Probability Learning: Contingent-Event Schedules with Lags

A contingent-event schedule with lag L is defined so that $P(E_{1,n+L}|E_{i,n})=\pi _{i1}\neq \overline{\pi}$, the marginal E1 event probability. The 16 subjects were run at L = 1, 3, 5, and 7 over four days, with $\overline{\pi}$ at either .50 or .75 throughout and with both perseverence and alternation sequences given each day. The effect of (i.e., memory for) the contingent event decreased with lag: at L = 1, P(A1,n+L|Ei,n) was slightly over π i1, and at L = 7 it was very near $\overline{\pi}$. The subjects also responded to the frequency and pattern of noncontingent events in the sequences.