A two treatment experiment is carried out within subjects to determine whether altruistic behavior observed in the laboratory is the result of preferences over certain outcomes, or preferences for being recognized as having created these outcomes. Using a unique design which creates a dichotomy between the way subjects truly behave and the way they are perceived as behaving, we are able to add greater nuance to the subject “types” determined by previous experiments. “Self-interested” subjects will behave selfishly regardless of how their behavior is perceived by others, even when a small sacrifice on their part would result in a great gain for their partner. “Appearance-concerned” subjects consistently exploit informational asymmetries in both treatments to obtain self-biased earnings distributions while maintaining a socially favorable appearance. “True-egalitarians” divide earnings equally, even if doing so makes them appear selfish in the eyes of their partner.