Heaviness or Moveableness? Is Perception Sensitive to High- Order, Action-relevant Physical Variables?

Since Weber's initial investigations into heaviness perception, psychophysics has primarily focused on the mapping between low-order, behavior-irrelevant physical variables to sensations. An alternative strategy is to evaluate perceptual sensitivity to higher-order, physical variables of direct significance to the perceiver's activities. Research over the past 15 years on heaviness perception illustrates the advantages of the latter strategy and the significance of such a strategy in accounting for the richness of experience. Psychophysics has historically focused on the mappings between simple (behavior-neutral) physical variables (e.g., mass, light amplitude/wavelength, sound amplitude/frequency) and their corresponding sensations (e.g., heaviness, brightness/color, loudness/pitch). This approach is motivated by the assumption that “such relations are fundamental for the apprehension of “secondary” properties” (Runeson, 1977, p. 175). Interestingly, humans are remarkably bad at judging simple physical variables and when perception does not reliably correspond to simple physical variables (e.g., when two different physical stimuli map to the same percept [metamers] or when two stimuli that are identical along a single physical dimension are perceived as different [i.e., illusions]) this is typically attributed to the subjective nature of perception (see Shockley, Carello, & Turvey, 2004, and Turvey, Whitmyer, & Shockley, 2001, for a more detailed discussion). Gibson (1979) claimed that because simple physical variables do not capture behaviorrelevant structure in our environment, they should not be the foundation of a theory of perception designed to account for the richness of behavior. He argued that perceptual systems must be sensitive to complex physical variables that are inherently meaningful with respect to the behavior of a perceiving-acting system (i.e., without interpretation or inference). In the spirit of this argument, Runeson (1977) suggested that perceptual systems should be conceptualized as “smart devices” that are directly sensitive to higher-order, behavior-relevant structure, rather that rote computational devices that operate over simple physical variables detected through the senses. He proposed a metaphor of the polar planimeter—a physical device that is designed to directly measure the higher-order variable surface area without any computation over lower order variables (e.g., extents). Implicit in Runeson’s argument was the assumption that the typical psychophysical strategy of evaluating the mappings between