Decision making Flashcards
Prediction-Choice-Outcome Loop
- We start out with a goal
- If there are several options, how to achieve this goal, we make predictions about the outcome of different options
- We then form a decision and make appropriate actions
- The result of these actions will be an action outcome that we observe
- This outcome will be subjected to internal monitoring processes where we evaluate, if our decision and the corresponding actions have actually achieved our goal. If we haven’t reached our goal, our brain seems to generate a prediction error
- Importantly, the result of this monitoring process, the prediction error, can be used to update our memory based on the experience we have just made. This updated memory content can then be used to make more precise predictions the next time when we face the same or a similar choice. - guide decisions next time
General features of decision making
Avoid harm
Minimise: cost of time, effort and missed opportunities
Maximise reward
Factors to consider before making a decision
- Difficulty of the action (effort)
- Probability of success and failure
- Fellows (2018)
- How valuable is the possible reward at this moment (-> context)?
- Hungry? Diet? —> goals
- Missed opportunities?
Fellows (2018); Frömer & Shenhav (2022)
Biases in DM
Stick with same option
Certain gains over gambles
Gambles over certain losses
Temporal discouting
Random-dot motion task
- Monkey maintains fixation
- Random dots are presented
- Certain percentage of dots moves coherently, other dots move randomly
- When monkey detects main motion direction → eye movement to corresponding target
- Noisy sensory signal is converted into discrete motor act
- Hanks & Summerfield (2017)
3 stages of perceptual DM
- Detection of sensory evidence; What are the alternatives that can be detected (left/right, red/blue, etc.)?
- Integration of evidence over time
-> because evidence is noisy - Checking if threshold has been reached
-> if so, elicit appropriate action
-> if not, accumulate more evidence
- Where does evidence accumulation take place:
- Brain areas responsible for encoding the relevant feature, e.g. area MT/V5 if motion is relevant for decision
- Parietal and dorsal prefrontal cortex
- Recent evidence: sensorimotor areas representing possible actions, accumulate evidence as well
Mental maps in decision-making
encoding of transitive relations to build mental map
Kaplan et al. (2017)
- Experiments by Edward Tolman: Rats experiencing a spatial maze
Schuck et al. (2015)
Study showing upcoming strategy switch in DM
- Identify arrangement of pattern within white square:
- Upper left or lower right corner = right button press
- Upper right or lower left corner = left button press
- (→ instructed S-R mapping)
- No instruction regarding colours.
- But colours could be used to determine response as well after association had been learned (→ learned S-R Mapping).
Schuck et al. (2015) showed that:
- Some aspects relevant for decision-making are represented in MFC
- MFC activity seems to be linked to memory formation in a decision-making task
- Activity in MFC predicts a shift in strategy based on a learned association
Representation of hidden states
- state space: cognitive map specific for a given task, grid-like representation
- state spaces represented in orbito- and medial frontal cortex
- hidden state: our position within the current task
- mental exploration: evaluation of potential outcomes for different choices
- state spaces help structure new experiences
- Kaplan et al. (2017)
Frontal lobe lesions and decision making
Patient studies
- Lesion in OFC-vmPFC disrupts value-based decision-making, even very simple preference judgments (Henri-Bhargava et al., 2012; Fellows and Farah, 2007)
- After ventromedial prefrontal lesions people show:
- inconsistent preferences
- deficient sense of guilt
- Kalat (2015)
Trust Game
(Kalat, 2015)
- A can decide how much of the money to give to the trustee (B):
☞ amount given to trustee will triple
☞ trustee can then decide how much money to give back to A - Patients with lesions in ventromedial prefrontal cortex:
- If in the role of person A: will give less to B
- If in the role of the trustee (B): will keep nearly all money instead of returning it
Lateral prefrontal cortex (PFC)
- Seems NOT involved in value-based choices
- But active in many decision paradigms (Mc Clure et al., 2004; Daw et al., 2005; Hutcherson et al. 2012)
Reward rate maximization and the urgency signal - Carland et al. (2019)
Aim of individuals: maximize subjective reward rate = reward rate maximization
maximise reward
minimise: time and effort