Welcome to our next lesson. In our previous session, we delved into the Prisoner's Dilemma, a game defined by a single, powerful Nash Equilibrium that, paradoxically, leads to a suboptimal outcome for all players. This illustrated a fundamental conflict between individual rationality and collective well-being.
Today, we shift our focus to a different class of strategic interactions: coordination games. Unlike the Prisoner's Dilemma, these games are characterized by the presence of multiple Nash Equilibria. Here, the primary challenge isn't to resist the temptation of a dominant strategy, but rather for players to successfully coordinate their actions to land on a mutually desirable, stable outcome. We will analyze two canonical examples, the "Battle of the Sexes" and the "Stag Hunt," to understand the strategic complexities that arise when more than one equilibrium is possible.
The Problem of Multiple Equilibria
In the Prisoner's Dilemma, the prediction is clear: rational players will defect. But what happens when there are several "regret-free" outcomes? This is the central question in coordination games. Players have a shared interest in coordinating, but they might disagree on how to coordinate, or face a choice between a safe, low-payoff outcome and a risky, high-payoff one.
This introduces the "equilibrium selection problem": if rational play can lead to more than one stable result, on what basis can players form expectations about each other's behavior to converge on a single one?
Let's explore this through two classic games. Below are the payoff matrices for the Prisoner's Dilemma, which we've studied, alongside the two coordination games we'll examine today.
1. Battle of the Sexes: Coordination with Conflicting Interests
The "Battle of the Sexes" game models a situation where two players wish to coordinate on an activity, but they have different preferences over which activity to choose.
The classic story involves a couple deciding on an evening's entertainment. Let's call them Player 1 and Player 2. They can't communicate, but must independently choose between going to the Ballet or a Fight.
- Both players prefer to go to the same event rather than being alone.
- Player 1 prefers the Fight over the Ballet.
- Player 2 prefers the Ballet over the Fight.
The payoff matrix for this game (similar to matrix B in the image above) illustrates this structure. The key feature to notice is that there are no strictly dominant strategies. Player 1's best move depends entirely on what they expect Player 2 to do, and vice-versa.
To see how we find the equilibria in this game, the following video provides a clear, step-by-step analysis.
Game Theory 101 (#10): Battle of the Sexes
This video introduces the Battle of the Sexes scenario and systematically identifies its pure strategy Nash Equilibria.
First, watch the setup of the game, which explains the players' preferences and the resulting payoff matrix. Then, follow the analysis that checks each of the four possible outcomes to determine which ones are stable, "regret-free" Nash Equilibria.
As the video demonstrates, the Battle of the Sexes has two pure strategy Nash Equilibria:
- (Fight, Fight)
- (Ballet, Ballet)
In either of these outcomes, neither player has a reason to unilaterally change their strategy. If they are both at the Fight, Player 1 is getting their best outcome, and Player 2, while not at their preferred event, is still better off than going to the Ballet alone. The situation is symmetric for the (Ballet, Ballet) equilibrium.
The outcomes (Fight, Ballet) and (Ballet, Fight) are not equilibria, because in both cases, each player would regret their choice and would prefer to switch to match the other's location. The dilemma here is purely one of coordination. How can the players ensure they end up at the same event, especially when they have conflicting preferences over the two equilibrium outcomes?
2. The Stag Hunt: Cooperation vs. Safety
The Stag Hunt, inspired by a passage from the philosopher Jean-Jacques Rousseau, presents a different kind of coordination problem. It models the tension between cooperation for a great reward and playing it safe for a smaller, but guaranteed, one.
The story involves two hunters who can choose to either:
- Hunt a Stag: This requires both hunters to cooperate. If they succeed, they share a large prize. If one hunter tries to hunt the stag alone, they fail and get nothing.
- Hunt a Hare: This can be done individually. A hunter who chooses to hunt a hare is guaranteed a small prize, regardless of what the other hunter does.
Let's see how this translates into a game and where the equilibria lie.
Game Theory 101 (#4): Pure Strategy Nash Equilibrium and the Stag Hunt
This video explains the Stag Hunt game and uses the formal definition of Nash Equilibrium to find its stable outcomes.
First, watch the description of the hunters' dilemma and its payoff matrix. Next, focus on the formal definition of a Nash Equilibrium as a stable state where no player has an incentive to deviate individually. Finally, observe how each of the four outcomes is systematically checked to identify the two equilibria.
As the analysis shows, the Stag Hunt also has two pure strategy Nash Equilibria:
- (Stag, Stag): Both hunters cooperate and achieve the best possible collective outcome.
- (Hare, Hare): Both hunters play it safe and achieve a smaller, but guaranteed, payoff.
This game elegantly captures a fundamental strategic tension. The (Stag, Stag) equilibrium is clearly better for both players than (Hare, Hare). In the language of our previous lesson, it is Pareto-superior. However, it is also risky. If you choose to hunt a stag, you are depending on your partner to do the same. If they don't, you get nothing. The (Hare, Hare) equilibrium, while suboptimal, is safe. By choosing to hunt a hare, you guarantee yourself a payoff, insulating you from the risk of your partner's actions.
The choice between these two equilibria hinges on trust and risk tolerance.
3. The Equilibrium Selection Problem
The existence of multiple equilibria in both the Battle of the Sexes and the Stag Hunt leaves us with an ambiguous prediction. To resolve this, game theorists have proposed several "equilibrium selection" criteria.
Focal Points
Economist Thomas Schelling suggested that in many real-world situations, certain equilibria act as focal points. These are outcomes that players seem to naturally gravitate towards because they feel special, logical, or obvious, often for reasons outside the formal mathematics of the game.
This excerpt from a textbook by Easley and Kleinberg discusses the problem of multiple equilibria and introduces Schelling's idea of focal points.
Please read the section titled Multiple Equilibria: Coordination Games. Start from the subsection A Coordination Game, which introduces the core problem, the concept of a focal point using the example of drivers on a road, and the structures of the Unbalanced Coordination Game and Battle of the Sexes.
A focal point could be established by:
- Social Convention: In many countries, "drive on the right" is a focal point that solves the coordination game of two cars meeting on a road.
- Salience: In the Stag Hunt, the (Stag, Stag) equilibrium might be focal because it is Pareto-superior. Players might reason that it's the "sensible" target to aim for.
- Precedent: If the two hunters have successfully hunted a stag together in the past, (Stag, Stag) is more likely to be the focal point for their next interaction.
Payoff Dominance and Risk Dominance
For a more formal approach, especially suited to your mathematical background, we can use the concepts of payoff dominance and risk dominance to compare equilibria.
Nash equilibria in games with continuous action spaces
This appendix provides formal definitions for selecting among multiple Nash Equilibria, contrasting Pareto (or payoff) dominance with risk dominance.
In the PDF, navigate to the Appendix section (page 19). Please read the subsections "Appendix – Pareto Dominance" and "Appendix – Risk Dominance." Focus on the formal definitions of a Pareto Dominated NE and a Risk Dominated NE.
Let's apply these to the Stag Hunt:
- (Stag, Stag) is the payoff-dominant equilibrium. It offers a strictly better payoff to every player than the (Hare, Hare) equilibrium.
- (Hare, Hare) is the risk-dominant equilibrium. The logic is that the penalty for attempting to coordinate on the payoff-dominant equilibrium and failing is high. If I choose Stag and you choose Hare, I get 0. The "safe" choice of Hare minimizes the potential loss from a coordination failure. It is less risky.
The tension between payoff dominance and risk dominance is central to the Stag Hunt and appears in many real-world scenarios, from technology adoption (choosing a superior but less-used standard vs. a safe, established one) to strategic decisions in multi-agent AI systems, where agents must "trust" each other to pursue a high-reward collaborative strategy.
Conclusion
This lesson has moved us from games with a single, inevitable outcome to the more complex world of coordination games with multiple stable equilibria.
Key Takeaways:
- Coordination games have multiple Nash Equilibria, creating an "equilibrium selection problem" for players.
- The Battle of the Sexes exemplifies coordination where players agree on the need to coordinate but have conflicting preferences over the equilibria.
- The Stag Hunt illustrates the tension between a high-payoff, high-risk equilibrium (payoff-dominant) and a low-payoff, safe equilibrium (risk-dominant).
- Concepts like focal points, based on social conventions or salience, and the formal criteria of payoff and risk dominance provide frameworks for predicting which equilibrium might be chosen.
In our analysis, we've informally compared the "goodness" of different outcomes, such as noting that (Stag, Stag) is better for both players than (Hare, Hare). In our next lesson, we will formalize this line of reasoning by defining and applying the concept of Pareto efficiency. This will give us a rigorous tool to evaluate and compare outcomes across all types of strategic games.