"Motion streaks" play a role direction discrimination, but not detection

D. Apthorp, D. Alais, P. Wenderoth

    Research output: Contribution to journalMeeting abstract

    Abstract

    Most models of motion perception assume motion is processed by a separate stream from other properties of an object such as colour and form, although recent evidence suggests interactions between the streams. In biological motion and structure-from-motion, shape information is available only when motion is present. Geisler (Nature, 1999) suggested form information could also be used by the motion system. Fast-moving elements leave neural “streaks”, due to temporal integration. These could be detected by orientation-sensitive neurons and combined with motion signals to improve direction discrimination. Our study used random and fixed-walk dots in a 2AFC direction discrimination task. Experiment 1: participants judged direction of motion of 120ms 50%-coherent RDKs. The number of fixed-walk frames controlled streak length of each dot (12min Gaussian blob). Above 2°/s, longer streaks produced lower discrimination thresholds, similar to Geisler’s “critical speed” of 2.5°/sec. Experiment 2: at each speed and streak length, contrast thresholds were measured for coherent motion detection, yielding a U-shaped function for speed, but no difference between streak lengths. Experiment 3 replicated Experiment 1 at low contrast (2.5 x threshold) and found no “streak facilitation”. Results support a role for streaks in direction discrimination, but not in coherent motion detection.
    Original languageEnglish
    Pages (from-to)15
    Number of pages1
    JournalAustralian Journal of Psychology
    Volume59
    Issue numberSuppl.
    Publication statusPublished - 2007
    Event34th Australasian Experimental Psychology Conference - Canberra, Australia
    Duration: 13 Apr 200715 Apr 2007

    Fingerprint

    Dive into the research topics of '"Motion streaks" play a role direction discrimination, but not detection'. Together they form a unique fingerprint.

    Cite this