Absentmindedness is a special case of imperfect recall which according to Piccione and Rubinstein (1997a) leads to time inconsistencies. Aumann, Hart and Perry (1997a) question their argument and show how dynamic inconsistencies can be resolved. The present paper explores this issue from a descriptive point of view by examining the behavior of absentminded individuals in a laboratory environment. Absentmindedness is manipulated in two ways. In one treatment, it is induced by cognitively overloading participants. In the other, it is imposed by randomly matching decisions with decision nodes in the information set. The results provide evidence for time inconsistencies in all treatments. We introduce a behavioral principal, which best explains the data.