M06 - Decision Making Flashcards
What processes are needed for decision making
- goal-directed behavior
- reward learning & memory
- executive control
- inference establishes causal links
- often in a social context
What are the two discipline based models in decision-making?
- economic models
- psychological models
What type of choice do you have in an economic model?
Rational choice but choice bias
What is the type of choice in a psychological model?
Choices are context-dependent (social, cultural, moral)
What are economic decisions?
- sophisticated rational beliefs
- stale preferences
- maximizes own payoff
- disregards other peoples well-being
What is the disadvantage of economic decisions?
Prone to biases
What disciplines make up neuroeconomics?
- psychology
- economics
- neuroscience
What decisions are made in neuroeconomics?
- value-based decision making
Is social value accounted like non-social value?
Are the underlying circuits similar?
Which brain chemicals are involved?
What are the core concepts in decision making?
- expected value
- risk & ambiguity
What is a risk?
Known variance of a distribution (you know all the possible outcomes of the distribution)
What is ambiguity?
Unknown variance (outcomes) (we don’t know the entire distribution)
What is risk aversion?
Prefer a lower, certain reward over a risky higher (on average) reward
What is risk seeking?
You need to be paid more than the expected outcomes to not take the gamble
What is the purpose of rewards?
To reinforce behavior
What brain region is largely destroyed in Parkinson’s disease?
Up to 60% of Substantia Nigra
What are the reasons of destroyed SNr in Parkinson’s?
- trauma, tumors
- encephalitis lethargica
- MTPT (frozen addicts)
What happens to dopamine in Parkinson’s patients?
Up to 80% of dopamine is lost because brain desperately tried to compensate (reducing degradation, increasing receptors)
What is the role of dopamine in learning?
- plays a large role in reinforcement learning
- acts as a teaching signal
- learning from (un)expected outcomes
Where does dopamine act as a teaching signal?
- Reward Prediction Error
- Rescorla-Wagner learning rule
- Basis of associative learning
What is the Rescorla-Wagner learning rule and prediction error?
→ mathematical rule used to describe and predict classical conditioning
→ basic idea: learning occurs when there is a discrepancy or prediction error between what is expected and what actually happens
→ rule suggests that learning occurs when actual outcome differs from what was predicted based on the existing associations
What is Neuroeconomics for?
concepts and methods of neuroscience could help resolve issues in economic and social sciences
What is an expected value?
The magnitude of an offer * its probability
What is utility?
Subjective valuation of outcomes (an apple could be worth more to me than you)
What does the certainty equivalent show?
It can show us the risk aversion level
How does a risk neutral individual approach a gamble?
There don’t prefer one outcome over another because they don’t have a risk attitude
How does a risk-averse individual approach a gamble?
The mean gamble payoff needs to be higher than the certainty equivalent
aka positive risk premium
aka happy to take a certain outcome even though it is lower than the expected outcome
Reward Signals influence processing in the whole brain. True or False.
True.
What is one of the most influential learning rules?
Temporal Difference Learning
What is temporal difference learning?
- reward prediction is not just dependent on what is happening to right now
Reward prediction error = actual reward at the time + predicted reward in the future - predicted rewards at this time
What is ambiguity aversion?
The difference what you would normally pick without ambiguity (nothing hidden) vs pick with ambiguity (hidden info)
What is the hyperbolic delay discounting?
You have small immediate reward (SS) vs large, delayed reward (LL). If you put the delay for the LL to a certain time, it’s value will become same as the SS and you “might as well” choose the SS. The value of future rewards decays hyperbolically.
Humans but not monkeys follow hyperbolic discounting. True or False.
False. BOTH humans and monkeys follow hyperbolic discounting.
How can you map out the discount curve?
By pushing out the reward and increase it in size you can map out the discount curve
Humans generally act pro-social (even in one-shot anonymous interactions). True or False.
True
What drives pro-social behavior?
Evolutionary account:
- stabilizes long-term social bonds
- increases group fitness
What are the 2 possible brain mechanisms behind pro-social behavior?
- High-valuation-specific
- Extended common currency
What is the Social-valuation-specific?
Network in the brain that is devoted to social value as opposed to non-social value (food etc.) are distinct
What is the extended common currency?
Social and Non-social processed separately but ending up in the same mechanisms
What are the main influences on trust?
- reputation matters (trust based on feedback, grows as you grow up)
- perception of moral character (stories about a person affect trusting behavior)
What is the reflection effect?
People become risk-seeking in the loss domain
In what situation does dopamine act as pleasure?
- recreational drugs activate DA receptors
- medial forebrain bundle electric stimulation prioritizes behaviour over basic needs
When does dopamine act as motivation?
- destroying DA neurons does not change “liking” of rewards
When does the dopamine level increase?
- when unexpected reward occurs
- when cue predicts reward fully
When does the dopamine level decrease?
- when outcome is lower that expected
What brain region is modulated during an unexpected gain?
Anterior cingulate cortex (ACC)
What brain region is activate during unexpected events?
Basal Ganglia
What brain region is activated during risk taking?
Insula
What brain region is active during ambiguity (aka incomplete information)?
lateral PFC
What are neural signatures of reputation?
- multi-round trust game (hyperscanning fMRI)
- players learned to build trust (activity correlated between brains)
- when reciprocated, players anticipated trust-reciprocation (potential signal of upcoming reward)
When does altruistic punishment occur?
- in a 2-person interaction (tit-for-tat)
- in a multi-person interaction (Public Goods games
How can you do common currency evaluation?
Through Padoa-Schioppa
What is Padia-Schioppa
- presented monkey with different rewards
- psychometric curves establishes “rate of exchange”
- neural activity reflected “integrated value”
How does value-based decision-making change in vmPFC (vento-medial PFC) patients?
vmPFC patients do not adapt their choice away from risky decks with learning/feedback. They retain their levels of risk -sensitivity in a betting game in the face of losses.
Explain drift-diffusion models.
- simulate the integration of evidence over time
- represent the integration of positive and negative information
- applied to various decision-making tasks
What brain area in mainly involved in drift-diffusion models?
dorsolateral prefrontal cortex (dlPFC)
- setting and maintaining the decision threshold
- influences the speed and accuracy of decisions
