In this paper, we consider a committee of experts that decides whether to approve or reject a proposed innovation on behalf of society. In addition to a payoff linked to the adequateness of the committee’s decision, each expert receives a disesteem payoff if he/she voted in favor of an ill-fated innovation. An example is FDA committees, where committee members can be exposed to a disesteem payoff (negative) if they vote to pass a drug that proves to be fatal for some users. Under the standard voting model, we show that information is aggregated in large committees provided disesteem payoffs are not overly large. However, we go on to document an empirically-relevant discontinuity in the standard model: if an arbitrarily large number of signals does not perfectly reflect the state of the world then, no matter how small the disesteem payoffs are, information aggregation fails in large committees and the committee rejects the innovation almost surely, providing an explanation for over-caution in committees. Finally, we extend the model to include a disesteem payoff when an agent votes to reject a beneficial innovation, and provide closed form expressions capturing the limit behavior of the committee for generic combinations of the two kinds of disesteem payoffs.