Vocabulary

antecedents: the event immediately prior to something antecedent to my stomach ache

denigrate: unfair critique. Sam denigrated the homeless when he said they deserved it for not paying taxes nor contributing to society.

bias: systemic error

precocious: a child having developed abilities at an earlier age than usual

Introduction to Thinking, Fast and Slow

A book about biases of intuition and how to create diagnoses and labels just like doctors have/do for illnesses.

Amos Tversky was one of the first in decision making research. Kahneman asked him in 1959 to speak at a seminar

A question was proposed. Whether humans are intuitive statisticians just like we are intuitive grammarians. The question lead to a survey proposal to figure out if researchers collect too few observations than necessary to draw conclusions, which lead to the conclusion that statisticians themselves are not intuitive and are not able to properly state how likely a result is to reproduce in a smaller sample. The goal was to judge the intuitive answers and not the actual answers to the statistical questions.

The Steve Problem

We can describe Steve as a person who resembles the “stereotypical” librarian and ask if Steve is more likely to be a librarian or a farmer. Most people choose librarian even though the ratio of male farmers to male librarians are 20:1. Personally, I did not fall for this trick because the stereotype that librarians are shy is not true, they are very outgoing and are the first to say hello to you.

The Resemblance Heuristic

The previous study highlights the resemblance rule of thumb that people make use of when asked with a difficult problem which leads to predictable biases. The takeaway here is to be aware of when we are employing this heuristic as to not make important decisions from them.

Ease of Memory Search

Two questions. One is questioning rate of divorce among professors which triggers searching for people trying to recall whether they know of such a situation. A more formal question is “Is the letter K more likely to appear in words of a text at the start or at the third position.” I personally think the third position as I could only think of a handful of words that start with K that are used often versus the word “like” which is even in the sentence! As I read further on, I am vindicated. I’m happy the book has been rubbing off on me.

The bias of the answers is that people are mostly thinking of words starting with the letter K than when K is in third position, so they will exaggerate the frequency of K being in the first position.

Availability Heuristic

When information pertaining to one situation is shown/presented more often than information pertaining to another, it is easy to place a higher probability on the former situation. The example in the book is that the author challenged his long-standing impression that politicians cheated more often than physicians and lawyers but it turned out his impression was just that and could be due to the choices of journalists. This is a very good topic that Noam Chomsky talks about in his books Manufacturing Consent, Media Control.

This heuristic also applies to what issues the public is vocal about and silent about. For example, vocal about housing crisis, not vocal about male circumcision.

Expert Intuition is Recognition not Magic

A remarkable feat of intuition, a fire fighter commander yelled to people to get out after they put out a kitchen fire as his ears were hot and fire was quiet so he could tell something was wrong. The floor collapsed soon after the fire fighters got out.

After studying chess master, Herbert Simon determined that the master saw the pieces differently after hours of practice since they could draw on information stored in memory rather than having to think and ponder over a position.

The parallel is drawn to how we are not surprised when a two-year old points to a dog with glee and says “doggie!.”

Not all Expert Intuitions are the Same

There is a stark difference between someone making a decision based on hours of experience and the example in the textbook of a chief investment officer buying shares in Ford because he was impressed at an automobile show he attended. A gut feeling is not the same as expert intuition, although a gut feeling that can be explained via hard facts like “hot ear” is different than the gut feeling based on feelings.

Conclusion by the author is that emotion plays a big role in intuition and the stock picker is an example of the Affect Heuristic; guided by feelings of like and dislike with little to no reasoning/deliberation.

Intuitive Heuristics

When faced with a difficult question that cannot be answered with intuition based on relevant experience, a similar and easier question is answered in substitute. When spontaneous thinking fails, slower thinking starts.

Publishing of the Article

After five years of investigating biases, they published an article on 20 or so heuristics/biases in the Science magazine. The article challenges the broadly held belief that humans are generally rational with normally sound thinking except when it comes to certain emotions (fear, affection, hatred).

Chapter 1 - Characters in the Story

Highlighted difference between how easy it was to identify a woman’s mood based on a glance of the image versus a multiplication problem 17 x 24, which most people are not tuned to answer using intuitive thinking. When doing the problem mentally, we keep the intermediate results in our brain, while our body is strained, and our eys focused. The answer of 408 finally comes after some time.

System 1: operates automatically and quickly, with little or no effort and no sense of voluntary control

System 2: allocates attention to the effortful mental activities that demand it, including complex computations. The operations of System 2 are often associated with the subjective experience of agency, choice, and concentration.

Performance of system 2 thinking falters when attention is taken away. Therefore, multi-tasking can only really work when there are no System 2 tasks involved.

Another takeaway is that when driving in an area that is of higher stress, a passenger can easily lower the performance of the driver. Example of the video of the invisible gorilla was given, where the conclusion is that not only can we be blind to what we ought to have seen, we are also blind to the fact that we are blind.

System 1 is running automatically, and system 2 is in low-effort mode. System 1 continuously generates suggestions for System 2: impressions, intuitions, intentions, and feelings. System 2 endorses the directive of system 1, these into beliefs or voluntary actions with little modifications. Whereas if System 1 fails, it calls on System 2. System 1 is basically surprised by an event that it determines not to fit in its model. System 2 is also used for continuous monitoring (e.g. polite when angry), and mobilizes when an error is about to occur. The “division of labour” between these systems leads to minimizing effort and maximizing performance. System 1 has biases though and the system cannot be turned off.

Aside on External Applications of Two Systems

The previous section I read was very good as it made me think of architecture designs when it comes to autonomous artificial intelligence as well as computer chips.

When designing an autonomous agent, we probably want it to also have these two systems. We might want to develop two different networks, somewhat connected, where the first network takes in input and tries to quickly generate something, which is then fed into a second system specialized in correctness, and then a third system depending on if the second system took displeasure.

When designing a chip, since most tasks are system 1 thinking, we want to pack as many efficient cores as possible that focus on minimizing energy usage and maximizing common tasks. For example, a computer is going to be used for two tasks most of the time: browsing the internet (network I/O), some sort of app that makes uses storage (productivity apps). Of course, a computer can also be designed knowing it will be used for a niche tasks more often than the others (e.g. gaming CPUs). So when mass producing these CPUs, 80% will be E cores, while maybe 20% will be performance. The 80-20 rule comes in handy a lot, but it too is a bias.

Obviously, we need to also determine what is the ideal number of cores and that comes down to total ideal energy usage as well as costs. Industry standard is 8 cores for mobile max, but I think 10 cores would be a game changer as two more E-cores can be added while maintaining 2 P-cores.

Conflict

This section starts by highlighting the difference in our efficiency when we are told to do a task where the contents the task pertains to have related or opposite meaning. The example tasks was read a column of words and whispering the casing of the words. The efficiency is lower when the words themselves are the words that could bew whispered. So when reading upper, you whisper lower, and when reading LOWER you whisper upper. Conclusion: system 2 is in charge of self control.

Illusions

The horizontal lines are in fact equal but you should’ve seen this illusion or some form of it enough time by the time you are 22 years old.

Even though we know that the lengths are equal, we still see that the bottom is longer, and this is system 1 at work. Although we can teach ourselves to ignore line length comparison when fins are attached, we cannot override system 1 thinking, for some reason.

It does seem that AI can be fed a lot of system 2 thinking without hindering what I call the system 1 thinking. So our own perception of what system 1 cannot overcome, can definitely be overcome by AI in a similar response time.

Cognitive Illusions. Apparently the professor of psychology at one of the lectures the author attended said therapists will meet a patient who claims this time will be different and how previous therapists have failed him and the professor says this person is most likely a psychopath so to skip them; Psychopathic charm. I don’t get it though. As a therapist, can’t you just use a psychopath for research then? Do moral psychopaths not exist? Plus you get paid anyways. Also some patients probably are hard on themselves and exaggerate the inefficacy of therapists.

Avoiding tedious effort, we should aim to learn to recognize when mistakes are more likely to be made and avoid significant mistakes in serious situations, as I mentioned earlier.

Chapter 2 - Attention and Effort

Add-1 Task

The Add-1 Task is to select random string of 4 digits and read out the digits with +1 added to each digit. So a 9 would become a 0. The person should wait for 2 beats of a metronome (I think they are supposed to hit the desk with their hand themselves) before reading it out. For better demonstration of system 2, the challenge can be made into Add-3.

Dilated Pupils

A dilated pupil can be interpreted as someone being interested. Whether in what they are looking at or in glee for the information they have processed. Pupils are sensitive indicators of mental effort. There is distinction between dilated pupils from mental effort and emotional effort. My girlfriend definitely recalled once when I was programming during a video call with her.

Jackson Beatty and Kahneman decided to do a test where participants would be asked questions and they would answer while pictures of their eye would be taken using an infrared camera on every beat of a metronome. The task of Add-1 was given where participants would speak each digit one by one, allowing the cameras to capture correlations of dilation with respect to many factors.

Add-1 with four digits caused a larger dilation than the task of holding seven digits for immediate recall. Add-3, which is much more difficult, is the most demanding that I ever observed. In the first 5 seconds, the pupil dilates by about 50% of its original area and heart rate increases by about 7 beats per minute.

When tasks become much harder, the pupils shrank or stop dilating. Therefore maximum dilation also means maximum tolerable difficulty, which if breached will lead to giving up.

After witnessing the pupils under a mundane conversation, where the pupils did not dilate, the conclusion drawn was that System 2 usually works at “the pace of a comfortable walk” and is sometimes interrupted by jogs/sprints. Priority to other tasks decreases as mental energy increases and so blindness also is coincident with mental strain.

The Law of Least Effort

As people are more skilled at tasks, less energy is require and this is a symptom of laziness being built into us. Effort is a cost so if we want to lower it, we acquire skills.

Prefrontal cortex associated with intelligence. Additionally, context switching is effortful. Ability to control attention is not just for general intelligence signals but performance in careers such as air traffic controller or air force pilots.

Time pressure makes the mind work uncomfortably hard.

We avoid mental overloading by committing intermediate memory into long-term memory or paper rather than overloading the working memory. Therefore, one might say to get better at writing/typing, but we can take it a step further that we can study the biological ways to expand the working memory as well as determining at what point a CPU cache is too big.

Chapter 3 - The Lazy Controller

Flow, A State of Effortless Attending

Mihaly Csikszentmihalyi studied the state of effortless attending, flow. Concentration so deep that you lose sense of time. I have attained this flow myself several times. The last time where I achieved flow was a week ago. It was Friday and I was working. Three hours or more went by just like that. A recent but more memorable example is the C# console application template I made. The purpose of the project is irrelevant, but it was definitely something that required no self-control and one that did require me expending energy. I was unable to simply pause the project, I was determined to make it work just the way I wanted it to work before I went to sleep. When there is no effort required in self-control, resources are free to be expensed on the task at hand. The question I have is did the task muster the flow or is flow a perk of some sort that improves the performance at the task?

Self-Control Requires Attention and Effort

When System 2 is cognitively busy, system 1 has more influence and so you can be swayed more easily and make choices that are easier (e.g. selfish choices, use sexist language, and make superficial judgments). Being cognitively busy is not the sole cause of these actions though, alcohol and sleep deprivation can have an affect too.

Ego depletion is when you are using effort on self-control so are unable to take on other challenges. Sounds a lot like what recovering addicts go through. “When there is ego-depletion, people succumb more quickly to the urge to quit.” One study was how physical stamina of a handgrip task would be less while the participant was told to inhibiting emotions and another study was about how resisting the temptation of chocolate cake by eating virtuous fruits would later lead to giving up earlier than normal when faced with a difficult cognitive task.

Ego depletion is not the same mental state as cognitive busyness, and so the incentives have to become stronger to overcome the depletion.

Blood glucose is the fuel of mental activity, and effortful mental activity corresponds to a higher gear and thus more fuel consumption. So what is the best way to restore glucose while trying to reduce weight?

It was found that after being ego-depleted by a task (watch this, ignore that), drinkers of lemonade with sugar were not depleted anymore compared to the drinkers of lemonade with Splenda.

Disturbing consequences: It was found that for parole judges in Israel, although the default decision is denial, with 35% approval rating, the proportion spikes after each meal, when about 65% of requests are granted. The approval rating drops to zero just before the next meal. What can we do? The default should be approval, and the judges have to write reasons to disapprove. Of course that means more criminals are on parole, but the point of the justice system is rehabilitation and reintegration, not punishment.

bat-and-ball Puzzle

Bat and ball costs $1.10. The bat is $1 more than the ball. How much does the ball cost?

Most people that just rely on intuition, will answer 10 cents, but the answer is 5 cents after working through it. When we answer the question with 10 cents, our system 2 would rather approve system 1 thinking that invest the small effort in double checking the answer. At renowned universities (Harvard, MIT, Princeton), 50+ % of students gave the wrong answers however at all the other universities, it was in 80+ %. The law of least effort is at work and many do not have active minds?

In Grade 12 philosophy, we did some exercises on logic. In this book we are asked to determine if this argument is sound as fast as possible

1. All roses are flowers.
2. Some flowers fade quickly.
3. Therefore some roses fade quickly.

I determined it was wrong because I re-read the first sentence to ensure I interpreted it properly. When I first read this, I made the relationship that flowers are the subset of roses, but when reading the third statement, I re-read the first statement to ensure that it wasn’t the other way around and it was the other way around so I knew the argument was not sound.

This experiment has discouraging implications for reasoning in everyday life. It suggests that when people believe a conclusion is true, they are also very likely to believe arguments that appear to support it, even when these arguments are unsound. If System 1 is involved, the conclusion comes first and the arguments follow

I wonder if this premise follows the conclusion reasoning can be applied elsewhere. *cough* *cough*.

A funny question asked was to guess the number of murders in Michigan which resulted in lower guesses than people asked about murders in Detroit, because people do not think about Detroit when they are told about Michigan. I don’t even live in the USA. I didn’t/don’t even know if Detroit has a murder problem!

Those who avoid the sin of intellectual sloth could be called “engaged.”

It’s funny the book mentions this, when I wrote about these sins just yesterday. However I called sloth lethargy, but clearly sloth is used for a reason.

Intelligence, Control, and Rationality

If people were ranked by their self-control and by their cognitive aptitude, would individuals have similar positions in the two rankings?

The experiment was one where four-year olds would be subject to staring at an Oreo and a bell for 15 minutes (nothing else in the room). They were told beforehand that if they waited for 15 minutes, they could get another oreo. However, they could at any time ring the bell and eat the oreo. The study found that those with self-control went on to have higher scores on intelligence tests.

Another study found that a game that trained attention (move the cat to the grassy area by avoiding the muddy areas), improved executive control as well as nonverbal tests of intelligence that was maintained for several months. Reminds me of that un-natural spatial IQ deficiency between women and men which is a result of differences in how they are raised but can be fixed.

Another study by the same group, identified specific genes involved in control of attention, showed that parenting techniques also affected this ability, and demonstrated a close connection between the children’s ability to control their attention and their ability to control their emotions.

Shane Frederick decided to study people who failed these anti-intuitive questions (including the bat and the ball),

prone to answer questions with the first idea that comes to mind and unwilling to invest the effort needed to check their intuitions.

These show up else where:

• 63% prefer to get $3,400 this month rather than next month compared to 37% of people who solved all 3 questions
• WIlling pay twice as much as solvers for overnight delivery of books that were ordered
• System 1 is impulsive. So when about to make an impulsive decision, keep track until system 2 naturally disagrees more often.

Keith Stanovich and Richard West are credited with System 1 and System 2 introduction, although they call it Type 1 an Type 2 processes. They stress that rationality and intelligence are separate as rationality prevent the biases whereas intelligence tests have not incorporated jumpy thinking (my words).

Chapter 4 - The Associative Machine

The book tries to gaslight you into thinking when you are shown Bananas Vomit that you will instinctively be disgusted and have a “temporary aversion” to both bananas and the book. When I saw the words, I was just confused because who puts those two words together. Incoherence was my first thought. Maybe I do not have much disgusting vomit memories to recall or maybe it’s just that I am looking forward to eating dinner. You see, I have to read 5+ chapters of this book today to catch up for my class and for today’s reading (Chapter 10).

It’s funny I call it incoherent because as I read on, the book talks about associatively coherent; where your cognitive, emotional, and physical responses are diverse but in sync (he calls it integrated).

From the psychologists perspective: each idea is a node that is linked to many others. Links can be causes, properties, categories. When people think of ideas, it is like an area blast, where many ideas are activated at once, but only few render to consciousness. I think of it like a video game where a grenade exploded, but clearly, the blast centre is significantly easier to notice than the edges.

We know far less about ourselves than we feel we do

Priming Effect

When we encounter words (or ideas), we are primed to think of the related ideas when faced with the next input. For example I’m hungry. What letter goes here: SO_P? Did you say SOUP? What if I said I plan on going to shower soon. Maybe SOAP? Khaneman says that the initial idea/word of EAT or WASH primes us to prefer one mental path over another. Psychological research pertaining to mapping these ripples is an exciting pursuit.

There is more to priming than concepts and words. Actions and emotion can be primed by events that you are unaware of.

In the Florida effect study done by John Bargh, students were asked to come up with sentences from a set of five words and the students who assembled sentences related to the elderly {Florida, forgetful, bald, gray, or wrinkle} walked slower in the task of being told to walk to another room.

The ideomotor effect:

…none of the students reported noticing that the words had had a common theme, and they all insisted that nothing they did after the first experiment could have been influenced by the words they had encountered. The idea of old age had not come to their conscious awareness, but their actions had changed nevertheless

Reciprocal priming also exists where walking slow leads to recognizing words related to old age!

In the book there is an interesting example of how people can be made to smile subconsciously by telling them to put a pencil in their teeth so that the eraser is facing the left and the point faces the right. This smiling is also associated with finding things funnier although the participant would not notice.

Habitual connections is all it takes. Heading nodding up and down can translate to being more agreeable. This applies even to voting when a vote occurring in a school makes voters partial towards school motions as well as if images are shown. Images have a greater effect than being a parent versus a non-parent.

We can even prime people to be money-primed and the effect will be significant enough to be noticeable. From being more selfish to prefer being more independent.

reminding people of their mortality increases the appeal of authoritarian ideas, which may become reassuring in the context of the terror of death

*cough* *cough*

Priming phenomena arise in System 1, and we have no conscious access to them, even if these priming examples lead to percents in changes.

A funny example of priming is an honesty box where there was a banner close by with either eyes or flowers and the money given per litre of milk consumed moved significantly. It goes to show you that we can make people behave more ethical not be actually violating people’s privacy, but my priming them with the idea that they are not in the clear to act unethically.

We can apparently prime people to be critical, so this can even translate into software development where people review PRs. We could possibly prime the reviewer by tinkering with the review interface.

Chapter 5 - Cognitive Ease

Cognitive Ease is measured from a range of easy to strained where easy indicate well going (no threats, no major news, no need to redirect attention) and strained indicates a problem exists. This is cognitive strain and it includes the current level of effort and the presence of unmet demands.

Illusions of Remembering

Recency bias but for words combined with cognitive ease can make those words or concepts hold greater value just because of recognition rather than fact.

A reliable way to make people believe in falsehoods is frequent repetition because familiarity is not easily distinguished from truth. It is most applicable when not remembering the source of the fact but the familiarity.

“We have always offered peace. They have always rejected peace.”

Illusions of Truth

• Bolding statements make them more believable
• In colour, bright blue or bright red are more believable
• couching familiar ideas in pretentious language is taken as a sign of poor intelligence and low credibility
• make the message memorable (in verse)
• Even a source has to have an easy name

Going back to the intuitive question study, the study also found out that student performance was better when the font was worse than when the font was good. Is this is a signal to publish important information is media that is hard to focus on?

The Pleasure of Cognitive Ease

People believe that stocks with easier to pronounce names will perform better than clunky names. IPOs of companies with easier to pronounce names do better in their first week.

The mere exposure effect indicates not based on frequency but based on flashing the words as someone, thereby leaving System 2 in the dark. Repeated exposure calms down the natural distress of novel stimuli.

Ease, Mood, and Intuition

• Remote Association Test
  • Give three words, ask for the shared associated word
  • Cottage Swiss Cake: Cheese
  • Dive Light Rocket: Sky (I said space)
• Apparently cognitive ease makes the words feel more related to each other. Mood plays a big role in intuitive accuracy. The words themselves can invoke emotions that lead to coherence.

Chapter 6 -Norms, Surprises, and Causes

The system 1 personal model is constructed by associations that link ideas of circumstances, events, actions, and outcomes that co-occur with some regularity,

There are events that happen that you don’t expect to happen and are not surprised when they do occur. Normal situational, but not probable enough to be actively expected. I guess a phone call would fit in here? All it takes is one incident to no longer be surprised. Anecdotes are given in the text of running into an acquittance on a vacation and then two week later somewhere else.

How many animals did Moses take into the arc? It was Noah not Moses. The moses illusion! I disagree though since this is a poor example.

Brain activity has a distinct pattern when it encounters words that are odd. When someone says a normal table, we know generally how it is, but we can’t give a specific definition.

Seeing Causes and Intentions

“Fred’s parents arrived late. The caterers were expected soon. Fred was angry.” We know that Fred is angry because of the parents arrived late and not because the caterers were expected soon. We know the link between punctuality and possibly anger.

A good example is when Saddam Hussein was captured, bond prices went up and then down but for both Bloomberg’s headline was that the capture lead to US treasuries being up and to them being down.

A statement that can explain two contradictory outcomes explains nothing at all. In fact, all the headlines do is satisfy our need for coherence:

The Limited information lets System 1 find a coherent casual story to link the fragments of knowledge at its disposal.

Belgian psychologist Albert Michotte argued in 1945 that we see causality not that we infer physical causality from repeated observations of correlations among events. An example of a 2D black square moving into another black square and that black square “starts moving.”

People are prone to apply casual thinking inappropriately, to situations that require statistical reasoning.

Chapter 7 - A Machine for Jumping to Conclusions

Jumping to conclusions: efficient when likely to be correct and occasional mistakes are tolerable. Risky when when the situation is unfamiliar, and the stakes are high.

The example provided to the reader that illustrates jumping to conclusion is hand drawn ABC but the vertical step of the B is spaced meaning that it could’ve been interpreted as A13C, On the right side is 12 13 14 but the 13 is drawn similarly to the B. The middle picture is a sentence using bank, but it could’ve been talking about a river bank. This example illustrated the point that jumping to conclusions can be simple and when it happens we never register that there was ambiguity to begin with. Daniel Gilbert, author of Stumbling to Happiness conducted a study on non-sense false sentence that participants had a hard to “unbelieve” when the sentence was shown while their system 2 was pre-occupied with another task (holding digits in working memory).

positive test strategy: the process in which system 2 tests a hypothesis by looking for examples that fit the narrative unlike scientists who try to refute the hypothesis.

Halo Effect

Underused term that describes common biases perfectly. The halo effect gives credibility where evidence is lacking or favourability over another due to the order facts are encountered. The halo effect suppresses ambiguity.

Are these words possible to describe a single person? intelligent—industrious—impulsive—critical—stubborn—envious

It seems that if we split the last three words to describe someone else, people believe it’s impossible that all six can be attributed to a single person.

Group estimates work best when each person makes an estimate individually and is not influenced by others in the group as the biases average out. Witnesses to an event should not be allowed to discuss it with each other before testimony.

Application to corporate executives: when having meetings discussing an issue, each person should be write a summary of their position which results in valuing the groups diverse knowledge rather than dealing with differences in how loud and often people talk.

To combat jumping to conclusions (What you see is all there is or WYSIATI), we may need to work backwards. So when asking if X is Y based on Z, we start by defining what it means to be Y and then assess if Z addresses all of what it means to be Y. Questions like is Mark an ethical person.

People who hear one side of the story will almost always have higher convictions of their judgement that those who hear both sides to a story / legal case.

Biases

• overconfidence: can’t tell if the evidence is missing
• framing effect: how you structure a sentence leads to different reactions and thoughts
• base-rate neglect: ignoring what the real world probability is

Chapter 8 - How Judgments Happen

System 1 continuously monitors the situation and gives basic assessments, which can form intuitive judgement. Substitutions of judgement is allowed. Translating values across dimensions (how tall would Sam be if they were as tall as their intelligence).

Any intension of system 2 to answer a question triggers system 1 again. Situations are constantly evaluated as good or bad, requiring escape or permitting approach. We have the ability to judge people as safe or dangerous off a single glance. It influences how people vote. Ratings of competence rather than likability are the predictors of voting choices, albeit it can only explain 70% of results. Competence is a combination of strength and trustworthiness. Strong chin, slight confident smile,

When people were asked how much they would contribute to help out birds that would otherwise drown, the number of birds to help didn’t matter because the groups still averaged the same amount of money. “The almost complete neglect of quantity in such emotional contexts has been confirmed many times.” Is it the neglect of emotional contexts, or does emotional contexts translate to a price for people and people would still answer with the max amount of money?

The mental shotgun is the inability of system 2 to direct system 1 to one train of thoughts. System 1 will always do more than just what is required of it. An example is asking someone who hates big tech if the companies in big tech are financially well off. Or better yet, asking someone who likes a product if the company is doing well (when the company is not in fact doing well such as Peloton).

Chapter 9 - Answering an Easier Question

The target question is the assessment you intend to produce.

The heuristic question is the simpler question that you answer instead.

The target question is the assessment you intend to produce. The heuristic question is the simpler question that you answer instead.

Substitution and heuristics in this sense stems from the mental shotgun.

For “How” types of questions, intensity matching is the process of converting the answer of a heuristic into an answer for the original question. If both things have a scale, then we can easily translate in our heads using System 1.

Going back to the happiness question when asked about happiness and more, the order matters. If a question that could be a heuristic for happiness is asked first, it severely influences the happiness question. The example is asking about dating life and then happiness.

Chapter 10 - The Law of Small Numbers

The title of this chapter is a play on words. The law of large numbers does not actually apply to small numbers, although intuitions believe so!

Why are the counties with the lowest and highest rates of kidney cancer rural? Rural life can’t explain both but being presented with one of these will make you use the rural life as an explanation of the statistic.

These are statistical facts so the actual possibility could simply be that there isn’t enough of a sample size to draw a solid conclusion. When population is smaller, the variability of any observation rate is high and cannot be taken as mere forward thinking fact. Ergo randomness.

The explanation can be done with marbles. Imagine there’s a pool of marbles of two different colours and two people sampling different sizes. The person sampling a smaller size is more likely to hit a homogenous sample than the person drawing more marbles. So if red marbles are kidney cancer and white marbles are not kidney cancer, then the person drawing smaller samples (for each county) is going to be hitting both all no cancers and all cancers.

Artifacts: results entirely produced by the method of research

psychologists commonly chose samples so small that they exposed themselves to a 50% risk of failing to confirm their true hypotheses

All this even though a sample size can be computed from the acceptable risk of failure.

Then when we discuss polls, people are not sensitive to the sample size!

There is a general bias that favour certainty over doubt which requires the hard work of sustaining doubt.

Cause and Change

In a hospital the sex of boys and girls are recorded. Are all three of these sequences likely?

BBBGGG
GGGGGG
BGBBGB

I already knew about this sort of bias somewhere before, so I said yes they are equally likely, because true randomness does not look as intuitive as we’d expect. The intuitive answer is that the third one is the most likely, but this is faulty logic, since each of the three sequences is a probability of 0.5^6.

We do not expect to see regularity produced by a random process, and when we detect what appears to be a rule, we quickly reject the idea that the process is truly random. Random processes produce many sequences that convince people that the process is not random after all.

The Gates Foundation invested money to find out the characteristics of the most successful schools. One of the conclusions of this research is that the most successful schools, on average, are small. In a survey of 1,662 schools in Pennsylvania, for instance, 6 of the top 50 were small, which is an over-representation by a factor of 4..

What are your thoughts on this finding based on what we discussed at the start of the chapter?

The first thing that came to my mind due to what was mentioned earlier was that the schools were small because of the inherent variability in smaller schools similar to rural counties. I’m surprised the book thinks that the readers have not learned how to recognize the bias yet. It turns out that even the worst schools were smaller so all these foundations investing in smaller schools might have backfired.

Takeaways of the chapter

• it is easy to pay more attention to the result than the reliability of the data used
• how certain are we the results are not chance?
• when drawing conclusions of good, we should also look at the conclusion of bad. To do good is also to avoid doing bad. (I added this)

Chapter 11 - Anchor

anchoring effect: consider a particular value for an unknown quantity before estimating that quantity

If you consider how much you should pay for a house, you will be influenced by the asking price.

Anchoring as adjustment

• Start at anchor and move away from it until no longer certain
• Therefore, we stopped too soon
• Insufficient adjustment

Anchoring as primer

When the anchor is absurd, we still are primed by it to think of items with similar characteristics.

Anchoring Index

The anchoring index is simply the ratio of the two differences (562/1,020) expressed as a percentage: 55%.

Real Estate agents will deny that the asking price influenced their decision of what the lowest price they would accept would be.

When asking for donations from people sensitive to the cause, even an extravagant number ($400) can lead to higher average donations ($143) compared to the average of $64 without an anchor, and an average of $20 with a low anchor ($5)

Example of Anchoring

Judges asked to roll a dice of 3 or 9 and then sentence the person; anchoring effect of 50% (5 or 8 months)

Rationing of campbell soup led to more soup purchased

Advice when encountering a ridiculous offer is to make a sene and leave rather than make an equally ridiculous offer which will hinder further negotiations.

To improve negotiations, think about lowest price that the opponent would accept as well as what the cost of losing out is.

Maybe due to anchoring effect, penalties should be capped based on the defendant’s cash position? e.g. (10% of Retained Earnings, 10% of net tangible value).

Chapter 12 - The Science of Availability

availability heuristic: substitute one question for another when wanting to estimate the size of a set by using the ease at which instances of the set come to mind

awareness of your own biases can contribute to peace in marriages, and probably in other joint projects.

When asked to rate assertiveness after listing six versus twelve examples, assertiveness rating is lower with more examples listed. This applies to lack of assertiveness (the more examples of lack of assertiveness asked, the more assertive the person rated themselves). The effect is not present if the participants are told beforehand that there is a reason for the diminished recalling.

Chapter 13 - Availability, Emotion, and Risk

Victims and near victims are very concerned after a disaster. After each significant earthquake, Californians are for a while diligent in purchasing insurance and adopting measures of protection and mitigation…However, the memories of the disaster dim over time, and so do worry and diligence.

Protective actions as done by individuals and governments are designed by tracking only the worst disaster experienced, assuming that future disasters could be no worse. It is hard to think that way though. This is a very good takeaway. We need to reassess some of our protective actions, especially for climate change, based on what could go wrong, not based on what has gone wrong.

• Strokes cause almost twice as many deaths as all accidents combined, but 80% of respondents judged accidental death to be more likely.
• Tornadoes were seen as more frequent killers than asthma, although the latter cause 20 times more deaths.
• Death by lightning was judged less likely than death from botulism even though it is 52 times more frequent.
• Death by disease is 18 times as likely as accidental death, but the two were judged about equally likely.
• Death by accidents was judged to be more than 300 times more likely than death by diabetes, but the true ratio is 1:4.

estimates of causes of death are warped by media coverage

affect heuristic: using how do I feel about it to answer what to think about it.

This heuristic makes the world seem more ordered than it is in reality. When people hear about benefits, they automatically think there is less risk (vice versa as well). In the real world, trade-offs do exist.

According to Paul Slovic, evaluation of risk depends on the measure itself (death per million, death per millions of dollars of product produced). “defining risk is thus an exercise in power.”

Cass Sunstein takes another stance which is that the existing USA regulations are about public perception and not rationality. He argues that we should focus on saving life-years, not even just life. By looking at life-years, we would focus on younger people more. Insert: this was a very important distinction, as I viewed adults and children as equal life before I read this. I still dislike when media, including government statements always include that women are affected even when men are affected more (e.g. homelessness in Canada).

availability cascade: biases flow into policy

On some occasions, a media story about a risk catches the attention of a segment of the public, which becomes aroused and worried. This emotional reaction becomes a story in itself, prompting additional coverage in the media, which in turn produces greater concern and involvement. The cycle is sometimes sped along deliberately by “availability entrepreneurs,” individuals or organizations who work to ensure a continuous flow of worrying news….anyone who claims that the danger is overstated is suspected of association with a “heinous cover-up.”

• Indigenous mass graves
• COVID-19 Lock downs and flight restrictions
• Monkey-pox?
• Book examples: Love Canal affair and the so-called Alar scare
  • Love Canal, buried toxic waste was exposed during a rainy season in 1979, causing contamination of the water well beyond standard limits, as well as a foul smell.
    • The legislation that mandated the cleanup of toxic sites, called CERCLA, established a Superfund and is considered a significant achievement of environmental legislation.
  • Alar is a chemical that was sprayed on apples to regulate their growth and improve their appearance. 1989
    • “when consumed in gigantic doses, caused cancerous tumors in rats and mice”
    • The manufacturer withdrew the product and the FDA banned it.
    • subsequent research deemed very small risk of a carcinogen
    • fewer good apples were consumed

It’s possible for the Canadian Cabinet to have non-MP members, and apparently even the majority, although its very unusual for the non MPs to not seek and win election.

Chapter 14 - Tom W’s Specialty

This chapter is about how dubious information still spoils rational thinking and that the imperfect choices made are due to System 2 laziness and not ignorance, meaning that if we keep System 2 engaged or ignore dubious information, we can think straight. The problem I have is that most of the time, the information isn’t dubious but rather arbitrary and the conclusions others have made due to quick thinking. How can we invoke other people to re-evaluate?

• Using the base rate is obvious when no other information is provided.
  • When a personality is given even when it should not be trusted, people will forego the base rates and rank the field of study according to stereotypes
• Stereotypes - judging via representativeness - can sometimes be true but also false
• Excessive willingness to predict low base-rate events
  • People with PhDs are more likely to subscribe to The New York Times than people with only a high school diploma
  • If someone is reading The New York Times, what is a better bet? That they have a PhD or that they do not have a college degree? The latter of course, since implications do not imply the vice versa

instructing people to “think like a statistician” enhanced the use of base-rate information, while the instruction to “think like a clinician” had the opposite effect.

• A method of improving predictive accuracy is to engage System 2 (e.g. via frowning)
  • Therefore, System 2 was failing due to laziness and not due to ignorance!
  • Individuals need to invest special effort to properly conduct analysis, it makes sense why comment sections are often filled with spam and ignorance

There is one thing you can do when you have doubts about the quality of the evidence: let your judgments of probability stay close to the base rate.

Constraining Yourself to the Logic of Probability

When you believe in an event, you need to confirm that the resulting logical probabilities (i.e. Bayesian statistics) are also what you believe

if you believe that there is a 30% chance that candidate X will be elected president, and an 80% chance that he will be reelected if he wins the first time, then you must believe that the chances that he will be elected twice in a row are 24%

1. Give high credence to the plausible base rate even with provided evidence
2. Question how applicable the evidence is (assurance)

Taking Advantage of Representativeness

The book mentions how to watch out for representativeness, but I re-arrange some examples for the beneficiary of the underdog.

• Even if a company/individual has a highly core competency, you need to present yourself well: well-trimmed lawn, competent looking receptionist, etc.
• Rare event predictions due to weak evidence, but what is weak evidence?
• A damning report, but how sure that the evidence is solid (e.g. Bear reports on certain stocks)

Chapter 15 - Linda Less Is More

Fictitious person.

Linda is thirty-one years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations.

• Linda is a teacher in elementary school.
• Linda works in a bookstore and takes yoga classes.
• Linda is active in the feminist movement.
• Linda is a psychiatric social worker.
• Linda is a member of the League of Women Voters.
• Linda is a bank teller.
• Linda is an insurance salesperson.
• Linda is a bank teller and is active in the feminist movement

Sets up a contradiction to the intuition that Linda is part of the feminist and a bank teller compared to just being a bank teller. “feminist bank teller” ranked higher than “bank teller” in both. Representativeness wins against logic even when asking Statistics doctoral students.

Problem was shortened to just “bank teller” versus “bank teller and active in the feminist movement” and people still answered with the second choice.

‘but she can’t just be a bank teller; read the description.’

When Detail Disrupts Perception

Adding detail to scenarios makes them more persuasive, but less likely to come true

To appreciate the role of plausibility, consider the following questions: Which alternative is more probable?

1. Mark has hair.
2. Mark has blond hair.

Which alternative is more probable?

1. Jane is a teacher.
2. Jane is a teacher and walks to work.

Less is More

John List economist experimented with baseball cards. When auctioning sets of ten high-value cards, versus sets where three modest cards were added, the larger set is valued less in single evaluation: where the group is shown only one of the sets than in joint valuation: where the observers can see both sets and then place a value.

WImbledon Survey

Rank from most to least probable:

1. Borg will win the match.
2. Borg will lose the first set.
3. Borg will lose the first set but win the match.
4. Borg will win the first set but lose the match.

72% ranked #2 lower than #3.

Multi-Faced Die

Regular six-sided die with four green faces and two red faces, which would be rolled 20 times. $25 if their sequence comes up.

1. RGRRR
2. GRGRRR
3. GRRRRR

Due to representativeness, we will gravitate to #2 even though it is simply just #1 but prefixed with G. It is actually less probable than #1.

Narrowing Down the Problem

Asking about hard numbers rather than percentages, reduces the error seen (65% to 25%). This is usually due to “frequency representation” which appreciates that one group is wholly included in the other. Forcing people think of individuals improves rationality

At the end of the day, try not to communicate in a way where probability is taken as plausibility.

Chapter 16 - Causes Trump Statistics

Statistical base rates are underweighted, but if you apply causality to them, it can be easily combined with the information in an individual case.

A cab was involved in a hit-and-run accident at night. Two cab companies, the Green and the Blue, operate in the city. You are given the following data:

• 85% of the cabs in the city are Green and 15% are Blue.
  • Implication: 85% of accidents involve green cabs (causal)
• A witness identified the cab as Blue. The court tested the reliability of the witness under the circumstances that existed on the night of the accident and concluded that the witness correctly identified each one of the two colors 80% of the time and failed 20% of the time.

What is the probability that the cab involved in the accident was Blue rather than Green?

Although our brain is tempted to write off the initial statistic, we need to apply Bayesian inference which results in a Bayesian estimate of 41%.

Solve for: Probability Blue given witness testimony

- P(B) = 0.15 (prior probability of the cab being Blue)
- P(G) = 0.85 (prior probability of the cab being Green)
- P(W|B) = 0.80 (probability of the witness identifying the cab as Blue if it was Blue)
- P(W|G) = 0.20 (probability of the witness identifying the cab as Blue if it was Green)
P(B|W) = [P(W|B) * P(B)] / [P(W|B) * P(B) + P(W|G) * P(G)]
P(B|W) = (0.80 * 0.15) / [(0.80 * 0.15) + (0.20 * 0.85)]
= 0.12 / (0.12 + 0.17)
= 0.12 / 0.29
≈ 0.4138
≈ 41.38%

Chapter 17 - Regression to the Mean

Useful information for betting, especially on golf. If someone does exceptionally well, is it skill or luck? If its skill, then they shouldn’t regress to the mean and do poorly the next round.

Chapter 18 - Taming Intuitive Predictions