Problem Solving and Decision Making Flashcards
What characterises “thinking”?
- Thinking is the s_ystematic transformation of mental representations of knowledge_ to characterize actual or possible states of the world, often in service of goals.
- Higher-order cognition is built on other aspects of cognition (perception, memory, language)
- Key aspect is creating and using knowledge, rather than extracting knowledge.
- Thinking is built on other aspects of cognition that are imperfect
- Recent in terms of evolution: mostly centered on frontal lobes
Who is Piltdown Man and what does he suggest about human cognition?
- 1908 to 1915: fossil parts found near Piltdown, UK; ‘Missing Link’
- Clearly human skull and ape-like jaw
- Appeared 500,000 years old
- Reconstruction suggested certain wear patterns on molars, then found
- Problems
- Suggested brain before jaw loss - didnt fit other skulls (ignored contradicting evidence)
- Piltdown man flouride tests - 50,000yrs old, artificial abraison = HOAX
- Why did we fall for it?
- confirmation bias: fit current theory, ignore contradicting evidence
- Bystander apathy: experts assume others did the work
- Emotion: national pride
How are heuristics, old information and analogies used in decision making?
- A heuristic is a short-cut used to make up for uncertainty in the environment (by attempting to use what you already know)
- rule of thumb - can be useful but lead to predictable errors like in Monty Hall Problem
- Substitutes hard questions for easy questions (will it rain today vs is the sky dark today)
- Past information: Often apply what we learned in one situation to new situations
- Analogies: solve electrical flow fault by analogy to water flow
- Priming people on past events influences their responses to current problems
- Analogies: solve electrical flow fault by analogy to water flow
What is problem solving? Is it a purely human capability?
- Problem solving is any goal directed behaviour that requires something that isnt immediately available
- Several species are known to exhibit problem solving behaviour
- Crows: demonstrated use of tools (not learned or modelled from humans). Successful fashioning and use of tools 10/17 trials for female.
What is the Gestalt approach to problem solving?
- Emphasized importance of changes in perspective, prior knowledge, and assumptions.
- Köhler (1925): studied chimps solving problems.
- Duncker (1945) emphasised analogy
- Approach demonstrated a number of important phenomena, especially restructuring representations
What are ‘functional fixedness’ and ‘set effects’?
-
Functional fixedness is an inability to break mental sets, leading to a failure to see things in novel ways
- Dunker Candle Problem: Need to use the box the materials came in to solve the problem
- 86% solved when box was empty, 41% when not
- Apollo 13
- Dunker Candle Problem: Need to use the box the materials came in to solve the problem
-
Set Effects are when previous solutions remain in the brain and block generation of new solutions
- Water Jug problems: given a series of difficult waterjug problems then given a simple one with easy and hard solutions. People overlook simple solution
- Can be very long lasting: Le Verrier used newtons laws to find neptune from anomilies in uranus orbit - thought mercury’s wobbles same cause until he died
What is the ‘Problem solving as a search’ (Problem Space theory) approach to problem solving?
- Newell & Simon (1970) proposed problem solving as a search through a problem space of possibilities
- Search can be physical or mental
- Can’t be exhaustive search (e.g., chess), so need strategies
- Start with situation we want to transform into something new. Do so by applying the different options we have.
- Terminology
- State: Specification of situation.
- Goal: The desired state.
- Operator: An action that changes one state into another.
- Solution: A sequence of operators that transforms initial state into goal.
- Constraints: Restrictions on what can be done.
What is the ‘generate test’ strategy in problem solving as a search?
- Generate Test: Randomly generate solution, then test it.
- Advantage: requires no knowledge
- Disadvantage:
- sometimes ineffectual (Generation may be hard, testing could be hard, search space may be very large).
- May reach solution too slowly
What is the ‘difference reduction’ strategy in problem solving as a search?
- Difference Reduction: Try to reduce difference between current state and goal state
- Advantage:
- Steps can be small,
- don’t have to know much, just what gets you closer
- Disadvantage:
- Requires some knowledge,
- it may not be possible (or desirable) to get closer on each step
What is Means-End Analysis?
- Involves dividing the task into subgoals in order of importance/requirements
- Compare current state to goal state and identify differences
- Select an operator to reduce the largest difference.
- If operator cannot be applied set subgoal of creating preconditions for its application.
- Return to (1) until goal is reached
- Example: Tower of Hanoi
- goal to move all disks to different arrangement
- set subgoals - move 1st disk
What is the evidence for Means-End analysis as a problem solving search strategy?
- Evidence from verbal protocols (Anderson, 1983)
- People often spontaneously set subgoals.
- Catrambone (1995): better performance in math problem solving when instructed to set subgoals.
- Egan & Greeno (1974): a problem becomes more difficult with more subgoals.
- Probability of error increased with the number of subgoals necessary.
- Patsenko & Altmann (2010) showed that people were not using detailed planning in Tower of Hanoi
- Changing number of disks during task did not change actions
What are the advantages and disadvantages of problem space theory?
- It’s a normative theory: specifies ideal
- People are often not so systematic
- Provides a framework for thinking about problem solving.
- Not refutable
- Disadvantage: underplays the role of knowledge in problem solving.
- Both representation and process are important
What is the knowledge and problem solving approach?
- Role of knowledge:
- Knowledge can eliminate problem solving. If know the solution, then no problem.
- Knowledge can provide us with the most creative solutions e.g., analogies
- But knowledge can be a barrier to problem solving if given wrong constrains
- Unecessary constraints can be part of the representation
- Insight problems: incubation reset
- Hints : 25% solve with hint, none without - hint is required to solve
- Unecessary constraints can be part of the representation
What is the role of transference in forming representations in problem solving?
- Transference is the application of old knowledge to a new problem
- Negative transfer: functional fixedness and mental set
- Positive transfer: using prior knowledge to help you solve a problem
- eg Analogies
- Analogies - positive transfer example
- used in standardised tests of intelligence, often a basis of creative solutions
- transference of analogies can be used over long periods of time
- 7 days after tutorial, 3 weeks if repeated
- Example: those who landed on ‘convergent’ solution, given analogy first, given analogy and hint
What is decision making?
-
Choosing between options on the basis of the available information.
- Not reasoning (which derives new conclusions from available information.) but often intertwined.
- Requires judgment: Could just calculate expected values (but we dont always have the info)
- Primary obstacle is uncertainty
What evidence is there that people tend to discount value for uncertainty?
-
Discount value: People tend to discount outcomes for uncertainty
- Green et al (1999) found people would take below expected value when uncertain (chance/value not linear).
- Can change for very low odds
- Discount $100,000 more than $200
-
Future uncertainty: Prefer $100 now to $100 next week
- Hyperbolic line: discount more in the shorter term than longer (years)
- Kids discount future more than young adults and than older.
- Level of trust in the future, percieved stability
- Men discount more after looking at pictures of “hot” women
- emotional - present grounding?
What is Expected Utility?
-
The Expected Utility Hypothesis:
- Utility is calculated by considering the value of each possible outcome and constructing a weighted average, based on probability of each option.
- Chose the outcome with highest expected utility
- Eg buying car: desired outcomes are economy and relaibility, car A has 60% e, 40% r, utility = .6e +.4r
- In practice
- Deal or No Deal: bank offers less than the EV
- Shows that people don’t tend to actually use EU theory
- Subjective goals = satisficing (aim for X amount not best amount)
- discount for uncertainty = more risks on higher amounts
- losers/winners take more risks = effect of past decisions
What are some problems with Expected Utility theory?
- Often doesn’t fit to empirical data.
- Leads to various paradoxes: EU shouldn’t be affected by the past but it is (deal or no deal)
- Doesnt explain “Sunk cost” fallacy
- doesnt fit with the effect of irrelevant info on decisions
- Probabilities and utilities may be subjective, based on our own experience.
- Could represent individual beliefs
- The outcome goal is not always the “best” option
- Satisficing = operating under cognitive constraints
- Savage (1954) developed subjective expected utility theory.
- Can think of expected utility theory as a normative theory
What is the overconfidence bias and how is it adaptive?
- Overconfidence bias is when a persons subjective confidence in their own judgements is substantially higher than their objective accuracy
-
Examples:
- CFOs asked to predict S and P index over a number of years
- no correlation between estimates and index
- Asked CFOs to give a 90% confidence range
- predicting a 20% ‘suprise’ rate, actually 67%
- CFOs asked to predict S and P index over a number of years
-
Advantages
- Underconfidence may be even more problematic (inability to make decisions, existential dread etc)
-
Disadvantages:
- Bias greater in more difficult tasks
- Bias grows with information aquisition regardless of accuracy
What is the Availability heuristic?
- Availability heuristic = Judgments based on ease with which relevant instances can be retrieved from memory
- Based on ease not number of recall
- effected by media
- Benefits
- memories strengthened by repetition, indicates importance
- ease of recall in 7s good indicator of total knowledge in 2m
- Disadvantages
- Can lead to systematic errors
- Judging anecdotal evidence over empirical evidence
- words starting with R vs words with R as 3rd letter
- Can lead to systematic errors
Give some examples of the availability heuristic
-
Ease vs Amount Schwarz, et al Assertiveness study
- Recall either 6 or 12 instances you were assertive, 6-group rated the task easier
- 6 group rated themselves more assertive
-
Probability and consistent errors (Slovic, Fischhoff, & Lichtenstein study):
- rated death by fire more probably than death by drowning
- rated plane-crash, cancer, earthquakes as killing more people than driving
- Consequences: People risk driving instead of ‘more dangerous’ flying
-
9/11: terrorism exploits the availability bias
- Severe rise in road fatalities after 9/11 due to driving instead of flying to destinations
What is the Representativeness Heuristic?
- representativeness heuristic: If something or someone appears to fit a category, you will use what you know about that that category to make judgments
- Examples:
- coin flip: which sequence is more likely, choose one that includes the other - representativeness over probability (shorter sequence always more likely
- Rate probability of Steve being a librarian or a librarian who also does yoga, or a librarian vs a sales person (many more sales people in US)
- Disadvantages
- Ignore base rate information: given probabilities, people will choose representativeness instead
What is the Loss Aversion Bias? How does it relate to the endowment effect?
- Loss aversion refers to the tendency for people to strongly prefer avoiding losses than acquiring gains
- Described by Kahneman & Tversky’s Prospect Theory**
- Eg many people reject a 50-50 bet in which they can win $200 but lose $100
- The Endowment effect: Sellers vs Choosers:
- Sellers given mug to sell, choosers asked how much they would pay
- Reframed: sellers ‘lose’ mug, choosers ‘gain’ mug
- Sellers place a higher price on “whats mine”
- Adaptive: losses can threaten survival
What is the Status Quo Bias?
- Status quo bias is an emotional bias; a preference for the current state of affairs.
- Related to loss aversion (which maintains status-quo)
- Example:
- Samuelson & Zeckhauser (1988): you have inherited a large amount money as blue-chip shares, risky shares, T-bills or bonds.
- Allowed to change to any other ‘form’ at no cost
- People show strong tendancy to stick with original form
What are framing effects?
- Framing effects change the representation of the problem and the way the options are evaluated
- Example: Context of coin toss:
- People took more risks to avoid ‘loss’
- assume you are $1000 richer, choose between sure gain of 500 or chance of 1000 vs assume you are $2000 richer, choose between loss of 500 or chance of loss of 1000
- Real world applications
- Organ donation rates: rates in opt-in vs opt-out programs
- Should we use this to ‘nudge’ people toward positive policy?
How can irrelevant infomation effect decision making?
- We are good at comparison, poor at absolute valuations. Contradicts EU theory which expects each option to be evaluated independently
- Simonson & Tversky (1992) offered two possible payments for doing an experiment:
- Take $6 or a nice pen: 36% nice pen, 64% take cash,
- Take $6, nice pen, or cheap pen: 2% cheap pen 46% take nice pen, 52% take cash
- Real World Applications
- SMH Subscription options: making a bad subscription option promotes choice
- Doctors more likely to prescribe drug if given 2 options
What is anchoring and adjustment?
- Anchoring is the tendency to rely too heavily on the first piece of information offered, choosing to adjust from the “base rate” by small variations
- Example:
- Tversky & Kahneman: you have Job A or Job B, but must change to Job X or Y
- job A is more similar to job X, Job B is more similar to job Y
- People consistently choose the job most similar to the prime
- Judges sentencing: roll dice and say whether sentence should be more or less in months - significantly changes initial sentence
- Tversky & Kahneman: you have Job A or Job B, but must change to Job X or Y
Give an example of how people ignore base rates and its implications
- Tversky & Kahneman Base rate neglect:
- Given the percentages of people at a party, will still decide on most likely profession ignoring this base likelihood
- HIV: Lots of low-risk people are tested each year
- If you are low risk and test positive what is the likelihood you have HIV?
- rate of false positives = 0.01%, HIV rate 0.01% in low risk group
- If testing 10million low risk chance of false postive = >50%
- if testing 10 million high risk chance of false positive = >1%
- If you are low risk and test positive what is the likelihood you have HIV?
- Implications
- Testing whole population for HIV is impractical
- Getting whole body scans initially, vs initial diagnosis and subsequent specific tests
What are the adaptive benefits of heuristics and biases?
- We have limited memory, cognitive capacity, and time, so make the best decisions we can rather best that are possible
- Bounded Rationality
- Use fast & frugal heurstics
- during emergencies, or risk assessment decisions need to be made fast
- We pick-up a lot of valid information from environment - assuming a city you recognise the name of is larger than one you don’t is usually accurate
- Making good enough decisions, not the best decision