Researchers describe an experiment in which 38 horses were invited to choose which of two lidded buckets contained a carrot. The horses had the benefit of two informants who pointed at the buckets — but only one knew which bucket contained the carrot. The other did not.