No abstract

Scripts are mental representations of activities in memory and are thought to be organized dimensionally in a temporal dimension. We investigated the cognitive strategies during the processing of temporal order of an event sequence to gain insight into the organization of scripts. Subjects were presented with triplets of script events (A - B - C). Fifty percent of the items included sequence violations at different positions within the triplet (late: A - C - B, or early: C - A - B). Reaction times indicate that subjects use an economical strategy by comparing the relative temporal positions of event pairs (e.g., A vs. B and if necessary B vs. C) and only attend to information that is necessary. Pupil data and error rates indicate that the temporal information of the complete sequence affects the decision process even if the first event pair indicates that temporal order has been violated. Results are seen as evidence of a dimensional structure of scripts.

Event sequences or scripts are the conceptual representations of activities in memory. Traditional views hold that events are represented in amodal networks and are retrieved by associative strategies. The embodied cognition approach holds that knowledge is grounded in perception and retrieved by mental simulation. We used a script generation task where event sequences of activities had to be produced. Activities varied in their degree of familiarity. In a regressional design we investigated whether amodal or perceptual variables best predicted knowledge retrieval to gain insight into the underlying representation. Retrieval depended on the familiarity of the activity. While novel activities mainly relied on perception-based simulation and to a lesser extent on associative strategies, moderately familiar activities showed the opposite pattern, and events of familiar activities were retrieved by association alone. We conclude that amodal structures exist in all representations that become stronger with increasing frequency and finally prevail over perceptual structures.

Event knowledge includes persons and objects and their roles in the event. This study investigated whether the progression of patients from a source to a resulting feature, such as the progression of hair that is cut from long to short, forms part of event representations. Subjects were presented with an event prime followed by two adjectives and asked to judge whether the adjectives were interrelated. Results showed that the semantic interrelation of two adjectives is recognized faster and more accurately when the adjectives denote source and resulting features of the patient of the primed event ("cutting": long-short). Furthermore, we found that presenting an event-related adjective in combination with an unrelated adjective makes it more difficult to recognize that the two adjectives are not interrelated, but only when the event-related adjective denotes a source feature. We argue that an inference mechanism automatically completes the representation of the event. We conclude that source and resulting features are represented in a goal-directed and chronological way.

The present study explores the processing of temporal information in event knowledge by focusing on the transition from an earlier, source state to a later, goal state. Participants were presented with an event verb followed by antonymous adjectives or adverbs denoting an earlier state and a later state. The states were presented either chronologically (to cook: cold -hot) or inversely (to cook: hot -cold) with regard to the denoted event. Participants were asked to identify either the earlier or the later state. We found that later states are identified faster and more accurately than earlier states. Later states presented chronologically were identified even more quickly than later states presented inversely. We attribute our results to the fact that directedness towards the goal state is a general principle of cognition which plays a fundamental role in language and in simulation, whereby language processing provides faster and more direct access to goals even than simulation.