Predictive Inference Generation as a Function of Working Memory Capacity and Causal Text Constraints

Abstract

Two factors that affect the predictive inferential process are examined: the working memory (WM) capacity of the reader and causal text constraints. A naming task was used in Experiment 1 to investigate predictive inferences that were based on texts containing low, moderate, and high causal sufficiency events at 2 time delays (250 and 500 ms). The results indicated that high WM capacity readers make predictive inferences when texts contain high causal sufficiency events, at both time delays, as evidenced by naming time differences between high-low and high-moderate causal sufficiency conditions. There is no evidence that low WM capacity readers make predictive inferences at either time delay. In Experiment 2, reading times for target sentences that either confirmed or contradicted predictive inferences in low, moderate, and high causal sufficiency conditions were collected. The results showed that high WM capacity readers make predictive inferences when causal sufficiency is high and target sentences confirmatory and that both low and high WM capacity readers make predictive inferences based on the comparison between high causal sufficiency texts in the confirmatory versus contradictory conditions. Thus, both experiments demonstrate that generating predictive inferences is a demanding process that requires a certain amount of WM resources and causal text constraints.