This paper reviews experiments that bear on the issue of binocular summation, the superiority of binocular over monocular viewing on various visual tasks covering studies published since the appearance of a previous review of this literature by Blake and Fox (1973). The experiments are grouped into three main categories-those that deal with the specificity of binocular summation (i.e., the extent to which inputs to the two eyes must coincide spatially and temporally), those that study binocular summation on suprathreshold tasks, and those that correlate binocular summation with other aspects of binocular function. The last section of the paper critically reviews several models of binocular summation.In a 1973 review of studies on binocular summation, the superiority of binocular over monocular viewing on visual threshold tasks, Blake and Fox concluded that (1) binocular summation occurs for many tasks, including increment detection, form recognition, contrast sensitivity, and flicker detection, and (2) this enhancement in binocular performance stems from genuine neural interaction between the eyes, not just probability summation. The present review examines experiments on binocular summation that have appeared since the 1973 survey, and, as before, attention is limited to psychophysical experiments comparing monocular and binocular visual performance, excluding investigations of binocular rivalry and stereopsis. The review examines (1) the specificity of binocular summation, that is, the extent to which inputs to the two eyes must be matched along some dimension in order to yield binocular summation, (2) the evidence for binocular summation in visual performance with suprathreshold stimulation, (3) individual differences in the magnitude of binocular summation, and (4) recent theoretical accounts of binocular summation.It should be noted that an improvement in binocular performance relative to monocular performance does not necessarily imply neural interaction between the two eyes, since statistical considerations alone dictate such an improvement: binocular viewing affords an observer two opportunities to detect a weakWe are grateful to Geoff Iverson, Joseph Lappin, James Hoffman, and Martha Teghtsoonian for their comments on an earlier version of this review. During preparation of this paper, R.B. was supported by a Career Development Award from NIH (EYOOI06). Requests for reprints should be sent to Randolph Blake, Cresap Neuroscience Laboratory, Northwestern University, Evanston, Illinois 60201.signal. This statistical property has been formalized as the probability summation hypothesis, and Blake and Fox (1973) discuss the various formulations of this hypothesis in detail. Recent work has generally recognized and sought to assess the potential contribution of probability summation, using either normative models (e.g., the integration model of signal detection theory) or empirical control conditions (e.g., temporal separation of the two monocular inputs). Moreover, very recently the logic of the probability s...