How are visual stimuli processed in the visual cortex?

(Essay plan notes)

Sanjay Manohar (Oxford 2015)

Intro


Localisation & identification: different kinds of problem
Muller's specific nerve energies; place-coding



Input from LGN: 6 layers;
  1-2=magno ->  layer IVc ->
  3-6=parvo (M/L) -> layer IVc
  konio=(S/ML) -> blobs layers 2/3
Lesion LGN: magno = motion; parvo = colour

V1

Primary V1 Area 17; Secondary V2-5 Area 18-19
Gordon Holmes - visual field defects
striate 4A/B/C - stellate cells 4C
Tootell 1991: 14C deoxyglucose autoradiography - retinotopy; calcarine fissure, 
 logarithmic (central 5 deg = 30%)
Allman Kaas 2000 - retinotopy of striate cortex in monkey single-cell
 

  Wong-Riley 1979, 
      Blob cytochrome oxidase - from koniocellular = colour contrast, double opponent
      interblob - parvocellular 4Cbeta, single-opponent, luminance responses
  Ocular dominance: inject one eye with radiolabelled amino acid. Autoradiography.

  hubel and wiesel - ocular dominance, anterograde radiolabelled proline 
                   - ice-cube diagram
                   - monkey patched eye = critical window; 
                    amblyopia and patch treatment
  Livingstone and Hubel  1983 single neuron microelectrode, cat 
  Spatial frequency selectivity (graph)
  Binocular disparity (graph, diagram)
    - stereopsis - near/far cells Pollen 1981
    - horopter, absolute vs relative
 
Simple cell RFs:
 - H&W 1959 graphs of responses with orientation, motion, gabors
 - static response to lines, location-specific
 - spatial & temporal linearity - DeAngelis study, reverse correlation
 - draw receptive field and possible wiring of 
    how orientation selectivity could arise
    how and motion detection arise
  H&W vertical penetration = similar orientation --> columns, pinwheel
Complex cell (H&W 1962)
 - dynamic responses, location invariance, nonlinear, binocular
 - LGN inputs -> simple cell RF; multiple simple cells -> complex RF
 - hierarchical; feedforward models
 
End-stopped cells
V2 - more complex RFs, feature conjunctions, curves, 3d corners
V4 - location invariance
Tanaka studies: RF size increases
Moran & Desimone 1985 top down input

Extrastriate / Higher areas


dorsal stream                     ventral stream
-------------------------------------------------------------
Parietal lobe                     temporal lobe
V5, IPL, SPL                      V4, FFA, LOC, VWFA, grandmother
Ungerleider and Mishkin 1982 - monnkey landmark task vs pattern discrimination
where                             what
spatial cues, allocentric         pattern detection, semantics




Dorsal stream 
  - MT/V5 Akinetopsia; active in imagined or illusory motion.
  - Goodale and Milner 1992 - DF: post-box task - "vision for action"
  - Salzman..Newsome 1992 microstim V5 -> altered report.
  - Bisley Pasternak 2000 - MT in WM / comparison of directions
Ventral stream "what" V4 IT 
 - colour, achromatopsia; 
 - form, prosopagnosia; Tanaka 1996 IT; 
 - apperceptive agnosia - can copy but not recognise
 - associative agnosia - identify but not semantically pair 
 - Fuster 1981 - IT in delayed response task

Both dorsal and ventral are polymodal
 

Weiskrantz 19174 blindsight - retinotectal / extrastriate
Anton Babinski syndrome confabulation

Big picture

Big picture:
  divergence & feature extraction vs integration & binding
  Gabor RFs as principal component analyis
  hierarchical processing, predictive coding, multimodal
  feedback as well as feedforward; attention, expectation

Why? Economical Olshausen and Field 1997 gabor
visual primitives
half cortex is "visual" 
predictive coding
grandmother cell

Does the visual system integrate or segregate information

Integration Segregation
orientation = sum of ON-centre different orientations in columns
colour (L/M; S/M+L) two separate chroma channels
colour + form (double-opponent, edges) blob vs interblob
disparity L-R ocular dominance columns
larger RF = combines information retinotopy, place-coded space
motion = T2-T1 area MT
IT cells complex arrangements of shapes dorsal vs ventral stream
top down input / expectation / polymodal fusiform face area; visual word form area