Best-Response Dynamics in Three-Player Vee Games

The Logic of Best Response

Once a game is defined by its payoff matrix, the central question for each player is: "Given what I believe the other players will do, what is my best possible move?" The answer to this question is called a best response.

A best response is fundamentally a player-specific optimization. The goal is to maximize one's own utility function, holding the strategies of others as fixed parameters. A Nash Equilibrium, which we are ultimately looking for, is a state where every player is simultaneously playing a best response to everyone else.

A simple, systematic way to find all best responses in a matrix game is often called the "underline" or "asterisk" method. This technique turns the search for equilibria into a straightforward visual exercise.

This method is powerful because it guarantees you find all pure-strategy Nash equilibria without having to check every single outcome individually.

Applying Best-Response Analysis to a Vee Game

Now, let's apply this systematic approach to the more complex three-player vee game. As we saw last time, we represent this with layered matrices, where the third player's choice determines which matrix the other two are playing in.

The logic of finding best responses remains the same, but the procedure requires an additional step.

• To find the best responses for Arm 1 (row player) and Arm 2 (column player), we perform the asterisk method on each matrix layer independently.
• To find the best response for the Hinge (the "matrix-choosing" player), we must compare their payoffs across the matrices for each given combination of the other two players' strategies.

This can be tricky to visualize. The following explanation provides a clear, text-based guide to this exact process.

Worked Example: The Disclosure Dilemma Revisited

Let's return to the "Disclosure Dilemma" from our previous lesson and solve it.

• Players: Hinge (H), Arm 1 (A1), Arm 2 (A2).
• Hinge's Strategies: Disclose Immediately (Top Matrix), Wait (Bottom Matrix).
• Arms' Strategies: React Supportively, React with Insecurity.

Here are the complete payoff matrices, with payoffs as (H, A1, A2).

Matrix 1: Hinge chooses Disclose Immediately

Matrix 2: Hinge chooses Wait

Now, let's find the best responses systematically:

1. A1's Best Responses (Rows):

   • Matrix 1: If A2 is Supportive, A1 prefers 8 to 2 (Supportive). If A2 is Insecure, A1 prefers 9 to 3 (Supportive).
   • Matrix 2: If A2 is Supportive, A1 prefers 5 to 0 (Supportive). If A2 is Insecure, A1 prefers 6 to 1 (Supportive).
   • Result: Supportive is a dominant strategy for A1 in this game. We can underline A1's payoffs in the top row of both matrices: (10, 8, 8), (4, 9, 2), (7, 5, 5), (2, 6, 0).

2. A2's Best Responses (Columns):

   • Matrix 1: If A1 is Supportive, A2 prefers 8 to 2 (Supportive). If A1 is Insecure, A2 prefers 9 to 3 (Supportive).
   • Matrix 2: If A1 is Supportive, A2 prefers 5 to 0 (Supportive). If A1 is Insecure, A2 prefers 6 to 1 (Supportive).
   • Result: Supportive is also a dominant strategy for A2. We can underline A2's payoffs in the left column of both matrices: (10, 8, 8), (4, 2, 9), (7, 5, 5), (2, 0, 6).

3. Hinge's Best Responses (Matrices):

   • If A1 and A2 are both Supportive (top-left cell), H prefers 10 (Disclose) to 7 (Wait). We underline the 10.
   • If A1 is Supportive and A2 is Insecure (top-right), H prefers 4 (Disclose) to 2 (Wait). We underline the 4.
   • If A1 is Insecure and A2 is Supportive (bottom-left), H prefers 4 (Disclose) to 2 (Wait). We underline the 4.
   • If A1 and A2 are both Insecure (bottom-right), H prefers 1 (Disclose) to 0 (Wait). We underline the 1.
   • Result: Disclose Immediately is a dominant strategy for the Hinge. We underline all of H's payoffs in the top matrix: (10, 8, 8), (4, 9, 2), (4, 2, 9), (1, 3, 3).

4. Identify Nash Equilibrium:
   We look for the cell where all three payoffs are underlined. This occurs only in one place:

   (10, 8, 8)

This corresponds to the strategy profile: (H: Disclose Immediately, A1: React Supportively, A2: React Supportively). This is the single pure-strategy Nash Equilibrium of this game. It represents a stable, self-enforcing outcome. Given that the hinge discloses immediately and the other arm reacts supportively, it is in each partner's best interest to also react supportively.

Dynamic Paths to Equilibrium

The analysis above identifies a static equilibrium. But how might players arrive there? One intuitive model is iterated best response, where players take turns adjusting their strategy. This mirrors a real-life process of negotiation or behavioral adjustment.

While we won't delve into the mathematics of potential games, the concept of iterated best response provides a powerful narrative for how a vee might find its footing. Starting from a state of misunderstanding, one partner might adjust their behavior, prompting another to adjust in turn, and so on, until a stable equilibrium (like the one we found) is reached.

An Alternative: Sequential Decisions and Backward Induction

Our analysis so far has assumed a simultaneous-move game. But what if decisions are made in sequence? For example, the Hinge decides to disclose, A1 observes this and then reacts, and A2 observes both actions before reacting. This is a sequential game, best represented by a game tree.

The tool for finding best responses in this context is backward induction. We start at the end of the game and work our way back to the beginning.

Backward induction finds a Subgame Perfect Nash Equilibrium (SPNE), which is a stronger equilibrium concept. It ensures that players' strategies are best responses at every possible stage of the game, not just at the beginning. This is a crucial concept for analyzing commitments and threats, as it rules out any threats that a player would not rationally carry out if called upon to do so.

Conclusion

Today, we developed the core analytical skill of a game theorist: solving a game. You learned how to find stable outcomes by systematically identifying each player's optimal choices.

Key Takeaways:

• A best response is a player's optimal strategy, given the strategies of the other players.
• A Nash Equilibrium is a strategy profile of mutual best responses, representing a stable outcome with no incentive for unilateral deviation.
• We can find pure-strategy Nash Equilibria efficiently by marking best responses for each player in the payoff matrix. This method extends to the 3-player vee structure.
• Iterated best response offers a dynamic model of how players might adjust their behaviors over time to reach an equilibrium.
• For sequential games, the equivalent tool is backward induction, which finds a more robust equilibrium by analyzing best responses from the end of the game to the beginning.

In our next lesson, we will move from the "how" to the "what." Having found the Nash equilibria, we will focus on how to interpret these pure-strategy Nash equilibria as potential stable outcomes in a vee game, exploring what these mathematical solutions tell us about the health, stability, and potential conflicts within the relationship.