Lecture 5 Flashcards
How can we study the influence of reward on selective attention?
– Behavioural studies (visual search)
– ERP studies
– fMRI studies
How do you learn from errors [correct action]?
If you perform correctly, there will be a zero error signal. No need to update.
If the network is working perfectly, no need to
change parameters.
How do you learn from errors [incorrect action]?
If you do make an error, error signal will be made. Some kind of sensorimotor signal to say it was not as expected.
Feedback continues until action is performed
correctly.
What are primary reinforcers?
Positive: water, food.
Negative: pain.
(Mostly used in animals due to ethics)
What are secondary reinforcers?
Positive: adding arbitrary points, money.
Negative: subtracting arbitrary points.
(Mostly used in people)
What must a reward be for learning to occur?
Unpredicted or surprising.
Status quo: there is nothing new to learn. Learning only occurs when something is different to what was expected.
We can also think of a reward as…
A prediction error.
Reward predicted IS NOT EQUAL TO reward obtained.
Planning and judgement are located in…
Prefrontal cortex
Frontal cortex.
Reward is located in…
Nucleus accumbens
Medial forebrain bundle
Ventral tegmental area.
Emotions and conditioned effects are located in…
Amygdala.
Explain dopamine activity slope in the reward process [not fulfilled].
Spikes in expectation, continues to rise, decrease when outcome is worse than expected.
Explain dopamine activity slope in the reward process [fulfilled].
The is a long slope towards the final reward. Starts to
increase until the moment of reward. Expectation increases signalling in the brain.
What is a rewarded non-movement?
Rewarded non-movement: get a reward for not performing an action.
Example is not talking in lecture. My example is dog staying.
What is an unrewarded movement?
Cued to make a movement and he does but we do not give him a reward. Eventual extinction.
How does reward influence ‘pop-out’ during visual search?
-People become a little bit faster when the
reward is higher.
-Magnitude gets bigger when
there are more repetitions.
Top condition, high reward, get 10 points 75% of
the time.
Bottom condition, low reward, get 1 point 75%
of time.
