In the description-identification relationship, memory for faces can be impacted negatively (overshadowing) or positively (facilitation). In a paradigm designed to elicit facilitation, theoretical predictions from the semantic processing, criterion shift, and transfer appropriate processing accounts were compared using conjunction faces and assessing contributions of recollection and familiarity. Participants viewed several faces sequentially and described each featurally, holistically, or not at all. Subsequently, participants made yes/no recognition decisions for studied, conjunction, and novel faces. Results demonstrate verbal facilitation as evidenced by higher hit rates for verbalizers, regardless of description condition. Despite equivalent false alarm rates across conditions, however, verbalizers also exhibited higher conjunction false alarm rates. This finding was complemented by the fact that the number of descriptors (featural or holistic) for each face positively predicted participants' likelihood to say "yes" to that face. These data support the semantic processing account such that verbalization provides a stronger memory trace that both elevates hit rates and conjunction false alarm rates.

Remember/Know/Guess results revealed that the featural condition was most likely to couple Hits with Remember judgments which contrasts previous research. These results are discussed in terms of the nature of the description-identification relationship and the contributions of recollection and familiarity.