Hidden In Plain Sight 9: The Physics Of Consciousness

by Andrew H. Thomas

Nowadays, neuroscientists have more sophisticated methods of watching the brain in action. One method is Positron Emission Tomography (PET) scanning, in which the patient is injected with a mild radioactive material which is dissolved in glucose. The radioactive source decays via beta decay, emitting positrons (positrons are antimatter, and the principle behind PET scanning was considered in detail in my fifth book about particle physics). When the emitted positrons reach any normal tissue in the brain, they are annihilated (via matter/antimatter annihilation). This annihilation produces energy in the form of gamma rays. The gamma rays can be detected by the surrounding scanner, and the source of the radiation can be pinpointed.

In the November 2017 cover story of Scientific American, Christof Koch described the promise of IIT: "The development of several technologies in recent years has raised real prospects for detectors that meet the criteria for consciousness meters — devices useful in medical or research settings to determine whether a person is experiencing anything at all. This ability to detect consciousness could also help physicians and family members make decisions about how to care for tens of thousands of uncommunicative patients."

the mind and the brain.

Neurons are the only cells which are able to transmit electrical signals. Neurons transmit those electrical signals to other neurons, forming a network (a network which eventually forms a brain).

The synapse gap is just one forty-thousandth of a millimetre in width.

The synapse gap is filled with chemicals called neurotransmitters.

Neurotransmitters play a vital role because they allow signals to bridge the gap of the synapse, thus allowing signals to be transmitted between neurons.

We know that messages are sent as electrical signals. It might be imagined that these messages would be formed of electrons passing down a conductive substance (like electricity passes down a wire). However, this is not the case. We will now see that a completely different — and apparently unnecessarily uncomplicated — method is used for sending signals down a neuron's axon to the next neuron.

Instead, the signals are passed in the form of ions.

Common examples of ions found in neurons are sodium ions and potassium ions.

Sodium has the chemical symbol Na, while potassium has the chemical symbol K. Sodium and potassium are both metals, which means they have a

single electron in their outer shells (it is that electron which allows electric current...

The body of each neuron is surrounded by a cell wall called a membrane. The membrane is punctured by many microscopic gates called ion channels which only allow certain ions to pass through them. These gates are opened or closed depending on the electric voltage (or potential) across them.

This

This fast moving spike of voltage represents the "firing" of the neuron and is called the action potential.

This mechanism is not just used by brain neurons — it is also used by nerves to tell muscles when to fire. Whenever I come home after running, I always have an electrolyte drink. Electrolytes are ions including sodium, potassium, and magnesium. After exercise, you can lose salt through perspiration. Salt (sodium chloride) includes the electrolytes sodium and chloride. It is therefore important to top-up your body's store of electrolytes to avoid muscle cramp, in which the ability to send correct firing signals to the muscles is lost.

So why is such a complicated chemical process used to send messages down an axon, instead of simply sending an electrical signal down a "wire"? In the next chapter we will see a reason why this is possibly the case, and it is all to do with the -70 millivolts threshold voltage difference across the membrane.

As Christof Koch said: "Two operations underlie information processing by neurons: the chemical transmission of information from one neuron to another at the synapses, and the generation of action potentials."

But that is not the challenge for us. It is undoubtedly the case that consciousness is not held in any individual neuron.

there may be 100 billion neurons in your brain, but each neuron is connected to as many as 10,000 other neurons.

If you do the arithmetic you find that gives the extraordinary total of 1,000 trillion synaptic connections in the brain.

Instead, there is a general consensus that consciousness arises as an "emergent" effect from the huge connectivity in the brain.

We will also be seeing that it is a property which is possessed by electronic transistors, a property which allows those transistors to make computers — which are, after all, "thinking machines". The property is called nonlinearity.

The reason that linear systems are easy to work with is because of the superposition principle. A description of this principle is presented on the Wikipedia page for the superposition principle: "If input A produces response X and input B produces response Y then input (A+B) produces response (X+Y)."

Simplify

The superposition principle suggests that we could simply this network down to a single component — a single resistor. And that is precisely what we find. There are simple arithmetic formulas which are well-known to any electrical engineer which can reveal that the previous network of seven resistors is equal to a single resistance value of precisely 3 ohms:

If you were actually designing an electronic circuit, you could never create an interesting complex system with just one resistor: a single resistor can do nothing useful. Therefore, logically, you could never create a complex system with any greater number of resistors — as the situation would be equivalent to the single resistor case.

Idea genial. Utilísima. Ejemplo para clases.

And the same principle applies to linear neurons: you could never create a complex system with any number of linear neurons.

Crucially, you could not build a brain with a single linear neuron, so, logically, you could not build a brain with 100 billion linear neurons (as the 100-billion-neuron brain could always be simplified to a single-neuron brain). For ...

So, if linear components on their own cannot be used to create complex systems, let us now move on to ...

As an example, the following graph shows the relationship between input and output when the output is the square of the input.

El cuadrado, no el doble.

Let us see if the superposition principle still applies in this nonlinear case. Considering the previous graph, an input of 2 produces an output of 4, and an input of 4 produces an output of 16. If the superposition principle applies then an input of 2+4 (which is 6) should produce an output of 4+16 (which is 20). But, from the graph, we can see that is not the case: an input of 6 produces an output of 36 — not 20. So, crucially, the superposition principle does not apply to nonlinear components.

"If input A produces response X and input B produces response Y then input (A + B) produces response (X + Y)." Aquí tenemos 2 (A), que produce 4, y 4 (B) que produce 16. Por lo tanto, si aplicara la superposición (A + B) tendríamos que 2 + 4 = 6, y en seguida (X + Y) sería igual a 4 + 16 = 20. Pero no es este el caso, pues 6 no produce 20 sino 36.

And, as you can see from the previous diagram, if the superposition principle does not apply then a network built from nonlinear components cannot be simplified (i.e., unlike linear components). It might sound like a bad thing to be unable to simplify a network, but if we want to create a large and complex network, it is very good news indeed. A network which cannot be simplified down to a single component is a network which can do very interesting things indeed.

resistor

transistor

semiconductor,

A semiconductor can operate as either an electrical conductor or an electrical insulator.

base.

If a sufficiently high voltage is applied to the base, a current can pass between the other two legs. However, if no voltage is applied to the base, the transistor acts like an insulator and no current can pass between the other two legs.

So the 0.7 volts acts as a threshold voltage.

For the neuron, this thresholding effect is called the "all-or-none" law.

So both transistors and neurons fire when the combined inputs exceed a certain threshold voltage — and this is no coincidence. The threshold voltage is essential. In both cases, the threshold voltage introduces the required nonlinear behaviour — and it is that nonlinear behaviour which allows us to build complex systems (computers or brains) from transistors and neurons.

The chemical method creates the crucial threshold voltage inside the neuron, and it is the existence of that threshold voltage which introduces the essential nonlinear behaviour.

There is one simple message I want you to take from this discussion: in its design and nonlinear functionality, a transistor can be considered to be an artificial neuron.

The vast majority of transistors are micro-miniaturised onto a semiconductor substrate to form an integrated circuit ("silicon chip").

This actually results in an individual transistor size which is rather smaller than a neuron, but it is clear that the principle of packing microscopic transistors onto an integrated circuit resembles the packing of microscopic neurons in a brain.

Here is an image of the Apple A11 microprocessor. If you have the latest iPhone then you are already in possession of one of these. This is an image of the entire external package — the actual "chip" and its 4.3 billion transistors is contained within it:

With 4.3 billion transistors in an integrated circuit, it is clear we are approaching the number of 100 billion neurons in the brain. So why do integrated circuits possess nowhere ne...

Three-dimensional chips are now starting to appear. Samsung have recently released computer memory chips featuring 38 layers of transistors, and the rumour is that this is seen as the future of the industry.

"The obvious choice for where to put these interconnects was the same solution in any sprawling metropolis; if you can't grow out, grow up. Three dimensional chips will be released. It is only a matter of time. There is simply no other way to increase the density of interconnects, the number of devices on a chip, or speed than by moving into a third dimension of silicon."