Can’t We All Just Be Prosocial? A Guest Post

David DeSteno, a psychology professor at Northeastern University, blogged here earlier this week about his research on moral hypocrisy. This is his last of three posts on the subject.

On Wednesday, I ended by suggesting that humans might just be good at heart. Sappy, I know, but nonetheless true (well, at least partially). For a long time, emotions have taken the rap for much of the dark side of human nature — violence, prejudice, greed. Yes, there is evidence to support each of these (some even from my lab).

However, just because emotions sometimes lead to undesirable behaviors does not mean that they must always be controlled for virtuous behavior to emerge.

A principal function of emotions is to provide an ongoing assessment of one’s environs and to constrain subsequent behaviors to increase adaptive responding. To use the classic example, the state that we call fear increases the likelihood of subsequent responses meant to avoid danger.

For humans, though, problems of predation or contracting communicable diseases may be no more important to flourishing than ones integrally involved in successfully navigating social life. Simply put, building social and economic capital is central to our success and depends on correctly dealing with questions like: “Who should I trust?” “Should I pay you back?”

Consequently, humans should possess a class of emotional responses that prod them to be “good partners.” As my collaborator Robert Frank has argued, many of these social emotions should help us control our greedy “id-like” impulses.

Now, the idea of such “moral sentiments” has been kicking around at least since Adam Smith, but supportive data has been hard to come by, at least until recently.

Consider the following experiment conducted by Monica Bartlett and myself. We brought people into the lab and set up 2 situations: One in which they confronted a problem which would require them to complete an onerous task and one where they didn’t face any problem. In the first case, a confederate, at some cost to herself in terms of time and effort, helped the participant solve the problem, which led to measurable feelings of gratitude. In the second, the confederate was just another person in the session.

After leaving the lab, all participants just happened to encounter someone asking for help on a different onerous task. This person was either the known confederate (labeled benefactor in the figure) or someone who was a complete stranger.

Looking at the first two bars, you can see that grateful participants helped the known confederate much more than neutral participants.

Ok, I know what you’re thinking. This doesn’t prove anything! They may just be following a reciprocity norm. Fair enough. But look at the second set of bars. If it were really reciprocity, then no increased helping should occur when a stranger requests help, as participants don’t owe this stranger anything. Yet, those who were feeling grateful still helped more. Simply put, gratitude functioned to push people to acquiesce to requests for help — even onerous ones from unknown others.


Importantly, another study showed that if we reminded the participants before they left that they were helped by the confederate, they didn’t help the stranger any more than control participants. By binding the emotional state so saliently to one person, it couldn’t be misattributed as a cue to help another, thereby indicating that the increased helping isn’t just adherence to a “pay-it-forward” norm. Yet participants still were paying-it-forward.

To me, this represents a spandrel effect for gratitude. Although its primary function is to make you feel grateful toward benefactors and thereby repay them at cost to yourself, there is a side benefit at the group level. If someone else just happens to ask for a favor while you’re feeling grateful, your odds of agreeing are higher.

As work by Nowak and Roch has suggested, mechanisms for such upstream reciprocity are fundamental to the development of cooperative society. But then again, they could also make you a sucker to a good con artist! On balance though, such emotional responses continue to lead me to believe that at heart, we’re designed to be prosocial.


This was an interesting post


For Ben, who believes that voluntary charity is better than 'coercive charity' because "charitable giving makes people feel unusually happy": I refer you to John Steinbeck's 'The Log from the Sea of Cortez': Giving builds up the ego of the giver, makes him superior and higher and larger than the receiver. Nearly always, giving is a selfish pleasure, and in many cases it is a downright destructive and evil thing. . . For giving can bring the same sense of superiority as getting does, and philanthropy may be another kind of spiritual avarice."

Tom Vellenga

Of course humans are prosocial. I wasn't aware this was doubted. There is a common misunderstanding that humans have evolved in the 200,000 years since the Savanna. That somehow we have become more civilized or intelligent or moral or whatever. This is a naive myth with no supporting evidence. Our ancestors were as intelligent as we are now, if we would have evolved since then we would no longer be considered homo-sapains. In my opinion, if you want to understand humans in the 21st century, you first have to remember that we all have prehistoric minds that were not designed to live in the world we now inhabit.

As to lying, every organism on this planet uses elements of deception to increase it's chances of survival and reproduction. From bacteria to Humans, we will all lie if we believe we can get away with it. Take the practice of women wearing make-up to artificially increase their sexual attractiveness by masking flaws in their appearance. This is a socially acceptable form of deception, but it is deception all the same. Morality is a matter of culture rather then of core psychology.

Humans are tribal animals, and as such we tend to view morality on an "in group" and "out group" basis. Many Americans seem to hold the belief that an American life (the in group) is worth more then the life of a non-American (out group). We are more likely to help people in our "tribe", and let them get away with more then we would a stranger.

There are many variables that could have explained the increase in the study such as how attractive the confederate was, weither he/she was of the same race and age group, etc.


David Kramer

In the Nicomachean Ethics, Aristotle says that virtuous behavior is a "Golden Mean" between two extremes, one excessive and one deficient. So, for example, courage is a mean between foolhardiness and cowardice; self-esteem is a mean between vanity and self-hatred.

In this case, charity is a mean between gullibility and stinginess. As has been posted earlier, if one is overly-generous with one's time or money in helping others, one will be duped by con-men. If one is overly suspicious, uncoooperative or miserly, one will develop a bad reputation and be ostracized by the group.

All this makes sense from an evolutionary perspective. The "Golden Mean" is the mid-point of the bell-curve of "fit" behavior. If your behavior is too extreme in either direction, you are less likely to pass on your genes.

Society uses rewards and punishments to encourage prosocial, cooperative behavior and discourage uncooperative, anti-social behavior. It tries to move the mid-point of fit behavior in the prosocial direction. However, it's debatable how far this can or should go, and how much influence this has on our evolutionary development.



It'd be interesting to know the difference between "prosocial" and "social".

Volt Rare

These are interesting observations.

Therefore, feeling grateful raises the bar for helping others? In effect, it raises a pro-social group norm?

But let me play devil's advocate by asking how does this apply to the real world where substantial money is involved?

For example, let's consider the employer-employee relationship: why in many cases, is good work & loyalty by employees not rewarded with gratitude & reciprocal commitment (ie: a more generous share of profits, mutual respect, & greater job security) by employers?

Of course, lack of reciprocity (in regard to "help-feelings"), in such economic relationships probably negatively impacts the overall business, and it would be interesting if there was further research-supported elaboration in this area.


The difference in the "stranger" data doesn't look statistically significant, but it does (barely) for the "benefactor" data. Follows the idea of reciprocity then?

steve long

Dave - Let's do a different analysis of the experiment. I believe this is much closer to what you are actually observing.

Subjects in a new situation are uncertain as to how they are expected to behave.

In their first encounter, they are either aided or left to fend for themselves.

Those who are aided have reason to believe that they are also expected to aid, and are more likely to do so in the second phase of the experiment.

Those who are not aided have reason to believe they are not expected to aid, and are more likely not to aid in the second phase of the experiment.

(The variation where the person rendering aid is praised for the aid, in turn, tells the subjects that the aid is exceptional and not expected.)

After reports of gratitude or lack of it are after-the-fact rationalizations. (I was not told to help, therefore it must have been gratitude that made me help.)

The most interesting groups in the study are those who refused to aid after being aided, and those who aided after not being aided. These guys were marching to different drums.

Of course, the chances are that all the subjects are strongly socialized, whether they helped or not. If you had conducted this experiment among misanthropic hermits or happily cooperative 19th Century Tahitians, the results would have skewed quite radically.

I'm afraid our social nature runs much deeper than this. If you speak a known human language, or go to the supermarket, or use an indoor toilet, your already indebted to thousands of humans who've contributed to these actually very socially-dependent activities.

If you want a REAL picture of raw human instincts, you're going to have to get your hands on some new-born infants, raise them entirely without human contact, and see how they turn out.

Nothing short of that will do, because your subjects are such social creatures that even when they go to the bathroom, they're counting on the cooperation of thousands of other humans who laid the pipe, pump the water, made the ceramics, designed the flusher and the toilet seat and cut down trees and transported the toilet paper, as well as all those who manage the sewage at the other end. Such a private event, but so intensely social.

We are not just pro-social. We are almost entirely social.



Dave DeSteno

Interesting points all. Much of the data and methods concerns raised can be clarified. I don't have the space here to go into details, but you can read the fuller account of methods and analyses in the published paper (Bartlett & DeSteno, Psychological Science, April 2006).

In a nutshell: (1) the differences are significant (i.e., main effect of emotion condition is significant and no hint of interaction, plus there are replications) (2) measures of gratitude are taken before the helping task, so they can't be post hoc rationalizations. Also, there is a series of regression analyses showing that variation in gratitude (even when controlling for whether the person was helped) predict subsequent helping.

Thanks again for your comments.

Lauren Auverset

This has been a fascinating series of posts, but I think this last is my favorite yet. It brings to mind a recent Scientific American article that dealt with the biological basis of trust (some hormone in the brain) and looked at how trust and the ability to work as part of a social unit were connected with economic prosperity. The June issue? July? I recommend the article for anyone interested in trust, emotion, and adaptiveness.

Statistical Sam

Why only upper bound error bars? While it's rather difficult to tell from the graph (which is something of a Tuftean nightmare) it appears that if the lower bounds were drawn in, the "gratitude" and "netural" error bars for the "Stranger" side of the experiment would overlap, leading me to question whether there's actually a statistically significant difference here...


Commenter #5, Sirdonic, poses an interesting question about grief. Why do we grieve?

I don't know that this question has been answered, and is certainly related to the topic at hand, though the lack of an answer doesn't necessarily disprove the profs argument.

I would point Sirdonic to Antonio Damasio's "Looking for Spinoza: Grief, Sorrow, and the Feeling Brain" for some insight, or Douglas Hofstader's "I am a Strange Loop", for some interesting thoughts on the subject.

...I tend to think that long-term grief, like grief over losing a loved one, is due to the fact that you become "addicted" to having feeling for, with, and about them for so long, that the grief is a chemical reaction to the idea of missing the joy.

Ian Gilbert

I feel good after I solve an onerous problem. When I feel good I am more generous and helpful than I am when I am in a bad mood.

If I had been one of the participants who solved the onerous problem, with or without help, I would be well-disposed to help someone who asked for help. While gratitude to someone not present might be involved, my good feelings about having solved the problem would also be important -- "I've accomplished something, so I'm glad to help someone who needs help."

Washington DC


I'm a little surprised at how low the threshold is for claiming an effect in psychology! The error bars there -- presumably one-sigma -- imply that about 1/3 of the time the results will flip the other way. About the same chance of getting two heads in a row!

It's not a strong result that's being shown here, and could very well -- especially once one worries about other sources of error not include -- be entirely fictional.

In more quantitative sciences, effects would have to be far stronger before these sorts of conclusions would be drawn.

maurie beck

Mr. Desteno mentions being set up for the next con. That is true, especially in a one play game such as the prisoner's dilemma. However, humans play the multi-play variant of that, which is Tit for Tat. If someone treats you well, you tend to follow suit. The con is usually only a one-play game with each individual. After being burned, the victim usually becomes more cautious.

Another idea is that cooperation is, in a sense, self-fulfilling. Individuals tend to interact with like-minded individuals. This results in a situation where individuals tend to interact cooperatively within a group of interactors. If someone breaks the trust, the cheater not only pisses off the victim, but gets a reputation as a cheater and is ostracized by the whole group. In other words, there is an enforcement mechanism for cooperative behavior. Without such enforcement mechanisms, the whole enterprise would quickly be overrun by cheaters, which would be the logical strategy. In large groups (e.g. societies), cheaters can often get away with cheating, by moving between groups. However, large groups also tend to have mechanisms to dissuade cheating (e.g. laws).

Finally, there are dominance hierarchies that can disrupt or enforce cooperation. In a small group a strong individual can impose his or her will on others in the group. This can lead to group stability. It can also lead to autocracy.


Andy B.

Hey, helping people just feels good.

I ran after a young lady in Midtown NYC yesterday to return the watch she had dropped.

I felt good for the rest of the day even though all I

objectively obtained was a quick "Thanks".

Go figure.

Danielle Zheng

Might there be a sample bias? I'm not sure how you recruit participants, but isn't there a likelihood that those volunteering to participate are the type that are more apt to help others, especially if they're feeling grateful? I suspect those that can't control their "id" impulses are the least likely to be in such an experiment, as they would either be behind bars (in the extreme cases) or wouldn't do volunteer for this study, since there is minimal benefit to them.


i'm always puzzled by psych researchers' habits of secrecy.... wouldn't we have an easier time making sense of this setup if we knew what the "onerous task" was? Are we talking about help screwing in a lightbulb? or about tax preparation?

Adam Hammond

I have enjoyed reading the articles and the comments. Thanks.

I suspect that the same increase in agreeing to help a stranger would be invoked by simple exposure to people who are helping each other. It is my perception that we are highly attuned to our immediate social setting. We subconsciously look for cues about how to act. I bet you could readily manipulate a persons willingness to help a stranger simply by changing the scenes on posters along the hall they walk down.

These are rarified circumstances that allow detection of changes to ultimately small decisions. A more important decision would likely fire up some more careful analysis and be less swayed by setting.


Somewhere back in one of these posts, someone suggested lying is an effective survival strategy. Perhaps, but only in the very short term. Besides the obvious that in the real world no one will want to do business with a known liar (or it will all be on a cash basis), this sort of experiment shows that when you get to know someone and then also learn how to acknowledge when they have done you a service, you will get better results (schmoozing them in the best sense).

This series started with a post on hypocrisy, noting that politicians call each other hypocrites with increasing frequency during elections. It is a very effective campaign technique since it gives voters an excuse to dislike the opposing candidate without having to think about it, but I think pointing out the mote in someone else's eye invites discussion of the beam in my eye (which as a rational person I have to admit is there).