One of the most replicable findings reported in the social psychological literature is the cross-race memory effect. We argue this effect derives from higher-order interactions among social cues that determine the perceived relevance of a face to an observer. The current research tested this hypothesis by examining the combined influences of eye gaze direction and race on face memory. The physical subtlety of eye gaze belies its powerful influence on social perception, and in this case helps specify the relevance of same-versus other-race faces. We found that only in faces making direct eye contact -not those displaying averted eye gaze-was the cross-race memory effect evident. Likewise, only in same-race faces did direct relative to averted gaze enhance face memory. These findings have implications for our general understanding of the combinatorial nature of social perception and help clarify the underlying cause of the cross-race memory effect.The proportionally large size of the human brain is sometimes explained as a consequence of social memory requirements (e.g., Dunbar, 1998;1995;1998). To keep track of relatively large social networks, people possess an extraordinary ability to remember faces. By the same token, social memory can not be entirely indiscriminate; for example, it is more important to remember the identity of a spouse than a cashier. Generally, people should better remember those whose actions are likely to be relevant to their own wellbeing (Maner et al., 2003;Mealey, Daood, & Krage, 1996;Oda, 1997;Rodin, 1987). Such a functional account has been invoked to explain several biases in face memory, including those favoring same-race faces (e.g., Shriver, Young, Hugenberg, Bernstein, & Lanter, 2007). Indeed, the cross-race effect (CRE) in memory is well chronicled (cf. Meissner & Brigham, 2001) and is often cited to explain alarming racial inequities in faulty eyewitness testimony (Innocence Project, 2007).If the CRE stems from the perceived relevance of ingroup faces, non-race facial cues that influence perceived relevance, such as eye contact, should impact the presence or absence of the CRE. Extending the functional approach in this way assumes that the combined processing of social cues is adaptive, facilitating the detection of self-relevant information signaled by the face (Adams, Franklin, Stevenson, & Nelson, in press Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. (Maner, Gailliot, & DeWall, 2007), possibility for affiliation (Mason, Tatkow, & Macrae, 2005), the presence of environmental danger and its source (Adams & Kleck, 2005), and ...