Heuristics and Biases

Substitution

A process that happens when someone gives a heuristic answer

 

Main elements of substitution:

1. Target Q = the assessment you intend to produce

2. Heuristic Q = the simple question that you answer instead

 

When faced with a difficult question, your S2 typically tries to solve it in a controlled manner

• BUT you end up subbing an easier question (heuristic) to replace a more difficult question (target)

 

Ex. T = "how risky is it?"

• H = "how fluent is it?" (what you end up answering)

 

Ex. T = "after 10 consecutive heads, what is the probability that the next coin flip will come up heads?"

• H = "what seems more random: 11 heads in a row or 10 heads and 1 tails?" (what you end up answering)

 

Substitution and the dual process model

What are the two points of intervention that help you from falling for heuristics?

1. Help system 2 find the answer to the target question

• Easier said than done…

• By showing you how to solve common fallacies, you can be supplied with the memory that you need, and thus you can recall the right answer

2. Prevent system 2 from signing off on the heuristic answer

• Help your S2 question heuristic answers provided by S1

• Must recognize that intuitive S1 answers are WRONG

Mood effects x life satisfaction study

Study details:

• Participants in a soundproof room

• Completed a filler sound task + questionnaire (real study hidden in the packet)

• Randomly assigned to write 3 pages about:

  1. A positive event (good mood)

  2. A negative event (bad mood)
     → Mood Induction task = IV

 

Key Question (DV):

"How satisfied are you with your life these days?"

 

Research Question:

Does current mood (IV) affect reported life satisfaction (DV)?

 

Hypothesis:

If participants answer the question rationally, mood shouldn't affect life satisfaction ratings

• Meaning they shouldn’t be more or less satisfied with their life just simply depending on their current mood

 

Findings:

Mood did influence life satisfaction!

• Writing about negative events → lower life satisfaction

• Writing about positive events → higher life satisfaction

The substitution process x life satisfaction study

POV: You just had to write about a negative event (now you are in a bad mood) + asked   "how satisfied are you with your life as a whole these days?"

 

1. Mental shotgun -> what do I need to know to make this judgement?

   • Basic assessment says that you feel like you are in a bad mood right now

2. Intensity matching -> I need to express my current bad mood in terms of life satisfaction

   • Mood = 2 / 10

   • So life satisfaction = 2 / 10 . . .

3. System 2 endorses heuristic answer

 

Basic idea = When asked a complex question like "How satisfied are you with your life as a whole?", people may unconsciously substitute it with a simpler question: "How do I feel right now?"

Mood effects x life satisfaction study (weird room manipulation)

Study details:

Participants wrote about a positive or negative life event (mood induction).

• Half were told: "This room has made people feel sad/anxious in the past"

• Other half were told nothing about the room

All participants then rated their life satisfaction

 

Research Question:

Does awareness of the room influence how mood affects life satisfaction?

 

Findings:

When told about the room → Mood had no effect on life satisfaction

When not told → Mood did affect life satisfaction

 

Why?

Attribution effects = When people have a plausible external explanation for their mood (like a strange room), they are less likely to let that mood influence their judgments (like how satisfied they are with life)

Participants attributed their mood to the room, not to their life

• "Maybe I'm just feeling off because of the room, not because my life is bad"

The substitution process x life satisfaction study (weird room manipulation)

POV: You just had to write about a negative event (now you are in a bad mood) + asked "how satisfied are you with your life as a whole these days?"

1. Mental shotgun -> what do I need to know to make this judgement?

• Basic assessment says that you feel like you are in a bad mood right now

2. Intensity matching -> I need to express my current bad mood in terms of life satisfaction

• Mood = 2 / 10

• So life satisfaction = 2 / 10 . . .

3. Attribution -> highlighting room and mood makes system 2 inspect system 1 heuristic answer

• "My bad mood is because of the room, not because of my life"

• System 2 does NOT endorse heuristic answer

Single people asked to answer two questions:

1. How satisfied are you with your life?

2. How often do you go on dates?

Is there a correlation between these questions?

NO correlation when asked in this specific order

Single people asked to answer two questions:

1. How often do you go on dates?

2. How satisfied are you with your life?

Is there a correlation between these questions? 

BIG correlation when asked in this specific order

Priming + substitution process

• When asked about dating frequencies first, it becomes salient in your mind

• So when asked about life satisfaction next, you substitute dating success as a proxy for overall life satisfaction

Single people asked to answer two questions:

1. How often do you go on dates?

2. How satisfied are you with your life ASIDE from your dating frequency?

Is there a correlation between these questions?

NO correlation when asked in this specific order + this manipulation

• "Aside from your dating frequency" draws participant's attention to the heuristic answer (system 2 inspects and does not endorse it)

• “Maybe my life isn’t bad—I'm just not dating much.”

How do you answer this question: How satisfied are you with your life these days?

Less diagnostic info = how do I feel right now?

• HEURISTIC ANSWER!

• Judging life satisfaction from biased observation (current mood is not indicative of life satisfaction)

• WYSIATI ("What you see is all there is")

 

More diagnostic info = are my needs met? How much stress do I experience in a give day? How much happiness do I experience in a given day?

• To answer this answer correctly you need to recruit more info and observations

 

Law of small numbers

Small samples produce more extreme, less reliable estimates

• Smaller sample = less precise estimates

Small samples have high variance, meaning estimates can be far from the true value

Ex. Voting % in samples of 30 can vary widely (e.g., 11%, 56%, 5%, 70%)

Law of large numbers 

Large samples produce more precise, reliable estimates

• Larger sample = more precise estimates 

Estimates tend to be closer to the true value with larger samples

Ex. Voting % in samples of 3000 stays consistent (e.g., 33%, 35%, 38%, 36%)

Why is the law of small numbers a problem? 

Small samples often produce extreme or misleading results due to high variability

System 1 ignores sample size and just tries to fit what we see into a simple story or pattern (a schema)

• Since it ignores that small samples can be unreliable, we tend to over interpret random fluctuations in small samples as being meaningful

POV: You are doing poorly in a class and you have not studied for the final. You can either choose to do a final with (1) 10 questions or (2) 50 questions. What final should you choose?

Choose the exam with 10 questions (even though most people assume the opposite)

• Small sample of questions = higher variability

• Your grade could swing up by chance (not reflecting your limited knowledge)

• Greater chance of getting lucky!

 

You don't want 50 questions because it would be a better assessment of your knowledge

• Large sample of questions = more accurate reflection of what you actually know

• Much harder to "get lucky" over 50 questions

 

*The Law of Small Numbers helps you here because it means your randomly high score is more likely with fewer questions

*BUT IF THE SCENARIO WAS FLIPPED AND YOU WERE VERY PREPARED FOR THE CLASS: take the 50 question exam, you do not want the 10 question because it has higher variability

Hot hand fallacy

False belief that after consecutive positive outcomes, another positive outcome is more likely

Ex. someone who has had success with a random event (like making basketball shots) has a higher chance of continued success — that they're "hot"

Stems from our tendency to perceive meaning in randomness (finding patterns where none exists)

Is there evidence of a hot hand in basketball? 

NO 

Probability of a hit (shot) after 3/4 previous hits = 50%

Probability of a hit (shot) after NOT making 3/4 previous shots = 57%

What is the difference between the hot hand fallacy vs. the gamblers fallacy

Hot hand fallacy = you expect runs to continue

Ex. You expect a player who just made 4 baskets in a row to get another basket on the 5th shot

 

Gamblers fallacy = you expect runs to end

Ex. You expect a coin that has just turned up head 4 times in a row will turn up tails on the 5th flip

T or F: Humans are bad at being random

TRUE!

 

Humans are bad at producing random sequences

If humans were good at being random, each of the bars would fall around~10% (more equal distribution)

Anchoring

Cognitive bias where people rely too heavily on the first piece of information (the “anchor”) when making decisions or estimates

Ex. “What is the answer to . . .”

1. 1x2x3x4x5x6x7x8 -> people guessed way lower numbers

2. 8x7x6x5x4x3x2x1 -> people guessed way higher numbers

Even though it's the same numbers, the starting number (anchors) influence guesses

Anchoring procedure

1. Anchor question (Q1):

• High anchor: “More or less than 80,000 students at Queen’s?”

• Low anchor: “More or less than 5,000 students at Queen’s?”

 

2. Estimate question (Q2):

• “What’s your best guess for the actual number of students?”

 

Result:

• People's guesses (Q2) were influenced by the anchor in Q1

• Higher anchor → higher estimates

• Lower anchor → lower estimates

 

Anchoring process x adjustment

Anchor becomes your initial intuition (S1)

• To reach a better estimate, you need to adjust (S2)

 

PROBLEM:  S2 is not strong enough to find enough reasons and facts…

• S2 = insufficient adjustment

• This causes your final answer to stay too close to the anchor

 

Ex. High anchor: "Are there more or less than 80,000 students at Queen’s?"

• Even if you know that seems high, your estimate still ends up higher than it should be

Intoxication x anchoring

Study details:

Students answered anchoring questions

• Ex. "When was George Washington elected?"

• Ex. "Who was the second explorer after Colombus?"

 

Students then asked: Had they been drinking?

 

Hypothesis:

• If adjustment relies on System 2, then drunk students (weakened S2) would adjust less from the anchor

 

Findings:

• Drunk students gave estimates much closer to the anchor

• Intoxication suppresses System 2, leading to insufficient adjustment

Anchoring processes x priming

Anchor activates associations/thoughts that are consistent with the anchor

S1 is biased to believe the anchor is true (just thinking about it makes it feel plausible)

Those activated thoughts then influence how you answer the second question

Comprehending = temporarily believing (S1 accepts info as true in order to process it)

Anchoring processes x priming

Study details:

Participants asked one of two questions:

1. "Is the average price of a car higher or lower than $10,000?"

2. "Is the average price of a car higher or lower than $50,000?"

Participants then complete a lexical decision task ("is this a word or not?")

1. Names of luxury cars (mercedes, audi, bugatti)

2. Names of common affordable cars (chevy, ford, toyota)

3. Regular words that have nothing to do with cars (digital, climate, ordinate)

4. Non-words (narmon, glowbe, bolm)

Findings:

The anchor participants got affected the speed at which they recognized words

• High anchor ($50,000) = faster recognition of luxry brand cars compared to other stimuli (common cars, regular words, nonwords)

• Low anchor ($10,000) = faster recognition of common brand cars compared to other stimuli (luxury cars, regular words, non-words)

These results illustrate priming

• $50,000 anchor → primes expensive/luxury car concepts → Participants recognize luxury car brands faster (e.g., Mercedes, Bugatti)

• Ex. $10,000 anchor → primes affordable/common car concepts → Participants recognize common car brands faster (e.g., Ford, Toyota)

Availability heuristic

Mental shortcut used to estimate the likelihood or frequency of an event based on how easily examples come to mind

Basing judgements upon the ease with which information comes to mind

Event is easy to recall = event perceived as more frequent/common

• If you can quickly think of many examples of an event (e.g., hearing about plane crashes on the news), you’re more likely to believe that event happens often

Event is hard to recall = event perceived as infrequent/uncommon

• If it's difficult to think of examples, you may assume the event is rare, even if it actually occurs frequently

Ease = Fluency = Familiarity ⇒ Perceived Frequency

Availability Heuristic in Terms of Target vs. Heuristic Question:

Target Question: How frequent or likely is this event?

• This is what you want to answer!

• Requires actual data or statistical reasoning

Heuristic Question (SUBSTITUTED): How easily does an example of this event come to mind?

• This is what your brain actually answers!

• Much quicker and easier

Availability heuristic x Priming

Highly accessible concepts are easier to retrieve and thus judged as more frequent or likely

T or F: we often overestimate our contributions to shared housework or group projections

TRUE

Availability heuristic = it is easy to recall the chores that you did + it is hard to recall the chores that your partner did (vice versa)

Assertiveness Study x Availability heuristic

OG Pilot study: "How many instances of your past assertive behaviour can you list?"

• Participants listed 9 instances on average

 

Study details:

Group 1: "List 6 instances where you behaved assertively"

• Easier task (lower than the average in pilot study)

Group 2: "List 12 instances where you behaved assertively" (other half of participants asked this Q)

• Harder task (higher than the average in the pilot study)

 

Both participants then asked "How assertive are you?"

 

Hypothesis:

If participants are using S2 (rational thinking)

• Participants who recalled 12 instances should say they are more assertive

• More instances recalled = more evidence of assertiveness

 

If using S1 (heuristic thinking / availability heuristic):

• Participants who recalled 12 instances should say they are less assertive

• Judgments based on ease of retrieval

 

Results:

Participants who listed 6 instances rated themselves as more assertive

• 6 instances is easier to retrieve → makes you feel more assertive

Participants who listed 12 instances rated themselves as less assertive

• 12 instances is harder to retrieve → makes you feel less assertive

*BUT if background music was playing, there would be no effect…

• People's difficulty in recall would be attributed to the music in that condition

• "It was hard to recall the instances because distracting music was being played"

How can you use the availability heuristic to convince your prof to round your grade to a 100?

Student: "Tell me 10 reasons why I should not get an A"

Prof: I can't think of 10 reasons, so I should just give them an A…

Availability and risk

People overestimate the risk of infrequent danger

• Ex. We think that tornados kill more people than asthma (even though the opposite is actually true)

Substitution process:

• Target Q: "How risky is it?"

• Heuristic Q: "How easily can I recall instances of this being dangerous? How easily can I picture this being dangerous?"

Why is showing cigarette packages with vivid photos effective?

Vivid + emotional things = easier to recall

Cigarette boxes w/ photos showing dangerous side effects = easier to recall negative things/risks associated with it

Affect heuristic

A mental shortcut where people rely on their current emotions to make a decision instead of using logic and objective information

 

Basing judgements about something on how you feel towards it

Affect Heuristic in Terms of Target vs. Heuristic Question:

Target Question: What are the benefits/drawbacks of X?

• This is what you want to answer!

• Requires actual data or statistical reasoning

 

Heuristic Question (SUBSTITUTED): How do I feel about X?

• This is what your brain actually answers!

• Much quicker and easier

Exemplar theory of categorization

1. Does X belong in category A?

2. Compare X to exemplars of category A

3. If X is similar to the exemplars = yes it belongs!

How does the mind represent categories?

Exemplar theory of categorization!

 

We store specific examples (exemplars) of category members in memory +  categorize new items by comparing them to known, specific examples

Representativeness heuristic

A mental shortcut where people judge the probability of an event or categorize something by how well it resembles a prototype or stereotype, often leading to errors like stereotyping and ignoring base rates

 

Basing judgements of likelihood upon the similarity/typicality of a stimulus to category exemplars 

Representativeness Heuristic in Terms of Target vs. Heuristic Question

Target Question: What is the likelihood that X belongs to category A?

• This is what you want to answer!

• Requires actual data or statistical reasoning

 

Heuristic Question (SUBSTITUTED): What is the similarity of X to exemplars of category A? Is it plausible that X could be in category A?

• This is what your brain actually answers!

• Much quicker and easier

T or F: Similarity equals likelihood

FALSE! Similarity and likelihood are NOT the same thing…

 

Ex. Which of these coin flip sequences are most likely?

• HTHHTTHTHTTT

• HHHHHHTTTTTT

• HHHHHHHHHH

• TTTTTTTTTTTTT

Correct Answer: All sequences are equally likely (each flips is independent)

Where people get it wrong: The first sequence looks more random, so people think it’s more likely (representativeness heuristic)

Conjunction fallacy

A cognitive bias that occurs when someone assumes that specific conditions that occur together are more likely / probable than either of the conditions occurring alone

 

Essentially CJ is an extreme version of the representativeness heuristic

Conjunction Fallacy in Terms of Target vs. Heuristic Question:

Target Question: Which is more probable?

• This is what you want to answer!

• Requires actual data or statistical reasoning

 

Heuristic Question (SUBSTITUTED): Which is more plausible/coherent?

• This is what your brain actually answers!

• Much quicker and easier

 

Conjunction fallacy example (feminist bank teller study) 

“Linda is 31, single, outspoken, and very bright.
She studied philosophy and cared deeply about social justice.
She protested against nuclear power.” 

Participants provided the character description above + asked “which is more likely?”

• A) Linda is a bank teller

• B) Linda is a bank teller and active in the feminist movement

  • Most people choose B, but this is incorrect…

Conjunction fallacy:

• A) = one event (just bank teller)

• B) = two events happening (bank teller and feminist)

• *The probability of two events together is always lower than just one, meaning the probability of Linda being a bank teller AND a feminist is lower than her just being a bank teller

*More details can seem more likely, but actually make it less likely

What is the difference between the representativeness heuristic vs. conjunction fallacy?

Representativeness heuristic:

• People rely on representativeness even when it ignores statistical information (ex. Base rates)

• Ex. Thinking you see Lady Gaga walking around Kingston (extremely low base rate of Lady Gaga being in Kingston, but you ignore it and assume it's her solely because of appearance)

 

Conjunction fallacy:

• People rely on representativeness, even when it goes against logical probabilities

• Ex. Thinking that people are more likely to like both bananas and oranges than they are to like bananas

While jogging around the neighbourhood, you are most likely to get bitten by (1) a dog or (2) your neighbour's aggressive and noisy Chihuahua? (conjunction fallacy)

You are most likely to get bitten by (1) a dog

 

*The second options tells a story, and those details makes It seem more plausible/likely

Conjunction fallacy x health survey study

Study details:

Participants told that a health survey was conducted in a sample of adult males in BC of all ages and occupations.

 

Participants asked to give their best estimate of the following values:

1. What % of men surveyed have has one or more heart attacks?

2. What % of men surveyed are both over 55 years old and have had one or more heart attacks?

 

Findings:

Participants estimated a lower % for the first question

Participants estimated a higher % for the second question

• This is because the details in the question fits with their mental schemas (older people are more likely to have heart attack) -> sounds more familiar -> feels more likely

 

Conjunction fallacy x health survey study (MAANIPULATION that DEBIASES THE CF)

Study details:

Participants told that a health survey was conducted in a sample of 100 adult males in BC of all ages and occupations.

 

Participants asked to give their best estimate of the following values:

1. How many of the 100 men surveyed have has one or more heart attacks?

2. How many of the 100 men surveyed are both over 55 years old and have had one or more heart attacks?

 

Findings:

Only 25% of participants committed conjunction fallacy in this manipulation

 

Rewording makes the target question easier to interpret

• "How many?" = easier for S2 to answer

• "What %?"= harder for S2 to answer…

 

Gricean conversational norms

1. Use common ground

- Speak based on shared knowledge and understanding with your listener

2. Only make relevant contributions

- Stay on topic and provide info that is relevant to the convo

How to mislead people without actually lying (breaking Gricean convo norms)

Study details:

Participants shown 4 different canola oils + asked:

1. "How much would you pay?"

2. "Which seems healthiest?"

Findings:

The oil labeled “No cholesterol”:

• People were willing to pay more

• Seen as healthiest

PROBLEM: Canola oil has no cholesterol anyway…

• This label is misleading by giving away irrelevant info

Base rate neglect

People ignore general info about how common something is (base rates)

• Instead people typically rely on specific but less reliable info

 

BUT if base rates are clearly linked to cause, people pay more attention

• AKA base rate is neglected unless given a causal role

Base rate neglect (hit and run study)

“Hit and run witness says with 80% certainty that the cab was blue”

+

Base rates:

• 85% of cabs in the city are green

• 85% of accidents in the city are caused by green cabs

Findings:

People tend to ignore the base rate of green cabs being more common and causing more accidents, and instead rely on the witness certainty

BUT if the base rate is given a causal role (e.g., green cabs cause 85% of accidents), people would be more likely to consider it

Why is it so hard to teach psychology

The identifiable victim effect

• "A single death is a tragedy, a million deaths is a statistic"

Singular, named individuals grab our attention — statistics don’t

We tend to ignore base rates when there's a vivid individual case

THUS psychology is hard to teach because our minds prefer vivid stories over abstract statistics

Regression to the mean

The tendency of results that are extreme by chance on first measurement (extremely higher or lower than average) to move closer to the average when measured a second time

 

Luck is random!

• Success = talent + good luck

• Failure = talent + bad luck

 

Great success is usually followed by less success

Tremendous failure is usually followed by less failure 

 

***Luck eventually balances out!

What is the difference between regression to the mean vs. gamblers fallacy?

Regression to the mean = Extreme things are more likely to eventually become normal

Gamblers fallacy = Extreme things are NOT likely to eventually become normal

Gamblers fallacy = Assumes events are dependent (when in reality they are independent)

*Double check this cue card