Perception Lecture Notes: LGN and V1

Professor David Heeger

What you should know about this lecture

• visual field
• vertical & horizontal meridia
• blind spots
• left & right hemifields
• Principle of lateralization (left-hemifield -> right hemisphere and vice versa)
• LGN (anatomy and physiology)
• magnocellular and parvocellular layers
• center-surround receptive fields
• retinotopic maps
• V1 (anatomy and physiology)
• simple cells, complex cells, and hypercomplex cells
• orientation- and direction-selectivity
• orientation and ocular dominance columns
• amblyopia

Retino-Geniculate Visual Pathway

Diagram of the visual pathways

The optic nerve leads from the eye to the optic chiasm. The optic chiasm is where some of the fibers cross. The optic tract proceeds from the optic chiasm to the lateral geniculate nucleus (LGN). The optic radiation leads from the LGN to primary visual cortex (V1).

There are two eyes, thus we illustrate two visual fields. The fixation point is the center of the visual field; it corresponds to the fovea. T he vertical meridian splits the visual fields into left and right hemi-fields. The horizontal meridian splits the visual field into upper and lower hemi-fields. The blind spot is the region of the visual field that corresponds to the optic disc.

Principle of lateralization: The right half of the brain receives sensory information from and sends motor commands to the left half of body. In the visual system, the right half of the brain receives information about the left half of the visual field (note: not just from the left eye).

Visual deficits due to lesions at different points in the pathway: Suppose someone comes in complaining of vision problems. Do you send them to an ophthalmologist or to a neurologist? Is the problem an eye disease? Or is it some central problem like a tumor pressing on the optic tract?

If the deficit is in only one eye's visual field, send the patient to an ophthalmologist. If the deficit is in corresponding parts of both eye's visual fields, then it is a central problem, and send the patient to an neurologist for an MRI or CT scan.

Lateral Geniculate Nucleus (LGN)

• 6 layers.
• Cells have monocular input.
  
• Layers alternate inputs from each of the two eyes.
• The top four are parvocellular layers, two layers from each eye. Parvo (small) LGN cells receive inputs from (small) midget ganglion cells.
• The bottom two are magnocellular layers, one layer from each eye. Magno (large) LGN cells receive inputs from (large) parasol ganglion cells.

LGN retinotopy

• The retinal (hereafter called "retinotopic") map is preserved. Axons from the retina preserve their order.
• There is an entire map of a visual hemi-field in each layer of the LGN.
• The maps are in register in each layer.

In the LGN, cells have center-surround receptive fields just like retinal ganglion cells. There is little or no information processing beyond that done in the retina, so what is the function of the LGN? Why not send axons directly to cortex? Nobody knows for sure. There are two compelling hypotheses that I know of:

1. The LGN brings retinotopic maps from both eyes into register to make it easy for cortex to combine inputs from the two eyes.
2. Only 10% of inputs to LGN come from the retina. 90% are modulatory inputs from cortex and the brainstem. The brainstem modulates the information flow from the eye to the visual cortex, for example, according to the sleep cycle. Cortical (feedback) inputs to LGN are not well understood but might have to do with attention. The LGN is a convenient bottleneck for these modulatory inputs from the brainstem and cortex. If you tried to send these projections from brainstem and cortex all the way back to the retina then you'd end up with a blind spot that was 10 times larger.

Primary Visual Cortex (V1)

The above figure shows the results of an experiment in which an anaesthetized monkey viewed a flickering bulls-eye pattern, and was injected with radioactively labeled glucose. The glucose was taken up by active neurons. The animal was then sacrificed, and V1 was surgically removed and flattened. The flattened V1 was then used to expose radioactively sensitive film. The result is a picture of regions of activity evoked by the bulls-eye. As you can see, V1 maintains a retinotopic map. However, it is distorted so that the central 10 degrees of the visual field occupies more than half of V1. This makes sense because of the poor acuity in the periphery (recall that peripheral ganglion cells have large dendritic trees and pool over many photoreceptors).

V1 physiology: Hubel and Wiesel won the Nobel prize for discovering the functional organization and basic physiology of neurons in V1. They discovered three different types of neurons that can be distinguished based on how they respond to visual stimuli that they called: simple cells, complex cells, and hypercomplex cells. V1 neurons transform information (unlike LGN cells whose receptive fields look just like those of ganglion cells).

• respond best to elongated bars or edges.
• are orientation selective.
• can be monocular or binocular.
• have separate ON and OFF subregions.
• perform length summation (they have bigger responses with increasing bar length up to some limit, at which point the response reaches a plateau).

Simple cell model/wiring diagram

A simple model of simple cell responses, suggested by Hubel and Wiesel, is that each simple cell sums inputs from LGN neurons with neighboring/aligned receptive fields to build an elongated receptive field that is most responsive to elongated bars or edges.

• are orientation selective.
• have spatially homogeneous receptive fields (no separate ON/OFF subregions).
• are nearly all binocular.
• perform length summation.

Hypercomplex cell receptive field

Hypercomplex cells are like complex cells except there are inhibitory flanks on the ends of the receptive field, so that response increases with increasing bar length up to some limit, but then as the bar is made longer the response is inhibited. This property is called end-stopping.  The upper graph shows the response of a complex cell as a function of bar length.  The lower graph shows the response of a hypercomplex cell as a function of bar length.

Direction selectivity: Some V1 cells are also direction selective meaning that they respond strongly to oriented lines/bars/edges moving in a preferred direction (e.g., vertical lines moving to the right) but not at all in the opposite direction (e.g., vertical lines moving to the left).

In class we showed a video that demonstrates how Hubel and Wiesel classified the various cell types and mapped their receptive fields. The video shows visual stimuli while recording from each of several V1 neurons.  The electrode was connected to an amplifier, and output to a loudspeaker. The audio track allows you to hear the loudspeaker - each click corresponds to an action potential. We showed examples of a simple cell, complex cell, direction-selective complex cell and a hypercomplex cell.

V1 functional architecture: Hubel and Wiesel also discovered that the neurons in V1 are arranged in an orderlyfashion. Neurons with similar response properties (e.g., the same orientation preference) lie nearby one another.

Columnar architecture: As one moves an electrode vertically through the thickness of cortex, one finds that most neurons have the same selectivity (e.g., the same orientation preference and eye dominance). Ocular dominance columns: As one moves an electrode tangentially through the cortex, one first finds cells that respond to left eye inputs, then binocular (responsive to both/either eye), then right eye, then binocular, then left again, etc. Orientation columns: As one moves the electrode tangentially in the orthogonal direction, one first find cells selective for vertical, then diagonal, then horizontal, etc. A hypercolumn is a chunk of cortex about 1 mm square by 3 mm thickn that contains neurons, all with approximately the same receptive field location, but with all different orientation selectivities, direction selectivities, both (left- and right-) eye dominances represented.

The notion of a functional columnar architecture is a big theme in cortical physiology. We'll see other visual brain centers with analogous organization: area MT has a columnar architecture of direction selectivity for visual motion perception, area IT has a columnar architecture of complex shape/feature selectivity for object recognition.

The functional architecture of V1 has been studied extensively.  Hubel and Wiesel's initial results have been replicated in recent years using optical imaging methods.

In optical imaging, you open a hole in the skull, point video camera at the brain, and collect images that reflect the relative amount of oxygenated ("red") versus deoxygenated ("blue") blood.

Top: picture of the brain while the animal views stimuli with the left eye (right eye patched/occluded). Then, one takes a picture of the brain while the animal views stimuli with the right eye. Bottom: one subtracts one image from the other to give a picture that looks like zebra stripes - these are the ocular dominance columns.

Analogous experiment have been carried out, but using orientation (e.g., vertical versus horizontal) instead of eye input (right eye vs left eye). This gives a picture of the orientation columns.

Amblyopia: Amblyopia, or cortical blindness, is a term that refers to a variety of visual disorders when there is no problem with the eye (the optics and retina are fine), but one eye has better vision than the other. Amblyopia can be caused by strabismus (wandering eye) if it is not corrected in infancy.  What appears to happen is that during development, if the signals from one eye are weak or out of register with input from the other eye, then the brain develops in a way that ignores the signals from the weak or misdirected eye.

Hubel and Wiesel performed a classic experiment on this topic.  They raised kittens with one eye sutured shut (monocular deprivation) and found that the ocular dominance columns did not develop properly.  Instead, primary visual cortex organized itself to respond almost entirely to the undeprived eye.    They found that there is a critical period during development up until the kittens are about 4 months old.  Monocular deprivation during this period messes up the ocular dominance columns.  Deprivation after the critical period has no effect. If the kittens were monocularly deprived throughout the critical period, then they could never regain the proper ocular dominance organization.

Consequently, we now know that we must intervene in infancy to prevent amblyopia.  A wandering eye can often be corrected surgically, but not until a baby is old enough.  In the meantime, one covers an infant's good eye with a patch for a few hours each day so that the brain must rely on signals from the bad eye.  This helps the brain develop properly to process signals coming from both eyes.