A Note on Mixed Strategies in Economics Experiments
Some writers have proposed that mixed-strategy experiments don’t reveal much about subjects’ decision processes, while other studies suggest that subjects behave as predicted in MSE games, even when those predictions are counterintuitive.[1] However, subjects cannot engage in truly random behavioral sequences without training[2] because the cognitive demands of playing a game interfere with the player’s ability to randomize effectively.[3] Another interpretation is that successful players don’t need to randomize as such in MSE, but rather need to avoid their opponent’s predicting moves effectively.[4]
[1] Camerer 2003, pg. 148.
[2] Neuringer 1986.
[3] Camerer 2003 at 138.
[4] Id. at 150.