All types of recognition errors are (at least partly) attributable to misleading memory evidence, even false alarms.

The error-speed effect, in which items are responded to more accurately if associated with a slow rather than fast erroneous response in a preceding task, is often interpreted as evidence that recognition errors are sometimes driven by systematically misleading memory evidence. However, recent observations challenge this interpretation demonstrating that the error-speed effect only occurs among previously studied items but not among non-studied items (Akan et al. Memory, 31, 1340-1351, 2023), which is not in line with most models of recognition memory. In the present study, we demonstrate that the error-speed effect can occur for non-studied items. In Experiment 1, we replicate the error-speed effect for studied but not non-studied items using picture stimuli. In the second experiment, we systematically manipulate the similarity between studied and non-studied items, thereby increasing misleading memory evidence and the rate of false alarms. As a result, the error-speed effect also emerges among non-studied items, indicating that false recognition decisions for these items are at least partially driven by misleading memory evidence.