhaps it's so effective that we (subconsciously) rely on it more than other methods of analysis, so it can overrule what might otherwise be a strong reaction to something. So if there's smoke coming from under the door but everyone else doesn't seem to mind, we don't say anything.

It's elegant, efficient and I would also add: socially polite. If everyone else knows what is going on, you don't need to add to the noise by making them explain it to you. For example, if the homeless person was trying to get some sleep, and not having a heart attack, it would be annoying if people kept waking him up to check on him. If you look around, maybe somebody knows which case it is.

Also I think performance anxiety is important. It's not just shyness -- I think a sense of inferiority is correct. I don't know how to do the Heimlich! And if I pull over, am I pulling over in the right spot or will I be in the way of the emergency vehicles? This isn't just rationalization -- I would be greatly relieved to find an unobtrusive place to park.

Funny, when I read this article I had to think about the current economic crisis and global warming. Maybe most people are apathetic about these because they look around and see that everyone else is acting like normal so they suppose it's not as big a deal as it really is and that everything is going to turn out ok.

That is frighteningly plausible. But how do you explain our political obsessions like abortion, gay marriage, or drugs? Those things aren't existential threats, but our political discourse treats them like they are and they often dominate the discussion.


Perhaps we differentiate between talking about a situation (e.g. "someone ought to do something about gay marriage") and "actually doing something" about it (e.g. writing your governmental representative).


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

Shyness is beeing discussed in analyses of the bystander effect. A psychology lecture i attended in 2001 mentioned it as the third important factor. This effect:

If I recall correctly, if the experimental subjects know each other, the chances of action also go up

was said to proof the role of shyness (beeing with a friend reduces shyness)

The wikipedia article on the bystander effect mentions it, too: "People may also experience evaluation apprehension and fear losing face in front of the other bystanders"

My intuition is the bystander effect is highly related to the fact that most crowds/groups in audiences will not speak, or even that most people at a social gathering hate to be the first to go to the buffet. At work I make a point of asking questions of presenters early on, of volunteering somewhat unfiltered interaction with the presenter when he asks for it. I believe my action greatly increases the rate of participation of co-workers in the interactive presentations in which I do this. My intuition is this is related to "warming up an audience," a process where if you ever go to a TV filming you will be encouraged to yell, scream, answer silly questions, and just generally abandon your inhibitions before the filming actually starts. They WANT filmable (at least audible) audience reaction and presumably they have learned that 'warmed up" audiences provide a lot more of that.

The reticence in all cases may have more to do with an evolved wariness of strangers than anything else. If someone I knew reasonably well were lying on the ground looking like he might POSSIBLY need help, I would not ignore him just because everybody else was ignoring him.

If I recall correctly, if the experimental subjects know each other, the chances of action also go up.

You do recall correctly according to the first guy you cited, and Experiment 2 in the paper you cited had 70% of friend pairs and 40% of stranger pairs helping.