Studying related pictures can reduce accuracy, but increase confidence, in a modified recognition test.

In 14 experiments, a dissociation occurred between subjects' accuracy on a forced-choice recognition test and their confidence in their choice. Nature pictures (e.g., Lake A) were shown. Later, the subjects were asked to choose the picture that they remembered, given the target (Lake A) and a novel picture (Lake C) as alternatives, and rated their confidence in their choice. When the subjects also studied a related picture (Lake B), their accuracy often decreased while their confidence increased. The dissociation cannot be explained by signal detection theories of recognition, which assume that strength determines both accuracy in a forced-choice test and confidence. Instead, familiarity with general themes may give people the illusion that they are accurately remembering details.