The effects of R-S interval duration on serial choice reaction time: preparation time or response monitoring time?

Abstract

Most laboratory choice reaction experiments have been poor simulations of everyday tasks because they have only considered cases in which signals occur at a constant, regular rate. Experiment 1 therefore examined performance in a 2-. 4- and 8-choicc task in which the interval between each response and the next signal (RSI) varied from trial lo trial, taking values of 20. 50. 100, 150, or 200ms with equal probability in random order. In Experiment 2, RSIs were either constant at 20 ms or 200 ms throughout a run, or might adopt either value, equiprobably, at random. In Experiment 3, 20 ms and 200ms RSIs occurred in unpredictable order. In Condition I RSIs of 20 ms occurred in 75% of trials and RSIs of 200 ms in 25% of trials. In Condition 2 this bias was reversed. In Condition 3 the two possible RSIs occurred equally frequently.

In all experiments RT fell as RSIs increased from 20 to 200 ms, irrespective of the number of choices, the repetition or alternation of responses and the predictability or equiprobability of RSIs. In none of these experiments did RT following a particular RSI in any trial, N, vary as a function of the RSI on the preceding trial, N - 1. In particular, RT was not faster when the same RSI occurred in successive trials than when trials with different RSIs succeeded each other. In sum, there was no evidence that subjects could improve their performance on any trial by correctly anticipating the precise moment when the next signal would arrive.

The fall in choice RT as RSIs increased from 20 ms to 200 ms in these experiments closely resembles data from simple RT experiments in which simple RT falls as the preparatory interval between a warning signal and a signal to respond varies over the same range. It seems that both in simple and in choice RT tasks subjects need at least 200 ms to maximise their preparation for any imminent signal. Thus, when all RSIs or preparatory intervals are shorter than 200 ms, they can gain nothing from successful predictions of the moment of signal onset, since the best they can do on any trial is to mobilise their preparation at the maximum possible rate. The relationship between CRT. and the maximum rate at which preparation can be mobilised thus becomes a critical factor when we attempt to predict the efficiency with which human beings can deal with rapid, irregular sequences of events in daily life.