Abstract
In a classic experiment on the detection of simple patterns, Campbell & Green (1965) showed that for both humans and an ideal signal detector, two eyes are only √2 better than one eye. If the ideal observer shows only a √2 improvement, then how can binocular simple cells (Ohzawa & Freeman, 1986; Smith, Chino, Ni, & Cheng, 1997) show a doubling in contrast sensitivity when given binocular input? We resolve this puzzle by showing that the ideal observer may differ from the Campbell & Green model, and we find that the binocular improvement for detection increases with spatial frequency to a level well above √2.