Abstract. Competent realistic drawings preserve viewpoint-invariant shape characteristics of simple parts, such that a contour in the object that is straight or curved, for example, is depicted that way in the drawing. A more subtle invariant öa V-shaped singularity of the occluding boundary, containing a T-junction and a contour termination öis produced at the junction between articulated smooth surfaces, as with the leg joining the body of a horse. 45% of the drawings made in 2007 by individuals with only minimal art education correctly depicted such junctions, a proportion that is not reliably different from the incidence (42%) of correct depictions in a large sample of cave art made 17000 years ago. Whether a person did or did not include the invariant in their drawing, all agreed that it made for a better depiction. European prehistoric cave art, dating back 17000 years, has, as its primary subject matter, large mammals such as horses, bison, oryx, sable, rhinoceri, and deer. An examination of these drawings reveals that viewpoint-invariant or `nonaccidental' prop- erties of simple parts expressed as variations of generalized cylinders (Binford 1971; Biederman 1987), such as whether the axis of the horns are straight (as with the oryx) or curved (as with the sable), as well as the presence of the parts and their relations, are accurately depicted (figure 1). In contrast to the nonaccidental aspects of depiction, informal perusal of the drawings leads to the conclusion that the rendering of metric properties, such as the aspect ratio of a body (figure 1), and the depiction of texture, such as a deer's antlers, are poorly represented. This greater sensitivity to nonacciden- tal compared with metric differences is evident in the performance of contemporary humans, not only in those immersed in the simple, regular artifacts that are character- istic of the developed world, but also in people of remote tribal areas with minimal exposure to such artifacts (Nederhouser et al 2005). Single-unit recordings of macaque IT cells also provide evidence that there is more sensitivity to nonaccidental than metric property changes (eg Kayaert et al 2003). In 1982, Jan J Koenderink and Andrea J van Doorn proved that another invariant must be present for articulated smooth surfaces: a V-shaped singularity of the occlud- ing boundary, containing a T-junction and a parabolic contour with a termination (figure 2). Koenderink and van Doorn noted a number of instances where this feature is omitted in otherwise competent drawings. For example, a line drawing of a front view of a breast should be a parabolic open curve, not an oval. The singularity is due to a negatively curved area which serves to bound `object-like' convex or concave areas (J J Koenderink 2007, personal communication). The transversality-regularity (Hoffman and Richards 1984) performs a similar function in defining a basis of segmenting parts. As the depiction of animals can be thought of as being largely formed by the join of smooth surfaces, there were plenty of opportunities for our prehistoric brethren to incorporate concave terminators in their portfolio. What was this incidence? And how does it compare with the incidence of this feature in drawings by individuals living today with little or no formal education in depictive art? LAST BUT NOT LEAST 17 000 years of depicting the junction of two smooth shapes Perception, 2008, volume 37, pages 161 ^ 164 Irving Biedermanô, Jiye G Kim University of Southern California, Department of Psychology (ô and Neuroscience Program), 3641 Watt Way, Hedco Neurosciences Building, Room 316, Los Angeles, CA 90089-2520, USA; e-mail: bieder@usc.edu Received 23 August 2007, in revised form 20 November 2007 doi:10.1068/p5907