Fig. 4
From: Human adaptation to adaptive machines converges to game-theoretic equilibria
Overview of co-adaptation experiment between human and machine. Human subject H is instructed to provide manual input \(h\) to make a black bar on a computer display as small as possible. The machine M has its own prescribed cost \(c_M\) chosen to yield game-theoretic equilibria that are distinct from each other and from each player’s global optima. (a) Joint action space illustrating game-theoretic equilibria and response functions determined from the costs prescribed to human and machine: global optima defined by minimizing with respect to both variables; best-response functions defined by fixing one variable and minimizing with respect to the other. Machine plays different strategies in three experiments: (b) gradient descent in Experiment 1; (c) conjectural variation in Experiment 2; (d) policy gradient descent in Experiment 3.
