The Guess Worker



  • Signals from dopaminergic neurons allow the brain to see colours

  • Each wavelength of light triggers the firing of a unique combination of dopaminergic neurons

  • Consciousness creates different colours by interpreting these combinations

  • Consciousness uses combinations to generate huge numbers of experiences

Our eyes don't see colours. Our consciousnesses do that. The eyes merely send signals about colours to the brain. The eyes send only one kind of signal - nerve impulses. But although there is just one kind of signal, we can see much more than just one colour. Our consciousnesses can distinguish many hundreds of thousands of different colours.*1 How does this happen?

Absorbing information

To answer this question we have to know a few things about the eye. The first point at which the eye gathers information about colour is the retina. In the retina there are colour sensitive sensory neurons called cone cells which contain pigments that absorb specific wavelengths of light.

Most human retinas have three types of cone cells, each of which is sensitive to a different range of wavelengths. Because of their wavelength sensitivities they are sometimes called “blue”, “green” and “red” cone cells or, more accurately, as “S”, “M” and “L” cone cells.*2 The graph below shows the change in sensitivity with wavelength for the three types of cone cell :


As we can see, the S cones are sensitive to wavelengths in the range 390 to 530nm, with maximum sensitivity at 442nm. M cones are sensitive in the range 410 to 650 nm, with maximum sensitivity at 546nm, and L cones are sensitive in the range 410 to 700nm, with maximum sensitivity at 570nm.

Dem cones

When light of the appropriate wavelength is absorbed by a cone, the cone responds by passing on a signal to a chain of neurons, linked by synapses, which leads into the brain. The following diagrams show how these neurons are connected:


Fig 2.

So - as we can see from the diagram - cone cells are connected to bipolar cells. The bipolar cells are connected to ganglion cells. The ganglion cells project through the optic nerve, out of the eye and into various regions of the brain.*3

No one knows how the brain processes information from the eye but we can be almost sure of one thing: either directly or indirectly, the ganglion cells are connected to dopaminergic neurons. This connection gives us our conscious experience of colour.*4

Incorrect assumptions, correct conclusions

Before considering how the brain converts impulses into colour, I'm going to simplify - and indeed distort – much of what scientists have discovered. Even though this simplification is incorrect, it shouldn't affect the accuracy of my conclusions.*5

First, let's imagine that when light activates one cone cell it causes one bipolar cell to send one impulse to one ganglion cell and, in turn, this ganglion cell sends one impulse to one dopaminergic neuron.*6 In other words, every time a single cone cell is activated a single dopaminergic neuron is fired.

Second, let's imagine that there are equal numbers of each type of cone cell evenly distributed in the parts of the retina responsible for colour.*7 In humans the number of cone cells in the fovea, which is the small area of the retina where light is focussed, is around 200,000. So I'll assume there are 70,000 of each of the three types of cone cell in the fovea.*8

Seeing as one dopaminergic neuron is linked to one cone cell, that gives a total of 70,000 dopaminergic neurons for each cone cell type. To show which type of cone cell the dopaminergic neurons are linked to, I'll label them with the letters S, M or L. That means there are 70,000 S dopaminergic neurons, 70,000 M dopaminergic neurons and 70,000 L dopaminergic neurons each linked to a cone cell in the fovea.

Proportional response

How do these cone cells respond when light shines on them? If the light has a single wavelength of 442nm, of the three types, the S cones will be activated the most. At this wavelength the s cones have their maximum sensitivity and so we could expect that within a very short space of time – let's say, 50ms – all the s cones in the fovea will be activated.

What, though, if the wavelength of light was 419nm or 410nm? At 419nm the S cones are half as sensitive as they are at their maximum. That means that within 50ms, 50% of the cones are activated. And at 410nm, 25% of the S cones are activated:


Because I have assumed that one dopaminergic neuron is linked to one cone cell, the number of dopaminergic neurons fired is the same as the number of cone cells activated at each wavelength. Therefore at 442nm 70,000 S dopaminergic neurons fire, at 419nm 35,000 fire and at 410nm 17,500 fire. Because the activation time is very short, it's not too far fetched to say that at each of these wavelengths these numbers of dopaminergic neurons fire simultaneously.

Seeing blue

So, different wavelengths cause different numbers of dopaminergic neurons to fire simultaneously. Could it be that consciousness uses the number of these signals to create colours? Could it be that if 70,000 S dopaminergic neurons fire, we see the blue of 430nm, if 35,000 fire we see the blue of 420nm, and if 17,500 fire we see the blue of 415nm?

Maybe. But there is a problem. 35,000 S dopaminergic neurons are fired at two distinct wavelengths. In fact, because the sensitivity curve is a parabola, the same number of neurons can always be fired by two distinct wavelengths (except at the maximum), as we can see below:


Colour combinations

How, then, could consciousness distinguish between the wavelengths 419nm and 475nm, or 410nm and 488nm? Even though the number of firing S dopaminergic neurons is the same for these pairs of wavelengths, what does differ is the number of firing M and L dopaminergic neurons. In the diagram below we can see the percentages of the three types of cone cells which fire at various wavelengths:


From this diagram we get the following table:


As we can see from the table each wavelength causes different numbers of S, M and L dopaminergic neurons to fire. The combination of these numbers for any given wavelength is unique. The uniqueness of each combination should give consciousness enough information to produce a wide range of colours.

But are there enough combinations available to account for the hundreds of thousands of colours we can distinguish? Yes, there are. With 70,000 of each of the three types of dopaminergic neuron, 70,0003 = 3.34 x 1014 combinations are possible. This number is hundreds of millions of times larger than the number of colours we can see. So each colour could be generated not by just one combination but by many millions of combinations.*9

Extra mechanism

I have suggested before that consciousness creates experiences based on the location of dopaminergic neurons.*10 So if a dopaminergic neuron connected to the visual cortex sends a signal, consciousness creates the sensation of sight. If, on the other hand, a dopaminergic neuron connected to the auditory cortex sends a signal, consciousness creates the sensation of hearing. Even though the two signals are physically identical, because of their different locations, consciousness is able to create two very diverse experiences.

Colour vision seems to show that consciousness has a yet another mechanism for producing experiences: through combinations of firing dopaminergic neurons.*11 Consciousness, though, is unlikely to use combinations solely for the experience of colour. Indeed, probably the majority of our experiences are derived from combinations – and probably very few are formed from information from single location in the brain. If consciousness uses a particular mechanism to produce one kind of experience, it is hard to imagine why consciousness would not use the same mechanism for other experiences.

Widespread application

On top of that, the number, variety and subtlety of our experiences seem to point to consciousness's wide use of combinations. Vision as a whole, for example, must depend on them. While observing an object our brains have to take account of factors such as distance, shape and brightness. Each of these factors probably trigger different sets of dopaminergic neurons which when combined give us a complete picture of the object.

The same is likely to be true of taste, too. What makes our experience of taste interesting is that it depends not only on taste but also on texture and smell. That means dopaminergic neurons linked to very different types of sensory neurons are combined to give the various experiences of taste.

Many of our emotions could well be caused by combinations too. Envy, let's say, may be caused by a combination of three types dopaminergic neurons each of which alone would give rise to a different feeling. These feeling might be: 1) a desire to possess 2) an anger from not being able to possess and 3) a realisation that someone else possesses what is desired.

However, we don't experience three feelings, just one: envy. Perhaps combinations like these explain why we have so many emotions and also why scientists have such difficulty classifying them.

Mega experiences

Combinations give an organism two advantages. The first is fairly obvious. Because a huge number of combinations is possible, an organism can have a huge number of experiences. The more experiences there are, the more refined the interpretation of incoming information can be.

If, let's say, consciousness receives a signal from a dopaminergic neuron which is fired by an S cone cell, consciousness is able to interpret this information as coming from light of a short wavelength. But that's as much information as the location of this dopaminergic neuron can deliver by itself. Combinations, on the other hand, give enough information to allow us to experience thousands of hues of blue.

Hidden benefit

The other advantage of combinations is perhaps less obvious. Having a vast number of potential, but as yet unused, combinations means an organism always has a chance of having new experiences.

During your life you are unlikely to see every one of the hundreds of thousands of colours your brain is capable of identifying. But if you do come across a hue you have never seen before, not only can you experience the new colour, but you can also distinguish it from other colours you have seen.

Similarly, you may eat an exotic fruit for the first time and experience an extraordinary new taste. Or perhaps you'll first feel what it's like flying through the air when you take up parachuting. Or perhaps you'll feel powerful sense of solidarity that you've never felt before when you join a protest march.

What is the benefit of having new experiences like these? By recognising new experiences we sense that situations around us are unfamiliar. Knowing that a situation is unfamiliar can help us react appropriately.

New experiences also have an additional benefit for humans in particular: they give us the ability to understand new concepts. Every new concept a person learns or discovers is a new experience. Without a capacity to create new experiences, we wouldn't be able to appreciate new concepts. And seeing as our brains can make unimaginably large numbers of combinations, the number of new concepts we can form is almost unlimited.

Constructive conclusion?

The mechanisms by which consciousness converts information about wavelength into colour can give us clues as to how consciousness converts information in general into experiences. In this post I have suggested that new concepts are new experiences. I have also come to the conclusion that consciousness uses both location and combinations to create new experiences. So could I use this conclusion to model how new concepts are built up from other concepts? Perhaps. We shall see in the next post.





*1. The exact number seems to be a mystery. One figure which is banded about in the literature is 10 million colours which is referenced as coming from a book written in the 1970's – Color in Business, Science and Industry by D.B. Judd and G. Wyszecki. Another often quoted figure is 1 million which is calculated based on the hypothesis that each cone cell can register 100 different colour shades. (A hypothesis which is at odds with the one in this post.) As far I as I can determine, none of the figures are a direct result of experimentation. In any case, as anyone knows who has tried to retouch paint on a wall, the number is very large

*2. The labels “blue”, “green” and “red” are not accurate because these cone cells are activated even by wavelengths which produce other colours. “S”, “M” and “L” stand for “short”, “medium” and “long” wavelengths respectively.

*3 I have left out two important kinds of neurons from this scheme: amacrine cells and horizontal cells. Amacrine cells interconnect bipolar cells while horizontal cells interconnect rods and cones. Both types of neuron send their signals laterally, that is between the neurons of the retina and not on to the brain. Their role is to either intensify or inhibit signals from the neurons they are connected to. They are thought to compensate for brightness and to sharpen contrasts. They probably have other functions, too, not related to colour vision.

*4. I have discussed in depth in previous posts the idea that dopaminergic neurons send information to the consciousness. (See Connections, Memory, Time Bridges, and Dopamine.) I believe pain and pleasure are conveyed to the consciousness by respectively falls and rises of dopamine in the synapses of dopaminergic neurons. With regards to colour, there are two undeniable facts: 1) we are aware of colours in the consciousness and 2) we find colours pleasurable. Therefore, in my view, information from ganglion cells about colour must be passed on to dopaminergic neurons.

There is some evidence to support this view. Dopamine in the prefrontal cortex of the brain has been shown to affect visual signals. (Control of visual cortical signals by prefrontal dopamine, Noudoost B., Moore T., Nature 474, 16 June 2011.) Experiments have also shown that retinal ganglion cells relay information to dopaminergic neurons via a region of the midbrain. (Cortical regulation of dopaminergic neurons: the role of midbrain superior colliculus, Bertram C., Dahan L. et al, Journal of Neurophysiology 2014 Feb 15 111(4) 755-767).

*5. By omitting amacrine cells and horizontal cells (see *3 above), by altering the ratio of neurons (see *6 below) and the numbers of cone cells (see *7 below) my simplification changes the proportions of signals coming from each cone type. But changing these proportions cannot affect the mechanisms the brain uses to interpret these signals.

*6. On average there are 100 cone cells for each ganglion cell. However this ratio varies greatly with position. In the fovea a single ganglion cell can be connect to a few as five cone cells, whereas in the periphery of the retina each ganglion cell can receive inputs from many thousands of cone cells. The ratios are also different for each cone cell type. In the macaque monkey, for example, in the central region of the fovea, on average each M and L cone connects to two ganglion cells and each S cone cell connects to just one ganglion cell. (See Anatomy of macaque fovea and spatial densities of neurons in foveal representation, Stein S.J., J Comp Neurol. 1988 Mar 22;269(4):479-505. )

*7. In reality, there is a 200 times greater density of cone cells in the fovea than elsewhere in the retina.

*8. The actual proportions of each cone cell are as follows: S cones 2%, M cones 32% and L cones 64%. There are no S cones at all in the central region of the fovea, but they rise in density in the area around the centre to about 12% of the total number. The low number of S cone cells suggests the eye may in some way amplify the S cone signals – perhaps through the increased numbers of ganglion cells communicating with each cell. (See *6 above.)

*9. You might wonder why consciousness would need many millions of combinations to produce just one colour. Isn't this unnecessary and wasteful? It is, and, on the contrary, the eye is probably very economical in its use of cone cells. Where has my calculation gone wrong? In my simplification I have assumed that all of the cone cells code for just one colour. But of course in reality we don't see only one colour, we can see many colours at once. Cone cells are therefore probably organised into groups so that small areas of the retina can signal for colour separately. To be able to see one million colours, such a group of cone cells would have to be made up of at least 100 of each type of cone cell.

*10. See Memory, footnote *2

*11. Scientists generally believe that colour vision is a product of two processes working together. One of these processes is described by the Trichromatic Theory and the other by the Opponent Process Theory. According to the first theory, signals from the three types of cones are combined and interpreted to create colours. According to the second, signals from pairs of cone cells sensitive to different wavelengths inhibit each other. In this last theory colours are registered by the differences in responses of cone cells to light.

The model suggested in this post is close to the Trichromatic Theory (but is based on the combinations of numbers of dopaminergic rather than on the strength of the response of an individual cone cell). Clearly the Opponent Processs Theory is not needed to explain colour vision – at least not at the stage in the brain when colours are actually created. This does not necessarily mean, however, that opponent processes are not happening in the eye.


Comments powered by CComment