The asymptotic distribution of response probabilities in the Audley-Jonckheere learning model