Brain operation and processes
A simplified description of the operation of a neuron is that it processes
the electric currents which arrive on its dendrites and transmits the resulting electrical
currents to other connected neurons using its axon. A simple explanation of the
processing step is that the cell sums up the incoming signals and produces an
output signal only if this sum exceeds some threshold; i.e. only if the total input
signal is big enough will the cell `fire' an output signal to its neighbors.
Once the cell fires, an electrical signal travels down the axon at a speed of around
100 meters per second (200 mph). These currents are very small by usual standards. The
typical voltage difference between the outside and inside of a nerve cell is 70 millivolts
(one millivolt is one thousandth of a volt). This is to be compared with the voltage at a
power socket in your home of 110 volts - a thousand times bigger. This signal is passed
onto other neurons at the synapse points in the following way.
The pulse reaches the end of the axon branch and causes the release of certain
chemicals called neurotransmitters. These diffuse across the synaptic gap (a
distance of some one-hundredth of a micron) to be picked up by so-called receptors
on the ends of the dendrites of a neighbor neuron. This absorption process on the new cell
changes its electrical state. If there are sufficient incoming signals to this neighbor
neuron, this change in its electrical state can be big enough to generate a new pulse in
the neuron. Thus the process repeats in this new cell. This cell is itself connected to
many others and in this way a wave of electrical activity can be set up. Different types
of brain activity correspond to different patterns of firings.
While we are born with a complete set of neurons, the connections between them are
determined in major part by a learning process; external stimuli coming in the form
of electrical currents from the sensory cells cause patterns of nerve impulses to be set
up. These impulses can alter the strength of the coupling between different neurons. While
the overall program for determining which neurons should be connected together is under
genetic control, it is external stimuli which are crucially important in determining what
network connections are made. Indeed, to some extent out brains are continually rewiring
themselves to cope with passing experience. This particularly true for small children who
are born with a full complement of neurons but a relatively primitive set of connections -
a useful set of network connections must be learnt during the early years.
It is also true that the levels of various neurotransmitters are, in part, determined
by early experiences. The overall functioning of the brain is strongly influenced by such
chemical balances. For example, the neurotransmitter seratonin plays a role in
regulating aggression; a lack of dopamine, another such chemical, reduces frontal
lobe activity and has been associated with schizophrenia. Endorphins play a role in
the system which produces sensations of pain and pleasure. It is becoming increasingly
clear that certain traits of personality may be determined in major part by biochemistry.
This opens up the possibility of a "chemically improved" society and all the
profound implications that implies.
The fact that our neurons can rewire themselves "on the fly" has the
consequence that our brains are amazingly robust - if a given neuron dies (which will have
happened to something like 20 percent of our original neurons by the time we die!), our
brain automatically undergoes a rewiring process in which new connections are made to
circumvent the defunct neuron. It is also the origin of the amazing diversity in peoples'
types and abilities. Intelligence is determined partly by genetics (the program that
governs the overall structure of what connections should be set up) and partly by our
experience which can influence very strongly the nature and quality of our neural
networks.
Two neurons are not merely joined or not - the nature of the synaptic connection
between them determines whether one neuron firing has a strong or weak effect on the other
- we talk of the strength of a connection between them. A strong connection between
two neurons means that it it more likely that one of the neurons firing will stimulate the
other to fire - with a weak connection it may only happen occasionally depending on the
state of very many more neurons for example. Of course, this connection strength feature
has much to do with the presence or absence of certain neurotransmitters since they are
crucial in determining the size of electrical signal which can pass between neurons.
This notion of dynamically changing connection strengths is thought to be important for
memory function - new memories are stored not on individual neurons but by adjusting
the strengths of connections between neurons. A simple rule appears to govern this
process: the connection between two neurons will strengthen if more often than not
the two neurons fire together. This is often called the Hebb rule. ![[take note!]](../images/note.gif)
Thus the operation of the cell depends on both electrical and chemical properties -
those of the neurotransmitter molecules. There are a number of types of these
neurotransmitters (research has identified at least fifty distinct types of
neurotransmitter). Some, referred to as exciter neurotransmitters act to trigger
the receiving neuron, while others called inhibitors act to damp out signals in the
neighbor neuron. For example, one particular such inhibitor chemical called GABA acts to
prevent abnormal or parasitic muscle movements. The degeneration of certain synaptic sites
rich in GABA provokes an illness called Huntington's Chorea, whose symptoms are almost
incessant involuntary movements. The famous folk singer Woodie Guthrie succumbed to this
genetic disease.
Much use has been made of the way that these neurotransmitter chemicals are used by
neurons. Tranquilizers act by modifying the natural chemicals in the synaptic gap
and drugs such as LSD by altering the balance of various neurotransmitters. This can cause
very dramatic effects; for example, sounds can be perceived as colors. The new drug Prozac
also acts directly to remedy chemical imbalances in the brain.
We may ask the question: what is about the structure of neurons and their
organization which determines the amazing computational power of the brain?
Certainly it is not the raw processing power of a single neuron - it takes about
one-thousandth of a second for a cell to return to a normal state after firing. This is
the minimum time before the neuron can process another incoming signal. While this seems
quite a short time it is ridiculously slow compared to even a modest home computer whose
silicon chip can perform operations in the incredibly short time of one-hundred-millionth
of a second.
The secret lies in the very number of neurons - many tens of billions as we have
said. If these neurons can be made to work efficiently and simultaneously on a
given task it is clear that the effective power of the brain is very much larger than
current computers. The hint that such a scenario may indeed be realized lies in the
detailed structure of the brain in terms of the connections between neurons. It is clear
that in order for neurons to cooperate in performing some function, they must be able to
talk to each other. We know that each neuron has many tens of thousands of connections to
other neurons that function as communication channels. These connections have an
incredibly complicated structure - different portions of the brain have different types of
connection pattern, while these different sectors of the brain are themselves linked
together by further specialized networking. We still have only the crudest understanding
of why these neural pathways are connected up the way they are. But it is surely the very
detailed way in which these connections are made that is at the heart of the power of the
brain as a thinking machine.
The very complexity of these neural networks poses a formidable barrier to
understanding. Nobody knows in detail how the individual firings of neurons coupled to
their interconnections can lead to all of the features observed - short and long term
memory, complex pattern recognition, logical reasoning, emotion and consciousness. Indeed,
it is not known how even the lower level unconscious functions such as those which
regulate breathing and heart rate emerge out of the complicated mutual interaction of
millions of neurons. Furthermore, it is generally believed that at least a partial
understanding will be necessary in order to build truly intelligent machines.
Nevertheless, we are making rapid progress in understanding some of the simpler aspects
of these systems in part through the study of computers whose architecture resembles that
of the brain. These computers go under the name of artificial
neural networks and will form a large part of this course.
|