Everything’s is an inference. We do not have access to our thought processes. When people tell us about why they behave a certain way or why they hold some opinion we tend to believe them. The truth is that they probably make all that up on the fly. It is certainly true for myself. When I come up with some thought I don’t have a clear thread that leads me to where that came from. Rather it is all made up as I go. Makes me wonder how much we really work like an LLM.
We make our judgments about the world based on stereotypes and schemas we learned at some point. These stereotypes color our perception.
There’s nothing either good or bad, but thinking makes it so.
Shakespeare
Through framing we can reach pretty much any outcome we like. Even obviously contradicting ones like the following: a monk asks the friar whether it is Ok to smoke while he prays. The friar is outraged. Obviously not. God demands our full attention. The other monk asks the friar whether he may pray while he smokes. Sure thing.
Imagine working yourself up a tricky mountain path. Somewhere up above a stone breaks loose and hurls past you. Now maybe you get some strong feelings toward the other hikers going along with you. You'll feel like the near death experience brings you closer. What's actually happening is that you attribute the racing heart to your fellow travelers rather than to the loose stone.
When we attribute our feelings to another person or event when they really come from other circumstances, psychologists speak of the Fundamental Attribution Error (FAE). We make this mistake all the time. When I meet someone on a sunny day I might forever store them as a warm person whereas when I meet them on a cloudy day they might seem glum to me. If someone hands me a warm cup of coffee I might perceive them as approachable. I’ll have no clue that my perception is influenced by this complete triviality.
What can we learn from that? We might not take ourselves and our opinions too seriously. When we say something today that is completely contradictory to what we said two years ago we might not be too harsh on the person who points out the fact. We might just take it as a fact of life.
The unconscious is a powerful processing tool for tackling hard problems involving lots of variables and hunches we might not be able to clearly articulate. The workings of the unconscious are inaccessible to us. But we can benefit from the work when the subliminal crosses the threshold into the conscious mind. One way to make use of the unconscious is to write down a problem we face in the evening and see the next morning if there’s something the brain came up with while we slept.
Economics discusses the choices we make in everyday life. Cost benefit analysis allows us to make better decisions. It consists of the following steps:
When we consider the earths health (like the atmosphere or ozone layer) I might get a benefit of +1 when I take a flight. If you take a flight you will also get a benefit of +1. But we will both only have a cost of 1/eight billion. The problem is that we don’t see as individuals when the costs outweigh the benefits because we’ll only ever see the +1 benefit for ourselves and neglect the minuscule cost for everyone. Since nobody sees that, we’re screwed at some point. Another analogy is that of the herders who shepherd their sheep on a common pasture. Everyone gets a benefit of +1 when they add one sheep of their own and a minimal cost due to the amount of grass available to the sheep declining. At some point over grazing occurs and everyone loses everything.
A pharmaceutical company saying they need to offset the costs for development are counting on you not understanding the principle of sunk costs. The company will charge as much as the market allows. The development costs play no role in that. They’re sunk.
Owning a car has opportunity costs. There’s the obvious financial cost. But there’s also the cost measured in time and well being when thinking about having to do maintenance, comparing your car to the one the neighbor has and so on. Cal Newport has this framed as the any benefit mindset. Owning a car surely has benefits. But just because something has any benefit does not mean you should get it or do it. Weigh the pros and cons for your situation.
Remember that calculating opportunity costs also has opportunity costs. Satisficing is a good method to avoid over optimizing things.
Losses weigh heavier than gains. This is why we tend to sell stocks that are going up rather than the ones that are down. We can't stand the feeling of locking in on our losses.
Owning a thing increases its perceived value. Take note of that next time you try to sell something online and are outraged when nobody wants to pay 200 € for your thirty year old armchair.
Large N increase our confidence in findings. As long as a sample is truly random we can get away with not making it very large and still get good results. See voting polls.
We vastly overestimate our ability to judge someone after meeting them for thirty minutes. Rather than basing our assessment of a person on a short encounter, we should look at their track record, like grades, endorsements of people who have worked with the person for a long time and such. Past behavior is the best predictor of future behavior. Just make sure you have a large enough N.
Were overly likely to think one exhibition of a trait is indicator for future behavior but underestimate the chances of many exhibitions of a trait being shown again in the future. For abilities we can better realize that a single exhibition is not telling for overall performance.
Reliability: the chances that two doctors say The same thing about a condition. Or that one doctor says the same thing about a condition on multiple occasions
Validity: the degree to which the statement is actually indicative of the real situation.
Don’t listen to the highest paid person in the room but get the data. People make assumptions which tend to be wrong. When looking at what provides the best outcomes for user retention or number of sales on a website don’t trust your assumptions but do A/B tests to figure out what’s true.
There’s a hierarchy of different kinds of experiments based on their ability to tell us something about the world.
First there’s anecdotal evidence. Then there’s MRA (multiple regression analysis). After that are natural experiments, where one looks at a large amount of data as it accumulates in the real world and tries to get some insight from that. And the gold standard is the double blind controlled experiment.
These controlled experiments can’t be done on humans for a lot of questions due to ethical reasons. As a proxy we can use animal studies.
When learning something about the world it’s useful to conduct experiments. Let’s see how we can conduct experiments.
We select a sample of a population. For that we first need to establish what the population is. Selection of the sample then is very important. Make sure the sample is selected at random. If there’s some self selection then the experiments results might be only mildly useful at best and misleading at worst. The sample should be sufficiently large to be able to have some confidence in the results. I’ll need to check the word confidence in that context as there are some words which have special meaning in this sphere. After randomly selecting a sample from a population one can assign conditions to them. In the simplest and presumably best case we have two experimental conditions which we’d like to check. For example some intervention vs no intervention.
Let’s look at confounding variables here. Time is one obvious aspect. There might be some global event or trend going on that affects both or one of our groups. To mitigate this we can maybe repeat the experiment at random intervals, switch up the conditions between the groups and have a within design, where we first assign condition A to the first sample and then condition B to the same sample. Same goes for the other sample.
People are highly malleable when it comes to asking them what they think about a certain topic. Ask people to vote for or against abortion and you can change the outcome of the survey by placing the voting booth inside a church or next to the abortion clinic. Or place it next to a prison to conjure up the image of a rapist. But you don’t have to manipulate the location to get vastly different results. Changing the wording can have just as great of an effect.
People don’t tend to be very consistent over time when asking them the exact same thing even. When looking at survey results ask yourself whether the creator of the survey might have some interest in getting a certain outcome. Then remember that the people designing the survey have great leverage to achieve the desired result.
When asking people about their opinions or thoughts consider that they’ll have some reference group against which they base their answer. Make this reference group explicit or else you won’t be able to compare the results with what you intend to compare them with. Also people won’t implicitly have the same reference group so comparison between people will be useless.
Asks one man: "How’s the wife?" Answers the other: "Compared to what?"
Eastern countries have a tradition of dialectical reasoning rather than formal logic. The following principles characterize eastern thought:
Note how this mode of thinking complements traditionally western modes of deductive logic. Nisbett finds some interesting differences in how people from eastern of western cultures think about certain things. Eastern people pay more attention to context and can more readily analyze a situation from different viewpoints.