a velocity representation model for mt cells
play

A Velocity-Representation Model For MT Cells Eero Simoncelli - PowerPoint PPT Presentation

A Velocity-Representation Model For MT Cells Eero Simoncelli Computer and Information Science Department University of Pennsylvania David Heeger Psychology Department Stanford University ARVO, 94 0 Input: normalized STE Input: image


  1. A Velocity-Representation Model For MT Cells Eero Simoncelli Computer and Information Science Department University of Pennsylvania David Heeger Psychology Department Stanford University ARVO, ’94 0

  2. Input: normalized STE Input: image intensities α v . . . . . . . . . • • . . . σ e-2 σ v-2 • • • . . . + • . . . + . . . . . . Output: normalized STE Output: normalized VE Two stages of computation, corresponding to: 1. V1 “simple” cells (tuned for Spatio-Temporal Energy). 2. MT “pattern” cells (tuned for “Velocity Energy”). ARVO, ’94 1

  3. Stage I Space-Time Weighting Fourier Tuning ω t ω y t y ω x x � Combines image intensities via space-time oriented linear Þ lters. � Tuning: localized in the Fourier domain. � A population of these mechanisms provides a distributed represen- tation of spatio-temporal energy (STE). ARVO, ’94 2

  4. Stage I is Not Velocity-Tuned ω t ω y ω x � Power spectrum of a translating 2D pattern lies on a tilted plane. � A model V1 unit will respond to many such planes. ARVO, ’94 3

  5. Stage II STE weighting Velocity Tuning v y ω t ω y v x ω x � Combines outputs of V1 units tuned for different orientations. � Tuning: localized in the image-velocity domain. � A population of these mechanisms provides a distributed represen- tation of velocity. ARVO, ’94 4

  6. Model Parameters 1. Stage I: � “Coverage” of Fourier domain. � “Order” of Þ lters. � Semi-saturation constant, e . � 2. Stage II: � “Coverage” of velocity domain. � Resting Þ ring rate, v . � � Semi-saturation constant, v . � ARVO, ’94 5

  7. Direction Tuning: Stage I Movshon et al., 1983 Model gratings plaids ARVO, ’94 6

  8. Direction Tuning: Stage II Movshon et al., 1983 Model gratings plaids ARVO, ’94 7

  9. Speed Tuning Maunsell & Van Essen (1983) Model 100 1 Response 50 .5 0 0 1/16 1/4 1 4 1/16 1/4 1 4 Speed Relative to Optimum spontaneous rate ARVO, ’94 8

  10. Correlation Response Function Britten et al. (1993) Model .75 150 Response .5 100 .25 50 0 0 0 25 50 75 100 0 25 50 75 100 Stimulus Correlation (%) spontaneous rate preferred direction non-preferred direction ARVO, ’94 9

  11. Response vs. Number of Dots Snowden et al. (1991) Model 80 1 Response 40 .5 0 0 0 64 128 192 256 0 64 128 192 256 Number of dots in preferred direction Number of dots in non-preferred direction spontaneous rate 0 16 64 256 ARVO, ’94 1 0

  12. Non-Preferred Suppression Snowden et al. (1991) Model 60 60 60 1 Response 30 30 30 .5 0 0 0 0 -180 -180 -180 -90 -90 -90 0 90 180 -180 -180 -180 -90 -90 -90 0 0 90 180 Mask Direction spontaneous rate preferred alone preferred + mask ARVO, ’94 1 1

  13. Model Behavior: Gratings Stage I Stage II Mean response plotted as a function of grating normal velocity. ARVO, ’94 1 2

  14. Model Behavior: Random Dots Stage I Stage II Mean response plotted as a function of dot drift velocity. ARVO, ’94 1 3

Recommend


More recommend