In psychophysical discrimination tasks, sensitivity often varies with the order in which a constant standard and a variable comparison is presented, a finding referred to as Type-B-Effect. Typically, difference limens are lower when the standard is presented before the comparison than when this order is reversed. This direction of the Type-B-Effect is defined as negative, while the opposite direction (higher difference limens when the standard precedes the comparison) is defined as positive.

The negative Type-B-Effect is predicted by the internal reference model (IRM), which assumes that the second stimulus is judged relative to an internal standard that is updated on each trial. Challenging this model, positive Type-B-Effects have been reported repeatedly for duration discrimination experiments.

Noteworthy, these experiments share the design choice of combining staircase procedures with very short standard durations. Here, we present simulations showing that measuring sensitivity under these boundary conditions may be biased. In detail, the procedure can asymmetrically inflate difference limens, in consequence leading to the appearance of a positive Type-B-Effect. In addition, data from two duration discrimination experiments show no positive Type-B-Effect when the method of constant stimuli is used, even with stimulus configurations closely matching those of earlier reports.

Taken together, our results suggest that previously reported positive Type-B-Effects do not provide convincing evidence against IRM. Rather, we suspect that positive Type-B-Effects arise as a procedural artifact from a specific combination of design choices.