Two perceptual variables in within-event learning