How to build thinking machines?

How to build thinking machines?

mediumThis post was originally published by Joe Molnos at Medium [AI]

The structure of the brain is based on Neurons.
In order to build thinking machines, we need to know the function and structure of our brain.
Neurons are individual nerve cells that conduct electrical impulses and the brain consists of a very large number of them. There are roughly ten billion neurons in the brain, each of which is connected, on average, to about ten thousand other neurons. That makes brains astonishingly complex. Imagine taking a country the size of India — which has a population of about a billion — and giving every man, woman and child a thousand pieces of string with instructions to find a thousand distinct people to hold the other end of each piece of string. When the whole country is connected up like this, with every person connected to a thousand other people by pieces of string, multiply the whole system in complexity by an order of magnitude and that’s how complex your brain is.

There are a number of quite distinct types of neurons, but that needn’t concern us here. We’re going to describe the structural features and operations of a paradigm neuron. Neurons have a cell body or soma which contains the nucleus of the cell. This is connected via an axon hillock to the axon, which is a protuberance that can extend as long as roughly a meter. These axons are coated with a myelin sheath which helps electrical signals flow more quickly and aids in insulation. At the end of the axon are axon branches that terminate at axon terminals. Axons are efferent connections — they carry signals away from the soma and along the axon towards the axon terminals. Incoming, or afferent, signals are carried towards the soma along the dendrites of the neuron. Dendrites are organized in a dendritic tree and there may be very many of them. When an axon terminal is in close proximity to a dendrite, a synapse will form (see Figure 4.5). These synaptic connections conduct signals from one neuron to another (strictly speaking, a neuron can form a synaptic connection with pretty much any part of a neighboring neuron, but we’re aiming to keep things as simple as possible). The operations of neurons are electrochemical in nature. An electrical signal flows along an axon to a presynaptic axon terminal where it is transduced into a chemical signal. This chemical signal is then carried across the synaptic cleft by neurotransmitters in synaptic vesicles. Once these synaptic vesicles reach the postsynaptic structure, the chemical signal encoded by the neurotransmitters is transduced back to an electrical signal which propagates along the postsynaptic dendrite and into the body of its neuron. The cell body of a neuron has a certain electrical resting potential. As it receives afferent electrical signals along its dendrites, the difference in electrical potential between the inside of the cell and the outside of the cell rises. When this potential difference is high enough, the soma will discharge an electrical impulse along its axon and return to its resting potential. This is something of a simplification but it suffices for our purposes. If you’re feeling a little overwhelmed with all these technicalities and all this new terminology, don’t fret. All you really need to take away from this chapter with respect to the operations of neurons in the following. There are very many neurons in the brain which are highly interconnected. These neurons function by passing electrical signals to each other. If a neuron receives sufficient incoming signals from other neurons, it will send out a signal of its own.

Spread the word

This post was originally published by Joe Molnos at Medium [AI]

Related posts