150 likes | 398 Views
Motion Perception: Theory and Experiments. (Pablo J. Velasco) New York University. Outline. Goal: Local cues: the aperture problem From local to global: INTEGRATION Hildreth’s model An interesting case: the rotating ellipse Hildreth’s model predictions Psychophysics experiment To do….
E N D
Motion Perception:TheoryandExperiments (Pablo J. Velasco) New York University
Outline • Goal: • Local cues: the aperture problem • From local to global: INTEGRATION • Hildreth’s model • An interesting case: the rotating ellipse • Hildreth’s model predictions • Psychophysics experiment • To do…
What we “detect”:The aperture problem • V • VII?
From local to global:INTEGRATION (output): GLOBAL perception of motion Small receptive fields (input): LOCAL information
Hildreth’s model1 a) b) v c) d) v • Finding the velocity field V = V(s) compatible with V Rigidity 1 Hildreth, E. C., The Measurement of Visual Motion, The MIT Press, Cambridge, MA (1983)
Hildreth’s model Assumptions (constraint): Q = Smoothness (Minimize)
Ellipse Rotating • Ambiguous stimuli illusions Change of perception with ellipse eccentricity e Give info about integration of local into global
What does Hildreth’s model predict? VROT Y Y uII uII u u X X VROT VDEF • V : minimum Q QROT QDEF
Experiment description • Goal: for a given e, find l. • Present ellipse (e) rotating (2.5 sec). • Add a bead on surface, rotating with a given ltrial. • Report whether the bead rotates faster or slower than the ellipse.
Experiment description (ahead) e Bead rotation (lagging) ltrial • Psychometric function: l for each e l
(Preliminary) results 1 0.5 0 1 0.5 0 1 0.5 0 Bead ahead Bead ahead e = 5.5 l e l e l l e e Subject 2: Subject 1:
To do: • Finish the psychophysics (capturing) • Bayesian model • 2D cues...