Visual Processing As Described By Contemporary Main-Stream Neuroscience

6.1 Retina

Eye. The light enters the eye at cornea. The optic flow then proceeds through the pupil, a circular hole with muscle-conditioned variable radius, then through the lens, which bends (refracts) and focuses (because of ciliary-muscle-conditioned flexibility) the rays, to the eye’s back where it impinges on the pigment-layer of photoreceptor cells¹ on the surface of the retina. A 2-dimensional projection of the 3-dimensional world emerges on retina.²

Retinal image. Stimulus intensities, which affect the activities of retinal cells, are mainly determined by: the light source, the surface reflectance and shape of the viewed object, the viewer’s standpoint and angle of view. Each location in the image can be described by two primary characteristics of the wave-modulated stimulus (Berne & Levy, 1993):

• 1.

  brightness (luminance), the intensity of light falling on retina, which determines the photoreceptors’ activation-level;

• 2.

  spectral decomposition (frequency or wavelength) which determines the differential activation of different cone types, leading to the perceived color.³

Derived information. Let us mention here that other characteristics can be in later stages (up to the cortex) derived from the primary retinal features 1 and 2. Such secondary characteristics are (De Yoe & Van Essen, 1988):

• 3.

  2-dimensional velocity of the retinal image;

• 4.

  spatial contrast;

• 5.

  2-dimensional orientation of the stimulus;

• 6.

  binocular disparity (which is also a source of stereopsis, i.e. 3-dimensional vision).⁴

Sensation. Photoreceptor cells are cones, which are color-sensitive and used for low-sensitive (for daylight), high-acuity vision, and rods, which are high-sensitive (for dark), “color-blind” and provide a rough “sketch” only. In the central part of retina, vis-a-vis to the lens-center (on the visual axis), lies fovea where visual acuity is the highest, because there is extremely high density of cones. Density of rods is, in contrast, high at all other parts of the retina except at fovea and at the location where the optic nerve starts to extend towards the visual cortex. In response to a flash of light within the receptive field⁵, a photoreceptor runs a biochemical cascade process and changes voltage. Over a limited range of stimulus intensity, the change in the light-modulated signal is proportional to the change in the light stimulus (Smith in Arbib, 1995). Retina realizes a sort of scale-invariant gaussian-profile sampling (details in Van Essen & Anderson in Zornetzer, Davis & Lau, 1990, pp. 45-47).

Retinal net processing. Retina is a multi-layer neural network. Details of its non-trivial hierarchical feed-forward⁶ processing, in spite of its importance for further visual processing, will not be presented here. However, description of processing has some remarkable similarities to those presented in Auxil. Chapter 11 for competitive cortical circuits even taking into account that the retinal net has only very partial feedback. Retina shares roughly several characteristics with later stages of visual processing (e.g., LGN), like the circular center-surround and similar receptive fields (Figure 1).

Figure 1.

Circularly-symmetrical receptive fields of typical LGN neurons. Left: A parvocellular neuron is excited (+) in the small centre by red light (R) and inhibited (–) in its surround by green light (G). / Right: A magnocellular neuron is excited by all wavelengths in the centre (R+G+B) and inhibited by all wavelengths in its surround (–R–G–B).¹⁸ (Redrawn from Livingstone & Hubel, 1988, p. 741)