Abstract Limited-lifetime Gabor stimuli were used to assess both first and second-order motion in peripheral vision. Both first and second-order motion mechanisms were present at a 20 deg eccentricity.

Second-order motion, unlike first-order, exhibits a bias for centrifugal motion, suggesting a role for the second-order mechanism in optic flow processing.

Introduction First-order motion consists of moving luminance-defined attributes. Second-order motion, on the other hand, consists of moving patterns whose motion attributes are not luminance-defined, e.g. moving contrast or texture borders [1,2]. The detection of first and second-order motion is thought to be mediated by different mechanisms, i.e. a quasi-linear (first-order) and a non-linear (second-order) mechanism [3,4].

Methods The stimuli consisted of linearly added Gabor patterns each consisting of a 1D sinewave carrier, oriented orthogonal to the direction of motion, enclosed by a 2D Gaussian envelope. The coherence, i.e. fraction of the Gabor patterns that moved coherently, and displacement were varied. Previous studies (eg. Baker and Hess [5]) using these stochastic moving Gabor patterns have shown that two mechanisms underlie the processing of these global motion patterns, a carrier-related (first-order) and an envelope-related (second-order) mechanism. First and second-order mechanisms can be isolated by different jumpsizes or by carrier-manipulations on alternate frames. The stimuli were presented at a 20 deg eccentricity in either the left, right, upper or lower visual field. Direction of motion was either vertical (up-down) or horizontal (left-right). A two-interval two-alternative forced-choice task was used, requiring two judgements from the subject: which interval contained the coherent moving stimulus, and the direction of motion. Due to the judgements which were required in each trial, change level is at 75% errors.

Results To reveal any centripetal/centrifugal anisotropies, the data from the different parts of the visual field were selectively averaged based in their directions relative to the fixation point.

Percent errors are plotted as a function of coherence (a & b) and displacement (c) for first-order (a & c) and second-order motion (b & c). The error bars indicate 95% confidence intervals. In each graph the data for four different directions relative to the fixation point are plotted separately, i.e. centripetal, centrifugal, clockwise (90 deg) and counterclockwise (270 deg) around the fixation point. The left and the middle panels show performance as a function of coherence for first (a) and second-order (b) motion. The right panel shows similar data as figure b (centripetal and centrifugal motion) plotted with error bars as a function of displacement. The four new lines indicate the perceptual judgements for first-order motion in all four directions for similar displacements. For second-order motion the subject's performance is worse for centripetal motion over a range of coherences and displacements. No consistent bias is found for first-order motion.

Conclusion Second-order motion, unlike first-order, exhibits a bias for centrifugal directions, suggesting a role for the second-order mechanism in optic flow processing and providing a dissociation between first and second-order motion processing.

References:

[1] C. Chubb, G. Sperling, (1988). Drift-balanced random stimuli: A general basis for studying non-Fourier motion perception. J.O.S.A., 5:1986-2007.

[2] P. Cavanagh, G. Mather (1989). Motion: The long and short of it. Spatial Vision. 4: 103-129.

[3] A.T. Smith (1994). The detection of second-order motion. In Visual Detection of Motion, Smith A.T. & Snowdon R.J., Eds. (Academic Press, London) pp. 145-176.

[4] C.L. Baker Jr. (1999). Central neural mechanisms for detecting second-order motion. Current Opinion in Neurobiology, 9: 461-466.

[5] C.L. Baker Jr., R.F. Hess (1998). Two mechanisms underlie processing of stochastic motion stimuli," Vision Research, 38: 1211-1222.