Two conflicting propositions concerning the nature of sensory discrimination have been advanced: (1) That the systems responsible are sensitive to changes in the magnitude of the stimulus, however small, but that there is interference with the signal because of noise in the system. (2) That sensation is stepwise or quantal in character, because changes in the stimulus below a certain magnitude are not detected. It is clear that these propositions tend to equivalence as the size of the postulated quantum is sufficiently diminished. In this paper, using experimental and mathematical models, we have: (a) confirmed that in a 'noisy' situation very small changes in the stimulus are appreciated on a statistical basis; (b) demonstrated that, as predicted theoretically, the appearance of quanta can be introduced into the experimental results by the use of techniques which permit information loss. It is further shown that, in earlier work purporting to demonstrate the existence of sensory quanta, the technique of experiment has always been such as to introduce the information loss necessary to create artificial quanta. It is concluded that there is, at present, no evidence for the existence of a genuine sensory quantum. These findings are of general application to sensory systems and to all situations, natural or experimental, involving choice. Their relevance to evolutionary theory is briefly discussed.