[Epistemic status: speculative. My goal here is only to articulate a plausible hypothesis given the data.]

The Moral Economy, which I mentioned in my post about my recent reading list, talks quite a bit about the crowding-out effect, in which placing external incentives on behavior decreases intrinsic drive for that behavior. This can cause incentives to have a smaller effect on behavior than classical economic theory would predict, or even have the complete opposite effect.

His main source of data is a large number of experiments in which participants across the world play simple games such as prisoner's dilemma, with modifications such as allowing players to punish each other or give bonuses for good behavior. The crowding-out effect is the most frequently observed phenomenon, but it isn't a predictable thing by a long shot: he observes a wide variety of behavior, even including what he calls "crowding in" (incentives increasing intrinsic pro-social motivation). So, there's one puzzle: what creates the different response patterns?

Another puzzle: the crowding-out effect suggests, heuristically, that capitalism would decrease pro-social motives. Incentivising behaviors with money made people more selfish in the experiments. Other priming experiments suggest that people get more selfish with the mere mention of money. Furthermore, experiments show that there is a "spillover" effect: playing games which crowd out non-selfish motives makes people play more selfishly in subsequent games. So, one might suspect that long-term exposure to capitalism would result in long-term crowding out, creating more selfish behavior in the games.

The opposite is true: exposure to capitalism highly correlates with altruistic behavior in these games. Hunter-gatherer societies which are based on mutual sharing of food are among the least cooperative, when it comes to the games. Why?

The book speculates that advanced economies engender a much higher degree of trust in strangers. If you regularly have economic interactions with strangers without being conned, you come to expect (and provide) a level of common decency.

Experiments also show that crowding-in, rather than crowding-out, can be engendered by allowing participants to discuss what actions should be punished before they start play, rather than making them anonymously dole out punishment without group consent. People from market economies are more likely to act like such an agreement already exists, though, while people from less market-based societies are likely to treat punishments as an insult and react by cooperating less.

It sounds to me like people from free-market societies, or people allowed to discuss the system of punishments before the game, internalize the incentives: they see it as consistent with their intrinsic motives, and so they feel guilty if they are punished, and respond positively by shifting their behavior to be more pro-social. (This is borne out by the data.)

When I say "guilty", I have in mind the technical distinction between guilt and shame; guilt is about bad actions, whereas shame is a feeling that you yourself are bad. Shame is low self-esteem.

The research in Self Theories (also discussed in my earlies post) suggests that a shame response is connected to goals such as impressing others, getting good grades, and looking smart. In other words, extrinsic motivators. The students in those studies who reacted well to failure didn't so much have high self-esteem; rather, they were not focused on self-esteem. They were focused on intrinsic motivators. This allowed them to take the feedback provided by failure well, converting it into information about what behaviors don't work (so, more in the realm of guilt) rather than information about them being good or bad (shame).

Both guilt and shame are internalizations of negative reinforcement. However, shame is more likely to lead to learned helplessness, while guilt is more likely to lead to adaptive behaviors.

This reminds me of Goals as Excuses or Guides by Fishbach & Dhar. They talks about a similar distinction with respect to positive rather than negative reinforcement. When we succeed, our subsequent behavior depends on what we think that success meant. If we interpret it as progress on a goal, moral self-licensing is more likely to occur. This is a phenomenon where you perceive yourself as good, and therefore, allow yourself to do more bad things. A classic example would be someone trying to quit smoking who feels good about going for a day without smoking, and rewards themselves with a smoke "because they deserve it". Fishbach & Dhar find that re-framing the achievement in terms of showing commitment to values, rather than progress toward goals, has a tendency to reinforce the behavior rather than the paradoxical self-licensing effect.

To make an analogy to the guilt vs shame distinction: moral self-licensing happens when we interpret our actions as meaning we are good, whereas reinforcement of the behavior happens when we think the actions are good. Again we can see the connection to self-esteem. High self-esteem seems as problematic as low self-esteem: it causes us to paradoxically work against ourselves. Concentrating on how our actions relate to what we value (IE, concentrating on our intrinsic motives) makes self-esteem less relevant and our actions less self-contradictory.

Human motivation seems to be really complicated! To summarize my hypothesis: extrinsic motivations crowd out intrinsic motivations, which can cause incentives to paradoxically have the opposite of the intended effect. However, this is not always observed in the experiments in The Moral Economy because sometimes people internalize incentives. It seems incentives can be internalized in at least two different ways: they can be connected to actions, or they can be connected to self-esteem. Connecting negative reinforcement to self-esteem is called shame in the psychological literature, and leads to learned helplessness. Internalizing negative reinforcement via a connection to actions is called guilt, and seems to be more adaptive. Connecting positive reinforcement to self-esteem could be called pride. It tends to make us work against our own goals via moral self-licensing. For the sake of rounding out the set of terms, I'll call the internalization of positive reinforcement as connecting our actions to our values joy. Joy reinforces successful behaviors, making our actions tend to be more consistent with our values.