Learning to discriminate interaural time differences at low and high frequencies