Palmer and Hemenway (1978 Journal of Experimental Psychology: Human Perception and Performance 4 691-702) reported that shapes with multiple axes of symmetry are processed faster than those with single symmetry even when trials are blocked so that the subject knows that any symmetry axis will be vertical. Because their model of symmetry detection postulated a two-stage process in which all orientations are searched crudely at first, in no particular order, followed by second-stage scrutiny, the continued salience of multiple over single symmetry with blocking could not be explained. They claimed that stimuli with multiple axes of symmetry have an additional 'goodness'. Four experiments are reported in which it is demonstrated that both sensitivity (d′) and response bias (β) vary considerably in symmetry detection, not just as a function of the positive (symmetrical) stimuli used but also as a function of the negative or conjugate instances selected. Although stimuli with multiple axes of symmetry may well have extra salience due to pattern 'goodness', this factor may have been confounded with response bias in Palmer and Hemenway's experiments. It is suggested that several of their - as well as other researchers - results could be due to some combination of the effects of type of positive stimulus, type of negative stimulus, and response bias directed towards responding positively to highly symmetrical stimuli in a mix of less symmetrical stimuli. Palmer and Hemenway appear to have been correct in suggesting that subjects are more sensitive to quadruple than single symmetry, but the experiments indicate that subjects are also more willing to respond "symmetrical" to stimuli with quadruple symmetry when trials are not blocked, as in Palmer and Hemenway's experiment 1. However, it is demonstrated that the latter effect can be removed by blocking trials so that only one class of symmetrical pattern and one class of asymmetrical pattern occur in any block.