what the other 85 of v1 is doing
play

What the other 85% of V1 is doing Bruno A. Olshausen Helen Wills - PowerPoint PPT Presentation

Aug 15, 2022 •241 likes •643 views

What the other 85% of V1 is doing Bruno A. Olshausen Helen Wills Neuroscience Institute School of Optometry and Redwood Center for Theoretical Neuroscience UC Berkeley The standard model of V1 R e s p ons e P o i n t w i s e R e s p

  1. What the other 85% of V1 is doing Bruno A. Olshausen Helen Wills Neuroscience Institute School of Optometry and Redwood Center for Theoretical Neuroscience UC Berkeley

  2. The “standard model” of V1 R e s p ons e P o i n t w i s e R e s p ons e Image R e c e ptiv e fi e ld nor ma li z a ti on non - li n ea r it y li n ea r - r e s p ons e + - or / r ( t ) I ( x , y , t ) - K ( x , y , t ) n e i g hb or i n g n e u rons

  3. Why I am skeptical of the standard model • Lessons from the retina • Lessons from invertebrates • Vast overcompleteness of V1 • Non-linearities of cortical neurons • Difficulty of predicting neural responses to time-varying natural images

  4. Lessons from the retina

  6. On vs. off cone bipolar cells

  7. Rod bipolar cell is of on-type only Net convergence of rods to bipolar cells

  8. AII amacrine cell links rod bipolar cells to ganglion cells

  9. Lessons from invertebrates

  10. Jumping spiders

  11. Jumping spiders

  12. Vast overcompleteness of V1

  13. 1 mm 2 of cortex analyzes ca. 14 x 14 array of retinal sample nodes and contains 100,000 neurons

  15. V1 output is overcomplete by a factor of 50:1 70 μ Parvo cell input fibers V1 output fibers (layer 2/3)

  16. Non-linearities of cortical neurons

  17. Hausser & Mel (2003) (a) (b) 2-Layer model � i Thin branch y 1 subunits (c) 3-Layer model � i Distal apical thin branches y 2 � j Perisomatic thin branches Current Opinion in Neurobiology

  18. Difficulty of predicting V1 neural responses to time-varying natural images

  20. Responses of V1 neurons are not well predicted by RF models 18 ms 53 ms 88 ms 123 ms 159 ms 194 ms 229 ms 264 ms receptive field: a219, group 3, cell 4 ρ =0.362 0.8 0.7 spike count (35 msec bins) 0.6 0.5 0.4 0.3 0.2 0.1 0 0 5 10 15 20 25 30 time (sec)

  21. Responses of neighboring cells are heterogeneous cell 1 30 20 10 0 0 5 10 15 20 25 30 cell 2 30 20 10 0 0 5 10 15 20 25 30 cell 3 30 spikes/sec 20 10 0 0 5 10 15 20 25 30 time (sec)

  22. What is the other 85% doing? 1. 0 Va ri a nce e x pl a ined ~ 0 . 4 ~85% of V 1 function not understood 0 . 3- 0 . 4 0 0 1. 0 Proportion of cells studied

  23. There’s hope.

  24. Silicon polytrodes (Swindale, Blanche, Spacek)

  27. What to look for • What does a “day in the life of V1” look like? • Explaining away (sparsification) • Phase • Figure-ground • Synchrony • Laminar distribution of function (microcircuit)

  28. Explaining away Sparsified Feedforward response ( $# ) response ( "# ) ! ! ! ! ! ! ! ! ! !

  29. Phase Iris recognition (Daugman) Phase-Quadrant Demodulation Code Im [0, 1] [1, 1] Re [0, 0] [1, 0] Figure 2: The phase demodulation process used to

  30. Time-varying phase encodes information about transformations amplitude phase coefficient index time time

  31. Modeling phase dependencies (Charles Cadieu) sparse

  32. Learned D (space domain)

  33. Learned D (frequency domain)

  34. Learned D

  35. Figure-ground

  36. V1 simple cells can represent amodal completion Sugita (1999)

  37. Synchrony LGN spikes are phase-locked to ongoing retinal oscillations (Koepsell, Sommer, Hirsch)

  38. Distribution of function across laminae

  39. The Unknown As we know, There are known knowns. There are things we know we know. We also know There are known unknowns. That is to say We know there are some things We do not know. But there are also unknown unknowns, The ones we don't know We don't know. Feb. 12, 2002, Department of Defense news briefing From: The Poetry of Donald Rumsfeld Hart Seeley, Slate Magazine

Recommend

More recommend