We report an experiment to study the effect of defaults on charitable giving. In three different treatments, participants face varying default levels of donation. In three other treatments that are paired with the first three, they receive the same defaults, but are informed that defaults are thought to have an effect on their donation decisions. The emotional state of all individuals is monitored throughout the sessions using Facereading software, and some participants are required to report their emotional state after the donation decision. We find that the level at which a default is set has no effect on donations, and informing individuals of the possible impact of defaults also has no effect. Individuals who are happier and in a more positive overall emotional state donate more. Donors experience a negative change in the valence of their emotional state subsequent to donating, when valence is measured with Facereading software. This contrasts with the self‐report data, in which donating correlates with a more positive reported subsequent emotional state. (JEL C91)