2 Multiple-Cue Probability Learning

12 Multiple-Cue Probability Learning (MCPL) is an experi-13 mental paradigm concerned with how well people can 14 learn imperfect relationships between cues and outcomes. 15 In a typical MCPL task, participants are shown an array of 16 cues each of which predicts a particular outcome with 17 some probability; usually this probability is less than 18 unity, mirroring the imperfect nature of cues in the natu-19 ral environment. The cues are usually instantiated as 20 simple perceptual stimuli, which can be either discretely 21 (often binary) valued, such as color – a given cue might be 22 a red or green, for instance – or stimuli can be comprised 23 of continuously valued dimensions – such as bars of 24 different lengths. The former case with discrete cues is 25 typically referred to as nonmetricmultiple-cue probability 26 learning (NMCPL), and the latter case with continuou