“…The answer probably lies in the slowness of improvement with extended practice. After 288 trials, overshooting is 0.03 (Friedman et al, 1964); after 400 trials it averages 0.05 (Myers et al, 1963); after 450 trials it is 0.08 (Gardner, 1957), though this was obtained with three choice alternatives, a manipulation that was found to increase overshooting systematically (Gardner, 1958). After 1000 trials, overshooting averages 12.6% (Edwards, 1961).…”