Studies of human memory show that performance can be uncorrelated, or dissociated, across various types of memory tests. Such dissociations are observed both in patients with brain lesions and in normal volunteers. In some cases, dissociations among memory tests are thought to reflect differences in the types of processing engaged during learning (encoding) and during memory testing (retrieval). Some memory tests emphasize conceptual processing, or the analysis of semantic meaning of to-be-remembered information, whereas other tests are more perceptually driven, requiring analysis of physical features of stimuli instead of semantic content. For example, a conceptual test of semantic cued recall might provide a word such as "terrorist" as a retrieval cue for the studied word "assassin." A perceptual test of word fragment completion, on the other hand, might provide the test cue "A _ _ A _ _ I N" for retrieval of the same word. Memory performance is best when the type of processing (conceptual or perceptual) performed during learning matches the type of processing performed on the memory test (conceptual or perceptual). When there is a mismatch between the types of processing engaged during learning and at test, performance will suffer. Thus, memory for the same studied material might appear to be good when tested with one type of test but would appear to be poor when tested by another memory task with different processing requirements.