The basic component of the brain is the neuron (there are other cells alongside the neurons to help them do their job, called glial cells). Neurons have electrical properties, known as firing or spiking rates. The neurons handle information and communicate using electrical signalling. They do different things: there are sensory neurons, motor neurons, neurons in memory systems and neurons in the emotion systems.
The brain gets its power as a thinking device because the neurons, doing relatively simple things on their own, are connected in complex networks. From the view of a computer scientist, the individual components may be simple, but when you wire them together into a large network, you get a powerful computational device. That’s the theory.
Neurons are connected to each other (you’ll hear words like axons, synapses, dendrites). The strength of the connection determines how much activity in one neuron influences activity in the other neurons to which it is connected. The main way that the brain acquires skills, abilities, and knowledge, is through changing the strength of the connections between neurons, mostly based on experience.
The brain is built out of neural networks because it was the path that evolution took for animals, a long time back. There might have been better ways to do it, of course. But as we saw, the way evolution works is to improve on what it already has.
Here’s the problem. Because knowledge is built into the structure of the brain – in the strength of the connections between neurons – it is very hard to move around. This is a stark contrast with the digital computer. The digital computer relies on abstraction and moving knowledge between general purpose processing mechanisms. How can the brain work if its knowledge cannot be moved around? It has to work by a different principle.
The brain comprises specialised systems that handle knowledge specific to their domain. The visual system handles visual information, the auditory system processes auditory information, and so on. These specialised systems store long-term knowledge in their connections. The shape of a cup. The sound of the word ‘cup’.
The specialised systems process the information they receive. Once activated, they maintain activation states about their current status by passing it around closed circuits (so-called ‘recurrent loops’: neuron A activates neuron B, which then activates neuron A, which then actives neuron B, and so on).
In a way then, each specialised system combines the hard drive (storing knowledge), working memory (keeping it in mind), and CPU (processing) into a single device that dedicated to handling a specific type of content.
These specialised systems need to be co-ordinated and controlled. This role is carried out by what we’ll call a modulatory system. This system is connected to all the specialised systems and its role is to modulate their activity. It needs to turn on (activate) the specialised systems or parts within them that are relevant to the current situation and goals of the individual, and turn off (or inhibit) the systems that are not relevant. It sets up the rest of the brain to do different tasks, switches between different tasks when necessary, stores the steps necessary in tasks that have multiple steps and keeps track of which step the brain is on. It monitors that plans are going as expected, and monitors for changes in the current situation that should prompt a change in task.
The modulatory system is at the front of the brain, while the specialised systems are at the back, on the outer layer (the cortex). You’ll remember that the front of the cortex is connected to the limbic system. This means that emotion is also inputting into which specialised systems are activated or inhibited. And the suite of systems activated or inhibited changes from moment to moment, driving the generation of appropriate behaviour.
You might think the modulatory system is a bit like a CPU, a central controller or an executive. Perhaps, but it knows nothing. The ‘executive’ can’t do any of the jobs on the factory floor. It has the specialised role of control – the content resides in all the specialised sensory and motor systems, stuck in the connections there. For example, have you ever wandered into a room and forgotten what it was you came in for? All you are left with is the sensation that you came in for some reason. That is the modulatory system (keeping the plan active), which has lost contact with the information stored in a specialised system (the details of the plan).
The brain has an episodic (or autobiographical) memory system, and this system offers a clear demonstration of how hard it is for the brain to move knowledge around. As we have seen, the brain contains a structure called the hippocampus, which brings together information from all the senses to create snapshot memories. The hippocampus fills up with memories eventually (it can store perhaps 50,0001; and there’s one on each side of the brain). Memories need to be transferred out and stored more permanently elsewhere in the cortex. In addition, the gist needs to be pulled out of the individual memories. You need to know the capital of France is Paris (as a fact), not remember details of the specific episode or moment when you were told that Paris is the capital of France (it was on a Wednesday, the wind was cold and the sky was grey, I was sitting next to my best friend Phoenix, and the teacher was droning on and on about capital cities).
Okay, so now the brain needs to move knowledge around. How does it do it? Here’s what happens. The brain needs to be shut down! It has to be taken off line for several hours each night. During sleep, the hippocampus spontaneously replays the events of the day, and cortical areas change their connections to store the information. The transfer of memories takes some time – days, weeks. If the hippocampus is damaged, not only is the ability to store new episodic memories lost (called ‘anterograde’ amnesia – memory going forward), but recent memories are also lost (called ‘retrograde’ amnesia – loss of previous memories). These are the memories yet to be transferred out of hippocampus, or in the process of being transferred as their strength is gradually built up in cortex.2 Full transfer from hippocampus to cortex may take several days or weeks. (See more in the later section Horses for courses).
Compare this way of working with how the computer works. In a computer, knowledge is moved between general-purpose devices in a fraction of a second. The faster the movement, the faster the computations. By contrast, the brain activates and deactivates specialised systems in a fraction of a second, but it takes hours to move knowledge around. These are two very different ways of working.
 Here’s a paper showing how this estimate was derived. It’s based on how many neurons there are in the hippocampus, how many connections there are between them, and computer science theory of the information that can be stored within neural networks. (See section 18.104.22.168.1. WARNING: may contain heavy-duty neuroscience!)  Once memories are stored in cortex, surgeons can reactivate very specific memories with pinpoint stimulation of cortex during open brain surgery.