Figure 1. Anatomy of the visual system. Light arrives at the eye and is focused by the lens on to the retina, where photoreceptors transduce the light into electrical signals that are processed by local retinal neurons. Axons of retinal ganglion cells, the output cells of the retina, leave the retina in a bundle called the optic nerve. At the optic chiasm, some axons cross over, so that axons representing the right half of visual space travel to the left lateral geniculate nucleus (LGN) and axons representing the left half of visual space travel to the right LGN (not shown). In the LGN, the axons segregate into layers according to eye of origin and other properties. LGN relay cell axons form a band called the optic radiations and project to the primary visual cortex, where LGN axons representing each eye ramify in an alternating fashion.

Figure 2. (a) The major cell types in the retina and their laminar organization. PRL=photoreceptor layer; ONL=outer nuclear layer; OPL=outer plexiform layer; INL=inner nuclear layer; IPL=inner plexiform layer; GCL=ganglion cell layer. (b) A centre–surround retinal ganglion cell that responds to light in the centre of its receptive field and is inhibited by light in the surrounding region. The stimulus is shown on the left, and action potentials in the cell relative to the onset and offset of the stimulus are shown on the right. Note that the cell responds most vigorously to a light spot in the centre surrounded by a dark annulus (second from the top), but the cell responds much less vigorously when stimulated by a large white spot (third from the top) because of the inhibitory surround. These properties are often denoted symbolically with the notation at bottom, with ‘+’ indicating a preference for more light relative to background and ‘–’ indicating less light. (c) Schematic diagram of retinal circuitry that mediates the centre–surround cell depicted in (b). Light in the centre hyperpolarizes a cone, which excites a bipolar cell, which in turn excites the retinal ganglion cell. Horizontal cells mediate the effect of the surround, providing inhibition to the bipolar cell in the centre when there is light in the surround. This figure is adapted from Werblin and Dowling (1969), who studied the salamander Necturus maculosus, and similar circuitry has been found in other vertebrates.

Figure 3. The lateral geniculate nucleus of the rhesus monkey. Midget cells of the retina innervate layers 3–6, and the parasol cells innervate layers 1 and 2. Koniocellular cells innervate K1–K6 but are also diffusely present throughout the entire LGN.

Figure 4. (a) Many neurons in the primary visual cortex respond to bars or edges at a particular orientation. The stimulus is shown at the left, and action potentials in the cell relative to the onset and offset of the stimulus are shown at the right. The neuron in (a) responds to a bar rotated 45° clockwise from vertical, but responds poorly to bars with other orientations. (b) Simple cells have separate regions of their receptive fields that respond to light increments and light decrements and thus respond to bars at specific locations. One example of such a receptive field pattern is shown. (c) In contrast to simple cells, complex cells respond to a properly oriented bar anywhere in their receptive fields. (d) Many cells in V1 respond to moving oriented bars. The arrows in the stimulus (left) indicate direction of bar movement. Most cells in V1 are orientation-selective and not direction-selective, responding to movement in both directions (upper), but some cells are direction-selective and only respond to bars moving in a particular direction (lower).

Figure 5. Hubel and Wiesel's model for the formation of simple and complex receptive field properties. (a) A simple cell (at right) that responds to a dark, oriented bar on a light background could receive input from several adjacent centre–surround cells that respond to light decrements in their receptive field centres. (b) A complex cell (at right) could obtain its indifference to bar position by receiving input from several adjacent simple cells (at left) sharing one orientation preference. The complex cell's receptive field properties cannot be represented in the same form as the schematics for the LGN cells and simple cell in (a) and the simple cells in (b). It responds to a properly oriented bar at any position in its receptive field.

Figure 6. Selected synaptic connections in the primary visual cortex. In each ocular dominance band, parvocellular LGN neurons from one eye project to cortical layer 4B, magnocellular LGN neurons project to layer 4A, and koniocellular LGN neurons project to the cytochrome oxidase blobs in layers 1–3. Neurons in layer 4 make connections with neurons in layers 2 and 3, both in the blobs and between the blobs, and these cells in turn project to layers 5 and 6. Cells in layers 2, 3 and 5 make connections with adjacent cortical areas, and cells in layer 5 also make connections with subcortical structures such as the superior colliculus. Finally, cells in layer 6 project back to the lateral geniculate nucleus. Adapted from Casagrande and Kaas (1994).

Figure 7. Horizontal organization of the visual cortex. (a) The topographic projection of visual space in the right visual hemifield on to an idealized, unfolded primary visual cortex (adapted from Hubel (1995) ). Note the large representation of the central region in V1, and the relatively small representation of the periphery. Within this topographic map is an alternating map of input from the two eyes. (b) A small section of V1 imaged optically using voltage-sensitive dyes by Blasdel and colleagues. Regions that respond to visual stimulation of the left eye are coloured black, while regions that respond to stimulation of the right eye are white. Woven into the topographic map and ocular dominance bands is a semi-regular map of orientation preference. (c) The same area of cortex as (b), except that the eyes are being stimulated with bars of different orientation. Each pixel in the image is colour-coded according to the bar orientation that evoked the largest response (see scale at right). For example, red regions in the image showed greatest activation by horizontal bars. (b) and (c) are reproduced with permission from Blasdel and Salama (1986).