Observing Unrevealed Preferences Ideally, it would be possible to design a system of incentives for the production and allocation of public goods with the following properties: (i) it would be budget-balanced; (ii) people would participate willingly because they would not be made worse off by doing so; (iii) it would be easy for each participant to find his or her optimal strategy; and (iv) the equilibrium solution would yield the optimal production of the public good. Sadly, this set of conditions cannot be satisfied simultaneously because it requires that self-interested individuals reveal voluntarily and truthfully how much they value the public good. Krajbich et al. (p. 596 , published online 10 September) ask whether a neuroimaging measurement can be used to circumvent this reliance on observed behavior by decoding individual valuations. A decoding accuracy of 55% would be sufficient, and, in a laboratory experiment, an optimal provision of public goods was indeed achieved.