Abstract
Whereas cue-combination theories classically articulate the benefits to discrimination of combining multiple sources of information, there is another class of theory that supposes that the mere presence of a signal in one channel or modality can actually enhance the processing of a signal in another by narrowing the range of possible values to be coded. Barlow's (1990) inhibitory interaction theory, for example, suggests that the perceptual value of one stimulus can be altered by the co-occurrence of another, normally correlated stimulus, even when that stimulus presently adds no information. Thus, for example, the visual velocity of the environment during self-motion is reduced relative to how the same visual stimulation appears to a stationary observer (Durgin, Gigone & Scott, in press). Motor-prediction theories make similar predictions regarding the muting of perceptual outcomes of actions compared to those experienced without concomitant and correlated motor signals. Here we show that discrimination thresholds for 3D velocity of structured flow-fields moving near walking speed are lowered, not only during active walking, where both motor-prediction and inter-sensory inhibition theory make a common prediction, but also during passive self-motion. Thus, even though the self-motion information contributes no discriminative information about the specific visual motion stimuli presented, visual motion stimuli experienced during self-motion are nonetheless represented in an altered coding space where discriminations of unusually low velocities are rendered nearly impossible (since they all appear to be zero), whereas those of higher velocities - especially those near the rate of self-motion, are better discriminated. The ability of sensory systems to adaptively alter their coding spaces in response to otherwise uninformative cues is of obvious importance, and must be considered as an alternative source of improved discrimination in tasks where multiple cues are combined.
Swarthmore College Faculty Research Fund, HHMI