Flashcards

Perception and Cognition TB9

Emily Hargreaves English 82 cards

Cards in this deck (82)

1What is problem solving?1. Two states: current position & goal
2. Route between two not clear
3. Route takes multiple steps
• Purposeful ‐ goal‐directed
• Controlled processes – not fully automatic (uses frontal lobes)
• Lack appropriate knowledge to generate immediate solution
o Example: making a cup of tea
2Planning and frontal lobes • Goel & Grafman (2000) o Architect with frontal lesion (white area) o Asked to design a lab/office space• Struggled to progress from problem structuring to problem solving
• Poor final design due to problem solving difficulties
o Understood information but could not arrange it in an appropriate way to create a suitable layout
3Well defined vs. Ill-defined• Well‐defined:
o Current position, possible moves, and goal well specified
 e.g., Chess, problems in the lab (easier to measure)
• Ill‐defined:
o Current position, possible moves and goal not well specified
 e.g., most problems in life!
4Knowledge rich vs. knowledge-lean• Knowledge‐rich:
o Only solvable via relevant knowledge
 e.g., chess, again!
• Knowledge‐lean:
o Can be solved without needing prior knowledge
 All information contained in presentation of problem
• e.g., most lab‐based examples
5Insight• That ‘ah ha’ moment
• Point at which solution to problem is suddenly seen, or more verbosely:
o “any sudden comprehension, realisation, or problem solution that involves a reorganisation of the elements of a person’s mental representation of a stimulus, situation, or event to yield a nonobvious or nondominant interpretation”
6Is insight special? • Metcalfe & Wiebe (1987)Participants rated “warmth” during problem solving task – proximity to goal/solution
o Increased activity in superior temporal gyrus
Insight provides feeling of being closer to the end point
• Really separate process, or just differs in terms of phenomenology – i.e., subjective experience?
7Insight provides feeling of being closer to the end point • Really separate process, or just differs in terms of phenomenology – i.e., subjective experience?• Ellis et al., (2011)
o Anagram task: KAFMS (solve 4 letter anagram – 1 irrelevant letter)
o Record eye‐movements during task
o Participants report sudden solution – ‘ah ha!’
o However, gradual decrease in fixations on non‐relevant letter
o Gradual accumulation of knowledge despite subjective experience of ‘ah ha’
o Unconscious?
• Insight is ‘real’ in sense that people experience it
• However, it may not be a separate cognitive process – higher order processes may gradually arrive at solution but we only become aware once threshold reached
• However, does ‘gradual accumulation’ concept make sense for all complex problem solving tasks?
8Facilitating Insight• ‘Hints’ – e.g., two‐string problem (e.g. ‘think about the pendulum in a clock)
• Incubation
o Stop directly thinking about the problem for a period of time
o Most have experienced the feeling of solving a problem during this problem
o Sio & Ormerod (2009) – meta‐analysis
o Incubation led to small but consistent improvements in problem solving, particularly for problems that required more ‘creative’ solutions
 e.g., consequences task – “what would happen if everyone suddenly lost the ability to read and write?”
9Facilitating insight - sleepo Often told to ‘sleep on it’
o Wagner et al. (2004)
 Participants given a series of numbers and asked to make a decision about what number to press based on the number on the screen and the number they previously gave
o Sleep increased insight – number strings where a “hidden rule” is present
 How often participants solve problem is how often they have insight
10What causes insight?• Simon (1966)
o Incubation leads to forgetting of control information about particularly strategy
o Leads to new strategy/approach
• Penaloza & Calvillo (2012)
o Incubation leads to forgetting of irrelevant/misleading information
o Beneficial effect of incubation only when participants forgot misleading information
11• Ohlsson (1992) o Representational Change Theoryo Representing the problem incorrectly
o Insight from a change in representing the problem
o Can occur in three ways:
1. Constraint relaxation
- Relaxation/removal on inhibitions on what is permissible
2. Re‐encoding
- Aspect(s) of problem representation reinterpreted
3. Elaboration
- New information is added to the problem representation
12• Knoblich et al. (1999) o Match‐stick problems (two distinct types)o Participants struggle with Type B due to implied constraint that operators don’t change (based on past experience)
o Need to ‘relax constraint’ to solve Type B problems
13• Reverberi et al. (2005) o Patients with damage to lateral prefrontal cortexo Patients BETTER at harder matchstick problems than controls (82% vs 43%)
o Lateral prefrontal cortex important in constraining solution space, inhibiting unlikely solutions?
o Important function – constraint useful in many situations however impairs in certain situations (Type A vs Type B)
o Paradoxical facilitation!
14• Representational Change Theoryo Experimental support in favour of ‘constraint relaxation’
o However, theory implies this is sufficient for insight, may not be the case – even when people told certain moves possible they still struggle to solve (Ollinger et al., 2014 – nine‐dot problem)
15Strategies • Newell & Simon (1972)o Problem solving from an information processing perspective (e.g., like a computer)
o Limited short‐term memory capacity
o Complex information processing serial
o Can’t hold all steps in mind at once
o Trade‐off: accuracy vs computational complexity (and vs time!)
o Rely heavily on ‘heuristics’ – computationally cheap rules of thumb that can produce reasonably accurate answers
16Heuristics • Means‐end analysis:o Form sub‐goal to minimise distance between current location and goal, move towards sub‐goal
o However, (1) requires info about location of final goal and (2) what if better strategy to move away from final goal (e.g., in maze)
o Hill climbing: – Change present state to one closer to end goal
o Simpler than means‐end, as no explicit ‘sub‐goal’
o Analogous to climbing hill by always moving to next highest point – can get caught in local maxima (climb the wrong hill!)
17Planning • Koppenol‐Gonzalez et al. (2010) o Tower of London tasko Planning before trying to reach goal
o Useful when reaching goal involves sequence of behaviour – e.g., making a cup of tea (again!)
o Tower of London task
o Participants who spent longer deliberately planning moves before acting performed better, with fewer errors than those who didn’t plan
18Progress Monitoring• Track progress towards goal and switch strategy if progress is slow
• MacGregor et al. (2001)
o Performance worse when participants thought progress being made
o When participants realised progress was slow, they were more likely to switch strategies
19Expertise• Lab tasks often knowledge‐lean
o Abstracted away from real world knowledge e.g., tower of London
• Expertise important for solving many real‐world problems
o Problems in real world often knowledge‐rich
• Definition:
o High level of knowledge and performance in a given domain acquired through a long period of systematic study/practice
20Expertise ‐ Chess• Chess experts dedicate 1000’s hours to practice/study
• De Groot (1965)
o Able to remember positions on chess board (but not better memory in general
 Highly specific!)
• Fast processes:
o Charness et al. (2001)
 Eye movements in first few seconds to more ‘relevant’ pieces in experts
• Burns (2004)
o High correlation (0.78‐0.90) between normal and ‘blitz’ chess
 (not a perfect correlation – something else going on – slow processes)
• Slow processes:
o Van Harreveld (2007)
 As time to make move decreases, skill differences less predictive of outcome
• Moxley et al. (2012)
o Longer time to make move leads to better performance
21Expertise – Medicalo Medical decisions really matter – sometimes life or death – unlike chess
o Serious problem – false negative (i.e., miss) <=30% in diagnosis of certain diseases via X‐ray
• Krupinksi et al. (2013)
o Breast biopsies by pathologists in training
o Reduction in number of fixations/image
o Fewer fixations on non‐diagnostic regions
• Kundel et al. (2007)
o Mammograms of possible breast cancer
o Mean search time ~ 27secs, but time to fixate on cancer ~1.13secs (<1sec for experts)
o Correlation between speed of fixation on cancer and overall detection performance
22Experience gone bad • Functional fixedness • Mental Set• Functional fixedness
o Any given object only has a fixed set of uses
o Can’t see novel use of object to solve problem, – ‘fixed’ on typical ‘function’ of object
• Mental Set
o Luchins (1942) – Water jar problem
o Infinite water supply, jugs need to be used to measure target amount of water
o If trained on easy examples previously, participants performed ~95% on other easy problems
o If trained on hard examples previously, participants performed ~35% on other easy problems (or solve with harder strategy)
23Reasoning• Inductive Reasoning
o Drawing general conclusions from premise
o Probably, but not necessarily true
o e.g., extrapolation
• Deductive Reasoning
o Draw definite conclusions if tenets are true
o Based on formal logic
• Informal Reasoning
o ‘everyday’ reasoning
o Relies on knowledge and experience of real world
24Inductive Reasoning• Drawing general conclusions from premise
• Probably, but not definitely, true
o e.g., All swans I have seen are white, so all swans are white
• Hypothesis testing
o Scientists use inductive reasoning: generate hypothesis based on limited data
o Most science is ‘Popperian’ (Karl Popper)
o Falsification of null hypothesis
 Confirmation – gain evidence confirming hypothesis is correct
 Falsification – gain evidence to falsify hypothesis
• Confirmation can never fully support hypothesis
• Falsification can prove hypothesis is wrong
25Wason’s 2-4-6 task• Given three numbers: 2-4-6
• Guess rule (generate hypothesis) that generated these numbers
• Give three further numbers to test your hypothesis
o 21% correctly guess rule on first attempt
o However, 28% never guess rule correctly!
 Actual rule: “three numbers in ascending order of magnitude”
26• Confirmation biaso Guess e.g., “goes up in twos”
o Generate sequences conforming to this rule
o Fail to falsify hypothesis
o Problem of overfitting in science!
• Cowley & Byrne (2005)
o More willing to falsify hypothesis if generated by someone else
o 62% abandoned other person’s hypothesis
o 25% abandoned own hypothesis
• Cherubinin et al. (2005)
o 2-4-6 & 16-18-20 30% correct rule
o 6-8-10 & 9-14-15 70% correct rule
 Latter example led to generation of broader (i.e., correct!) hypothesis
27Wason’s 2-4-6 task • Task provides insight in inductive reasoning and how scientists reason • However:o Not “real world”
 Immediate feedback doesn’t often occur in real world
 Feedback in real world often not fully informative (i.e., not binary ‘correct’ / ‘incorrect’)
o Rule is very general – confirmation testing not appropriate
 If rule was more specific, confirmation testing may be more appropriate
o Confirmation bias not always present (e.g., someone else’s hypothesis)
28Do Scientists hypothesis test?• Often, but not always
• Unusualness heuristic
o Guided by “unusual” results
• “What if”
o Hypothesis generation and simulation without experimentation
29• Dunbar (1993) o Genes controlling other geneso Focus on “activation” vs “inhibition” hypothesis
o If adopted confirmation strategy, often failed to solve problem
o 20% who solved problem focussed on discrepant results (unusualness)
o Many went through every possibly activation hypothesis before changing
o Few engaged actively in falsification!
30• Fuselgang et al. (2004) o Molecular Biologyo 417 results, 223 inconsistent with predictions
o 88% inconsistent results blamed on methods
o 12% modified theories based on discrepant results (not good!)
 However: When discrepant findings replicated – 61% change theory
 Sensible strategy if strong prior (Bayesian) on previous theory?
31Deductive Reasoning• Draw definite conclusions if tenets are true
• Based on formal logic
Conditional reasoning (propositional logic)
o Symbols stand for sentences (e.g., P = study, Q = good grade)
o Logical operators (e.g., if, and, or) applied to reach conclusions
 e.g., if you study, you will get a good grade
 If P then Q
32Conditional reasoning• If P then Q, P = true - antecedents (or premises)
• Therefore Q - consequents (or conclusion)
• Binary – either affirm or refute the consequent based on antecedents
• Modus ponens
o If P then Q – P therefore Q
• Modus tollens
o If P then Q – not Q therefore not P
 Both statements above correct, however more refute MT than MP
33• Affirming the consequent • Denial of the antecedent• Affirming the consequent
o If P then Q
o Q therefore P?
• Denial of the antecedent
o If P then Q
o Not P therefore not Q?
 Often appropriate in real world
 If you study hard (P) you will get a good grade (Q)
 You don’t study hard – unlikely you will get a good grade
 “Unlikely” – probabilistic, whereas logic is binary
34• Logic can construct false conclusionso If Sarah in Rio (P), then she is not in Brazil
o Sarah is in Rio therefore not in Brazil
o Clearly false, but logically consistent
o Real world knowledge irrelevant in conditional reasoning
35o Why?  Probability of relationship in real world? Probability of relationship in real world?
• E.g., close 1-to-1 correspondence between P and Q in (2) than (1)?
• “no smoke without fire”
 Generate counterexamples
• If counterexample successful, than reject conclusion
• e.g., A window could be broken by a bird flying into it, a hammer smashing it…
o We aren’t very good at conditional reasoning, and often use real world knowledge (inappropriately!)
36Wason Selection Task• Four cards: R G 2 7
• If R on one side, then 2 on the other side
• Only turn over cards needed to tell if rule is correct
Informal Reasoning
• Reasoning we do every day (supposedly)
• Knowledge and experience useful (unlike informal logic)
o e.g.,
1. The drug is safe because we have found no evidence that it is not
2. Ghosts exists because no-one has proved they do not
o Many say (1) more likely than (2) despite ‘logic’ being identical
• Deductive reasoning – binary (true or false)
• Informal reasoning – probabilistic (80% chance in being correct)
37Informal Reasoning • Straw-man fallacy• Straw-man fallacy
o Select weaker points of opponents argument and focus on them to decrease probability of their point being correct
o Your argument becomes RELATIVELY more probable
38• Motivation o Myside biaso Myside bias – evaluate statements with respect to one’s own beliefs
o Stanovich & West (2007)
1. College students who drink alcohol while in college are more likely to become alcoholic in later life
2. The gap in salary between men and women decreases substantially (~1% relative to >20%) when they are employed in the same position
o Students who drank alcohol rate accuracy of (1) lower than those who didn’t drink alcohol
o Women rated accuracy of (2) lower than men
39Climate Change • Kahan et al (2012) • Howe & Leiserowitz (2013)• Kahan et al (2012)
o Best predictor of denial of climate change was cultural values
o “egalitarian communication” vs “hierarchical individualists”
o Little to do with actual scientific evidence, or scientific literacy
• Howe & Leiserowitz (2013)
o Those who were dismissive of climate change remembered previous (very hot!) summer as colder than those concerned about climate change
• Informal Reasoning based on probabilities
• More akin to ‘real world’ reasoning
• Knowledge, experience, cultural values, motivation all contribute to conclusions – very different from formal logic!
40Bounded rationality• Idea that we are rational, within the limits of our cognitive capacity
• We produce workable solutions to real world problems in spite of limited processing resources
• Hence correlation between performance in reasoning tasks and IQ – more processing capacity = better at reasoning
41What is a judgement?• Calculating likelihood of events using incomplete information
o e.g., given my past experience, how well will I do in my next exam?
• Accuracy matters!
• Different from ‘decision’ – actively choosing 1 from number of possible actions.
o e.g., in multiple choice exam – ‘judgement’ about likelihood that each option is correct, ‘decision’ about which option to choose.
42Disease diagnosis • Some terminology:o Hit
 Positive result if you have disease
o Miss
 Negative result if you have disease
o Correct rejection
 Negative result if you don’t have disease
o False alarm
 Positive result if you don’t have disease
43Bayes TheoremBayes Rule
• Posterior ∝ likelihood x prior
o e.g.,:
 prior = probability of having the disease (base rate)
• p(H) – probability of hypothesis being correct
 likelihood = probability of data (e.g., test result) given the hypothesis
• p(D|H) – probability of data GIVEN hypothesis
 posterior = belief that you have a disease
• p(H|D) – probably of hypothesis GIVEN data
44Bayes Theorem • Specifically the odds ratio version of Bayes theorem• Ratio of probability for one hypothesis over another
o e.g., “I have disease X” vs “I don’t have disease X”
• Ratio of probability for data given hypothesis 1 vs hypothesis 2
o e.g., given I don’t have disease X, how likely is it that I will receive a positive test result?
• Ratio of probability for hypothesis 1 vs hypothesis 2 given data
o e.g., given I have positive result, how likely is it that I have disease X?
45• Odds ratio – how likely one hypothesis is over another, given current data and prior oddso p = don’t have disease; d = have the disease
o Disease X prevalence = 1/100
o Pr(Hp) = 0.99 Pr(Hd) = 0.01
o Prior odds = 99
 99 times more likely to not have disease than have the disease
o Evidence = positive test result
o False alarm rate = 0.05
 5% probability of getting positive result if you don’t have disease
 Pr(Evidence | Hp) = 0.05
o Hit rate = 0.95
 95% probability of getting positive result if you have disease
 Pr(Evidence | Hd) = 0.95
o Likelihood ratio = 0.05
o Posterior odds
 = Likelihood ratio x Prior odds
 = 0.05 * 99
 = 4.95
46Are we optimal? – Base rate neglect • Are humans optimal? Kahneman & Tversky (1973) • Lawyer/Engineer problemBayes provides ‘optimal’ way of calculating probability of events
o Group of 100 people: 70 Lawyers, 30 Engineers
o “Jack is a 45-year-old man. He is married and has four children. He is generally conservative, careful, and ambitious. He shows no interest in political and social issues and spends most of his free time on his many hobbies, which include home carpentry, sailing and numerical puzzles”
 90% said ‘engineer’, regardless of 70:30 or 30:70 split
• Ignored base-rate!
• What if group was 99 lawyers, 1 engineer? 100 lawyers, 0 engineers?
47Cascells et al (1978) • Real world example! o Faculty & advanced students at Harvard Medical Schoolo “If a test to detect a disease whose prevalence is 1/1000 has a false positive rate of 5%, what is the chance that a person found to have a positive result actually has the disease, assuming that you know nothing about the person’s symptoms or signs?”
 Actual answer: ~2%
 45% of participants said 95% (i.e., 1-false positive rate)
• As in Lawyer/engineer problems, neglected base rate
48Are we optimal? – conjunction fallacy Tversky & Kahneman (1983) • Bank teller problemo “Linda is 31 years old, single, outspoken & very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations”
 Is Linda:
• A bank teller
• A bank teller active in the feminist movement?
o Many choose statement (2), despite (1) being inclusive of statement (2)
o Conjunction fallacy – mistaken assumption that probability of conjunction of two events > probability of one of them
• However:
o Maybe we value precision/specificity – i.e., balance between probability of being correct, and specificity of answer?
o We may ignore redundant information in answers – i.e., given ‘bank teller’ in both answer, that must be true, so we focus on extra information
49Are we optimal? • Lots of evidence suggests we aren’t optimal: o Base rate neglect o Conjunction fallacy • However:o Including information about causal structure improves performance (Krynski & Tenenbaum (2007)
o Presenting in frequentist terms improves performance (Cosmides & Tooby, 1996)
o Personal relevance of problem improves performance – e.g., “do you have disease X” (Ditto et al., 1998)
• Often when presented in ‘correct’ format, people do much better!
o Giving people knowledge about why there might be false alarms/misses, people do better
o Presenting data in a particular way improves performance
• But still not perfect…
50Heuristics – again! • Representative heuristic • Availability heuristic• Representative heuristic
o Assume object/individual belongs to specific category because it is representative
• Availability heuristic
o Frequencies of events estimated by ease of retrieval
 e.g., estimate probability of contracting disease based on number of people you know with disease
51• However (Oppenheimer, 2004):o Effect can be reversed
o Which surname is more common? – one famous, one non-famous
o Participants more often chose non-famous, despite ‘availability’ of famous surname
o However, availability of famous name perhaps reduced as only retrieving instances related to specific individual, versus all individuals with non-famous surname?
52• Heuristics important in explaining possible reasons we aren’t always ‘optimal’ or ‘logical’ • However:1. Somewhat vaguely defined
2. Doesn’t define WHEN specific heuristics used
3. Not necessarily biased processing, but poor information
4. List of heuristics doesn’t equate to ‘theory’
53Theories • Several theories attempt to explain human judgement o Dual-process Theory (Kahneman)o Dual-process Theory (Kahneman)
 Judgements based on two distinct systems
o System 1 – fast, automatic, effortless, implicit
o System 2 – slow, serial, effortful, controlled
• De Neys (2006)
o Correct answers took 40% longer than incorrect – System II
o Really evidence for 2 distinct systems though?
• Penyock & Thompson (2012)
o Base rate vs availability – set up conflict
o Initial vs slower judgement
o 30% took account of base rate for slower judgement, however 23% took LESS account!
54• Likely to be two systems, one faster and more automatic, one slower and more effortful• Conforms to ‘intuitive’ sense of fast solution, versus slow effortful solution
•Often System 1 is used – heuristics!
o Emphasises System 1 as not optimal – i.e., prone to errors
o ‘Cognitive misers’ – we can use System 2 to produce correct answer, but often use System 1 because it is easier/less effortful
• However
o System 2 can also leads to errors – not clear when
Really evidence for HARD distinction?
o Both relatively ill-defined – hard to predict a priori which will be used and what answer it will give
o Some evidence for intuitive use of base rate (i.e., System 1
55Theories • Several theories attempt to explain human judgement • Fast-and-frugal heuristics (Gigerenzer)• Fast-and-frugal heuristics (Gigerenzer)
o “if we’re so dumb, how come we’re so smart?”
o Heuristics are useful
o Allow for rapid processing with little information
o Correct high percentage of time (but are susceptible to errors)
o Trade-off between time and accuracy!
56• “Take the best, ignore the rest”o Which city is bigger New York or Guangzhou?
1. Search rule (e.g., do I know this city?, does it have a cathedral?)
2. Stopping rule (e.g., if I know one city, but not the other, stop search)
3. Decision rule (e.g., choose city I know)
57Recognition heuristic• Select object that is recognised
• e.g., which if bigger – New York or Guangzhou?
• Most (in UK) would choose New York due to familiarity
Goldstein & Gigerenzer (2002)
• US students chose between 2 German cities – chose one they recognised 90% of time – but really recognition?
Pachur et al (2012)
• Doesn’t make sense in some situations – e.g., which city further North?
• 0.64 correlation between validity and usage
Richter & Spath (2006)
• German students chose US city based on recognition 98% of time when that city also had international airport
• Only chose 82% of time when other city had international airport
• They used other information they deemed valid
58What is a ‘decision’?• Choosing a specific option from > 1 options
• Judgements emphasise accuracy – often proceeds decision
• Decisions emphasise consequences – what happened after I made this decision
59Theories• Normative theories
o Concerned with how people SHOULD make decisions
o Assumption that people act rationally
• Descriptive theories
o Concerned with how people ACTUALLY make decisions
• Utility theory
o Assumption that people act (rationally) to maximise ‘expected utility’
o Expected utility = p(given outcome) x utility of outcome
o Utility = SUBJECTIVE value we attach to a given outcome (this might not be rational!)
 Calculate expected utility of each outcome and choose option with greatest expected utility
 Emphasis on final amount, not whether gain or loss (i.e., starting point irrelevant)
60Utility theory• Calculate expected utility of each outcome and choose option with greatest expected utility
• When decision based entirely on monetary gains/losses
• Assumptions:
o Maximise chances of gains
o Minimise chances of loss
61Prospect theory • Kahneman & Tversky (1984) • Main assumptions:1. Individuals identify ‘reference point’ representing current state (starting point critical!)
2. Individuals more sensitive to potential loss than gains (loss aversion)
3. Individuals overweight rare events
• Individuals only likely to take 50:50 bet when gains x2 of loss
• Based on how sensitive individuals are to gains/losses, can calculate a value function
 relationship between utility (i.e., subjective value) and actual gains/losses (i.e., objective value)
o Losses – much steeper line – higher subjective value than a gain of the same amount
62Utility vs Prospect theory• Prospect theory:
1. Emphasises starting/reference point – e.g., rich vs poor
2. Provides a value function
3. Emphasises loss aversion (can only discuss gains/losses by having a starting/reference point)
4. Emphasises overweighting of rare events
• Prospect theory better descriptive account of decision-making than utility theory?
63Framing effect• Decisions influenced by irrelevant aspects of situation
• Example – “Asian disease problem”
o Tversky & Kahneman, 1981
o “Imagine that the US is preparing for an outbreak of an unusual Asian disease, which is expected to kill 600 people. Two alternative programmes to combat the disease have been proposed. Assume that the exact scientific estimate of the consequences of the programme are as follows:”
64o Gain-frame condition:o If programme A adopted, 200 people will be saved
o If programme B adopted, there is a 1 in 3 probability 600 people will be saved, and a 2 in 3 probability that no people will be saved
 72% chose programme A, despite fact that (if implemented several times), both would save on average 200 people
65o Loss-frame condition:o If programme C adopted, 400 people will die
o If programme D adopted, there is a 1 in 3 probability nobody will die, and a 2 in 3 probability that 600 people will die
 78% chose programme D (opposite to Gain-frame condition)
o Loss aversion to high probability that 600 people will die?
66• However: o Mandel & Vartanian (2011) o Almashat et al (2008)o Mandel & Vartanian (2011)
 Ambiguity in option A – interpret as AT LEAST 200 people saved?
 When programmes completely described – framing effect disappeared
 Similar to judgements, when fully explained?
 Medical scenarios involving cancer treatments
 When participants explicitly listed advantages/disadvantages of each option and justified their decision – framing effect disappeared
 i.e., engaging in deliberate weighing of options reduces/eliminates effect
 2 systems?
67Sunk-cost effect• “tendency for people to pursue a course of action even after it has proved to be suboptimal, because resources have been invested in that course”
o i.e., “throwing good money after bad”
• Dawes (1988)
o $100 deposit for hotel room
o On route to hotel, become ill
o Think you will probably have more pleasurable time at home
o Do you keep going or turn back
 Many decide to keep going – sunk-cost
o Loss aversion?
• However
o Baliga & Ely (2011)
o Give MBA (business) students full information about investments
o Showed opposite of sunk-cost i.e., more likely to switch
o Certain groups don’t show the effect, so not universal?
68Overweighting rare event• Think rare events are more likely than they are
• Explains national lottery? Very low chances of winning, but people still buy tickets
• In lab, when events described, participants often overweight
o Hertwig (2004)
 What about real world decisions, that involve experience?
 When decision based on descriptions – overweighted
 When decision based on experience – underweighted
 Due to low sampling = never experienced rare event?
 Availability heuristic?
69Loss neutrality• Choose between:
• 50% £1 or 50% £-1
• 50% £5 or 50% £-5
o Loss aversion – should pick (1) not (2)
o However, 50/50 split between choosing (1) and (2)
o Neutral to loss in this situation (different from coin toss example earlier)
o Breaks down when using extreme amounts (e.g., +/- £1 vs +/- £1000)
70Returning to prospect theory • More detailed than utility theory • Introduces concept of value function• Introduces concept of value function – non-linear mapping between objective and subjective value
• Predicts ubiquitous phenomena – loss aversion, overweighting
• However
o Doesn’t explain why value function exists – i.e., why are we loss averse?
o Many predicted phenomena can disappear/be reversed in specific experimental situations
o Doesn’t account for individual differences:
 High self-esteem – prefer risky gambles
 High narcissism – high sensitivity to reward, low sensitivity to loss
71Emotional and Social factors• Anticipated/actual loss may -> negative emotions (e.g., anxiety) – may explain loss aversion
o Kermer et al (2006)
 $3 loss predicted to have greater impact on happiness than $5 gain
 No actual experienced difference – anticipation differs
 Impact bias – overestimation of intensity/duration of negative emotions to loss
o Giorgetta et al (2013)
 Decisions made by participant or computer
 Loss felt as regret if decision made by participant
 Loss felt as disappointment if decision made by computer
 Agency (i.e., whether you did it) affects emotional response to loss
72• Omission biaso Preference for inaction when engaged in risky decision making
o Brown et al (2010)
 Parents making vaccination choices (very important!)
 Parents willing to accept higher risk of child having disease than child suffering adverse reaction to vaccination
o Wroe et al (2005)
 Higher anticipated responsibility and regret if child experienced adverse reaction due to vaccination – i.e., “it’s my fault”
o Aberegg et al (2005)
 Pulmonologists treating lung disease
 Less likely to choose best management strategy when given option to do nothing (40%) vs when not given option (59%)
73• Status-quo biaso Prefer to accept status-quo than change decision
o Samuelson & Zeckhauser (1988)
 Retired people keep same allocation of retirement funds, despite no financial cost in changing
 However – cost in terms of effort in changing?
74• Accountabilityo Simonson & Staw (1992)
o Increased accountability -> increased sunk-cost
o Experience greater need to justify initial decision, so stick with it for longer
75Complex decision making How do decide when faced with wealth of options?• Lab-based decisions often involve selecting between 2 options
o Real world – many options
• Decision makers should go through following steps:
1. Identify attributes relevant to decision
2. Decide how to weight attributes
3. List all options under consideration
4. Rate each option on each attribute
5. Obtain total utility -> select option with highest utility
 e.g., renting/buying a house
76• Simon (1957) o Bounded rationalityo Bounded rationality – decision making ‘bounded’ by
 environmental constraints (e.g., information costs)
 cognitive constraints (e.g., short-term memory capacity)
o We are as rationale as permitted within these constraints
o Satisficing – choosing first option that satisfies individual’s minimum requirements
o “perfection is the enemy of good enough”
77• Elimination-by-aspects theory o Tversky et al (1972) Serial elimination based on specific criteria until one option remains
 Order can matter!
 Can’t handle trade-offs (e.g., distance from university vs rental cost)
78o Kaplan et al (2011)  Two-stage process1. Elimination (as per elimination-by-aspects theory)
 re: somatic marker hypothesis
2. Detailed comparison of small number of remaining (as per Utility theory)
 Useful for filtering when many options available
 Still entails detailed comparison to make ‘optimal’ decision from small subset
79Somatic Marker HypothesisBrain simulates consequences of any possible decision/response option
Body reacts to simulation, causing appropriate responses – ‘somatic markers’
Brain interprets somatic markers for possible options
Attention/resources focussed on ‘good’ options for further processing
Quick, ‘rough-and-ready’ way to assess various response options
Provides ‘gut-feeling’ about what to do in a given situation
mPFC simulates consequences/interprets somatic markers
- Skin conductance response (anticipatory)
80Memory guided decision making• Often use past experience to make rapid, pressured decisions
• Klein (1998)
o Recognition-primed decision model
o Experts needing to make rapid decisions
o Retrieved previously ‘similar’ situation and evaluate whether previous decision appropriate – i.e., no need to generate new decision
o Experts typically characterised novel situations as examples of familiar situations
o Retrieved and utilised previous decision
o Emphasises role of expertise to rapid decisions – memory useful
o Elimination-by-aspects – more typical for non-experts making considered decisions
81Are we rational?• We’re good at reasoning and decision-making!
1. Heuristics are useful and often accurate (particularly under time pressure)
2. Do use base-rate info when made explicit (see Lecture 3)
3. Often “errors” relate to accountability for one’s own actions
4. Certain tasks (e.g., deductive reasoning tasks) artificial and unlike those made in real world (see Lecture 2)
5. Real world rarely binary, but probabilistic in nature (see Lecture 2)
82• But maybe not…1. Motivation in reasoning tasks leads to (at most) modest improvements in performance
2. Complex judgement tasks performed better by people with higher IQ
3. Even when fully described, performance in certain tasks still poor
4. Dunning-Kruger effect – people with poor reasoning ability have less insight into their competence (i.e., think they are good when they aren’t!)

Information

Description: Flashcards on Perception and Cognition TB9, created by Emily Hargreaves on 29/05/2019.
Created by: Emily Hargreaves
Views: 0
Rating