When is motion 'motion'?

Examples of visual motion have become more and more abstract over the years, leading up to 'third-order' stimuli where direction is actually determined by the observer through top-down attention. But how far can this be pushed are there motion stimuli that are yet more arbitrary and abstract? Actually, there is a broad class of 'conceptual motion' stimuli - things like a moving grating of faces, or a shifting pattern of words - that are perfect analogs to traditional 'perceptual motion' stimuli, solvable by the same motion computation and for which observers can readily make direction-of-motion judgments. Interestingly though, these do not produce a sensation of motion (among other automatic consequences of motion detection). Here we compare a luminance-based perceptual motion stimulus to a semantic-based conceptual motion stimulus to contrast the psychophysical hallmarks of these motion categories.