Source attributions for detected new items: Persistent evidence for schematic guessing

Performance in source-monitoring tests is not only determined by source memory but also by source guessing. Source guessing is not random as it is informed by two distinct mechanisms. (1) People may show a schema-based guessing bias and rely on cross-situationally stable world knowledge. (2) They may apply probability matching and rely on the specific item-source contingency experienced at encoding. According to probability matching theory, source guessing is based on probability matching when a specific contingency representation is available. This conclusion is derived from a source-monitoring paradigm in which no source judgements for detected new items are required. Here, we extend this paradigm to examine source guessing not only for detected old items but also for detected new items. The results suggest that participants take the old–new recognition status of the items into account when making source attributions. Probability matching is used only for detected old items: Source guessing sensitively reflects the item-source contingency for these items. For detected new items, participants resort to schema-based guessing. Using schema-based guessing rather than probability matching when judging detected new items may have the advantage that a newly acquired contingency representation that may only be locally valid is not generalised too readily at the expense of a schematic expectation that reflects a larger and more comprehensive learning history.