In search of an elusive hard threshold: a test of observer's ability to order sub-threshold stimuli

The contrast transducer function (d' vs. contrast) for sine gratings was claimed to come up from some non-zero contrast value rather than from the origin. This implies that there is a point (a hard threshold) on the grating contrast axis below which observers could not distinguish between presentations containing gratings and those containing a homogeneous field. We studied the ability to order sub-threshold square wave gratings and found, to the contrary, that observers were able to do this no matter how low the contrasts. At the same time, the observers failed to order the sub-threshold gratings when they were of the same contrast. The latter is inconsistent with signal detection theory which predicts that an observer's judgements are based on the same ordered set of sensory states irrespective of whether the stimuli differ or are the same. On the other hand, these data can be reconciled with the notion of a threshold if the latter is thought of as a fuzzy rather than a sharp margin on the contrast axis.