Vision in detail Archives | Page 2 of 2

Retinal input

Visual input is initially encoded in the retina as a two-dimensional distribution of light intensity, expressed as a function of position, wavelength and time. This retinal image is transferred to the visual cortex where primary sensory cues and, later, inferred attributes, are eventually computed in the process of actualising our perceptions. Parallel processing strategies are employed from the outset to overcome the constraints of the individual ganglion cell’s limited bandwidth and anatomical bottlenecks as data approaches the optic nerves that connect each eye to the visual cortex.

References: DeYoe and Van Essen (1988): https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2771435/

Rods and cones

Both the photosensitive rods and cones form a regularly spaced mosaic of cells across the entirety of the retina – bar the absence of rods in the fovea centralis. Because there are 100 million rod receptors and 20 million cone receptors in each eye, rods are packed more densely per unit area. The synaptic connections of both rods and cones vary in function in different locations across the retina, reflecting the specialisations of different regions. This, for example, allows the eyes to deal with daylight and darkness and with what we see at the centre and periphery of our field of view.

Rods and cones are easily distinguished by their shape, from which they derive their names, the type of photo pigment they contain and by the distinct patterns of synaptic connections with the other neurons around them.

Neurons (nerve cells) are present throughout the human central and peripheral nervous systems and fall into three main categories: sensory, motor and interneurons. Rods and cones are both sensory neurons. Rods don’t produce as sharp an image as cone cells because they share more connections with other types of neurons. But a rod cell is believed to be sensitive enough to respond to a single photon of light whilst cone cells require tens to hundreds of photons to be activated.

The principal task of rod and cone cells alike is photo-transduction. This refers to the type of sensory transduction that takes place in the visual system. It is the process of photo-transduction that enables pigmented chemicals in the rods and cones to sense light and convert it into electrical signals. Many other types of sensory transduction occur elsewhere within the body enabling touch and hearing for example.

References: Functional Specialization of the Rod and Cone Systems: https://www.ncbi.nlm.nih.gov/books/NBK10850/

Trichromatic colour model

Trichromatic colour models (and the trichromatic colour theory that underpins them) provide methods for visually matching and mixing colours from combinations of three primary colours – red, green and blue ( or cyan, magenta and yellow). The information about how much of each primary colour is needed to produce a target colour is stored as tristimulus values. Tristimulus values are simply codes that can be used to record and pass on colour information.

The LMS colour model (long, medium, short), is a trichromatic colour model that represents the response of the three types of cones of the human eye, named for their responsivity (sensitivity) peaks at long, medium, and short wavelengths. It is used to systematize the response of the three types of cones of the human eye to different visual stimuli, that is, different wavelengths of light. The strength of the LMS colour model is its concern for the connection between the physiological aspects of vision and the everyday visual experience of an observer. L M and S refer to the bands of wavelengths that each cone type within the retina responds to.

Let’s look more carefully at this connection between trichromacy and tristimulus systems of which the RGB colour model provides a good example.

We start with the premise that trichromatic processing within the retina reduces all colours an observer sees to responses corresponding with the spectral biases of L, M and S cone cells.

This premise can be demonstrated experimentally by positioning an observer in front of three different light torches, each covered with a red, green or blue filter, that project light onto the same area of a neutrally coloured surface. The effect of each filter is to block all wavelengths except one. If the three torches project light at equal intensities the surface appears white. If the intensity of light or the colours of the filters are not exactly matched a colour cast will be apparent. If one light is turned off, then a secondary colour appears. Depending on which colour is absent the result will be cyan, yellow or magenta.

The reason the surface appears white to the observer in this experiment when all three torches are turned on is that each of the three cone types in their retina is being triggered evenly. This means that each of these types of photosensitive neurons are registering the presence of the wavelengths of light they are tuned to.

In the second part of the demonstration, a calibrated dial is used to alter the intensity of each torch. By setting each dial to a component of a tristimulus value for a known colour, it is possible to test whether the resulting stimulus causes the observer to see the intended colour.

This experiment corresponds directly with the way all RGB devices such as TVs, computer monitors, phone screens and projectors work in so far as tristimulus RGB values are used to stimulate the L, M and S cone cells on the retina to produce the intended experience of colour.

Opponent processing does not play a determining role in this experiment. We know from opponent-processing theory that after trichromatic processing takes place, the signals will be processed based on whether the cone responses indicate that the stimulus is bright or dull, more red-or-green, and at the same time, more blue-or-yellow. The output of this process will be fed into the million-or-so fibres of each optic nerve encoded into two channels of chromatic information and one dealing with the perception of brightness.

Experiments by several generations of scientists and artists have confirmed the connection between trichromacy and tristimulus systems. Opponent-processing cannot be demonstrated quite so directly but visual illusions and unexpected consequences of different attributes of colour perception have been used experimentally to unravel what is going on with extraordinary success.

One of the outcomes of research into tristimulus systems is the requirement, when choosing primary colours, that two of them cannot be combined to produce the third. Each must be unique so far as the human eye is concerned.
Research into the opponent-process has established that there are in fact four unique colours, red, green, blue and yellow, each of which shows no perceptual similarity to any of the others.

The implications of the fact that human vision can be stimulated by three distinct colour inputs are:

In normal conditions, any particular colour seen by an observer is produced by complex patterns of different wavelengths and intensities of light from across the visible spectrum as they enter the eye and are absorbed by cone cells within the retina in real-time.
The complex pattern of wavelengths and intensities of light being emitted by a light source at any moment is called its spectral power distribution. A spectral power distribution can be plotted on a graph and always appears as a wavy line with red at one end and violet at the other. The profile of the plot rises for high and falls for low intensities of light.
The colour notation used to record tristimulus values can, in principle, describe any human colour sensation.
If tristimulus values corresponding with the full range of human observable colours are plotted on a graph, with three axes drawn perpendicular to one another, they can produce an inclusive representation of colour perception in the form of a 3-dimensional colour solid.
The three axes correspond with the range of responses of the three cone types and so can be labelled L, M and S. A scale along each axis can be added to correspond with a minimum cone response at one end and a maximum at the other. This is the basis of the LMS colour model, which is one of a number of colour models devised to quantify human colour vision.
Colour models such as RGB colour and the Munsell colour system also use tristimulus notation to record colour information. The implications are that LMS, RGB and Munsell are all grounded in the trichromatic nature of human vision and take advantage of the resulting opportunities in terms of systems that use additive colour.
Other colour models such as HSB colour, HSV colour and HSL colour which are all variants of RGB colour do not use forms of notation that correspond directly with tristimulus value.

Trichromatic colour vision

Trichromatic colour vision (Trichromacy)

Photo-transduction by cone cell receptors is the physiological basis for trichromatic colour vision in humans. The fact that we see colour is, in the first instance, the result of interactions among the three types of cones, each of which responds with a bias towards its favoured wavelength within the visible spectrum. The result is that the L, M and S cone types respond best to light with long wavelengths (L =biased towards 560 nm), medium wavelengths (M =biased towards 530 nm), and short wavelengths (S = biased towards 420 nm) respectively.

Trichromatic processing

To get some sense of what is going on during the trichromatic processing of visual information collected during phototransduction within rod and cone cells, we need to imagine the light-receiving layer within the retina of each eye as being composed of around 120 million points. Each point is the location of a rod or cone within a mosaic of cells stretching to the boundary of the retina in every direction. Each point also corresponds with a different location within the overall image of the world outside that is projected onto the retina by the lens as light enters the eyes through the cornea, and pupil.

Every rod and cone cell within the human eye is an independent photo-sensitive neuron that responds independently to incoming photons of light. In the first instance, each one carries out its task of responding to the stream of different wavelengths of light without reference to the receptors around it. We must imagine all these neurons as being capable of functioning simultaneously and being able to fire several times a second. Each time a cell responds, phototransduction takes place, which is a chemical response to light that produces electrical impulses ready for transmission.

What precisely happens depends on lighting conditions.

When light levels are low, each photosensitive rod cell may be hit by a handful of photons of a given wavelength and in some cases, this is enough to trigger the chemical response. But at very low levels of lighting, the response of cone cells is very limited. The outcome is a typical night scene that appears blue-purple-black with very little detail.

When light levels are high, each cone cell may be bombarded with tens of thousands of photons of a given wavelength within the briefest fraction of a second, producing a powerful chemical response. In these conditions, rod cells tend to be overwhelmed by the sheer quantity of light but can still provide useful information in peripheral vision. The outcome is a world rich in colour, full of detail and with contrasting highlight-detail and deep well-defined areas of shadow.

The trichromatic process involves the sampling of this mosaic of rod and cone photoreceptors by bipolar and horizontal cells. Individual photoreceptors are connected in small groups to bipolar cells that receive their electrical output. The bipolar cells compare the response of receptors with the response of neighbours, whilst horizontal cells help to aggregate the result and encode it into three independent data channels.

The comparison process involves what is termed centre-surround antagonism. This refers to the way bipolar neurons organize their receptive fields. A close-up view reveals that centre-surround antagonism relies on small groups of cells being arranged around a centre point where one rod or cone synapses on the dendrites of a bipolar cell. Around this centre are other photoreceptors that also synapse onto other dendrites of the same bipolar cell. The signal received from the centre is compared with a summation of the signals from neighbours to establish to what extent they agree or disagree. This process goes on in real-time as bipolar cells receive successions of signals and horizontal cells modulate the information to improve fidelity. The scale of this enterprise as it takes place across the surfaces of the retina in each eye and in real-time is extraordinary.

The role of horizontal cells in both trichromatic and centre-surround antagonism is best conceived as signal conditioning and particularly with globally adjusting visual information ready for opponent processing. However, ongoing research suggests that horizontal cells may also be involved in an early stage of signal interpretation and so contribute towards things such as the detection of edges and movement.

Trichromatic colour vision is governed by the fact that there are three distinct types of cone cells within the retina, each tuned to respond to a different band of wavelengths of light. As light floods in, each type of cone outputs a signal if it picks up the presence of photons with wavelengths it recognises. Every identifiable point on the retina contains all three types of cone cells and these are tightly packed into a random mosaic pattern across the entire surface. The result is a photosensitive film containing millions of receptors capable of responding to all wavelengths across the visible spectrum.

General descriptions of trichromatic vision often suggest that the three types of cone cells in the human retina are responsive to wavelengths corresponding to red, green and blue. It is more accurate to say that the peak sensitivity of these L, M and S cones types respond with biases towards different regions of the visible spectrum and have a loose correspondence with red, green and blue:

L cones: Respond to long wavelengths (peak sensitivity around 560 – 580 nanometres) so with a region of sensitivity that includes red, orange, green and yellow but with a peak bias between red and yellow.
M cones: Respond to medium wavelengths (peak sensitivity around 530 – 545 nanometres) so with a region of sensitivity that includes orange, green, yellow and cyan but with a peak bias between yellow and green.
S cones: Respond to short wavelengths (peak sensitivity around 420 – 440 nanometres) so with a region of sensitivity that includes cyan, blue and violet but with a peak bias between blue and violet.
Rods: Rod cells are most sensitive to wavelengths around 498 nanometres, so with a peak sensitivity towards green-blue, and are insensitive to wavelengths longer than about 640 nanometres (red).

This arrangement suggests that:

At any specific moment, all three types of cone cells at any specific location on the retina may fire multiple times per second in response to streams of photons that have constantly changing wavelengths.
The assessment of the distribution of wavelengths by every individual cone depends on both its range of sensitivity and the wavelength at which its response peaks.
Centre-surround comparisons check for consistencies and variations within the responses of cone groupings at every location.
As soon as a clean and noise-free consensus is achieved, the data specific to each cone type is output on one of three separate data channels.
Once the composite of trichromatic and centre-surround encoding of colour information is complete it is sent onward for opponent processing within the retina.
Recent research suggests that the three output channels do not correspond directly with the L, M and S cone types. There are indications that two channels contain chromatic information whilst the other contains achromatic data.

Trivariance

The term trivariance is used to refer to this first stage of the trichromatic process. It refers to both the phototransductive response of the cone cells themselves and to the three separate channels used to convey their colour information forward to subsequent levels of neural processing.

Each channel conveys information about the response of one cone-type to both the wavelength of the incoming light it is tuned to and to its intensity. In both physiological and neurological terms this process is exclusively concerned with trivariance – three discernible differences in the overall composition of light entering the eye.

It is the separation of the signals produced on each channel that accounts for the ability of our eyes to respond to stimuli produced by additive mixtures of wavelengths corresponding with red, green and blue primary colours. But more of that later!

By way of summary, the rod and trivariant cone systems are composed of photoreceptors with connections to other cell types within the retina. Both specialize in different aspects of vision. The rod system is extremely sensitive to light but has a low spatial resolution. Conversely, the cone system is designed to function in stronger light. As a result, cones are relatively insensitive compared with rods but have a very high spatial resolution. It is this specialisation that results in the extraordinary detail, resolution and clarity of human vision.

Rod System	Cone System

High sensitivity, specialized for night vision	Lower sensitivity specialized for day vision
Saturate in daylight	Saturate only in intense light
Achromatic	Chromatic, mediate colour vision
Low acuity	High acuity
Not present in the central fovea	Concentrated in the central fovea
Present in larger number than cones	Present in smaller number than rods

Caption

Visual processing

Visual processing is a complex and dynamic process that involves interactions between various retinal cells, neural pathways, and brain regions, ultimately leading to conscious visual perception.

Visual processing begins the moment light enters the human eye. It then progresses through multiple stages as signals travel towards the visual cortex, where the neural activity is integrated, resulting in conscious visual experience.

As visual processing begins the retina starts to process information about colors, as well as basic information about the shape and movement associated with those colors. By the end of this stage, multiple forms of information about a visual scene are ready to be conveyed to higher brain regions.

Let’s examine two major forms of processing, trichromatic and opponent-processing, which occur within the eyeball as visual information is gathered from light entering our eyes.

Trichromacy, also known as the trichromatic theory of colour vision, explains how three types of cone receptors in the retina work together with bipolar cells to perform their role in the initial stage of colour processing. Rod cells also play a significant role in this form of processing visual information, particularly in low-light conditions.

Opponent-processing, also known as the opponent-process theory of colour vision, explains the second form of processing. Opponent-processing involves ganglion cells that process the data received from trichromatic processing and combine it with other intercellular activities.

It is interesting to note that as both trichromatic and opponent-process theories developed over the last century, researchers and authors have often pitted one theory against the other. However, both processes are crucial for understanding how colour vision occurs.

Trichromatic theory explains the encoding of visual information when light hits the retina, while opponent-processing explains a subsequent stage of information convergence, assembly, and coding before the data leaves the retina via the optic nerve.

Note that:

Both trichromatic and opponent-processing occur independently within each retina, without comparing with the other.
Each eye gathers information from a specific viewpoint, approximately 50 mm to the left or right of the nose.
The two impressions are later compared and combined to provide us with a single three-dimensional, stereoscopic view of the world, rather than two flattened images.

We can consider the layers of retinal cells involved in trichromatic and opponent-processing as examining, interpreting, and transmitting visually relevant information. However, it would be incorrect to view this as a straightforward linear process due to the intricate neural networking, cross-referencing, and feedback loops within the retina.