Decision Making 2

to know the expected value/utility of a potential outcome, we have to weigh

how good/bad it will be, and how likely

reward magnitude and probability are integrated in

vmPFC/vStr

expected value signals appear even for

highly abstract outcomes

we are ___ for gains

risk-aversive

we are ___ for losses

risk-seeking

we place less value in a reward if it comes with a cost, like

an actual price

having to wait to get it (delay discounting)

having to work for it (effort discounting)

(price costs) vmPFC often tracks

overall (discounted) value of the reward

(price costs) insula, ACC often track

costs of reward (price)

(price costs) vStr (NAcc) nucleus accumbens

reflect product value

(time costs) vStr NAcc reflect

future $$

(time costs) LPFC reflects

cost (delay)

(time costs) vmPFC reflects the

delay-discounted subjective value of future $$

“tagging” a delayed reward with a specific episode (e.g., vacation) can make people

discount it less, via vmPFC interactions with MTL

(overall) vmPFC tracks discounted reward value across

delay, effort, probability

individual differences in delay and effort discounting have been used to predict

impulsivity (e.g., substance use) and motivation (e.g., apathy) outside of the lab

weighing choice attributes

vmPFC reflects combined value of attributes, weighed by importance

(weighing choice attributes) lateral OFC reflects

low-level features (e.g., of foods)

LPFC was linked to

health-related focus in foods

risk

the outcome is uncertain but the chances are known

ambiguity

outcome and chances are uncertain

people tend to prefer risk to

ambiguity

often people choose more uncertain options

(explore) so that they can learn something new (vs. exploit known ones)

anterior LPFC is associated with

exploration and evaluating counterfactuals (e.g., value of the option you didn’t choose)

decision-making

process of accumulating information supporting each of the possible choices (evidence)

decision threshold

once you have accumulated enough evidence for one choice over the alternatives

drift diffusion model

popular form of evidence accumulation model

evidence we accumulate is

noisy (affected by random factors)

decision vary in

the amount of evidence we have in support of each of the choices, determining the evidence accumulation rate

how much information we require before making a decision (threshold)

whether we start out the decision process closer to one threshold (“thumb on scale”/response bias)

decision unrelated processes (e.g., perception, action)