First of all, let’s just clarify what we mean by “parallel” and “sequential”. Sequential processing is processing one after another. Parallel processing is when processing is done at the same time. So, we then get what’s called the “parallel advantage”, which is that parallel processor when processors are all working at the same time, the processing occurs a lot quicker as long as we know what we’re doing. So, the example is Ali Baba — he is hiding in one of 40 jars. What’s the best way to find him? The sequential way is to pick the fastest slave, and he or she runs to every jar, looks in it, and comes back and tells you where he is. The parallel way is you pick 40 really slow slaves, who may be quite dopey, and each one of the slaves just looks in the jar and reports. So, obviously, the 40 slaves — the slow slaves — are going to be faster than the one fast slave. So, this is basically an analogy — a comparison between computing and the brain. Now, the computer has a parallel port. It’s got 25 lines on it. The brain has a parallel connection between the two hemispheres which has 800 million lines in it. And the optic nerve has over a million lines. So, when we’re talking about the brain's parallel processing and our computer's parallel processing, there’s a huge difference. So, what we’re going to say is that the brain is *massively parallel*. So, the computer is sequentially very fast — like the fast slave — but it’s minimally parallel, while the neuron is pretty slow — like the slow slaves — but massively parallel. This is how slow neurons can give fast responses. So, computers can do things very fast, but they tend to do things in sequence — i.e., one at a time; one after the other. Computer events take about a billionth of a second. A neuron takes about a thousandth of a second, and so that’s a *lot* slower. So, the brain makes up for that by being massively parallel. People can recognize faces in about 1/10th of a second. If you can’t remember someone’s name in 1/10th of a second, you probably won’t remember it for the next 10 seconds. It’s biologically very important to get things *quickly*. Now, if the brain was processing faces sequentially, you would as a programmer have to write brain recognition software in 100 lines of code, which is clearly not possible. From a structural or system point of view, you have a fixed number of neurons let’s say, or transistors — it doesn’t really matter — or nodes in a network And this processing structure can be broad. In other words, it can have a very wide base. Or it can be tall, in which case you have a pyramid — very high levels with lots of vertical levels. So, this is obviously a trade-off for the given number of neurons.