Plasticity, Self-organisation and Hebbian Learning

Question 1

What is the difference between neural plasticity and learning?

Answer 1

Learning is a cognitive process, while plasticity is a neural mechanism.

Question 2

What is homeostatic plasticity?

Answer 2

The capacity of neurons to change their parameters to regulate their own excitability, a compensatory mechanism.

Question 3

Describe a conductance-based synapse

Answer 3

I_syn = gbar_syn n (V-Vsyn)

When a signal arrives from the presynaptic neuron, a current is generated in the postsynaptic neuron which is a function of the synaptic conductance, the proportion of open receptors (n) and the driving potential for that type of synapse.

There is a flow of ions modulated by the membrane potential of the postsynaptic neuron and the synaptic conductance.

Question 4

Describe a current-based synapse model

Answer 4

A model that approximates the conductance-modulated flow of ions as the injected current that is not dependent on the membrane potential.

This simplification is known as a current-based synapse and it has less biological plausibility than the conductance-based synapse model.

Question 5

What is the conductance-based model of a synapse?

Answer 5

Just as in HH ion channels, it is equal to

total conductace * receptor activation * driving force

Question 6

This graph represents the time course of synaptic potentials for different receptors. In terms of temporal characteristics, GABA_B postsynaptic current is slow or fast?

Answer 6

Slow.

Question 7

This graph represents the time course of synaptic potentials for different receptors. Which one is the fastest receptor?

Answer 7

AMPA

Question 8

What does η define in this equation for synaptic plasticity?

Answer 8

The learning rate

Question 9

What type of plasticity is represented in this equation?

Answer 9

Hebbian Learning

Question 10

What is the problem with this learning rule?

Answer 10

The weight growth is unbounded. They will grow indefinitely when the neurons continue to be activate together.

Question 11

What is the difference between supervised and unsupervised models?

Answer 11

An unsupervised model (such as Hebbian learning) self-organizes and tries to make sense of the data independently without a predefined label or error. No indication of “what should be learned”.

In a supervised learning model, errors are predefined, there is a cost function and the algorithm uses gradient descent to reduce the error.

Question 12

What are the properties of reward/reinforcement learning?

Answer 12

(1) Success is predefined
(2) There exists a reward function
(3) There is a memory trace to remember previous steps

Question 13

Explain the terms in the Hebbian learning formula means.

Answer 13

At a certain time (implicitly defined here), neurons will change their weights as a function of the correlation between their activities and the learning rate ($\eta$).

If we assume that a_i and a_j are bounded sigmoid functions such that they only have values between 0 and 1, the neurons are maximally co-active when both of them are 1 (maximally coactive).

Question 14

How can we stabilize Hebbian models?

Answer 14

One way is to subtract the average expected activity from the activities of both a_i and a_j.

If we consider now a scenario where activations exceed the expected average, this stabilization will lead to depression of the output.

We see long-term potentiation as well as long-term depression in the same function.

Question 15

What is the relationship between spike-timing-dependent plasticity and Hebbian learning?

Answer 15

Spike-timing dependent plasticity is the “spiking version” of Hebbian learning. It transforms the Hebbian idea that neurons have to be co-active to the idea that neurons have to be co-active with a certain time in between them.