Lecture 27 - Reinforcement Learning and Motor Sequences Flashcards

1
Q

What is classical conditioning? What is the famous example?

A

a learned (reinforced) reflex/ response that is evoked by stimulus

Pavlov’s Dog

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is reinforcement?

A

increase behaviour

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is punishment?

A

decrease behaviour

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Give an example of positive reinforcement.

A

Sweet treat

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Give an example of negative reinforcement

A

take away homework

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Give an example of positive punishment.

A

writing lines

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Give an example of negative punishment

A

take away recess

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What does the process of reinforcement learning involve?

A

Learning to link reward with specific actions (and their outcomes) so they become repeated

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is binary reward feedback?

A

Action is rewarded or not

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is a scalar quantity eward feedback?

A

relative to the utility of action/reward outcomes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is the goal of reinforcement learning?

A

to maximize reward and minimize loss

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is cummulative reward?

A

it might be better to sacrifice immediate reward for long-term reward

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What are some examples of cummulative rewards?

A
  • chess
  • investements
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What happens to actions that are associated with reward?

A

they become strengthened/repeated

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

processes of reinforcement learning:

What is exploration?

A

the trial and error process of aquiring more information about the envrionemnt by searching possibilities

searchhing many action possibilites to determin which action tends to maximize reward

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

processes of reinforcement learning:

What is exploitation?

A

capitalize on known information to maximize reward

actions associated with past history of reward tend to be repeated to maximize future reward

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

What is the tradeoff between action exploration and exploitation?

A

shift emphasis from exploring to exploitation to maximize reward

18
Q

What is a hockey example of the exploration, exploitation trade-off?

A

exploration - you find out goal tender is weak low
exploitation - shoot low to score goals (maximize reward)

19
Q

What is the basal ganglia?

A

Collection of subcortical structures in the brain

20
Q

Where is dopamine produced?

A

In the substantia nigra

21
Q

What is dopamine?

A

a neurotransmitter that is part of the brains intrinsic reward system

22
Q

basal ganglia:

What are the 2 parts of the striatum?

A
  • caudate nucleus
  • putamen
23
Q

What is dopamine input into the striatum critical for?

A

learning from reward and strengtheing the representation of specific actions

24
Q

Learning the piano example:

What happens to serial actions with learning?

A

can produce faster sequences with less errors. Key presses become smoother and linked together

25
Q

What can practice do in terms of sequential actions?

A

Can link sequential actions into a single movement pattern

26
Q

Chuncking of individual elements in a motor sequence

What happens in early practice?

A

shifting gears requires several dinstinct movements

27
Q

Chuncking of individual elements in a motor sequence

What happens in middle practice?

A

independant actions are ‘chunked’ into a larger subunit of movement sequences

28
Q

Chuncking of individual elements in a motor sequence

What happens in late practice?

A

actions become ‘chunked’ together into a cohesive movement sequence where successivce actions are ‘coarticulated’

29
Q

What is chunking?

A

fusing a series of individual elements into a larger subunit of a movement sequence

30
Q

What is co-articulation?

A

adjacent movement elements influence each other

31
Q

What does co-articulation cause?

A

a blending of discrete movements within a sequence and is associated with an ncrease in speed and smoothness of movement sequences

32
Q

What does sequence learning alter?

A

processing in several sensory and motor brain areas

33
Q

Early stages of learning:

What is the role of the dorsolateral prefrontal cortext?

A
  • working memory, attention, action selection
  • conscious planning and processing of movements
  • likely involved in intial stages of chunking motor sequences
34
Q

Early stages of learning:

What is the role of the inferior parietal cortex?

A

processing visual and somatosensory information

35
Q

Early stages of learning:

What is the role of rostral premotor areas?

A
  • planning motor actions
  • processing visual and somatosensory feedback
36
Q

Early stages of learning:

What is the role of basal ganglia?

A

processing reward during action

37
Q

Later stages of learning:

What is the role of supplementary motor area?

A
  • selecting and planning previously learned motor responses
  • likely involved in later stages of motor chunking and coarticulation
38
Q

Later stages of learning:

What is the role of the dorsal premotor area?

A
  • processing visual and somatosensory information
  • involved in selecting learned actions
39
Q

Later stages of learning:

What is the primary motor cortext involved in

A

producing motor actions

40
Q

Early stages of learning:

What is the cerebellum involved in?

A

detecting and correcting errors

41
Q

How does brain activity evolve during motor sequence learning?

A

Early stages require high cognitive demand and conscious prcoessing.

Later stages require low cognitive demand and automatic processing.