The Impact of Social Value Orientation on Nash Equilibria of Two Player Quadratic Games

Dan Calderone,Meeko Oishi

2024-11-14

Abstract:We consider two player quadratic games in a cooperative framework known as social value orientation, motivated by the need to account for complex interactions between humans and autonomous agents in dynamical systems. Social value orientation is a framework from psychology, that posits that each player incorporates the other player's cost into their own objective function, based on an individually pre-determined degree of cooperation. The degree of cooperation determines the weighting that a player puts on their own cost relative to the other player's cost. We characterize the Nash equilibria of two player quadratic games under social value orientation by creating expansions that elucidate the relative difference between this new equilibria (which we term the SVO-Nash equilibria) and more typical equilibria, such as the competitive Nash equilibria, individually optimal solutions, and the fully cooperative solution. Specifically, each expansion parametrizes the space of cooperative Nash equilibria as a family of one-dimensional curves where each curve is computed by solving an eigenvalue problem. We show that both bounded and unbounded equilibria may exist. For equilibria that are bounded, we can identify bounds as the intersection of various ellipses; for equilibria that are unbounded, we characterize conditions under which unboundedness will occur, and also compute the asymptotes that the unbounded solutions follow. We demonstrate these results in trajectory coordination scenario modeled as a linear time varying quadratic game.

Optimization and Control

What problem does this paper attempt to address?

### What problem does this paper attempt to solve?

This paper aims to explore the influence of **Social Value Orientation (SVO)** on the Nash equilibrium in a two - person quadratic game. Specifically, the author focuses on how to quantify and analyze the changes in the Nash equilibrium under different degrees of cooperation within a cooperation framework. The following are the main problems and goals of the paper:

1. **Introducing the concept of social value orientation**:
- Social value orientation is a psychological framework used to measure the degree to which an individual values the interests of others when making decisions. Each player incorporates the other's cost into their own objective function according to the preset degree of cooperation.
- Expressed by the formula:
\[
J_i(u; \theta_i) = \cos(\theta_i) J_i(u) + \sin(\theta_i) J_{-i}(u)
\]
where \(J_i\) is the cost function of player \(i\), and \(\theta_i\) is the social value orientation angle of player \(i\).

2. **Studying the influence of SVO on the Nash equilibrium**:
- By introducing SVO, the author redefines the Nash equilibrium in a two - person quadratic game and calls it the **SVO - Nash equilibrium**.
- Analyzes the changes in the SVO - Nash equilibrium under different degrees of cooperation (i.e., different values of \(\theta_1\) and \(\theta_2\)).
- In particular, studies the characteristics of the SVO - Nash equilibrium in bounded and unbounded cases and gives the corresponding conditions and formulas.

3. **Providing a theoretical framework and practical applications**:
- Proposes a theoretical framework for describing competitive behaviors at different cooperation levels.
- Demonstrates the practical applications of the SVO - Nash equilibrium through specific examples (such as the collision avoidance scenario in the linear quadratic trajectory planning problem).
- Discusses the significance of these results for the future design of cooperative autonomous systems (such as self - driving vehicles).

### Main contributions

1. **Theoretical framework**: Proposes a competitive framework for a two - person quadratic game that includes the degree of cooperation.
2. **Equilibrium characteristics**: Describes in detail the characteristics of the SVO - Nash equilibrium, including bounded and unbounded cases.
3. **Comparison with related games**: Analyzes the relationship between the SVO - Nash equilibrium and other related games (such as the competitive Nash equilibrium, the individual optimal solution, and the fully cooperative solution).

### Structural overview

- **Section 2**: Introduces the background of social value orientation in two - person games.
- **Section 3**: Reviews the basic two - person quadratic game and its method for calculating the Nash equilibrium.
- **Section 4**: Presents the main results, including the parametric variation of the SVO - Nash equilibrium.
- **Section 5**: Analyzes the bounded and unbounded cases of the SVO - Nash equilibrium.
- **Section 6**: Demonstrates the behavior of the SVO - Nash equilibrium through a simple linear quadratic trajectory planning problem.
- **Section 7**: Discusses future research directions and conclusions.
- **Appendix**: Contains detailed proofs and extensions.

Through these efforts, the paper provides an important theoretical basis and practical tool for understanding and designing more complex cooperative systems.