Sem2 Week 5

Rods are sensitive to low light and handle dark/night vision. Cones are concentrated in the centre and handle colour vision in daylight. The retina contains about 120 million rods and 6 million cones.

The output neurons of the retina. Their axons form the optic nerve and carry visual signals from the eye to the rest of the brain including the LGN.

Changes in light rather than steady illumination. They signal the onset or offset of light in their region of the visual field making them sensitive to contrast and edges rather than absolute brightness.

A retinal ganglion cell that fires when light turns on in its receptive field and reduces firing when light turns off.

A retinal ganglion cell that fires when light turns off in its receptive field and reduces firing when light turns on.

A retinal ganglion cell that fires briefly both when light turns on and when it turns off. It is particularly sensitive to changes and movement.

The region of sensory space that when stimulated causes a neuron to change its firing rate. For retinal cells this is a small patch of the visual scene.

The receptive field is divided into a central zone and a surrounding ring that have opposite effects. Light in the centre excites an ON-centre cell while light in the surround inhibits it and vice versa for OFF-centre cells.

A mathematical model of the centre-surround receptive field. It subtracts a broad Gaussian (surround) from a narrow Gaussian (centre) producing a filter that responds strongly to edges and contrast changes but weakly to uniform illumination.

It acts as an edge detector and contrast enhancer. Uniform regions produce little output while boundaries between light and dark areas produce strong responses. This also compresses the image since most of the signal is discarded.

Under low luminance the surround inhibition is weakened. Strong edge enhancement would amplify photon noise in dim conditions so a weaker surround gives a better signal-to-noise trade-off.

The central region of the retina densely packed with cones and with very small receptive fields. It provides the highest spatial resolution at about 300 dpi and is where we fixate when looking directly at something.

When viewing adjacent bands of different grey levels the edges between them appear brighter or darker than they really are. This happens because the retinal DoG filter enhances contrast at boundaries making edges look more pronounced than they physically are.

Dark blobs appear to flicker at the intersections of a white grid overlaid on black squares even though the intersections are just as white as the lines. The effect disappears when you look directly at an intersection.

In peripheral vision receptive fields are larger so the inhibitory surround covers both the line and the adjacent black squares at an intersection reducing the response. In the fovea RFs are tiny so the surround cannot span both and no suppression occurs.

A circuit motif where active neurons inhibit their neighbours from the side. In the retina horizontal cells spread inhibition from surround cones onto the central bipolar cell creating the OFF surround of the receptive field.

It enhances differences between neighbouring regions of a stimulus such as edges in vision or frequency contrasts in hearing. By suppressing uniform responses it allows the system to highlight changes which carry more information.

Receptive fields apply across sensory systems. In touch they describe the patch of skin that activates a neuron (small at fingertips large on the back). In hearing they describe the frequency range a neuron responds to. In smell they relate to the chemical properties of odourants.

Before signals even reach the brain the retina performs compression contrast enhancement and edge detection via its layered circuitry. It discards redundant uniform information and emphasises boundaries making it fundamentally unlike a passive camera.

Retina → LGN (Lateral Geniculate Nucleus) → V1 (Primary Visual Cortex)

The ventral pathway (‘what’ pathway) identifies objects; the dorsal pathway (‘where’ pathway) processes spatial location

A plot of neural firing rate (spikes/sec) against stimulus orientation (degrees), showing the neuron’s preferred orientation at the peak

A smooth bell-shaped curve peaking at the preferred orientation (e.g. 0°) and falling to zero ~±40° away

Hubel and Wiesel, 1968