Search

Scripts store the temporal order of component events of everyday activities as well as the temporal position of the events within the activity (early or late). When confronted with an activity, predictions are generated about how the component events will unfold. Thereby, an error-detection mechanism continuously monitors whether they unfold as anticipated or not in order to reveal errors in the unfolding activity. We investigated whether the temporal position "early" or "late" influenced the detection of errors using the pupillary response as an index of cognitive resource consumption. An event triplet consisting of three events was presented in a chronological or non-chronological temporal order. Crucially, the triplet focused either on the beginning (temporal position "early") or the end (temporal position "late") of an activity. We assumed that these position codes would be used to facilitate error detection when a non-chronological event was presented. Results showed that errors in the temporal order were detected more successfully in early than in late triplets. Results further suggest that strong predictions are formed about how an activity begins. Violations of this prediction must be overcome by zooming into the representation and allocating attention to the temporal position that consumes cognitive resources. Only after zooming in has taken place successfully may the position codes be used to anticipate temporal violations in unfolding event sequences.