I’m just reading Dowling’s “Understanding the Brain : From cells, to behaviour, to cognition”. Sometimes I’m surprised by what I’ve forgotten, and sometimes he surprises with facts I hadn’t been aware of – so just in case you hadn’t heard of them either, I’ll mention a few here as I come across them – and paraphrase portions of the book for their relevance to vision therapy and reconstruction 🙂
It’s a great book : so why not buy a copy !
I shall paraphrase in regular test, and add my comments/thoughts in italics.
Recent genetic studies have shown that there may be as many as 86 billion neuronal genetic subtypes. Couple this to the fact that basically no new neurons are created in the brain after the first year of life, and given that there are 86 billion neuronal cells in the brain, this implies that each and every neuron may be inheriting its specific function. My take on this? Coupled with epigenetic processes, it implies that an extremely fine-grained transfer of structure may be possible. Interestingly, the only area of the human brain to create new neurons is the hippocampus (associated with memory) so we may not be transferring memories through the generations, but perhaps we are transferring a far more finely grained set of psychological potentials and likelihoods than has hitherto been recognised.
Clusters of neurons are called nucleii. These are clusters of short axon (association) neurons which are connected to other clusters via long axons. This has long been a hypothesis of mine, that in computing we ought to be creating highly specific “modules” (clusters/nucleii) and connecting them via genetic algorithm approaches to create emergent behaviour. This would allow learning networks to become more robust, simple to understand, and more able to learn around new classes of problem. Nucleii are typically clustered or layered (as in the retina).
We know that the visual system is organised in processing pathways of increasing abstraction. Dowling points out that : “severe damage to the primary visual area (V1) brings about complete and permanent loss of visual perception related to that part of the visual field served by that portion of the cortex. Yet similar damage to area V5 may lead primarily to a loss of movement perception; other aspects of vision are maintained. It is particularly interesting, in the case of amblyopia, to consider the situation in which multiple structures are simply inactive, not having been wired into their surrounding environment: I’m thinking specifically of structures for fusion, depth perception, and coordinated eye movement.
A generalization emerging is that more recently evolved parts of the brain, those concerned with higher neural functioning, are more flexible or plastic than are older parts of the brain. This plasticity is reflected in the ability of a part of the brain to reorganize itself after damage and to recover function. This type of brain plasticity is being specifically and often studied in relation to amblyopia – and it is being shown that higher level visual structures are particularly plastic. This is the basis for the hypothesis, that given access to adequate training environments, the brain can effectively rehabilitate binocular vision in many cases.
The section on how the brain emerges from the neural plate as mesodermal cells in generated in the second week of an embryo’s development cause the ectodermal cells on the outside of the embryo (cells which normally turn into skin) to differentiate into neural cells through a process of bootstrapping from 2D into 3D is just awesome.
Fascinatingly, virtually all neurons are generated in the first 4 months of gestation… but the brain continues to grow in volume until the age of 20, after which is steadily declines in mass. The factor driving this growth is the development and elaboration of the neurons themselves. Their cell bodies increase in size as they extend new dendritic branches and form new synapses. We may not grow new neural cells, but we are very very busy thickening the bush of connections.
Having understood that kidney failure kills, not because of increased levels of Urea, but of increased levels of potassium ions (K+) affecting neurological control (causing uncoordinated and uncontrolled twitching of the heart muscle) this became the standard way to kill people through death by lethal injection. Go science?!?
The process of information flow alternates between receptor/synaptic potentials and action potentials. In the first form, modulation is by amplitude. The stimulus at the receptor or postsynaptic membrane causes an opening of channels which admit positively charged ions (Na+). This gradually increases the internal voltage of the synaptic/receptor cell from its resting state of about -70mv.
The transmission of information, however, along the axon involves a conversion from amplitude modulation to frequency modulation. The mechanism is quite straight forward at first sight : beyond a given threshold (around -50mv) the cell becomes depolarized, leading to a cascading opening of Na+ channels to cause a voltage spike around +50mv as Na+ rushes in. This is synchronised with more slowly opening channels which allow K+ (which leaves rapidly) to rush out, causing the spike to normalise and the cell to repolarize (although, I am not clear on when the cell re-accumulates K+ to be prepared for the next spike). Thus, the amplitude of the receptor/synaptic potential will alter the time required to reach depolarization -> hence higher amplitudes trigger a higher rate of frequency modulated firing along the axon then lower amplitudes (in the case of excitatory cells -there are also inhibitory cells which do the opposite).
The signal travels down the axon (at a speed of between 100 and 200 mpg in vertebrates) as each action potential opens adjacent Na+ channels on the axon, causing a transmission of the signal. It is unidirectional as it takes time for activated gates to be ready to re-fire, by which point the “wave” of electrical propagation is far enough along the axon to not cause a re-triggering. There are two specific types of axon, those which are myelinated, an those which are not. Myelinated axons transmit signals particularly quickly but if the myelin is damaged, it can lead to illnesses such as multiple sclerosis (the reason myelin is damaged appears to be caused by autoimmune problems).
It’s interesting to note that myelinated axons transmit signals further, faster, and with smaller cross-sections than non-myelinated axons, as the core electrical transmission factor is the difference in electrical resistance between the membrane of the axon and its internal resistance. In invertebrates, internal resistance may be reduced by increasing the axon cross-section size, whilst in vertebrates, myelination (the wrapping of glial cell membranes around the axon) increases the resistance of the membrane relative to its insides.
Excitatory synapses are most commonly found on dendrites, while Inhibitory synapses are found on the cell body. Inhibitory synapses typically transfer Cl- which moves the neuron as a whole away from a sufficiently positive charge to pass the threshold required for an depolarization.
It is worth noting that neurotransmitters are stored and released from vesicles within synapses. In response to the increased positive charge of the synapse as it is transmitted via the axon, the vesicles bind to the synapse membrane, where the neurotransmitter is released, flowing to channels on the postsynaptic membrane, thus causing them to open. These neurotransmitters are then broken down by enzymes, or taken back up for reuse by transporter proteins which bring them back into the synapse. This must happen quickly for the synapse to be ready for the next firing, although some drugs act to inhibit transporters, and thus effectively increase the amount of neurotransmitter released at the synapse thus creating psychological effects.
Other drugs affecting (preventing) the release of neurotransmitters (by preventing vesicles from connecting to the synapse membrane) include the Botulinum toxin – a toxin that thrives and multiplies under conditions of low oxygen and is commonly found in soil and on fruits. This was a problem in the early days of canning, when fruits and vegetables were insufficiently heated to kill the bacterium.
Myasthenia gravis, a disease of the synapses between nerve and muscle, is characterized in its early stages by muscle weakness and fatigue. Diplopia (double vision), due due to a weakness of the ocular muscles, is another early symptom, along with droopy eyelids.
As well as synaptic excitation and inhibition at the synapse, there is neuromodulation – a process related with remembering and learning amongst other things. Neuromodulation is related to the relatively rare situation in which the synapses of two neurons are so closely connected that a protein channel forms to transfer ions directly from one neuron into the next.
Chemical synapses may involve dozens of different neurotransmitters, creating the possibility to inhibit or excite the same neuron depending on whether the neurotransmitter is +ve or -ve, and potentially even altering behaviour across entire regions of neurons in a similar manner. Chemical synapses are also very quickly acting. A synapse may be activated and result in a voltage change within half a millisecond.
Neuromodulation are more “meta”. Neuromodulators are released at synapses and interact with membrance proteins called receptors, which are linked to intracellular enzyme systems. These receptors mediate the behaviour of the cell (switching on/off behaviours via phosphorylation), from properties of the channel proteins in the membrane, to the expression of genes in the nucleus. Neuromodulation is a meta-programming mechanism to dynamically alter the behaviour of the cell’s role in signalling.
Some substances released at synapses are exclusively neurotransmitters or neuromodulators, whereas most actually affect both. Their role depends on whether the postsynaptic membrane has appropriate channels or receptors. Different drugs may be designed to target the receiving platform, thereby limiting the role of a particular messenger to being a neurotransmitter, or neuromodulator. The paramaterisation of a neuron through phosphorylation is incredible : from modifying its sensitivity to a neurotransmitter, the length of time a channel remains open after activation by a neurotransmitter, or even its ionic specificity. These represent a range of features desperately lacking in computing based models of neural networks- which are so over simplified and lacking in this ability to “dynamically reprogram themselves” that it is no wonder another AI Winter is coming. Not mentioned here are “second messengers” – which quite frankly, make neurons pretty much a complete turing capable machine in their own right.
Photoreceptor cells are examples of membrane receptors linked to intracellular enzyme systems. e.g. altering the behaviour of the photoreceptors to respond to different light levels over time? Is it similar at the end of the optical nerve? Is it purely a chemical inhibition, or is there a neuromodulator in play when it comes to eye dominance?
With relation to amblyopia : Two … amino acids, glycine and y-aminobutyric acid (GABA) are the major inhibitory neurotransmitters in the brain. GABA inhibitors are used to block the end of the optical nerve from retransmitting signals further into the brain. GABA can be readily created from glutamate (the brain’s major excitatory neurotransmitter) by chopping off a carbon and two oxygen atoms.
GABA and glycine mainly activate channels that allow Cl- to flow into neurons. This results in inhibition of the neurons. The GABA-activated Cl- channel is of particular interest because its properties are specifically altered by three kinds of drugs: barbiturates, benzodiazepines (e.g. Valium), and alcohol. This is particularly interesting. We’ve come across people two times who say that after a glass of wine, they can work with their eyes more easily when trying different exercises.
Dopamine and serotonin are key neuromodulators (types of Monoamines), derived from amino acids. They are critical in regulating the brain’s mood – affective and arousal states. LSD interferes with membrane receptors activated by serotonin, preventing serotonin from interacting with the monoamine receptors. Amphetamines are stiumlants which increase levels of catecholamines (a type of Monoamine) at synapses, particularly dopamine, by releasing these substances from nerve terminals and interfering with the reuptake of catecholamines into the presynaptic terminals. Given the relation to learning, what would happen when a person is making progress in learning to use both eyes, and wants to cement this with the measured application of neuromodulators to help “fix” those new experiences? We have also had a report from an amblyope who experiences depth perception when they occasionally take LSD. This may, very much, be a space in which a university partner would be interesting to explore a cross-over pharmaceutical approach to managing vision-rehabilitation.
Although neuromodulators can simultaneously trigger multiple secondary message channels (e.g. to alter a cells genes, produce enzymes, alter membrane properties, or cause self-triggering) they are actually a bridge-head to detecting properties of the world which are “non-electrical”. For instance… Salts and some amino acids interact directly with channels on the receptor cells, resulting in the depolarization and activation of the taste cells, much as excitatory neurotransmitters activate neurons. Other substances, including sugars and bitter compounds, activate G-protein-coupled receptor molecules on the receptor cells that link to either adenylate cyclase (sugars) or other synthetic enzymes (bitter substances). In other words, if I understand it correctly, salts and acids affect the polarization of the cell directly (+ve/-ve) whilst G-Proteins enable the receptor to recognise anything a protein can, and convert that into the kind of potential, over secondary messengers, it needs to create activation potentials, or do set in motion other processes (perhaps shared between multiple pathways).
The process is similar in the way that olfactory receptors bind to specific airborne molecules (the different olfactory epithelium effectively form an array [about 400 types large] of molecule specific sensors), using secondary messengers to trigger the action potentials which are then eventually recognised by the olfactory cortex.
Now lets talk eye basics. There are many types of eyes, but Dowling starts by covering vertebrate eyes, with the classic rods and cones. In a simplistic way, cone and rod photoreceptors are neural cells specialised for detecting in dim-light conditions (rods), and brighter light (cones). At their heart are light sensitive molecules (visual pigments) which lead to the excitation of the photoreceptor cell in the manner of a neuromodulator.
A visual pigment consists of a large protein bound to a slightly modified form of vitamin A (called retinal). To maintain an ability to respond to both dim and bright light, the visual pigments are being constantly broken down and replaced when light strikes them. As you enter dim light conditions, less pigment is broken down, causing a gradual increase in your sensitivity to light.
The sensitivity of the cell is also regulated in different light conditions as the phototransduction process balances the cell’s polarization – causing initially an highly sensitive initial response to a light source to quickly drop and sustain itself as the cells internal balance of messengers is adapted.
Dowling then moves away from behaviour at the cellular level, towards systems neuroscience – the study of networks of neurons, which is still a relatively poorly understood arena.
Bloch’s law : a short, intense flash looks identical to a longer, dimmer flash is both flashes have the same number of photons.
Lateral inhibition. First discovered in Limulus. This results in edge contrast enhancement. Neighbouring eccentric cells (which receive light from a photo receptor) laterally inhibit their neighbours. Therefore, two adjacent cells receiving photons will interfere with each other and make a lower average response… but as we are talking about many cells, this all averages out. At a dark/light boundary, however, the cell on the dark side is then aggressively suppressed by the cell receiving light from the light side (hence the border appears darker on the dark side next to the light area) and conversely the light receiving cell on the edge of the light to dark side, is less inhibited by its neighbour receiving (less) light from the dark side, hence that edge appears brighter.
In cold-blooded animals the midbrain is much more prominent… the tectum receives visual and other sensory input and tectal neurons project to the spinal cord, where they snapse onto motor neurons. In higher mammals sensory processing, sensory-motor integration, and the initiation of motor activity are in the cerebral cortex. The midbrain and tectum in the mammal mediate noncortical visual reflexes like pupillary and eye movements. In higher mammals the tectum also helps to coordinate head and eye movements. This is particularly interesting to us. It is an example of processing which we simply don’t consciously control – are these structures as amenable to change as visual structures in the cerebral cortex? How relevant are they to vision rehabilitation? How much emphasis ought to be placed on training at this level (eye movement)? We already have many designs for experiences at this level.
The medulla links the spinal cord to the brain – sensory neurons ascending, and motor neurons descending. Half of the cranial nerves enter the brain in the medulla.
The hypothalamus mostly regulate basic drives and acts such as eating, drinking, body temperature, and sexual activity. Lesions in the hypothalamus, or stimulation of parts of it, can lead to irritability or even aggressive behaviour, or placidity. The hypothalamus also regulates the pituitary gland, which in turn releases hormones regulating other glands in the body. Hypothalamcic nuclei, along with nuclei in the medulla, help to regulate… the pupil of the eye. I see an interesting map of visual function is needed!
The cerebellum coordinates and integrates motor activity (although it doesn’t issue the higher level commands saying what needs to happen, it does analyse the various inputs from the senses, integrating them, and sending signals to the motor neurons to accomplish smooth movements. Lesions of the cerebellum typically result in jerky, uncoordinated movements. Movement initiation may be delayed, or movements may be exaggerated or inadequate. The cerebellum is also responsible for the learning and memory of motor tasks, such as riding a bicycle. Thus , cerebellar lesions can also interfere with learning and remembering motor skills. How relevant is this to low motility in some patients who have strabismus? The eye movements are jerky, inaccurate… is this the area of the brain that we would be targeting for refining muscular control over eyes which have been semi-dormant for many years?
The thalmus has numerous nuclei whose role is to relay sensory information to the cerebral cortex…. the lateral geniculate nucleus, for example, transmits visual information to the cortex. Most optic nerve axons terminate in this nucleus, and the neurons in the lateral geniculate nucleus then project to the area of the cortex that processes visual information. This is, AFAIK, the location at which the optical nerve synapses are inhibited – resulting in amblyopia.
The central (foveal) region of the retina, which mediates high-acuity vision, encompasses a large proportion of the primary visual area in the central cortex.