Exam, computational neuroscience

Question 1

what is the law of effect?

Answer 1

It was invented by Edward Thorndike and indicates how the associations between stimuli and responses (S-R) vary depending on the outcome of an animal’s actions

Question 2

which formula best capture the notion of prediction error?

rå=prediction error at time t
r=reinforcement at time t
p(t)=prediction at time t
p(t-1)= prediction at time t-1
y=temporal discount parameter

Answer 2

rå=r+y*p(t)-p(t-1)

Question 3

Name some fact about the association between a stimulus and an action [Q(s,a)] in Q-learning!

Answer 3

The association between a stimulus and an action in Q-learning is incremented when a positive prediction error occurs

Question 4

Where in the brain are the Q-values represented?

Answer 4

In putamen

Question 5

what does dopamin represent in computational neuroscience?

Answer 5

It represents rewards prediction errors

Question 6

What is the learning rate (alpha) in Q-learning?

Answer 6

The learning rate (alpha) is a value between 0 and 1 which in order for learning capable of integrating information over several events to occur, should be grater than zero and less than 1.

Question 7

What happens if an action is followed by firing of dopaminergic neurons?

Answer 7

This action will tend to be be repeated in the same context

Question 8

which region strengthen their functional link to the putamen when people learn a new stimulus-response (S-R) association

Answer 8

Sensory and motor cortexes