1.5 Making Connections: Synapses

Loading...

From the course by University of Washington

Computational Neuroscience

431 ratings

University of Washington

Computational Neuroscience

431 ratings

This course provides an introduction to basic computational methods for understanding what nervous systems do and for determining how they function. We will explore the computational principles governing various aspects of vision, sensory-motor control, learning, and memory. Specific topics that will be covered include representation of information by spiking neurons, processing of information in neural networks, and algorithms for adaptation and learning. We will make use of Matlab/Octave/Python demonstrations and exercises to gain a deeper understanding of concepts and methods introduced in the course. The course is primarily aimed at third- or fourth-year undergraduates and beginning graduate students, as well as professionals and distance learners interested in learning how the brain processes information.

From the lesson

Introduction & Basic Neurobiology (Rajesh Rao)

This module includes an Introduction to Computational Neuroscience, along with a primer on Basic Neurobiology.

1.5 Making Connections: Synapses20:18

Meet the Instructors

Rajesh P. N. Rao
Professor
Computer Science & Engineering
Adrienne Fairhall
Associate Professor
Physiology and Biophysics

Hello again and namaste to all of you.

In the last lecture we talked about ion channels and

how to give rise to action potentials or spikes.

In a later lecture we will describe this mathematically

using a model first proposed by Hachkin and Hagsli in 1952.

Hachkin and Hagsli got a Nobel prize for

their computational model, now what does that mean for you?

It means that you too could get a Nobel Prize someday

by solving problems in computational neuroscience.

But before you start making plans for that trip to Sweden, we first need to learn

about what happens to a spike when it reaches the end of an axon.

This is where we meet the Synapse.

So, what is a Synapse?

A synapse is a connection or junction between two neurons.

There are two different kinds of synapses.

The first kind are called electrical synapses and

the used something called gap junctions.

The second kind of synapses are chemical synapses.

And they use chemicals known as neurotransmitters.

Let's first look at electrical synapses.

Now here is an example of an electrical synapse between two neurons,

neuron A and neuron B.

Now electrical synapses function in a manner very similar to the way that

the connections between components in your cell phone or in your computer function.

They allow the activity from one side, in this case neuron A,

to be directly propagated to the other side, in this case neuron B.

So they changed the voltage on one side,

given some activity and changes on the other side.

And the way that neuron a communicates with neuron B, or vice versa,

is through what is known as a gap junction here.

And a gap junction is depicted here in terms of these.

So these are essentially ionic channels.

So here's one, here's another and so on.

And you notice that these ion channels span the membrane

of both neuron A and the membrane of neuron B.

And the result of this particular arrangement is that if you have

excitation on one side, due to for example, an action potential.

And maybe you have sodium ions on this side that are in

higher concentration than on the other side.

Then these channels allow these ions to migrate to the other side.

And the result of this movement of

ions from one side to the other is that you're going to have a change

in the membrane potential of neuron B as a result of some activity in neuron A.

So these types of electrical synopsis

are really useful when you want to have fast connections between two neurons.

And typically they're found in the case where you need to synchronize,

which is you want to make neurons fire simultaneously together.

It turns out to be a useful mechanism when you want to synchronize sets of neurons.

The other case where you would like to have these types of

fast electrical synapses or fast connection between neurones is when

you have to implement something like an escape reflex.

And that's something that's found for example in the crayfish.

So that's an example of an electrical synapses and what about chemical synapses?

So here's a depiction of a chemical synapse.

So suppose we have a neuron A and

a neuron B and here's the action potential, or spike coming in.

Now what we have on one side, on the side of neuron A are these bags which are known

as vesicles, so these are bags of neurotransmitter molecules.

So these are chemicals that are stored in these bags.

And so when an action potential spike comes in along the axon of neuron A it

causes these bags, these bags called reticules to

fuse with the membrane and in doing so this bags

release the neurotransmitter molecules into the gap between the two neurons.

Now this gap is called the synaptic clef.

And so when these neurotransmitter molecules stand

fuse with the receptors on the other side.

So these receptors are nothing but the chemically gated

ionic channels that we talked about in the previous lecture,

then these chemically gated channels start to open.

And so you know what happens when these channels are to open?

Well, they're going to allow some ions to either come inside or

go outside according to the concentration of the ions on the inside or the outside.

So suppose that these are channels that allow sodium ions to come in,

then you're going to have sodium ions coming in, into this neuron.

And as a result, you're going to have an increase

in the membrane potential of neuron B.

So you can see how a spike from neuron A, causes this

cascade of chemical events, and then that in turn causes these channels to open.

And that finally causes some changes in the membrane potential.

So you're going from electrical activity to chemical and

back to an electrical change again, on the side of neuron B.

Now you might ask yourself why evolution go to the trouble of constructing

such a complex electrical chemical an electrical connection when it could have

just use the electrical connections with gap junctions to begin with.

Any thoughts on that?

So why would something like this, like in a chemical synapse,

be useful compared to just a simple electrical synapse?

So here's a possible answer.

A chemical synapse allows you to change the way that neuron B is

affected by the seam spike, by simply changing the number of

ionic channels or the density of ionic channels that you have on this side.

And so, if for example for the seam spike you want

to decrease the amount of excitation in B, then you could,

for example, take out or remove some of these ionic channels and

then just leave a few of them and, in that case, for the same spike from neuron A,

you would have a lesser amount of excitation in B.

So this is the reason why chemical synapses have been suggested as

being the basis for learning and

memory in the brain, as we will look at a little bit later.

Here's a wonderful picture from the Kennedy Lab at Cal Tech.

Now what does this picture remind you of?

Well, it looks like it could be the picture of a city taken at night

from an airplane flying overhead perhaps.

Well actually it's the picture of a neuron and all of its synapses.

Each of these bright spots corresponds to a synapse.

And typically a neuron has up to 10,000

synapses on its dendrites and on the cell body.

Now these synapses can be either excitatory or inhibitory.

So what does that mean?

An excitatory synapse is one that tends to increase

the postsynaptic membrane potential.

So what does that mean?

Well, if you have a neuron A and a neuron B and you have a synapse between them,

then we call neuron A the presynaptic neuron, and

we call neuron B the postsynaptic neuron.

So pre and post, and so an excitatory synapse is

one that tends to increase the postsynaptic membrane potential.

Which means that it tends to excite the neuron B.

On the other hand, an inhibitory synapse is one that

tends to decrease the postsynaptic membrane potential,

which means that it tends to decrease the membrane potential of neuron B.

Now, how does that happen?

Let's go through the sequence of events governing the action of an excitatory

synapse.

So first we have an input spike.

So here's a spike.

And the spike is going to cause neurotransmitters to be released.

So in the case of an excitatory synapse, the neurotransmitter could be Glutamate.

And when these neurotransmitter molecules are released into the synaptic cleft,

they are going to bind with ion channel receptors.

So in this case the ion channels could be selective for sodium,

so you might have these sodium selective channels opening.

And that in turn is going to cause sodium ions to come into

the cell, in this case cell B if this was cell A.

And now what's going to happen on the side of cell B is something

called depolarization, so you might remember the word from a previous lecture.

And that intern causes an EPSP or excitatory postsynaptic potential.

That's again something we encounter in the previous lecture.

And so, the sequence of events is basically spike.

And then you have a release of the neurotransmitter, Glutamate.

And then the ion channels opening, causing sodium to come inside.

And that in turn increases the membrane potential of neuron B.

Synapses are thought to be the basis for memory and learning in the brain.

This is called the Synapse Doctrine.

This is quite deep actually.

Think about what this means.

This means that all of your memories from the first time you rode your bike and

crashed into your neighbor's brand new car,

to the first time you met the love of your life is all stored in your synapses.

What's more all of your skills from typing, dancing,

driving to reading and writing.

Guess where it's stored?

It's in your synapses.

Synapses plays such an important role in our lives that perhaps

we should have a world synapse day.

What do you think?

Well okay maybe not, but

how do synapses play such an important role in memory and learning?

Let's find out.

Synapses allow learning in the brain through mechanism called

Synaptic Plasticity.

Let's look at one particular example of synaptic plasticity and this is

Hebbian Plasticity, named after Donald Hebb who was a Canadian psychologist.

In 1949, Hebb proposed the following rule for plasticity.

If a neuron A repeatedly takes part in firing neuron B,

then we should strengthen the synapse from A to B.

Now, here's a diagram that represents Hebbs rule for plasticity.

So here's neuron A and when neuron A fires it participates in firing neuron B.

So here is a single spike from neuron B.

Now according to the rule for plasticity, we need to strengthen

the synapse from A to B, and that's shown by this, bigger red triangle.

So, what happens?

So, the next time A fires.

So, here's A and here's the input from A.

You're going to get a bigger and perhaps a faster response from neuron B.

Now why is this computation useful?

Well, suppose you are in a jungle and it is dark.

And then you hear a growl.

Let's say the growl is conveyed by this firing of neuron A.

Now right after that, neuron B fires and you see a tiger.

And you start running and barely escape with your life.

Now, if you increase the strength of the connection between neuron A and

neuron B as we did here using Hebbian Plasticity.

Then the next time that you hear a growl, and that's given by neuron A firing,

you're going to predict that you're going to see a tiger.

And you're going to start running even before the tiger appears.

And you have Hebbian Plasticity to thank for it.

Isn't that great?

Hebbian Plasticity is so

useful that there's in fact a mantra that goes along with it.

Here it is.

Neurons that fire together wire together.

Some say that if you repeat the mantra three times every morning,

you're sure to have a great day.

But I'm skeptical.

Is there any evidence for Hebbian Plasticity in the brain?

In fact, there is.

There is something called Long Term Potentiation

that researchers have observed in several areas of the brain,

most prominently in the area called the hippocampus.

So LTP is defined as an experimentally observed increase

in the synaptic strength that lasts for hours or even days.

And the way that you can demonstrate that experimentally is

by measuring the size of the EPSP caused by a single input to this neuron B.

So if you demonstrate that the size of the EPSP increases

as you pair neuron A with neuron B and you give

lots of stimulation to neuron A which in turn causes some excitation in neuron B.

Then, what you'll observe is an increase in the size.

So, here is the final size.

Here is the initial size.

So, you can see that the effect of a single input

to this neuron B has increased from being this little bump to this really big bump.

And that is taken to be evidence for

an increase in the synaptic strength from A to B.

What about the opposite?

Is there something that can decrease the synaptic strength between

two neurons and which can last for hours or days.

Well, there's something called long term depression.

That's not the same as the kind of depression you get when you fail

an exam or get a rejection letter that's a different kind of depression.

This is an experimentally observed decrease in the synaptic

strength from one neuron to another that lasts for hours or even days.

Now how do we experimentally observe long term depression or LTD?

Well, you can do the same thing that you did with LTP.

You could look at the EPSP that's generated when you stimulate one neuron,

you observe the EPSP in the second neuron B.

Now, if over time, there's a decrease in the size of EPSP

until you get a smaller sized EPSP for the same input from A to B.

Then you have evidence that there has been a decrease in

the synaptic strength from A to B.

There's a recent twist to the story of synaptic plasticity that I thought

I should mention.

And that is that synaptic plasticity depends on spike timing.

So what do we mean by spike timing?

Well it turns out that whether you get LTP or

LTD depends on the relative timing of input and output spikes.

So remember that we have A neuron A that has a synaptic connection to neuron B.

And so when you have an input.

So we're calling A the input and B the output neuron.

So when we have the situation that the input spike is before the output spike,

which means that neuron A fired and then you have neuron B firing.

Then what you have is a situation such as this.

So here is the input-output pairing.

So you have the input happening before the output spike,

which is given by this spike over here.

Then what we observe in this case is an increase in

the size of the EPSP after you have repeatedly paired

the input-output in the sequence that the input occurs before the output.

Now the interesting case is where you have the reverse situation

where the input spike occurs after the output spike.

So you have B firing, and then you have A firing.

So in that situation, here's the situation.

Here's the pairing.

So you have the output spike here, and

then you have the input spiking right after that.

And so if you keep doing this pairing for several repetitions,

you're going to observe that the EPSP size diminishes from being

like that to something that looks like this.

And that's what we mean by long term depression.

So we have both LTP and LTD.

So if you summarize in this particular result for different intervals between

the input and the output spike you get a window of plasticity that looks like this.

So we have one portion of it has LTP the other LTD, and

here is the delta t between the input and the output spikes.

So the input happens to be before the output.

So the delta t is bigger than 0, then you are in this portion of this plot.

And you can see that if the input happens before the output, you have LTP or

an increase in the synaptic strength.

If the input occurs after the output where the delta t is negative, then you have

LTD or a decrease in the synaptic strength from neuron A to neuron B.

So this is actually quite amazing because you can see that there is

a very short interval of about 40 milliseconds between the input and

output and depending on where the input spike occurs with respect to

the awkward spike, you could get a dramatic shift from LTD to LTP.

So the neurons in several brain areas, including the cerebral cortex,

seem to be very sensitive to the timing of the input versus the output spike.

And we learn later that this type of sensitivity to the timing of

spikes is very important for learning sequences to make predictions.

Okay, to summarize the last couple of lectures,

we seem to know a lot about ionic channels, about neurons, about synapses.

But what do we know about networks of neurons

give rise to perception behavior and maybe even consciousness?

Well, not as much.

This is actually one of the primary motivations for

the recent large scale brain project that has been announced in Europe and the US.

In Europe, the Human Brain Project led by Henry Markram,

will attempt to construct a large scale computer simulation of the human brain.

While in the US, President Obama has announced the brain initiative to map

the activities of hundreds of thousands of neurons simultaneously in order to

understand how large scale networks of neurons give rise to perception,

action and cognition.

In the next lecture we will briefly cover

what we do know about networks in the brain and their function.

Until then, bye-bye.

Explore our Catalog

Join for free and get personalized recommendations, updates and offers.

Coursera provides universal access to the world’s best education, partnering with top universities and organizations to offer courses online.

© 2018 Coursera Inc. All rights reserved.

Download on the App Store

Get it on Google Play