Fostering Trust and Cooperation in Health Care:
Learning from the Prisoner’s Dilemma

Digest of a presentation by Robert Axlerod, Ph.D.
Monday, April 26, 1999

Answers the questions:

• How can we attend to the damaged trust that exists within and between health care organizations?
• How can we strategically employ cooperation and competition in nurturing relationships in health care?

Introduction: The Prisoner’s Dilemma

In the health care industry, many are operating in a con-text where trust is damaged — whether it is within an organization, or between organizations and communities. Dr. Axelrod offers some intriguing insights into the nature of this trust, and the effects of competition.

Even in the simplest of relationships, the dynamics of competition can be complex. This is famously illustrated in a scenario known as the prisoner’s dilemma. Here’s how it works. Imagine that two men are captured, taken prisoner and then separated by their captors. Each prisoner is then made an offer by his captor: Betray the other prisoner (per-haps by sharing information) and receive a lighter punish-ment. Unable to communicate with one another, each pris-oner is now tempted to betray his partner in exchange for more favorable treatment.

The possibilities and their results might be summarized like this:

Thus, if both prisoner A and prisoner B remain loyal, both refusing to betray the other, they each receive three “points.” (Points in this case are used only symbolically, to indicate advantage.) Consider this scenario from the point of view of pris-oner A:

• “If I remain loyal to prisoner B, my best-case-scenario is 3 and my worst-case scenario is 0.”
• “But if I betray prisoner B, my best case-scenario is 5, and worst is 1.”
• “Therefore, either way I am better off betraying pris-oner B. Defecting is always better than cooperating.”

Meanwhile, prisoner B is coming to the same conclu-sion that it is better to defect. In this structure that encour-ages both parties to defect, both end up losing.

This scenario plays out in television crime dramas fre-quently. It can also be seen in real-life arms races between countries, price wars and in natural systems in which two groups must share a limited food resource. Prisoner B Prisoner B

The Roles of Time and Learning in the Prisoner’s Dilemma

In actuality, Prisoner’s Dilemma scenarios are rarely “one-move” games. Rather, both parties typically have opportunity to “play” over and over (as in escalating arms races.) As they do this, they begin to explore strategies for cooperating and defecting, and each begins to learn about his opponent.

Over time, a scenario might play out like this:

 

Round	You	Other	Your score	Other’s score
1	C	C	3	3
2	C	C	3	3
3	C	D	0	5
4	D	D	1	1
5	D	C	5	0

For example, in rounds 1 and 2, both parties choose to cooperate (C), and both are rewarded. But then, in round 3, one party chooses to defect (D) thereby profiting even more. However, this move produces consequences in round 4, with damaged trust provoking a mutual defection that causes both to lose. And on it goes in a continuing dance of trust and cooperation, collaboration and conse-quences.

“Tit-for-Tat:” The Simplest Strategy for Cooperation and Competition

What are some strategies, one might wonder, for inter-acting in this competitive context? When should we coop-erate? When should we defect? Is there a balance between cooperating too much (and being a “sucker”) and defecting too much (thereby becoming a “bully”)?

In exploring this, Dr. Axelrod asked mathematicians, behavioral theorists and political scientists to offer strate-gies for the best way to play this game. They developed many complex strategies that took into account different variables such as learning styles, etc.

However it turned out that the single most effective strategy was also the simplest, developed by a computer. The strategy is called “tit-for-tat,” and it consists of just two simple maneuvers:

1. Cooperate on the first move.
2. On following moves, do whatever the other player did.

The strategy is distinguished by the following characteris-tics:

1. It is nice. Those who employ the tit-for-tat strategy are never the first to defect. This is critical to building a trustful environment, and is the biggest predictor of success.
2. It is provocable. When the other party defects, the first party responds immediately and clearly.
3. It is forgiving. If the other party backs off, then the first party responds in kind. “Let bygones be bygones.”
4. It is simple. This is not computer chess. The tit-for-tat strategy requires little strategic planning.

Basic Principles about Cooperation

1. Cooperation can evolve even among parties who are moti-vated by self-interests. The tit-for-tat strategy of reci-procity is powerful even between “egoists,” or parties who are motivated by self-preservation. However, two conditions must be present for the tit-for-tat strategy to work:
   1. Individuals must be able to recognize those they have interacted with, and be able to recall what happened.
   2. Both parties must be able to see that their rela-tionship will continue for some period into the future. After all, if the relationship is near an end, one might as well defect. But if both parties are “in this together” for awhile, both have incentive to build a cooperative relationship.
2. Cooperation does not require foresight or communication. Surprisingly, it’s not necessary to talk to the other guy in order for the tit-for-tat strategy to work. In effect, you response becomes your communication. This could be observed in World War I, where enemy units tended to be very fixed in their position for long periods. Over time, these enemies learned to “livewith” one another by developing a cooperative “live and let live” strategy without ever communicating or having relationships.
3. Tit for Tat is a robust strategy. The strategy of reciproc-ity spreads quickly. As one party continually models it (beginning with a cooperative move), the other party can quickly discern the pattern, and adopt the strategy themselves. If there are a lot of players in the world, it doesn’t take long for reciprocity to spread throughout the population. All that is required is that a few people model it at the outset.
4. A reciprocity strategy can never score better than the other player. From a short-term perspective, use of the strat-egy does not win any rounds. But from a long-term perspective, it allows both parties to “win the tourna-ment.”

Application to International Politics

The tit-for-tat reciprocity strategy, Dr. Axelrod illustrates, has played out many times in international politics. A review of military threats over the last 100 years reveals that most were successfully deterred when the tit-for-tat model was employed. Specifically, the threats were deterred when two conditions were present:

1. Military responses to threats were delivered at the same level of power as the threat, rather than exceed-ing or failing to match the challenger.
2. Diplomatic responses to threats were firm, and car-ried an offer to reciprocate compromise. Successful responses were never unyielding, nor were they uni-lateral accommodations.

Reciprocity is a natural strategy that is reinforced by international practices such as trade agreements. During the cold war, it can be observed that the U.S. and Soviet Union quickly reciprocated each others’ actions.

Questions from Bosnia

Dr. Axelrod was invited to the United Nations during the Bosnian conflict, with the intent of bringing the Serbs, Croats, Muslims, Macedonians and Hungarians together, and see if they could talk to each other and build mutual respect. After presenting the prisoner’s dilemma, partici-pants asked some important questions:

1. Can the reciprocity strategy work if one side is stronger than the other? Yes. Provocability is possible, as long as each side perceives that the other can punish them — even if the degree of power is not symmetrical.
2. What if there are more than two players in the prisoner’s dilemma? In the Bosnian conflict, for example, there were seven local and five international players. However, in such cases the players typically form alliances. Once they organize and polarize in two groups, the dynamics are the same.
3. What if there is so much hatred that the parties do not even accept each other’s existence? Again, there needs to be a sense of shared future in order for the two groups to begin interacting. In a game of Monopoly, one’s goal is to bankrupt the opponent and then end the game. A reciprocity strategy does not work in this case. But in a non-zero sum game (which is what our existence in the world is), the goal is not to defeat the other player. There must be a shared understanding among both parties that they will live in the same world together.

Noise and Misunderstandings

While working with the United Nations to develop a strategic approach to arms control, Dr. Axelrod found that large organizations (in this case, countries) often had mis-understandings. In effect, one party says “I cooperated, but the other side saw it as a defection.” This is “noise,” or mis-understanding of the response. (This was the case when Soviets shoot down an intruding Korean airline.)

This presents a problem: When such misunderstanding occur, there needs to be forgiveness — but forgiveness invites exploitation. This happened during the end of the Gorbachev period, when the Soviets were very soft and for-giving... and as a result were exploited.

There are two approaches to dealing with noise:

1. Be contrite. Do not be provoked by a “justified” defec-tion by the other.
2. Be generous. Cooperate some of the time when you would otherwise have defected.

Over time, employing these responses during misunder-standing will help to further engender a context of trust.

Opportunities in Health Care

What are some arenas in health care for fostering coopera-tion via a strategy of reciprocity? There are many:

1. Between organizations: Competition for patients; Alliances among health care orgs., such as HMOs, hospitals, partnerships; Health care orgs. and regula-tors.
2. Within an organization: Between doctors; Between doc-tors and nurses; Between management and unions.
3. Between doctors and patients.

Summary

• Cooperation among egoists (players who care only about their own success) can succeed on the basis of reciprocity.
• For this to work, there must be recognition of the other player, and a continuing interaction.
• Tit for tat is a robust strategy; it self-perpetuates and spreads within a system.
• A balanced level of generosity or contrition is helpful.
• Trust derives from the shadow of the future. It requires continuing interaction.
• The dynamics of this strategy are the same among fish, people and nations. Maybe it can even work with health care.

Q & A with Dr. Axelrod

Q: “In our group, there are a few of us stuck in defecting. We are an independent hospital with escalating problems with a small clinic. This has been going on for three years. Its hurting us, the hospital and the whole community. The doctors are making money, and they don’t see any nega-tive consequence. What can we do?”
A: This is a deep rut you’re in. The problem is that the participants don’t see the consequences. As a result, the doctors are just exploiting the situation. You need to introduce consequences for defecting doctors. They must see how this is hurting them — for exam-ple, another group threatens to take them over. Someone needs to initiate those consequences.

Q: In health care, we tend to have people sit around a table and talk, talk, talk. What’s the difference between finding one answer and using the tit-for-tat strategy?
A: Remember the trench warfare scenario, where there was no verbal communication at all. But if their snipers killed one of our guys, then we would recip-rocate. Action is important; You don’t really need for-mal communication.

However, communication with words can help by shortening the cycle. You can say “this is my inten-tion, to be reciprocating. And if you do X, I will respond by doing Y.”

Also, you can say “here’s the kind of behavior that I find unacceptable. If you do this, I will be provoked.” Clarify what counts as defection, what counts as cooperation.

Q: What about management and staff relationships in a non-union context? How can reciprocity work there?
A: When the staff is not formally organized, it is more complex to employ this strategy. A lot of times, the management would prefer to have a central person on staff to deal with. But without that, norms do devel-op. “Here is what we expect of you as a staff member, and here are the consequences.” It has to be fair, con-sistent and impartial.

Q: What is the threshold of provocability?
A: I was surprised to find that it pays to be provocable right away. I figured it would make sense to be nice for a while at the beginning, but that is not the case. If you let lots of defections build up on the other side, it’s harder to “punish them enough.” Also, the more you wait, the more you look like a sucker. So it pays to respond right away and at roughly the same level.

Q: What if the two sides have different ideas on the length of the future? Say, one side will be bought out or bankrupt soon.
A: It doesn’t have to be the same future... but there does have to be a shared sense that each group will be around for awhile. This was the case when Kissinger negotiated between Israel and Egypt. He took lots of little steps. Israel would withdraw a little, then Egypt would form trade agreements, and so on. They did this bit by bit, and not all at once.

Q: There are lots of health care providers blaming each other: Nurses blame labs, labs blame the doctor, who blames radiologists, and so on. I’m hopeful that one small group can get reciprocity started. But how do I start that?
A: The problem here is, each side does not know what the other side did. There needs to be a procedure so we can know who did what, and what went wrong.

Q: Even if two parties want to agree, there’s often a third party in healthcare. Physicians might get together in joint venture with hospital. But there’s no incentive from the HMO for the hospital to do this.
A: There are several levels at which the third party can involved. If the HMO will get the benefits from the doctors working with the hospital, they need to share that with the doctors, so they too have incentive. The HMO should say “we really have an incentive (mon-etary or otherwise) for doctors to cooperate with the hospital.” There are also situations where the third party is a parent to the two orgs. (such as a parent organiza-tion.) They have to help the players by structuring incentive for both players to cooperate, and then define what counts as cooperation and defection. Sometimes, mediators can help the parties under-stand the consequences of what they’re doing.

Q: What about when a patient sues?
A: That’s a bad situation, because it’s a one-shot game. The patient says “I don ‘t like the results I got, and now I’m going to be selfish and take all I can.” There are no opportunities for reciprocation. So the best you can do is deal with it early. When the other side is provoked, you should say right away “yes, the out-come is bad... but there are better ways to deal with this without escalating.” One possibility is to cooper-ate and reciprocate with the lawyers: If the legal com-munity has long-term interaction with hospitals, they may have an interest in cooperation.

About the Presenter
Robert Axelrod, Ph.D. is the Arthur W. Bromage Distinguished University Professor of Political Science and Public Policy at the University of Michigan. His work in the area of cooperation has received awards from the American Association for the Advancement of Science, the MacArthur Foundation and the National Academy of Sciences.