The eye’s most basic function: light detection
The retina: a computational machine. Its job: to form an economical, precise and sensitive neural representation of the retinal image.
Arrangement of light-sensitive cells; very high pixel density in fovea, but nonuniform distribution; retinal ganglion cells, and their blood supply (why are the retinal vessels not seen?); pressure blinding.
The retina is thick (0.25 mm) but transparent since light must traverse it; many neurons, of many types, arranged in well-defined circuits.
It is a neural net: actually two (inner and outer plexiform layers).
Two kinds of light-sensitive cell: rods (night vision) and cones. The fovea has no rods, only cones. So: the periphery is more sensitive at night, and people with no cones have a central blind spot.
Visual pigment in the outer segment absorbs photons. Initial effect of absorbed photons: to straighten out (isomerize) the bent (regenerated, 11-cis) form of the retinal that forms a part of the molecule of visual pigment that absorbs the photon.
The resulting signal delivered by the photoreceptor: an increase (hyperpolarization) in its membrane potential, and a reduction in its rate of release of the neurotransmitter glutamate.
The hyperpolarization is graded with stimulus intensity..several million charged particles (e.g. sodium ions) are kept out of the photoreceptor cell per photon absorbed.
The amplification cascade: each isomerized molecule inactivates many molecules of the intracellular messenger cGMP; each cGMP molecule assists in opening one channel that can admit many sodium ions.
The retinal signal pathway through the bipolar cells to ganglion cell (=optic nerve). Role of horizontal cells (outer plexiform layer) and of amacrines (inner plexiform layer).
Electrical behavior of these cell types (Werblin and Dowling): receptors are local light detectors; horizontal cells collect from moderate-sized neighborhoods; bipolar cells are driven by contrast between center and surround; amacrine cells respond transiently.
A classic experiment on vision: Hecht, Shlaer and Pirenne (1930s)—(1) how many photons do we need to see? (2) Can individual rods respond to a single photon?
Around 10 absorbed in a patch of 10000 rods is enough
Chance of single hit on any given rod around 0.1%
Chance of double hit on any given rod around .0001%
Chance of double hit on some rod is around 1%
So: we don’t need double hits to see the test flash
Later confirmation with suction electrodes (Baylor and Schnapf)
Why can’t we see just a single photon, instead of needing 10?
False alarms due to spontaneous (thermal) isomerization..one every several minutes per rod. Implied stability of visual pigment: spontaneously isomerizes once in 10 years (in a warm human) or once in centuries (in a cool toad)
Signal detection analysis
Relation to convergence in retinal pathways
Recommended optional reading: Schnapf and Baylor, “How photoreceptors cells respond to light”, Scientific American, April 1987, pp32-38