In Progress. We experimentally study how people acquire information in a sequential search problem and find systematic departures from optimal stopping that stem from a failure to think ahead about what further evidence could tell them.