Abstract
In a matching task an adjustable stimulus needs to be matched with a standard stimulus. This can be done within or across a class of stimuli. In contrast to for instance yes-no paradigms, the matching task yields exact numerical values of the property estimated. As a result some effects become more directly accessible such as perceptual biases. Unfortunately, it is difficult to make a distinction between perceptual and decisional factors. We propose a method to disentangle perceptual and response biases. Oriented lines within a frame of orthogonal lines were used as stimuli. We compared sensitivity measures for yes-no, 2AFC and matching tasks. Both within and across participants the agreement was good. However, we also find non-perceptual factors that strongly affect the matching tasks. We will discuss regressions towards the mean, categorical biases and effects that depend on response time. In addition, short-, intermediate- and long-term memory changes as well as memory intrusions will also be discussed. Contrary to expectations, perceptual experience did not lead to a more veridical response. When oriented lines are used as probes for slanted grids, we argue that intrinsic properties of both sorts of stimuli, such as their specific variances, should be considered.