Most thinking is quick, automatic, and 'good enough'. A slower, effortful mode can step in and correct it, but usually it just signs off on the gut answer. Knowing which mode you are designing for changes how you do it.
Let's start with the distinction everyone agrees on: a great deal of behaviour is handled by processing that is fast, automatic, and effortless. You read a word, recognise a face, recognise the texture of your sandwich, or sense that something is off like your chair is wet, an object is broken, the light is dimmer than usual, all without trying. A smaller share is handled by processing that is slow, deliberate, and effortful: holding two figures in mind while you compare them, or working through a logical step. Stanovich and West (2000) gave these the now-famous names System 1 (the fast one) and System 2 (the slow one), and Kahneman's Thinking, Fast and Slow (2011) put them in everyone's vocabulary. Evans (2008) reviewed the wider evidence and found the fast/slow split holds up across many separate research traditions.
The relationship between the two is the part worth getting right. The fast mode is the default: it is always running and it produces an answer to almost everything. The slow mode sits behind it as a monitor that can intervene, but intervening is effortful, so most of the time it simply accepts whatever the fast mode served up (Evans & Stanovich, 2013). And the fast mode has a particular trick. Faced with a hard question, it quietly swaps in an easier one and answers that instead, a move called attribute substitution, then reports the result with full confidence. The cleanest demonstration is the Cognitive Reflection Test (Frederick, 2005). Asked "a bat and a ball cost $1.10 in total, and the bat costs $1 more than the ball; how much is the ball?", the figure "10 cents" leaps to mind. It is wrong (the answer is 5 cents), and yet large fractions of students at MIT, Princeton and Harvard gave it. The wrong answer is not a failure of intelligence so much as the fast mode getting there first and the slow mode not bothering to check.
You can even feel the fast mode refuse to switch off. In the Stroop task (1935), you are shown colour words printed in the wrong colour ("RED" in blue ink) and asked to name the ink colour. It is strikingly hard and slow, because reading is so automatic that the word fights the answer you are trying to give. That interference is the fast mode running whether you want it to or not.
Two corrections matter, and the first is bigger than most popular accounts admit. It is not at all clear that there are two literal systems. Critics point out that the qualities we casually bundle together (fast, automatic, unconscious, intuitive on one side; slow, controlled, conscious, rational on the other) do not reliably travel together, so "two systems" may be a tidy story rather than a real piece of mental architecture (Melnikoff & Bargh, 2018; Evans himself raised the worry in 2008). The defenders have retreated to a narrower, sturdier claim: there are two kinds of processing, autonomous (Type 1) versus the kind that loads working memory and supports "what if" thinking (Type 2), without the rest of the baggage (Evans & Stanovich, 2013). Treat "System 1 and System 2" as a useful map, not two little people in the skull. The second correction: the popular shorthand that the fast mode is irrational and the slow mode is rational is wrong. The fast mode is usually accurate and well-adapted, and the slow mode rationalises at least as often as it reasons.
Much of the evidence comes from reasoning puzzles and tightly controlled lab tasks, so the two-mode picture is a working model, not a brain scan. And the Cognitive Reflection Test, useful as it is, also picks up general intelligence and a personal disposition to reflect, so it is not a clean meter for "System 2."
The live questions are architectural: how the effortful mode actually gets recruited, whether the switch is a discrete event or a matter of degree, and what makes a person check the fast answer on one occasion and wave it through on the next.
Almost every choice a customer, voter, or citizen makes is made on the fast pass. The slow, deliberating mind you tend to write for is mostly not in the room, and when it is, it often just defends the gut. Design accordingly.
Write for the skimmer whose gut decides, not the careful comparison shopper who rarely shows up. Things that are easy to read, familiar, and quick to process feel truer and better (a quirk called fluency), so clarity and recognisability do real persuasive work. When you genuinely want people to slow down and reconsider, for example before an impulsive cancellation, a little friction or a simple "are you sure?" recruits the slow mode on purpose. The rule is to match the mode to the moment: smooth the path when you want the easy yes, add a speed bump when you want a considered one.
Voters process politics fast: gut feeling, identity, and familiarity carry the message, and the deliberate mind, when it engages, mostly builds a case for the conclusion the gut already reached. A message has to win the fast pass first. An elegant argument aimed at the reasoning mind tends to be received by a lawyer, not a judge.
Match the design to the kind of decision you want. For choices that should be considered, like pensions, consent, or medical options, actively recruit the slow mode: slow people down, force a step, ask them to state their reasoning. For behaviour you simply want to make easy and reliable, make the desired action the fast default and strip out friction. The mistake is using a slow-mode tool (dense information) for a fast-mode decision, or the reverse.
Ask one question before you design anything: is this a fast-mode or a slow-mode decision for the person on the other end? Default to designing for the fast mode, because that is the one almost always answering. Recruit the slow mode deliberately, with friction or a prompt to reflect, only when the stakes are worth the effort. And on a tricky number of your own, do not trust the first answer that arrives. That is exactly the one the bat-and-ball is built to catch.
.png)