Abstract
I design an experiment to analyze the effect of unknown preferences in a strategic communication game. The experimental environment is according to Li and Mararasz (2008) where the conflict of interests between the expert and the decision maker is constant in its magnitude but uncertain in its direction. Using a best response dynamic approach, I show that as players' level of sophistication increases, the outcome converges to the strategy profile prescribed by the most informative equilibrium when the expert's bias is unknown to the decision maker. Social welfare, as a result, is higher when the bias remains uncertain than when it is disclosed to the decision maker. Experimental evidence will be provided.