The reported reversals of a rotating Necker cube, which changes direction of rotation when a perspective reversal occurs, were examined under a number of conditions. These permitted comparisons of reversal rates within viewing periods, across successive viewing periods within an experimental session, and across successive weekly sessions. In addition, observers viewed either one or two rotating cubes simultaneously within each of the various viewing periods. Clear evidence for a learning effect was obtained in the form of significant savings across successive viewing periods and sessions. At the same time, results from the multiple-cube conditions and from the pattern of reversals within individual viewing periods appeared to be more consistent with a process of neural fatigue. A two-stage model of reversible-figure perception is proposed which is characterized by (1) fatiguing with extended viewing of the two sets of neural channels that underlie the two percepts of the reversible figure, and (2) learning, which helps to establish the organization of the cortical channels as well as moderate channel activity via such processes as attention and strategy.In the last decade there has been a resurgence of theoretical and empirical activity related to a particular class of visual illusions most commonly referred to as reversible figures. This class of illusions includes such well-known figures as the Necker cube, Rubin's vase-faces, the Schroeder staircase, the Mach folded card, Fisher's man-girl, Boring's young girl-old woman, the Maltese cross, and several others (see Attneave, 1971, for examples of these and other reversible figures). Although empirical research with these figures can be traced back to the last century, there is still extensive interest in them because of the insights they are thought to provide into the nature of the perceptual process by their curious multistable character.Two competing classes of theories have evolved in an attempt to explain the perceptual fluctuations reported for the Necker cube and other reversiblefigure illusions. I The older theory, that of neural satiation (cf. Attneave, 1971;Howard, 1961;Kohler & Wallach, 1944;Orbach, Ehrlich, & Heath, 1963), proposes a passive process of fatigue as the basis for perceived reversals. In this view, the cortical organization underlying one perspective of a reversible figure satiates or fatigues with extended viewing. It then gives way to a second, fresher cortical organization underlying the other perspective. This