The First Human Images

These mostly palm-sized statuettes appear to depict nude obese women with faceless and usually downturned heads, thin arms which commonly end or disappear under the breasts (but occasionally cross over them), an abnormally thin upper torso carrying voluminous and pendulous breasts, exaggeratedly large or elevated buttocks often splayed laterally but sometimes distended rearward, a prominent, presumably pregnant or adipose abdomen with a large elliptical navel, and what often appear to be oddly bent, unnaturally short tapering legs which terminate in either a rounded point or disproportionately small feet. Although readily recognizable, these anatomical details do not add up to an accurate image of the human figure. 

I contend that it is the fixed angle of self-regard which accounts for both the odd "realism of parts considered independently one from another" observed by Saccasyn-Della Santa (1947:96) and Leroi-Gourhan's conclusion that the figures appear "centered on the torso, breasts, thighs and abdomen," with the rest "attentuated" or "dwindling away" above and below (1968a:207). The latter researcher christened the collective result of these distinctive distortions, anatomical omissions, and general disproportion of parts the "lozenge composition" (1968:90; 1968b). The structural nature of these distortions has often been overlooked by scholars who see gender or variations of feminine morphology and reproductive histories in the style of these works (Dolores 1992b; 252); Duhard 1991, 1993a, b; Nelson 1993; Pales and de St.-Pereuse 1976; Rice 1981; Soffer 1987). In fact, the consistent departures from nature seen in these early images involve basic structural alterations in the normal vertical and horizontal proportions of the human body (Pales and de St.-Pereuse 1976:68-73). 

In human beings, half the body's length typically lies below the level of the hip joint or crotch and half above. For the average woman, this vertical midpoint of the body also coincides with its greatest horizontal or lateral width. In the typical "lozenge composition," however, while the vertical midpoint and greatest horizontal width continue to occur together, their intersection is unnaturally elevated to the level of the navel. This effect results from a general atrophy of the lower body wherein the distance from the cratch to the ground is typically represented as about one-third of the total body length instead of half (Pales and de St.-Pereuse I976:7I).⁶ 

Women today, regardless of race, weight, or reproductive history, do not have such disproportionate structural relationships between body parts. While Delporte recognizes the critical importance of understanding this generalized atrophy of the upper and lower body (1993a:244, 275), he perpetuates an unfortunate assumption by seeking the explanation in "a psychological imperative which corresponds to a conception of women in the life and behavior of prehistoric men" (1993c:10). Why speculate about psychological mechanisms before experimentally examining the material evidence of human vision? We should not simply ascribe the "violation of certain body proportions" to the deliberate "accentuation" or "willful distortion" of female body parts (Gvozdover 1989b; Delporte 1993a:259) before asking if a physical mechanism could be responsible for the "violations" observed. I contend that their origin lies in what all humans and especially expectant mothers can and cannot see when they look down at their own bodies. 

The distortions in these first images are produced by three structural regularities inherent in the body as directly self-inspected but not necessarily observed from the point of view of other human beings. First, because it begins with the same fixed point of view, everyone's experience of self-generated visual information has the same structure, including a distinctive canon of proportions, despite variations expressive of individual physiognomy, age, and gender. Second, because of the oblique angle of self-regard, self-generated information is always strongly foreshortened, and body parts close to the eyes project a proportionately larger image on the retina than those farther away. Both an invariant order of proportional relationships and foreshortened shapes are imposed upon human anatomy viewed egocentrically. In addition, many objective relationships between regions of the body cannot be directly apprehended, among them the true length of the lower extremities and the thickness of the torso, while otherwise prominent anatomical features such as the buttocks are virtually or completely absent from the visual field. Finally, since one cannot visually apprehend one's own body as a whole, any image of self as an independent three-dimensional entity must be the mental combination or integration of the multiple viewpoints possible in direct visual self-inspection.⁷ Multiple viewpoints, having more or less finite if overlapping boundaries, are an inherent requirement of all (technologically unassisted) human self-inspection. Operating together, these structural regularities provide a material origin for the "lozenge composition." Moreover, the discontinuous nature of the visual information thus produced about the human body and the sequence or order in which it is experienced may be relevant to the content and fabrication processes seen in other categories of female representations from the Upper Paleolithic such as "sketches" (ebauches) and "buttock" images.