Many studies have demonstrated that newborns prefer upright faces over upside‐down faces. Based on this evidence, some have suggested that faces represent a special class of stimuli for newborns and there is a qualitative difference between the processes involved in perception of facelike and non‐facelike patterns (i.e. structural hypothesis). Others suggest that there is no reason to suppose that faces are different from other patterns, because faces, like any other class of visual stimuli, are subject to filtering by the properties of the visual system (i.e. sensory hypothesis). The core question that will be addressed in the present paper is whether, to manifest itself, face preference requires the unique structure of the face, represented by the relative spatial location of its internal features, or rather some more general properties that other stimuli may also possess. Evidence will be presented supporting the idea that newborns do not respond to facelike stimuli by ‘facedness’ but, rather, by some general structural characteristics that best satisfy the constraints of the immature visual system. Copyright © 2001 John Wiley & Sons, Ltd.