This research investigated the hypothesis that both the serial order of events and their temporal organization influence the internal representation of a sequence and thus influence its learnability. Dual-task serial reaction-time conditions were simulated in a recurrent network (P. F. Dominey, 1995; P. F. Dominey, M. A. Arbib, & J. P. Joseph, 1995) by introducing random pauses in the response-stimulus interval (RSI). Behavior in transfer between single-task (no-pause) and dual-task (pause) conditions corresponds to that observed by T. Curran and S. W. Keele (1993) in humans and is explained by a single learning mechanism, consistent with the hypothesis that attentional shifts can interfere with the learning of a sequence by disrupting its temporal organization (M. A. Stadler, 1995). The network can also learn specific RSI structures and displays performance deterioration for RSI increases as observed in humans by P. A. Frensch and C. S. Miner (1994). These results support the proposed hypothesis
| 
