Human curriculum learning of a cue combination task

Humans often learn better when problems are broken down into parts, but this phenomenon has eluded explanation at the computational level. Here, we study how differing training curricula help or hinder learning in a classic probabilistic cue combination task. Training curricula that ‘divide and conquer’ by presenting one cue at a time facilitate later performance on test trials involving multiple cues. We show that this can be explained by a boundedly rational learning model in which credit is (erroneously) assigned to each input independent of every other. We use this model to generate new ‘skewed distribution’ multi-cue curricula that should, and should not, successfully promote human learning. The model makes accurate predictions, demonstrating that we can use cognitive models to accelerate human probabilistic learning.