ii. Some behavioral models obtain asymmetric updating or overconfidence via "self- signaling". That is, they assume an individual selectively generates a signal that she is the desired type, then interprets this signal as impartially generated (Quattrone & Tversky 1984, Bodner & Prelec 2003, Bnabou & Tirole 2004). If the assumption that the sender cannot verify that she did not receive evidence is driving asymmetric updating and overconfidence, what does this imply for self-signaling models? Is this reasonable