Why Did the Bees Commit Suicide?
TWO RESEARCHERS BASED IN WASHINGTON, D.C., PHILIP KEEFER AND Stephen Knack, set out to ascertain the extent to which people trust strangers in a research study that they published in a leading economics journal in 1997.1 Thousands of individuals in dozens of countries were asked to rate their trust in people they did not know well, ranging from their car mechanic and primary care physician to government officials responsible for public services. One of the more interesting findings in that study was a strong correlation between the trust people are willing to give to strangers and the GDP of the country in which they live.2 Countries with high levels of trust in strangers have correspondingly higher GDPs. The study did not reveal a direct causal link between trust and economic development, but subsequent research studies, some of them using laboratory experiments, have been able to uncover convincing underlying reasons for the correlation.
Trust is an engine of cooperation between individuals. Cooperation, in turn, is an engine of economic growth and social welfare. Trust cannot be sustained in a society without credibility, the behavioral trait that fosters trust. On the other hand, just as trust cannot survive for long without credibility, credibility is eventually destroyed without trust. If trust is virtually nonexistent in a social setting, then there is no point in trying to develop or sustain credibility; in that situation you are better off adopting selfish and unreliable behavior. Societies and nations can be in one of two equilibria: a “good” equilibrium in which individuals trust each other and behave in a reliable and cooperative manner toward others (justifying the trust), or a “bad” equilibrium in which individuals do not trust each other, with that lack of trust becoming self-justifying as people act without any sense of a need to be trustworthy or reliable. It is easy to guess, even without empirical data, which of these equilibria leads to greater economic growth.
Economists are divided on the question of whether these equilibria emerge from random processes or are dependent on initial conditions. If the first opinion is true, then the difference between contemporary Angola and Switzerland is due to long-ago random events that caused Angola to be trapped in a bad equilibrium, while Switzerland found itself in a good equilibrium. According to this view, there was once an equal chance that an alternative history could have developed leading Angolan society to look like Switzerland, while the Swiss would today be living like Angolans. Those holding the opposing opinion claim that there were initial conditions (such as a prevalence of natural resources, or a certain mix of cultures, etc.) that determined which country would be lucky enough to end up with a good equilibrium and which would find itself in a bad equilibrium. Those initial conditions could involve climate, geography, or cultural elements.
Of course, none of this matters if it is possible for a society to shift from one equilibrium to another (hopefully, from bad to good). Economic researchers are divided even more harshly on this matter, which has been labeled “convergence.” Supporters of the convergence theory, who are apparently optimists by nature, claim that it is only a matter of time until Angola moves to a good equilibrium that will grant its citizens a living standard equal to that enjoyed by the Swiss. Their opponents claim that such equilibria are “ergodic” or “absorbing,” meaning that it is difficult to move from one equilibrium to another (because the bad equilibrium “absorbs” the change rather than being overturned by it). The scenarios for going from a good equilibrium to a bad equilibrium are somewhat easier to imagine: food or water shortages, disease outbreaks, collapse of the government—any of these might precipitate a breakdown of a country’s social order. But it seems to be especially difficult to go from a bad equilibrium to a good equilibrium. Imagine, for instance, that you are asked to move a large trunk from one room at your friend’s apartment to another with the help of three other people. The trunk is so heavy that it can be lifted only with the full effort of the four of you. Following several failed attempts to lift the object, it would be pretty hard to get it moved. Each of you is likely to be suspicious about how much effort the others are exerting and whether they themselves believe that the job can be done. It would take quite a bit of discussion to get the job done, as by the time mistrust builds up it would require changing the behavior of the four of you to move to a better equilibrium. If at some stage you managed to pool your efforts together and get the trunk lifted, you’d be moving to the good equilibrium, but this new equilibrium is pretty fragile. It would take only one of you to change behavior (shirk a bit) for the object to fall and for the trust to collapse.
Although both camps in the dispute over convergence theory make use of extremely complex mathematical models, no resolution of the matter has yet been attained.
Research in economics differs in some ways from research in the natural sciences. Much contemporary economics research is theoretical, making use of mathematical models. This theoretical work is not dissimilar to theoretical work in physics, which also uses mathematical models. But unlike physics, where the ultimate test for determining whether a theoretical hypothesis is true is in the supportive empirical data, there are many economic theories that are widely accepted even without being tested empirically. In many cases there is simply no way to create the empirical results that could support or refute a given theory. How in the world could we use empirical tools to settle on the truth or falsehood of a theory that claims that within the next 1,000 years the living standards of Angola will converge to those in Switzerland?
This type of theoretical research is nevertheless very important in economics; human behavior is much too complex to describe precisely using mathematical models. Instead, the role of such models is more often the clarification of a claim or insight that could also be described without a model. Models in physics are the essence of the science; models in economics are tools. There are some fancy models in economics that show why a monopoly makes a higher profit compared with a firm that operates in a competitive market. These models provide many important insights, including ones that are relevant for policy making, but these models are far from describing the entire picture, and they are rather useless for purposes of making predictions about the economy.
In the 1990s, three American economics researchers suggested using a simple game called the trust game, suitable for laboratory experiments, for studying the extent of the trust and credibility people are willing to extend to others.3 There are two players in the trust game: the first player (the proposer) is given a sum of money, say $100. This player can keep this money for himself or alternatively propose giving some of it to the second player (the receiver). For every dollar that the proposer gives the receiver, the experimenter adds two more dollars to the amount given to the receiver. For example, if the proposer grants $20 (out of the original $100) to the receiver, then the receiver will end up holding $60 (three times as much). At that point the receiver has the option of giving some of the money she is holding back to the proposer, as generously (or stingily) as she wishes.
Try to put yourself in the shoes of the players of this game and imagine what you would do. Your behavior as the proposer clearly depends on the amount of trust you are willing to put in the receiver. If you choose to keep the entire original amount for yourself, you will go home with $100 and the receiver will go home empty handed. On the other hand, if you give her some of that money, which will then be multiplied by three, and she in turn gives you half of that tripled amount of money, then both of you will end up better off. If you are really daring and you give her the entire $100 sum, then she will hold $300 in her hands; if she then gives you half of that back, both of you will end up with $150, a tidy profit all around.
But the receiver has no incentive to share the money she gets, other than good will, a sense of generosity, or a sense of shame for acting in an ungrateful manner. You, as the proposer, find yourself faced with a dilemma. If we assume selfishness and rationality on the part of both players, then game theory would predict that the proposer will never offer a penny of what she initially gets to the receiver because she can be certain that the receiver will end up giving her nothing.
Like the ultimatum game, the trust game rapidly became one of the most prominently discussed games among behavioral economists. Unsurprisingly, from the start laboratory experiments on behavior in the trust game showed that proposers were typically willing to give a significant amount (usually about one third) of the money they had to receivers. Receivers, in turn, usually rewarded this generosity by returning to the proposers the original amount given to them by the proposers in addition to a small bonus.
The significance of the trust game, however, is not that it shows that people are willing to trust others to some extent but is its ability to measure and compare the extent of trust across different cultures. Several interesting experiments have been conducted relating to this.
To take one example, Uri Gneezy and Chaim Fershtman, two Israeli researchers, set out to study the effects that ethnic origins have on people.4 They included in their experiment students at Tel Aviv University and Haifa University whose ethnic background—either European or Middle Eastern—could easily be identified by their family names. The participants in the experiment played the trust game via computer terminals, with the proposers located in Tel Aviv and the receivers in Haifa, about sixty miles away.
Each player was informed of the name of the player against whom he was playing. Players were paired in all the possible combinations: proposers of European background with receivers of Middle Eastern origin, proposers of Middle Eastern background with receivers of European origin, two European players, and two Middle Eastern players. The surprising and socially disappointing conclusion was that receivers with Middle Eastern family names were given significantly less than receivers with European names when proposers were called on to decide how much they were willing to give receivers. This discrimination against players of Middle Eastern background was mainly due to the behavior of European players, but Middle Eastern players also exhibited a degree of discrimination against players who shared their ethnic origin. Men tended to discriminate on this basis more than women. In other words, men systematically trusted players who had a European name more than they trusted players with a Middle Eastern name.
Discrimination, it turns out, is alive and kicking, as revealed by a simple experiment. We no longer see blatant discrimination around us because society strongly frowns on openly discriminatory behavior. But when we are far from the social spotlight, latent discrimination can rear its ugly head. In this experiment, the source of discriminatory behavior was an intuitive (perhaps even subconscious) feeling that many proposers had, leading them to believe that European receivers would be more likely to reward them for generosity than Middle Eastern receivers would be. Even Middle Eastern proposers apparently felt the same way about people in their own ethnic group, given the discrimination that they exhibited.
One might ask at this point if this intuitive feeling—that Middle Eastern players would be stingy in rewarding generosity—was justified by what the experiment revealed about the behavior of the receivers in the trust game. The answer is not at all. Regardless of their ethnic background, all the receivers tended to reward generous proposers to the same extent. In fact, Middle Eastern receivers were slightly more generous than European receivers!
How did the stigma against people of Middle Eastern background originate? In the previous chapter we mentioned the distinction that Robert Aumann proposed between rule rationality and act rationality. A rule rational action, as its name implies, is an action that is based on an instinctive rule that is favorable for us on average when we are involved in many different interactions over a lifetime, while an act rational action requires much more cognitive attention and is appropriate for a specific interaction.
Trust and mistrust are governed primarily by emotional rules. But while rules are effective in allowing us to reach quick decisions, they also have a major downside in their reliance on overgeneralization. The discrimination displayed in the experiment described above is an example of this type of misleading generalization. It stems from the perception that we should not trust people who are different or less well-off than us. While this might be a reasonable way of behaving under some circumstances, it could be detrimental to our interests in others. Such rules often form after just a very few instances in which we trusted the wrong people, and they are hard to modify. Indeed, they tend to survive for a long time even when they are proven to be harmful and wrong.
In this respect humans are not much different from bees, which also rely heavily on rules they find very hard to uproot. An interesting experiment on this was carried out several years ago in Germany with the aid of “artificial flowers.” An artificial flower is a colored round box containing nectar that is very attractive to bees. The experimenters created a field of artificial flowers painted in different colors, yellow or blue. Nectar was placed in the yellow flowers, but the blue flowers were left empty.
A swarm of young bees was released over the artificial field. The bees immediately began to flit between the flowers. A bee landing on a yellow flower was able to fill up on nectar while the bees who found themselves on blue flowers were quickly disappointed and moved on to another flower. Over time the number of bees landing on blue flowers gradually decreased until all the bees learned to avoid the blue flowers and flew directly to the yellow flowers every time researchers released them over the artificial field.
At that point the experimenters switched the rules on the bees: they placed nectar in the blue flowers and left the yellow flowers empty. The expectation was that the bees would gradually learn that they should switch to the blue flowers and would desert the yellow ones. But no, that did not turn out to be the case. The bees stubbornly continued to visit only the yellow flowers, maintaining their previous behavioral pattern. Locked in a false stigma, they avoided the blue flowers despite repeated frustration in every visit to an empty yellow flower. This persisted even as the bees continually lost strength from lack of nourishment. Eventually the entire bee population died. In a sense, the bees committed suicide on the altar of the “stigma” they had applied against blue flowers.
The bee experiment teaches us about the perils of our unconscious biases but also suggests a way in which humans are able to resist them. As the trust game illustrates, our willingness to place ourselves in the hands of others can be changed by social conditions. As these experiments show, that kind of environment is only possible where emotions supersede pure, logical self-interest.