In general, the detectability of a signal preceded and followed by noise maskers is less than the prediction based on a simple addition of the effects of the maskers (i.e., an intensity sum). Data verifying and extending this finding were collected in a variety of conditions in which the two maskers either surrounded the signal in time or preceded it. These data are used to support two related claims. First, the failure of the intensity‐sum argument is likely to be due to nonlinearities in the coding of intensity. Second, the signal threshold is a monotonic measure of the internal percept at essentially the same time relative to the signal for different masking conditions. This latter result negates the integrator‐movement hypothesis (which holds that the time varies) and provides important support for the chain of derivations obtained by Penner and Shiffrin [J. Acoust. Soc. Am. 67, 617–627 (1980)].