Abstract
Motion perception is an important function of the human visual system (e.g. Braddick 1974). Experiments on motion have typically employed categorical tasks, where subjects had to either detect the presence of motion (Van Doorn & Koenderink 1982) or decide whether motion was going in one out of a few discrete directions (Newsome & Paré 1988). Only very recently focus has shifted towards continuous report tasks (Luck VSS2018, Barbieri VSS2018) as they reflect the natural inherent continuity of low-level sensory features and might allow a more precise identification of the underlying perceptual processes (Smith 2016). Here we describe a perceptual decision-making experiment in which six groups of participants were tested on combinations of different stimuli and response modalities. In a two-by-three factorial design, we assessed behavioral performance of three types of feature-continuous RDKs: Brownian Motion (BM), Transparent Motion (TM) and Limited Lifetime White Noise Motion (WM), judged by using two alternative response methods. The first consisted of a ‘direct’ response, which required participants to move a trackball in the desired direction. The other, was an ‘indirect’ response, in which a single button had to be pressed when a self-rotating line matched with the chosen direction. By using a five parameter von Mises Mixture Model (vMMM), we were able to evaluate the properties of accumulated distributions of trial-wise response deviations, including detection, guessing, systematic errors, and a specific type of mis-perception (reporting the opposite direction) that would go unnoticed in standard categorical task designs. We further show that the detection frequencies are largely independent of stimulus and response modalities, while the precision parameters are more sensitive to both factors. Taken together, our findings underline the benefits and caveats of feature-continuous task designs, and validate the use of a diagnostic tool that allows to estimate the underlying parameters of the response distributions.