THE BOOK IN A NUTSHELL
We are lousy decision makers – we make faulty judgements.
The original views of humans being ‘econs’ whereby we make perfect decisions through weighing up the pros and cons of each decisions has been shown to be palpably wrong. Instead we make decisions upon limited information (What you see is all there is – WYSIATI). Furthermore we can easily be seduced by emotion and seemingly irrelevant information in our decision-making.
In the previous millennia, social scientists and economists assumed that people were generally rational in their thinking and that most of the time our thinking was sound. However, recent work by neuroscientists and behavioural economists have shown that we don’t think logically. Instead we are unconsciously biased in our thinking which means many times our decisions are flawed. Some of the key reasons for this are:
1) We rely on short-cuts and general rules of thumb (heuristics) which are sometimes inaccurate for the situation at hand.
2) We are heavily influenced by what we can immediately recall when making decisions – and only what is ‘in field’ at the time is used to make decisions (A concept we will return to called WYSIATI – What you see is all there is).
3) Emotions and cognition (i.e. perception) easily derail our rational decision making.
4) We often answer an easier question than the difficult question posed (e.g. ‘Should I invest in Ford?’ gets answered instead by the simpler question, ‘Do I like Ford cars?’)
The problem is these happen unconsciously. Raising them to consciousness can help a bit but we are still prey to many of their effects (a bit like visual illusions, we know what’s going on but still get caught by it).
To help understand the weaknesses with our cognition, Kahnemen devised a metaphor of two parts of the brain – System 1 and System 2 (NB you cannot dissect the brain and find these two parts neatly divided into the different hemispheres).
System 1 & System 2 Thinking
The brain has two ways of working: System 1 and System 2.
System 1 is primitive, unconscious, runs automatically (and cannot be turned off). It’s emotional, intuitive, powerful, fast, impatient and impulsive. It can work on many issues/levels at the same time. It uses little energy, quickly creates meaning out of things and is easily influenced.
System 2 is linked to the Neo Cortex. It has conscious attention (normally running at a low level but then gets attuned onto specific issues). It is rational, methodical, cautious, has small processing power, limited capacity and is single focused. It is slow, a heavy energy user, is not able to control System 1 well and is lazy. System 2 can override System 1 under normal conditions but if System 1 is highly fired up (e.g. through the power of emotions) or System 2 is tired or pre-occupied then it fails to be able to control System 1. Furthermore, System 2 is very good at making comparisons but does not perform well under pressure.
System 1 – the state of unconscious flow – Most of our actions happen unconsciously. Csikzentmihalyi has defined the state of ‘flow’ as a place of unconscious competence, where the rest of the world melts away and you lose the sense of time and become lost in the experience. It seems that we perform our best when we move from conscious control to unconscious effortless competence.
System 1’s flawed instant response – Q: A bat and ball costs $1.10. The bat costs one dollar more than the ball. How much does the ball cost?
In cases like the above System 1 jumps to an intuitive answer ($0.1) when the real answer is $0.05. System 2, being lazy, does not intervene unless steps are taken to activate it (e.g. when told $0.1 is wrong). Even very intelligent students fail this test.
Hoodwinking the brain by distracting System 2 – If you distract the conscious System 2 brain, then it is less able to control the impulses from the System 1. When System 2 is distracted, it allows other statements of falsehood to sneak through the System 1 brain (System 1 will believe almost anything!). This is why storytelling works so well as System 2 is engaged in the logic of the story allowing the moral to sink straight into System 1. System 2’s ability to control System 1 also diminishes with tiredness, drinking or conscious mental effort. Furthermore, the trouble with self-control is it’s tiring – being ‘on guard’ needs a lot of energy.
When people are offered two desserts: a chocolate cake or a virtuous fruit salad, they are more likely to choose the chocolate cake when they have to complete some mental arithmetic.
In a study with eight parole judges, they found no prisoners were granted parole just before lunch (when their System 2 energy was low) but the rate after lunch was 65% of cases (versus an average of 35%).
Another way to increase plausibility is to wrap your statement in other known truths – a silly example is ‘A chicken has 4 legs’ takes a little while longer to decipher its falseness than ‘A chicken has three legs’. Or ‘How many animals did Moses take with him on the Ark?
Also if you use high quality paper and clear print it makes it cognitively easier to read and so less likely to be judged. Also try to make the language more memorable by building rhythm, alliteration or repetition (We shall fight on the beaches etc).
Finally, quotes are more credible if it comes from a name that is easy to pronounce.
In experiments, ‘A fault confessed is half redressed’ was more easily recalled and thought more meaningful than ‘A fault admitted is half redressed’.
Easily pronounced words create more favourable reactions. For the first week after floating on the stock market, Companies with easier to pronounce names do better than those with less easy to pronounce names.
How to kick System 2 into action – System 2 is lazy and will often not get involved until triggered to – for example by making the font difficult to read.
The Criterion Referenced Tests (CRT) fool people because they suggest an intuitive System 1 answer. However when told they are wrong, System 2 jumps in and quickly finds the real answer.
Q: If it takes 5 machines 5 minutes to make 5 widgets, how long will it take 100 machines to make 100 widgets?
When the questions were set in light grey, System 2 jumped in and the respondent’s wrong answers dropped to 35% whilst in normal black font 90% of student made at least one mistake.
(A: 5 mins)
Some people have better developed System 2’s to control their System 1 thinking – Scientists have found specific genes that influence attention and our ability to control our emotions. For some they have a more developed System 2. However, System 1 dominant people instead tend to be more easily influenced (with little critical judgement), are impulsive, impatient and keen to receive immediate gratification (thus leading to ill-considered decisions and judgements). They are also prepared to pay twice as much for overnight delivery of a book than others!
Mischel offered four-year olds a choice between a small reward now (One Oreo) or a larger reward (two Oreo’s) in 15 minutes. About half the children managed the task (mainly by distraction techniques). 15 years later, they found those who resisted had greater cognitive and personal self-control (as well as having developed improved IQ scores) (http:// www.youtube.com/watch?v=x3S0xS2hdi4).
Cognitive bias (What You See Is All There Is)
The way our two thinking systems work can lead to a wide range of flawed perceptions and hence decision making:
WYSIATI – We often make decisions on just a small subset of information available. If we do not retrieve useful information at the time of decision-making it may as well not exist. The trouble is our brain will only use the information it has immediately available to make its decision (A concept called WYSIATI – What you see is all there is).
We do not use all the information available to us to arrive at our assessment of a situation. Research into painful medical procedures found patients assessed the overall experience of the operation by two elements – the peak level of pain experienced and the level of pain at the end of the operation (with the length of the procedure not being a key factor).
When the Valdez oil spill happened, people were approached for donations to save oil covered birds. In one sample they were told the fund was to save 2,000 birds, another 20,000 birds and the third 2m birds. The number did not dramatically affect average donations ($80, $78 and $88).
Availability – Depending how easy it is to retrieve information on a subject will influence the quality of our decision (as WYSIATI). For example we believe there is greater divorce rates amongst actors and actresses because we hear about them more often. We think dying in a plane crash is more likely than it is – especially after a much publicised one. Furthermore, we recall our own activities and effort better than other people’s (a key problem in marriage
or work based disputes is we under-emphasise the activity of the other person, and feel our part has not been recognised enough by the others).
If you ask a person to list 6 times when they were assertive and then ask them how assertive they are, they will believe they are more assertive than if not primed. Interesting if you ask for 12 examples, people see themselves to be less assertive because it’s more difficult to recall 12 than 6. Thus it’s fluency of recall rather than absolute number that is the key.
Cognitive ease – The easier it is to tangibly recreate the event in our minds (especially visually) the greater the chance it will be ‘in play’. Vividness increases its saliency. We find it easier when things are more tangible and less easy when kept conceptual – for example it’s easier to answer ‘Out of 100 people how many…’ than ‘What percentage…’ as we get a clearer representation in our heads.
Example: ‘A vaccine that protects children from a fatal disease carries a 0.001% risk’ vs ‘One of 100,000 vaccinated children will be permanently disabled’. The second statement conjures up a clearer image in your mind of a disabled child (and we choose not to picture the 99,999 healthy children). Such presentation significantly influences the decision of parents to treat.
Denominator neglect – Focusing on one thing takes our eye off other things. Thus what we focus on is in the ‘field of play’ and other things are consequently pushed ‘out of play’ and hence out of influence (as WYSIATI). The invisible gorilla video is a good example of how our conscious mind when focused on one thing can easily miss other things (http:// www.youtube.com/watch?v=vJG698U2Mvo).
Rare events – We overestimate the likelihood of improbable but highly emotional events (and often underestimate more common events), as the intensity of emotions make us recall something more vividly. This then over-influences our decisions.
When people are asked to assess the frequency of deaths, they over-inflate events like tornadoes, lightening attacks, and under-inflate deaths from diabetes and asthma.
Conversely if an improbable event is not easily recalled then we tend to under-estimate its likelihood).
What’s on the field is in play – A further extension of WYSIATI is ‘What’s on the field is in play’ i.e. seemingly irrelevant data gets used. For example, if you put things side by side System 1 will try to create a connection. If you add more detail you increase plausibility.
A mock jury were given a greater, more vivid description of the events from one side (with no extra relevant factual information) and it dramatically influenced the verdict.
Less is more – Too much information distracts us from the important – Decision-making is usually easier with less information than more because System 2 cannot cope with lots of information and easily gets confused. Evidence suggests that seemingly few criteria can account for a large degree of predictiveness. Dawes amusingly suggested that marital stability could be predicted accurately by just two factors: Frequency of lovemaking minus Frequency of quarrels – simple rules are best!
Paul Meehl demonstrated that clinical predictions based on statistical analysis of a few key metrics were more accurate than clinician’s subjective impressions based on a 45 minute interview, backed up with a raft of other information. This study has been replicated in numerous other situations. In 60% of studies, algorithms have been shown to be more effective at predictions than people.
Dr Apgar develop a simple 5 test rule to assess the health of a new born child – the now famous Apgar score.
In an experiment where a mock jury was shown one side of a dispute (even though they could easily have worked out the counter argument), just hearing the one side dramatically influenced their decision-making. Furthermore there were more certain of their decision than those who heard both sides.
We know that seemingly irrelevant stimuli can influence us, whilst a computer is not distracted, and instead remains steadfastly focused on only the critical issues. Experts place too much weight on their intuitions, downgrading the importance of the key pieces of evidence – emotions steer us away from a purely rational response.
How to hire a person – select up to 6 key traits that are key to success in the role. Ideally they should be independent with little overlap. Develop some factual questions to assess each. Upon answer, convert them into a 1-5 scale. Then add up the scores and cross compare versus the other candidates. Vow to take the person with the highest score irrespective on how you feel. Resist the urge to ‘re-evaluate’ to change the rankings!
The wisdom of crowds – In some tasks people are very bad at predicting but the wisdom of the crowd is better.
Get a large number of people to independently guess the number of pennies in a jar, and the average is often close to the real answer.
Priming – We can easily be nudged to think in certain ways as we draw upon our unconscious associations.
People were asked to recall a situation they were ashamed of. When asked to fill in the word S_ _ P they typically said ‘soap’ not ‘soup’.
Priming does not just happen with words – your actions and emotions can be primed as well.
Psychologist, Bargh gave students a set of words to turn into a sentence. For one group he scrambled words associated with the elderly (such as Florida, forgetful, bald, grey, and wrinkled). They then had to take their test scores down the corridor. He found they walked slower than the other test group (yet none of them were aware of the ‘old’ theme).
Encouraging people to nod or shake their heads will prime then to agreeing or disagreeing to a statement.
Voting for improved school funding goes up when the polling station is held inside a school.
In a university kitchen, they operated an ‘honesty box’ system for for the teas and coffees. Over a ten week period a different image was placed above the honesty box. Alternate weeks showed either flowers or eyes. When eyes were displayed, the contribution to the honesty box was three times higher than flowers.
The symbols in a culture will unconsciously prime certain behaviour traits. So a culture full of statues of leaders prompts a different behaviour from one full of Christian images or symbols of wealth.
Decision-making by frames of comparison – We make judgements via comparisons. When buying something we find it easier to assess it alongside something else in the same category. Our world is broken into categories – and we use norms in that category to help us make decisions. However, we can easily be misguided depending upon the frame and comparative context.
What do you prefer? Apples or Peaches? Vs What do you prefer? Apples or Steak?
A man was shot in the arm in a convenience store. Consider two scenarios: 1) It was his usual store, 2) his usual store was closed for a funeral, so he went to this other shop instead.
Q: should compensation be different?
Whilst in theory the compensation should be the same, when people are given either scenario, we see a different level of compensation being awarded than if they see both. Because we live in a world of WYSIATI, not seeing the other side means we only make a decision based on that limited amount of information. !
Imagine a situation where two people change to more fuel efficient cars:
A: Adam switches from a car that does 12mpg to one that does 14mpg.
B: Beth switches from a car that does 30mpg to one that does 40mpg.
We all instinctively think Beth is doing a better job for the planet, but over 10,000 miles, Adam’s actions has saved 119 gallons whilst Beth has just saved 83 gallons. Thus we get hoodwinked by the frame of mpg rather than the more relevant frame of gallons saved.
Context determines meaning – In the example below the image (I3) in the middle is the same, but in one we read it as a B (as it is surrounded by A and C whilst in the other we read it as 13 (as surrounded by 12 and 14). Thus we can easily jump to the (wrong) conclusions.
First impression bias – First impressions are the most influential and we find it much more difficult to change our minds once initially set upon a certain course of thinking. Furthermore, the mere sequencing of information can affect our perceptions.
What do you think of Alan and Ben?:
Alan: Intelligent, industrious, impulsive, critical, stubborn and envious.
Ben: Envious, stubborn, critical, impulsive, industrious and intelligent.
Research shows that Alan is liked more than Ben.
When marking exam papers, if the first question the candidate answers is good, then we are more influenced to mark the rest of their questions on average higher than if their first question was poor.
Anchors – Random information can ‘anchor’ our thinking. For example, in purchasing something (like a house), we will be heavily influenced by the first price suggested.
Half the respondents were asked ‘Is the tallest redwood less or more than 120 feet?’ The other half, ‘Is the tallest redwood less or more than 1200 feet?’ When asked to guess the height of the tallest redwood, those primed with a low number (120 feet) guessed on average of 282 feet and those anchored to the 1200 feet guessed 844 feet – a difference of 562 feet based on a spurious piece of data.
When an oil tanker spilled its load in the Pacific Ocean, an experiment was run asking for donations. When no primer was given the average donation was $64. But when asked, ‘Would you be willing to donate $5 to the cause?’ the average offered was $20. When set a higher anchor of $400, the average sum raised was $143 – i.e. a $123 difference.
In a supermarket, Campbell’s soup was on offer with a sign above it. On some days it said, ‘Offer limited to 12 cans per person’. On other days it read, ‘No limit per person’. On the days with the limit, they sold on average 7 more cans per person – twice as many when there was no limit.
Even experts can get fooled.
Estate agents were asked to price a property. One group of agents were shown a previous much higher value – and the other group were shown a previous much lower value. This resulted in a 41% difference in the valuations.
In negotiations, it is recommended that rather than counter with an equally outlandish offer which creates often an unbridgeable gap, instead threaten to storm out unless that opening offer is taken off the table (as need to make it clear to them and most importantly yourself that that will not be a figure the negotiations will be anchored around).
The familiarity effect – We find that familiarity makes it much easier to accept something (our critical faculties drop with familiarity – hence the power of advertising or how a politician keeps repeating a denial – after a while a repeated falsehood becomes the accepted truth).
Answering an easier question – Human beings are to independent thinking as cats are to swimming. We can do it, but we prefer not to. The brain is lazy and rather than answering the real question, will often answer the question it has the answer to. System 1 finds an easier answer so System 2 does not have to work out the real answer (and because System 2 is lazy it does not impose much scrutiny on the veracity of the answer given by System 1).
‘How much am I prepared to pay for this?’ is replaced by an (easier) emotional one such‘How much am I prepared to pay for this?’ is replaced by an (easier) emotional one such as ‘How much do I like this?’
‘How happy are you in your life these days?’ becomes ‘What is my mood right now?’
‘How popular will the President be in 6 months time becomes ‘How popular is the President now?’
Moods affects our cognition – What we like/dislike affects our beliefs (because we let how we feel about something influence what we think about it). When happy we are more intuitive and System 1 is in control. When distracted by negative emotion System 2 becomes more controlling. !
System 1 – Making associations and meanings – System 1 creates associations to help us make sense of our world. This can lead on some occasions to making false associations.
Correlations ≠ causation – In Black Swan, Nassim Taleb mentions how bond prices increased on the day of Saddam Hussein’s capture. The two events were linked (when it was not the case in reality). System 1 does not try to assess all possible information or all possible explanations but jumps at the first one that makes sense to it.
Heider & Simmel used a video of animated geometric shapes that seemingly interact. People developed a story from the movements of a bullying scenario. Interesting Autistics do not create the associative story but see it for what it is (http://www.youtube.com/watch? v=VTNmLt7QX8E).
Kidney cancer is lowest in rural communities – by presenting these two facts together we create a causal relationship such as healthier living, better quality food etc – when in reality this is not the reason.
Under normal frequent situations, when System 1 jumps to conclusions it can be accurate. However, it becomes more risky when we are in infrequent/unusual situations as we often draw upon our experiences in other situations and mistakenly try to apply it to the new situation.
Storytelling – System 1 loves storytelling as it helps makes sense of things. Storytelling is therefore not a culturally imposed phenomenon, it’s one of our basic core programs we run.
Confidence does not come from the data – it comes from the story we create around the data. The better the story ‘fits’ the ‘facts’ the more believable the story. Furthermore, we will then ignore data that conflicts with our story (as the story becomes the ‘truth’).
Nassim Taleb introduced the concept of narrative fallacy in his book Black Swan. A narrative fallacy is a past flawed story that still shapes current perceptions. The more concrete the story, the more believable it is (even if it’s not true). What tends to happen is a story is created around a few ‘nodes’ which help explain those activities. We then chose to ignore other bits of information that do not fit the story. Furthermore we extrapolate via the halo effect (i.e. if the story is a positive one we then want to see all other aspects as positive – good people only do good things….as this helps keep the story coherent and simple). Such
is the desire of System 1 to make sense of things it can too easily create causal narrative that is factually untrue. But the power of the story seduces even System 2.
The Mathematics Of Cognition
Misleading by numbers – Numbers feel authentic and credible, but we can easily be misled by them. Large samples are more compelling due to their data size – but the trouble with them is they flatten out the data and create less anomalies. Thus the outliers in data gets suppressed (unlike in smaller samples). Often we ignore the impact of sampling variance on research studies (especially if the data tells a good story). System 1 suppresses doubt – it is naive and accepts things at face value. Likewise, if it makes sense, then System 2 rarely intervenes.
We are pattern seekers, trying to find order in the world. Our love of finding causal relationships forces us to often make poor decisions. When we detect a pattern in random data (such as a gambling machine) we will believe there is an order as we find it difficult to accept the concept of randomness and no order. In reality randomness does create random order.
Research into sports people so called ‘hot hands’ has been shown to be false.
Research into successful schools found a correlation with size of schools. This makes intuitive sense as we can build a strong narrative why this should be the case. This led the Gates Foundation to invest over $1.7bn into developing small schools. If the statisticians had investigated the worst performing schools they would have found that bad schools are also more likely to be smaller size as well (i.e. this is not the causal factor).
Base-line data – In an experiment people were asked to rank the likely course a student would do at university. Because they did not know anything about him, they defaulted to the overall popularity of courses. Next they were given a pen portrait of his personality. Drawing on this evidence they re-shifted their answer according to his traits. In theory this all makes sense. However, Kahneman argues that this is a mathematical error as the likely course will still remain most determined by the popularity of the courses overall – i.e. we make a suboptimal more emotional decision based on System 1 and not a logical System 2 decision.
We fall into the stereotype heuristic trap which dominates and clouds our logical thinking (as System 1 wants to make things tidy and organised). When in doubt we should therefore make decisions based on what is called base line data – i.e. probability data (as System 1 is more than likely to lead us astray).
The film, Moneyball demonstrated how professional baseball scouts made poor decisions based on irrelevant information (such as how someone behaved off-field). Billy Beane of the Oakland A’s ignored this and instead relied on pure data of percentage to reach first base. This allowed him to buy good performing players at discounted prices (https://www.youtube.com/ watch?v=PX_c7W5RNJ8).
Regression to the mean – Our performance on anything tends to aggregate around an average – i.e. sometimes we perform better and other times less well, but over a longer period of time we will level out at our average. Thus when things are going much better than expected, we often then see a dip in performance. Likewise when we see a significant underperformance against our norm, we tend to move back closer to the line of average.
This phenomenon of regression to the mean has been observed in organisational performance as well (the under-performing companies do better and the over-performing companies drop). The trouble is we are bad predictors and ignore this principle. So if one golfer shoots 66 and another 77, when asked who will score better the next day, we assume a continuation of form (as we make the misguided judgement that the scores are reflective of talent – but we ignore the regression to norm). Instead the best predictor is reference to their base line – so 66 could well do worse on day 2 and 77 do better.
The Sports Illustrated jinx (that a sports person featured on the cover will under-perform the following year) is not a jinx but the reality of regression to the mean. You need to have massively over-performed to be on the front of Sports Illustrated so….
Our Misguided View On Performance
Luck – Luck plays a BIG part in success. Kahneman’s model for success is ‘Talent + luck’. Great success = ‘A little more talent’ + ‘A lot more luck’.
Confidence bias – the myth of control – We tend to exaggerate our ability to control events around us and hence be overly optimistic about what we can achieve, discounting the other external factors that can slow down a project (especially competitor’s activity). Many organisations fail as they have an over-inflated belief in their ability to control situations (luck again plays a big part in a company’s success or failure – i.e. a company is never in total control of its destiny). Likewise the CEO. Just because on paper a company has a ‘strong’ CEO does not mean s/he will be successful as a large proportion of their fate (i.e.
luck/no luck) is outside their control. Most of the examples of excellent companies studied in books like ‘Built to last’ and ‘In search of excellence’ have receded over time (again down to the principle of regression to the mean). Thus we have consistently over-exaggerated the effect of a CEO’s leadership ability and therefore followed illusory actions to mimic their so called leadership traits and processes – ignoring the real impact of timing and luck.
Optimism – the driving force of capitalism – Optimism is good. In life optimists are more cheerful, more popular, more resilient to dealing with hardships, and less likely to become depressed, and hence tend to be healthier (to the point that they even live longer). Optimists are the achievers, the visionaries, the entrepreneurs and the leaders in life – they make things happen around them. Their self-confidence breeds faith and following in others. When there is an issue, rather than giving up, they believe it can be solved and this perseverance helps them win on through.
Most people are born optimists. Amusingly, most people genuinely believe they are superior to most others on many different traits.
90% of drivers believe they are better than average (as do teachers).
Supporters of basketball teams in the playoffs were asked to rate the probability of their team winning the playoff. As to be expected each team over-rated their chances. When all the scores were added up they came to 240%.
The trouble is optimism is a highly socially valued way of being – it is more acceptable to agree and support than to challenge and pull down an idea – leading to a ‘collective blindness’. It can also lead to bold forecasts and timid decisions.
So how to overcome this disease of overconfidence? Get the team together and start with the premise: “Imagine that we are a year in the future. We implemented the plan as it now exists. The outcome was a disaster. Now write a brief history of how and why this happened.”
The illusion of predictions – Nassim Taleb in Black Swan points out our tendency to build and maintain a narrative that makes sense of the past. This makes it difficult for us to accept our limits of forecasting ability (i.e. because we have ‘made sense’ of the past, we assume we can predict the future). !
System 1 jumps to conclusions with very little evidence – and we hold to these conclusions with a staggering degree of confidence!
When assessing the soldiers in the Israeli army, Kahneman and his staff independently observed the behaviours of the soldiers and then together agreed who were the excellent soldiers. The trouble was their ability to detect who did become the excellent soldiers was little better than random luck yet they still remained locked in the process (as they believed in their personal ability to predict success).
Likewise we see this in the selling and buying of stock. In theory everyone has access to the same information, so what makes one trader think the price is under-inflated and another think it is over-inflated? Why does one investor predict that the price will go up and another that the price will go down? Again this is the illusion of belief as in theory the market works on perfect price (i.e. all there is to know has already been taken account into the prevailing price).
Odean, from UC Berkeley studied the trading records of 10,000 brokerage accounts over a 7 year period (163,000 trades). On average the shares the traders sold did better than those they bought by 3.2% per year. They found the most active traders had the worst results – i.e. doing nothing would have yielded a greater return!
It seems the illusion (especially amongst men) that they know better than the market is wrong (indeed women who are less prone to this effect tend to be on average better investors).
The evidence from more than 50 years of research in this field is that for the large majority of fund managers is no better than rolling dice – i.e. luck. Typically 2/3rds of mutual funds under-perform the overall market in any given year. Thus so called ‘experts’ do not seem to have mastered the ability of future prediction.
Tetlock interviewed 284 people who made a living by being political and economic pundits. He asked them to assess the probabilities on certain future events in their specialised fields. In all he gathered 80,000 predictions. Comparing the results with what actually happened he found that the experts performed worse than chance – i.e. ‘dart throwing monkeys’ would have performed better. The reason is when a person becomes ‘an expert’ (based on past experience) they assume an unrealistic level of confidence in predicting the future. The reality is the world is unpredictable and we are therefore all weak at it – especially over longer time frames and dynamic situations.
When radiographers were given the same X-rays they contradicted themselves on 20% of occasions. Overall 41 studies in different areas have drawn the same conclusions that experts are inconsistent.
The power of loss – We are genetically wired/evolutionarily adapted to be more aware of risk than opportunities. Hence we do not like loss, so we also dislike risk (and are prepared to pay a premium to avoid loss).
When people are asked, how much they would need to win in order to risk losing $100, the average is $200 – i.e. we value a loss at twice that of a gain.
We also know this level of unease over risk increases dramatically the greater the amount at stake (cf the ‘Who wants to be a millionaire’ effect – the greater the prize the less likely you take risks with your answers). Thus the psychological value of a gamble is not focused on the possible positive uplift but more focused on the potential loss – i.e. it’s more about what is at stake to lose than what is to gain. Thus to encourage people to ‘bet’ more you need to focus more on minimising the loss rather than boosting the gain.
The threat need not be real to attract out attention.
Experiments show that negative words are picked up quicker than other more benign words. Negativity is noticed more, influences us more and stays with us longer.
Emotions are processed rapidly and unconsciously – hence why we can have irrational fears as the conscious System 2 brain is never consulted!
Animals and humans fight harder to defend their territory than to gain new territory.
Loss aversion is a powerful holding force that drives stability and certainty in our lives, our relationships, our employment and our societies.
Gotman, a German psychologists estimated that a stable relationship needs good interactions to outnumber bad ones by a ratio of 5 to 1.
We are driven more powerfully to avoid losses than to achieve gains – the aversion of failure is stronger than the desire to excel (hence people often just hit their goals rather than strive to exceed them).
Economic theory would suggest that a taxi driver would work many hours on a rainy day and then take time-off on slower days. Instead on slower days taxi drivers stay out longer to hit their daily target and then on rainy days go home early.
Professional golfers putt more accurately for a par than for a birdie (they analysed 2.5m putts to prove this).
Prospect theory – When all options are bad, then we are more prepared to gamble. For example when offered the following: Lose $900 for sure or take a 90% chance to lose $1000, then people take the gamble. But when offered the choice: Get $900 for sure or 90% chance to get $1000, people opt for the sure reward of $900. The disadvantages looms larger than the advantages. Thus the focusing on the negative prospects influences our decision (i.e. nothing to lose versus everything to lose).
The endowment effect – What we own we value more highly than things we do not own. Endowment effect is not universal and is more extreme on rare items.
Avid fans for a concert who have a ticket for a sell-out concert will only be prepared to sell the ticket for six times the original cost.
In a series of experiments in the buying and selling of unique, specially decorated university coffee mugs, they found the average selling price was twice the average buying price.
Finally, be aware that in negotiations, it’s often what people sense they have to ‘give-up-on’ that causes the sticking points in the discussions.
Probability versus certainty – When we make a complex decisions System 1 unconsciously assigns different weights of importance to the different factors we need to take into account. Changing the weight of importance can therefore affect the final decision.
The problem is we get seduced by different levels of probability. 0% -> 5% (i.e. ‘nothing’ to a possibility of ‘something’) and 95% -> 100% (i.e. ‘uncertainty’ to ‘certainty’) are both more highly valued that say 5%->10% or 85% -> 95%. We therefore overweight our chance of winning when the probabilities are low but underweight our chance of winning when they are high. We are therefore prepared to pay a premium to eliminate risk altogether.
You have inherited $1m but your stepsister is contesting the will in court. Your lawyer believes you have a 95% chance of winning. The day before it goes to court, a risk adjuster contacts you to buy the case from you for $910,000 – most people take it because they fear the loss.
At 95% we fear the loss more than we do if we just have a 5% chance of winning. We will pay this premium to reduce risk.
Hence life insurance plays of these fears of loss and risk, as does plea bargaining. The lottery is a very small chance of winning but you are prepared to gamble the relatively small sum for a big win.
In an experiment they found that mothers were prepared to pay three times more for a product that eliminated risk versus a product that just had a minor level of risk.
We tend to attach values to gains and losses more than to what is at stake. This led Tversky and Kahneman to propose the four fold pattern within their Prospect theory:
Spreading the risk – When Paul Samuelson, a famous economist was asked if he would accept a bet on a toss of a coin in which he could win $200 or lose $100. His response was to decline on the grounds that he would ‘feel’ the loss of $100 more than the benefits of $200. However, in a perfect twist, he said he would take the bet on for 100 goes.
When we spread our exposure to risk over a greater range of activities, we will more likely come back to the expected norms. With one toss we could more easily lose $100 but over 100 we would undoubtedly win a lot of money on these odds. Thus Kahneman recommends we see these ‘little decisions’ not as isolated risk events but instead see them as all part of the 100 throws. One of the reasons for this is we do not compute multiple issues well – instead we tend to focus on just one aspect and allow that to influence our decisions (a form of finding an easier question to answer).
Thus taking risks pays-off in the long run even of it does not pay-out on each event.
Day to day examples are to never pay for extended warranties (accept you will loose on some but gain in the long run by saving the premium) and not to take the add-ons when buying insurance.
When 25 top managers in one company were asked if they would be prepared to support a project that had equal chance of success or failure, with the upside if successful to double the capital. None of the executives were willing to take the risk. Interestingly, the CEO said he would accept the risk from all of them as he saw the broader frame and knew that if everyone took the risk, over the 25 projects they would win out.
Mental accounting – We mentally hold money in different compartments in our brain. and so treat money differently (when in truth a $ is a $). For example we may be happy to run up credit on a card yet not take money out of the school savings account (even though we are paying higher interest on the card than we earn in the savings account).
A: A women has bought two $80 theatre tickets. When she gets there, she discovers she has lost them, Will she buy two more?
B: A women goes to the theatre with $160 in her wallet, intending to buy two tickets. When she gets there she discovers she has lost the cash. Will she still buy the tickets?
Most think A will not but B will as she treats cash differently from the purchased tickets.
Sunk costs – We are tied down by sunk costs rather than seeing that any future decisions are free of those sunk costs (as any decision made now will not bring those sunk costs back).
A company has already invested $50m in a project that is overrunning in time and costs. An additional $60m is needed to secure its completion (even though expected returns are likely to now be lower than originally expected). There is another project that requires the same level of funding that suggests larger returns. Most companies are so blinded by the sunk costs they cannot value the two projects rationally and so will usually invest more money in the failing
project (as they cannot be seen to have ‘failed’).
The sunk costs fallacy keeps people in poor jobs, in bad houses, under performing investments and unhappy marriages.
Fear of financial loss – You would assume we would be more rational about financial decisions but reality suggests otherwise. We hate losing money and will try to mitigate against it – which can mean we make poor decisions. Take for example the holding of two stocks. One is over performing and has exceeded the purchase price, whilst the other you would have to sell for a loss. The logic is to hold onto the over performing share, and ‘cut your losses’ and sell the under performing one. However, we get ‘anchored’ to the original purchase price of the share so we hold onto the poor performing share in the hope of eventually getting our money back (and instead sell the good share). Yet again we view each share independent of each other and thus make poorer decisions than if we viewed them all from a broader frame. We see this also in the field of gambling as well where people try to win back their losses, and merely compound their losses.
Costs are not losses – Framing something as a loss creates greater emotional feelings. So when offered either a gamble that offers a 10% chance to win $95 and a 90% chance to lose $5 versus to buy a lottery ticket for $5 with a 10% chance to win $100 and a 90% chance to win nothing, most people opt for the lottery ticket as our brain reframes the same $5 as a cost not a loss.
Life as a story – It seems that when making judgements on broad issues like life our assessment of the quality of our life is not based on a thorough examination but only a small section of information available. As with the peak and end rule, a life well lived is often judged using the same criteria – we tend to look at the big events and our most immediate history (as those are the key ones we can remember – WYSIATI).
In an experiment describing a person’s life, one where an extra 5 years were lived, but were described as ‘pleasant, but less so than previous years’. They found that respondents judged the person’s life as being less complete when the extra 5 years were added on.
Subjects were invited to fill in a questionnaire about life atisfaction. However before they began they were asked to photocopy a piece of paper. Half of them found a dime beside the copier. This small (irrelevant) bit of positive luck caused a marked improvement of their ratings of their life!
Taking photos is increasing our modality of storing our memories – yet in many ways the stories around events (maybe triggered by the photos) is realistically the way we categorise and makes sense of our lives. The remembered life is rarely the sum of all the points in our life but rather the stories we create to make sense of it all. If we choose to tell a negative (or positive) story (irrespective of the complete facts), then that tends to imprint the overall experience as negative (or positive).
Thus we have two selves – the remembering self (who is making sense of our past through our stories we have created) and our current experiencing self (often just adding further chapters to the stories already created).
Defining happiness – The remembering self is not very accurate with the truth, so maybe it is a misguided notion to even ask people how satisfied they are with their lives. Instead it is better to focus on the here and now – i.e. ‘How happy are you right now?’ Kahneman suggests a simple measure of happiness as ‘Time spent doing things we want to do’ minus ‘Time spent doing things we do not want to do’.
In an experiment to map people’s happiness throughout a day they developed the ‘U’ index (higher the percentage the greater the time spent in an unpleasant state). For American women their U index differed throughout the day: 29% for the morning commute, 27% for work, 24% for looking after the children, 18% for housework, 12% for watching TV and 5% for sex. One of the key predictors of positive feelings during a day is contact with family or friends. On Gallup’s,
‘Ladder of life’, some of the key influencers of a good life include educational attainment, and religion, whilst the key negative was ill health.
We are born with a pre-disposition for well-being. People who appear equally fortunate in life vary greatly in how happy they are. Thus we return to our base line irrespective of our ills or fortunes.
Immediately after the accident Paraplegics feel understandable low, but as life returns to a new equilibrium they focus more on the differences than their new reality and so return to a similar level of happiness they had before the accident. Likewise people who win the lottery, eventually return to close to their old state of happiness.
Setting (and then achieving) of goals appears to also be a key determinant of our happiness.
A study of 12,000 students found that those who had rated highly the statement ‘being very well-off financially’ as essential were shown to be more likely to have achieved financial success in their lives. It also helped with their life satisfaction if they had achieved their desired goals.
That said, above a certain amount of money ($75,000 in 2011) more money does not increase happiness. Whilst money does buy better pleasures, it does not buy greater happiness as people rapidly get used to new higher standards. It also reduces our ability to enjoy the small things in life.
Likewise, a new car (or any big purchases) tends to only have a temporary lift as we quickly subsume it into our everyday life. Thus once we get used to something, it loses its ability to lift our emotions. One of the reasons cited why rich people are no happier is there is little major differential in the extra class of items they can buy.
However, joining a club or learning a new activity (be it learning the cello or playing tennis) where you constantly have new challenges or interactions is different and so is often more uplifting in the longer term as it cannot disappear easily into the backdrop tapestry of life.
A Few Final Thoughts
Just because we are not rational in our decision making, does not mean we are irrational. We are trying our best but sadly easily misguided. Thus we often need help when making bigger decisions (e.g. though use of computers or a number of people to filter the decision making process).
Whilst unscrupulous people can take advantage of people thorough the fault lines of our decision-making, one can also use this to encourage positive actions such as pensions saving, organ donation and healthier eating (cf Nudge).
System 1 is sadly the origin of many of the errors we make. Unfortunately it’s difficult to educate and control. Furthermore just because we become more aware of these limitations does not stop us falling under their misguided spell. It’s easier with System 2 – when you become aware that you are in a cognitive minefield, slow down and ask for help.
This book is dense. It is full of great insights into human frailty of decision-making. What is so good about this book is we have the architect of the theories and the experiments rather than a third party reporter (like Gladwell or Lehrer). But that closeness to the subject comes with a few problems.
Maybe quite harshly, this book feels a bit self-indulgent in that it goes into far too much depth – It gets lost in minutiae and over-labours points repeatedly. I believe he could easily have edited out 100 pages of unnecessary detail of discussion.
I also worry that people will read this book and assume they can manipulate others. Many of the experiments are isolated events which bear little reality to life (for example in the era of ease of access to information any time anywhere, we are less likely to be as negatively influenced). Thus some of the biases may have been over played.