Our brains have evolved over hundreds of thousands of years to be the most advanced pattern matching systems in the known universe. This extraordinary feat is what allows us to do many of the complex interactions that make up our daily lives. Furthermore, it allows us to do many of these really difficult things without thinking about them (such as driving a car.)
Lately, I’ve been catching myself going too far. In an attempt to improve efficiency I catch myself making generalisations from a single interaction/event.
Where the problem comes from
When doing any sort of scientific study, especially around psychology, the sample size is an extremely important factor in how valid the findings are.
If I was to ask five of my friends which flavour of ice-cream they prefer, and then after analysing the results find that “humans prefer chocolate ice-cream,” there would be very little scientific weighting to my conclusion.
However, if instead I interviewed hundreds-of-thousands of people across multiple ethnic, income, gender, and age groups, and still found the same results, there would be much more credibility to my statement that chocolate ice-cream is superior.
This probably isn’t anything new to anyone, yet this isn’t how we as humans come to make opinions about the world.
Judgements from small sample sizes
In Thinking Fast and Slow, Daniel Kahneman splits the human consciousness into two parts; System 1 and System 2. These two systems are not anatomically distinct parts of the brain but rather a useful (knowingly simplified) model to understand how our mind makes decisions.
- System 1 – is mostly what we consider our rational, conscious mind. It makes slow decisions and is very inefficient. In many ways, science is the embodiment of System 1 thinking, as it meticulously searches for answers with a focus on accuracy and logical reasoning.
- System 2 – is generally the unconscious, pattern-matching mind which makes snap judgements about the world through heuristics that it has been acquired and passed down over generations.
So when we make a snap judgement about something from a small sample size (nearly every judgement we ever make) it is largely coming from System 2. We then (through System 1) often create a narrative or “rational explanation” about why we chose that answer.
Patterns of one
This all leads us to the major problem. Sometimes our sample sizes get so small that we are making snap judgements from a single experience.
Having been travelling a lot lately I often find myself falling into this pattern at lot. I’ll interact with a single person or city, and then make a broader snap judgement about an entire group.
- That city is dangerous (because I got mugged there)
- Italy is a tourist trap (because I visited Venice)
- Germans are blunt (…no this is pretty accurate actually 🤷♂️)
When we do this, we using the single interaction that has happened and often inflating the sample size with things we’ve heard, read, remembered from the past.
The first problem with this is that our memories are notoriously inaccurate, being extremely influenced by our current opinions and state of mind.
Secondly, the sources that we’re using are dubious at best. We rarely scrutinize where we got our information from; “was it a tweet or a scientific peer-reviewed journal I heard it in?”
So we’re making judgements on shitty information that we inaccurately remembered 🤦♀️
But what about intuition?
In Blink by Malcolm Gladwell, he argues that often the snap judgements from our System 2 are actually more accurate than our slow and deliberate System 1 judgements (which often get misled by too much information). However, this is only true when you are have had enough exposure to the field so that your snap judgements are being influenced by a vast amount of underlying data.
In other words, snap judgement are sometimes more accurate when you are an expert, but we’re not experts in 99% of the things we do.
The obvious lesson
This all just leads to the obvious ending that you’ve probably heard thousands of times in your life:
Don’t make System 2 snap judgements on things that you’re not a domain expert in, especially when they are based on a sample size that you can count on a single hand… or finger 😬
Psst. Some books that I recommend if you’re interested in this stuff is (ordered by quality):