Ψ Sensation & Perception
Ψ -Discussion Section- Session 4 – Visual Cortex I Icecube (Hubel & Wiesel) Pinwheel (Issa & Stryker)
Administrative stuff
Presentation 1: Receptive fields and functional architecture of monkey striate cortex (1968) presented by John Scott-Railton Presentation 2: Spatial Frequency Maps in Cat Visual Cortex (2000) presented by Jasmine Kwong
Next week Week 7: 11/08/2004 Higher visual perception: Contours Paper 1 (Classic): von der Heydt et al., 1984 (Perla) Paper 2 (Modern): Bakin et al., 2000 (?)
BACKGROUND
This weeks issue:
How can one effectively represent 6 stimulus dimensions on a 2 How can one effectively represent 6 stimulus dimensions on a 2.5 dimensional sheet? Orientation Ocular dominance Spatial frequency Motion Depth Color
Essentially a mathematical optimization problem Similarly: How to put as much of the sheet as possible into a sphere? (Cortex into skull) Solution? Folding!
That’s why cortex is folded It optimizes area while minimizing volume So far, so good. But what about the basic functional unit of cortex organization? How to represent all stimulus dimensions effectively in all locations?
Two proposals:
1. The “Iceblock” model (Hypercolumns) David H. Hubel Torsten N. Wiesel Nobel prize in Physiology, 1981
2. The “Pinwheel” model Naoum P. Issa (UofC) Michael P. Stryker
Optical imaging Rationale: Deoxygenated blood is darker than oxygenated blood. This is particularly true in the range of red light. Active neurons use up more oxygen than neurons that are less active. This leads to a local, spatial distribution of blood saturation levels. Regions of active neurons should appear darker.
Optical Imaging
Pro and Con Pro Con Very high spatial resolution Allows to visualize large scale activity patterns Relatively direct link to electrical activity of neurons Better time course than fMRI Con Still relatively bad time course (linked to oxygen dynamics) Invasive Very low signal to noise need to average many, many trials to see signal. Hemodynamics indirect, nonlinear.
BOLD time course
The case for spatial frequency Orientation Ocular dominance Spatial frequency Motion Depth Color
Picture representation in V1
The concept Luminance Space Amplitude = Contrast Wavelength = Spatial Frequency Amplitude = Contrast Space
Looks like:
Fouriers theorem EVERY waveform can be decomposed into simple sine-waves with the right amplitude and frequency. EVERY waveform can be synthesized by adding sine-waves with the right amplitude and frequency.
V1 neurons are apt to represent sine waves of different frequency
Philosophical implication The representation of the visual world by neurons in V1 is VERY different from our phenomenological experience. Each neuron only represents a tiny slice of oriented spatial frequency!