Abstract
Three experiments, based on Mani and Johnson-Laird's (1982) study, tested memory for spatial descriptions. In Phase 1 of each experiment, subjects judged whether diagrams matched verbal descriptions that were either determinate or indeterminate. In Phase 2, subjects attempted to recognize the descriptions studied in Phase 1. Mani and Johnson-Laird reported a crossover: Gist memory was better for determinate descriptions and verbatim memory was better for indeterminate descriptions. This crossover suggests that mental models are remembered. All three new experiments failed to replicate the crossover, challenging whether models are preserved in episodic memory. It was hypothesized instead that episodic memory records the mental operations by which models are constructed. This hypothesis accounted for the findings of all three experiments. In Experiments 1 and 2, performance varied with the construction-trace overlap between recognition test descriptions. In Experiment 3, performance was depressed if the order of sentences within descriptions was altered between study and test.

This publication has 18 references indexed in Scilit: