Abstract
60 Ss practiced 6 days on the stabilometer. No reminiscence or warm-up decrement was found; loss of skill occurred from 1-day layoffs which lessened as learning progressed. A 3-component exponential equation fitted the learning trend (including relearning). Both individual differences and intra-individual variations decreased exponentially with practice, but the ratio of individual differences to mean score increased. Practice had little influence on adjacent trial rs. Between scores in any particular pair of separated trials, r decreased when increasing number of trials separated them. Averaging 8 initial and 8 final trials gave optimum learning score reliability (r = .95). Individual rates of learning were not appreciably correlated with amounts of learning (r = .23). Final performance levels could only be predicted 50% from initial skill and cumulated amounts of learning unless more than half of the total practice (i.e., more than 90% of the potential learning) had been accomplished. The pattern of zero order rs between amount of learning and initial and final skill levels differed markedly from that previously observed in learning the ladder climb.