Task 7: Adding affect Flashcards
A1: Revise the location of the amygdala
Aymgdala (AMY)
- diencephalon
- medial temporal lobe
- part of limbic system
- end of hippocampus (HC)
A1: Why is the AMY relevant for goal-directed decision making & neuroeconomics?
Neuroeconomics assumed rationality, but people show non-rational behaviour/heuristics-based choices/biases/…
- E.g. framing, emotional impact of winning or losing
- > AMY is crucial in this
- > also “tags” events/objects/actions with affective value (positive & negative) -> influences stimulus-action-outcome associations made by the cortex
De Martino (2006) - AMY activation is related to risk-aversion & risk-seeking -> framing effect
A1: What are the two major misconceptions about AMY function?
1) negative or positive?
+ Monkey evidence
1) AMY is only involved in negative emotions, as “protection device”
- Fear conditioning is commonly used to model emotional learning
- But evidence for positive processing too
Monkey evidence
Paton (2006):
- single-neuron measurement in monkeys conditioned to associate visual stimuli with positive/negative valence which was swapped after some time
- If AMY only represents negative -> activity should change
- Results: Activity for both reward & punishment, activity just reversed
-> single cells in AMY respond to stimuli valence independent of positive/negative valence
-> AMY activity reflects stimulus-valence learning & later reflect switching
Braesicke:
- viewing high-/low-incentive food, anticipatory (AP) vs. consummatory periods (CP)
- before lesion: skeletomotor responses in AP & cardiovascular in both AP & CP
- after AMY lesion: no cardiovascular response in AP & still other 2 responses
- > AMY contributed to positive affective association with conditioned stimulus
Study on reinforcer devaluation
- no lesion -> sensory-specific satiety, food preference available
- AMY lesion -> no sensory-specific satiety, intact preference
- > AMY role is limited to updating monkey’s estimation of current value of food
- > once updated AMY is not needed for choice (representations of expected food value stored elsewhere)
A1: What are the two major misconceptions about AMY function? 1) negative or positive? \+ Rat evidence \+ Human evidence \+ conclusions
Rats learn S-R associations -> later spend more time near associated stimulus
- > tendency to associate physically with positive stimuli
- Central AMY nucleus lesion -> no approach behaviour
- BLA lesion -> no approach behaviour
- > both regions of AMY mediate positive affect
Somerville
- humans learning face + name + valence associations
- right AMY
- > selectively sensitive to emotional descriptions (both positive & negative)
- > generates nonspecific arousal signal
- > could also encode stimulus-valence associations
Johnsrude
- subjects preferred images paired with high-reward probability (vs. low/mid)
- AMY lesion -> no more preferences
- > AMY mediates an association between sensory inputs & their affective valence
- > People can be unaware of the associations but behaviour is the basis of them
- > Role of AMY is just as important for positive emotions as for negative ones
A1: What are the two major misconceptions about AMY function?
2) S-R?
- Reasoning
- Monkey evidence
2) AMY contributes to S-R learning & emotions are by-product of reinforcement
Rationale:
- AMY lesion -> impaired reward-based object selection
-> interpreted as: AMY forms S-R & by-product emotions
-> BUT: inaccurate lesions
Monkey studies for emotional reaction (without S-R)
- Task: grab food from box with spider or neutral object
- Healthy -> intact emotional reaction (hesitating + defensive reaction to spider vs. neutral)
- AMY lesion -> no hesitation or defensive reaction
Evidence against involvement of reward processing
- precise AMY lesion -> no impaired “object-reversal learning” and “win-stay lose-shift” task
- > both are S-R tasks that do not depend on AMY
- > lesion had no effect on S-R but on emotional responses -> need to be differentiated
A2: Specify the role of AMY in goal-directed decision making
Keyword: AMY-OFC network (Murray)
AMY has reciprocal connections with OFC & sensory areas (IT, PRh)
AMY -> OFC interaction: affective info
1) AMY updates reward values
- > only temporarily necessary for acquiring S-R
2) OFC stores reward values
- > always necessary
- AMY lesion -> OFC can’t store S-R properly
- OFC -> AMY: feedback, important for updating
- Decision-making is guided by these outcome values
IT/PRh -> OFC interaction: processing visual info
- implementing visually guided rules (e.g. object reversal learning)
- AMY not involved in object-reversal learning tasks
AMY -> IT/PRh interaction: attentional modulation
- enhanced sensory processing of significant stimuli/events
A3: Briefly explain the following terms:
- Rationality
- Framing effect
- Loss aversion
Rational decision-making: Invariance/consistency in choices regardless of how choices are presented
Framing effect: Same information given different meaning based on its framing
- possible explanation: incomplete information -> relying on heuristics
- Gain frame -> risk-aversion -> chose sure over gamble option
- Loss frame -> risk-seeking -> chose gamble over sure option
- > subjects are unaware, frame does not affect RTs
Loss aversion: Tendency to prefer avoiding losses to acquiring equivalent gains
e. g. of 50€ you lose 30€ or get 20€
- > stronger meaning given to loss than to win
Risk-aversion vs. loss-aversion
- LA -> deals with difference between gain vs. loss frame without choice for gambling
- RA -> deals with difference between safe vs. gamble choices within gain frame
A3: Describe De Martino’s neuroeconomics experiment
- AMY
- MCC
fMRI:
- Choices in line with framing effect (G(sure) + L(gamble))
- > increased activity in bilateral AMY
- > decreased activity in MCC (+ even SMA)
- Choices opposite to framing effect/more rational (G(gamble) + L(sure))
- > decreased activity in bilateral AMY
- > increased activation in MCC (+ even SMA)
- -> AMY activation is related to risk-aversion & risk-seeking in line with framing effect
- independent of valence
- framing effect is driven by heuristics by emotional system
- -> MCC (+ even SMA) activation suggests opponency between 2 systems
- MCC -> detecting conflicts between “analytic response tendencies” (rational choices)
- AMY -> detecting more “emotional” tendencies (heuristic choices)
A3: Describe De Martino’s neuroeconomics experiment
- OMPFC
OMPFC activity correlated negatively with subjects’ susceptibility to framing effects
- not predicted by AMY activity
- strong reciprocal connections between AMY & OMPFC (Murray)
Current theory:
- OMPFC incorporates inputs from AMY
- OMPFC represents value of stimuli -> to guide future behaviour
More rational individuals:
- may have better representations of own emotional biases
- > ability to modify behaviour appropriately (e.g. when biases are suboptimal)
- -> evidence for a model in which OMPFC evaluates/integrates emotional & cognitive information
- -> this underlies more “rational” behaviour + ability to notice own heuristics + susceptibility to framing effect
A3: Discuss how De Martino’s study confirms Murray’s hypothesized role for AMY
Reminder:
Murray:
- AMY updates outcome values -> OFC then stores outcome values
De Martino:
- AMY active for heuristic decisions
- OMPFC active for susceptibility to framing effects
Current theory!:
- OMPFC incorporates inputs from AMY
- > evidence for a model of OMPFC evaluating/integrating emotional & cognitive information -> leading to more rational behaviour
- > Murray describes exactly such a model
A4: Explain the Hampton (2007) experiment on reward expectancy & rule switching
- Aim
- Methods
Aim:
- AMY & OFC/vmPFC activate to rewarding/punishing outcomes & subsequent behavioural decisions
- Hypothesis: AMY lesion -> OFC/vmPFC deficit in neural representations of reward expectancy
Methods:
- 2 patients with focal bilateral AMY lesion, SM & AP vs. 41 matched healthy controls
- Reversal learning task
- > learn choice-reward association
- > flexibly switch choices when rule is changed
- > probabilistic (p(reward)<1) vs. deterministic (p=1) task variety
- analysed switch trials vs. stay trials
A4: Explain the Hampton (2007) experiment on reward expectancy & rule switching
- Results in healthy participants
- Behavioural choice signal
Switch trials: greater activation in
- anterior frontal insula
- > reflects increased cognitive control in response to negative feedback prior to response scheme switch (in the next trial)
- posterior lateral OFC (PLOFC)
- > S-R storage + updating
- > negative feedback (prior to switch trials) on previously rewarded stimuli is important update trigger!!
- MCC
- > monitoring outcome + response error
Stay trials: decreased activity in
- mPFC
- > less cognitively demanding
A4: Explain the Hampton (2007) experiment on reward expectancy & rule switching:
- Results in healthy participants
- Expected Reward Signal
- significant correlation with signal in mOFC & mPFC
- > time-locked
- > increased linear with increasing expected reward value
A4: Explain the Hampton (2007) experiment on reward expectancy & rule switching:
- Results in healthy participants
- Responses to reward/punishment
- Reward -> significant activation in mPFC & mOFC
- Punishment -> significant activation in anterior lPFC & lOFC
A4: How do Hampton’s results fit with the two previous tasks (cortex reward & error monitoring) & Camille (2011)
- Insula, vmPFC, lPFC, lOFC, MCC
Insula
- Task 6 -> error + negative feedback
- Here -> error + negative feedback leading to switch
vmPFC
Task 5 -> Default network
Here -> less cognitive energy needed
lPFC
- Task 5 -> state representation
- Here -> specific activation to punishment
lOFC
- Task 5 -> Value representation (S-R) + updating (e.g. for satiety)
- Here -> S-R storage + updating due to error + negative feedback (expected reward, rule-switching)
MCC
- Task 5 -> action selection, outcome evaluation (A-O)
- Task 6 -> error detection
- Here -> error detection, leading to rule-switching
Camille (2011)
- similar reversal learning task
- > OFC damage -> more errors during initial learning/fewer reversal -> for stimulus value
- > MCC damage -> more errors during initial learning -> for action value
