2274Preparing for a future event involves expectations about both the what and the when of that event. In the present study, we considered two very different perspectives on the role of temporal contexts in eliciting such preparatory expectations. A classic view derives from the foreperiod (FP) literature, in which responses to single time intervals have been studied. Different FP paradigms permit examination of the relationship between probabilistic uncertainty and reaction time (RT). A common assumption is that low uncertainty about the what and when of a forthcoming time interval, conferred probabilistically by prior trial context, facilitates preparatory activity. A different perspective derives from research on sequence perception. It is concentrated on relational (rather than probabilistic) properties of time intervals within sequences that facilitate preparation of forthcoming time intervals.In the present study, we outline contrasting hypotheses about the role of temporal context that emerge from these two perspectives. To evaluate these hypotheses, we modified a standard FP paradigm by incorporating each of three FPs (250, 500, and 1,000 msec) as the final time interval of rhythmical auditory patterns that varied in degree of temporal coherence.
Uncertainties in FP ContextsClassically, temporal preparation for a forthcoming target event has been studied by varying the duration of a single time interval (i.e., an FP) that occurs between a warning signal and a target on each of a series of trials. 1 Faster RTs to the target presumably indicate better temporal preparation to a particular FP. Conventionally, an RT to a given FP depends on two features: FP duration and the intertrial context in which the FP occurs (e.g., Niemi & Näätänen, 1981).Intertrial (or global) context differentiates the two major paradigms used to study temporal preparation: In a variable-FP design, FPs vary from trial to trial within a block; in a constant-FP design, a single FP is used within a trial block. In the variable-FP paradigm, the probability distribution of FPs over trials is assumed to determine the uncertainty of an FP on an individual trial. In the simplest case, a uniform probability distribution of different FPs typically yields RTs that get shorter as an FP lengthens: a decreasing RT-FP function. The most parsimonious explanation for this function is that uncertainty decreases as time elapses from a warning signal, and this decreased uncertainty results in shorter RTs to longer FPs (e.g., Klemmer, 1956;Niemi & Näätänen, 1981;Woodrow, 1914). A flat RT-FP function is also consistent with a uniform FP distribution, but suggests that the subjects acquired little knowledge of FPs from the global context-that is, that uncertainty was uniformly high for all FPs.The constant-FP paradigm yields an RT-FP function that is characteristically different from that of variable-FP profiles, with RTs lengthening as FP lengthens. Such ascending RT-FP profiles are attributed to subjects' reduced abilities to estimate longer (albeit pred...