← Concept Vault
Concept · Cognitive Science · Maps to the Law of Cause and Effect

System 2 Thinking

The slow, effortful, deliberate mind — and the uncomfortable truth that it's lazy, expensive, and almost never the one actually driving.
Read · 23 minPillar · Cognitive Science

Most people who've absorbed a little pop psychology think they know this one. "System 1 is fast and emotional, System 2 is slow and rational, and the goal is to use System 2 more." That sentence is half-right and dangerously incomplete. The real story is stranger and more useful: System 2 is not the wise CEO of the mind. It's an expensive, easily-tired consultant that mostly rubber-stamps whatever System 1 already decided, and only occasionally does the hard work it's famous for. Worse — the single most famous mechanism that was supposed to explain why System 2 runs out of gas (ego depletion) largely collapsed under replication. So let's pull this all the way down: where the idea came from, what the evidence actually says, how the machinery works, and exactly where the popular version lies to you.

01

What it is — and where the idea came from

The terms "System 1" and "System 2" were not coined by Daniel Kahneman, even though his 2011 book Thinking, Fast and Slow made them household words. The labels come from the psychologist Keith Stanovich and his collaborator Richard West, who popularized "System 1" and "System 2" as terms around 1999–2000 (notably in their 2000 Behavioral and Brain Sciences target article) as deliberately neutral placeholders. They chose bland, numbered names precisely to avoid smuggling in claims — they didn't want to call it "the emotional brain" or "the rational brain" and prejudge what these systems were. Kahneman borrowed the terminology and popularized it, but he was explicit that the names were Stanovich and West's.

Behind the labels sits a much older and broader research program usually called dual-process theory — the idea that human cognition runs on two broadly different modes. This wasn't one person's invention. It emerged across several subfields in the 1970s–1990s, with major contributors including Seymour Epstein (cognitive-experiential self-theory), Steven Sloman (associative vs. rule-based systems, 1996), Jonathan Evans (reasoning research, going back to the 1970s and still actively refining the theory today), and others. Kahneman's own foundational work — the heuristics and biases program he built with Amos Tversky starting in the early 1970s — predates the System 1 / System 2 framing entirely. Tversky and Kahneman were documenting the errors (availability, representativeness, anchoring) long before there was a tidy two-system architecture to hang them on.

So when we say "System 2," here's the careful definition Kahneman used, and it's a cluster of properties, not a brain region:

System 2 is the mode of mental activity that is slow, serial, effortful, deliberate, and rule-governed. It is what you experience as agency — the felt sense of choosing, reasoning, computing, checking. It's invoked when you multiply 17 × 24, when you fill out a tax form, when you compare two phone plans, when you override your first impulse, when you deliberately hold your attention on something boring. Its operations require attention and are disrupted when attention is drawn away. It's tightly bound to working memory — the small, temporary mental workspace where you hold and manipulate information — and to the prefrontal cortex, the brain's executive-control machinery.

System 1, by contrast, is fast, automatic, parallel, associative, and effortless. It's pattern recognition, intuition, the snap judgment, the emotional reaction, the answer that arrives rather than the one you work out. You don't choose to read a word you see; you don't choose to detect hostility in a voice; you don't choose the answer "12" when you see 2 × 6. System 1 is always on. System 2 is usually idling.

The crucial point that gets lost in the pop version: these are not two homunculi, not two brain regions, not two literal things. Kahneman was emphatic — and slightly apologetic — that "System 1" and "System 2" are useful fictions, characters in a story he tells because the human mind understands agents better than it understands statistical tendencies. There is no anatomical "System 2" you could point to on a scan. There's a family of slow, effortful, attention-demanding processes that behave as if they were one agent. Hold that. The map is not the territory here, and the originators said so out loud.

02

The real science — the evidence base

What's the actual empirical content? Strip away the storytelling and you find a set of robust, repeatedly-demonstrated findings about effortful thinking. These are the load-bearing pillars.

1. Effort is measurable, and it's physiological. This is one of the oldest and most solid results, and it predates the whole System 1/2 framing. In the 1960s, Kahneman and Jackson Beatty showed that pupil diameter dilates in proportion to mental effort. Give someone progressively harder tasks — hold three digits, then four, then five, in memory — and the pupils widen in lockstep, peaking at the moment of maximum load and contracting the instant the task is done. The pupil is, almost literally, a meter for System 2 engagement. This wasn't a metaphor for "trying hard"; it was a continuous, physiological readout. Effort also shows up as elevated heart rate and increased skin conductance. Established. This is some of the firmest ground in the whole field.

2. Effortful thinking is capacity-limited and serial. You can walk and chew gum, but you cannot multiply 17 × 24 while turning left through heavy traffic — and you'll notice you instinctively slow the car or stop the conversation when the math gets hard. The classic demonstration: Kahneman's "Add-1" and "Add-3" tasks, where you hear a string of digits and must report each one incremented. As difficulty rises, performance and the ability to do anything else collapse together. Effortful operations compete for one shared, scarce resource. Established.

3. System 2 is a "lazy controller." This is the most counterintuitive and best-documented finding, and it deserves its own emphasis below. The core experimental tool is the Cognitive Reflection Test (CRT), developed by Shane Frederick (around 2005). Its most famous item:

A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?

The intuitive, System-1 answer — 10 cents — leaps to mind for almost everyone, and it's wrong. (If the ball were 10¢ and the bat is $1 more, the bat is $1.10 and the total is $1.20.) The right answer is 5 cents. What makes the CRT diabolical is that the wrong answer is fluent and confident. Getting it right doesn't require advanced math — it requires System 2 to wake up, notice that System 1's answer needs checking, and do the small computation. Most people, including a startling share of students at elite universities, don't bother. The answer feels right, so they ship it. That's the lazy controller in one sentence: System 2 endorses System 1's output without checking, unless something forces it to engage. Replicated extensively and held up well — though see the debates section for important nuances about why people fail it.

4. Cognitive load degrades self-control and judgment. The landmark demonstration is Baba Shiv and Alexander Fedorikhin's (1999) "cake vs. fruit salad" study. Participants were asked to hold a number in memory while walking to another room, then offered a snack. Those holding a seven-digit number (high load) chose the chocolate cake far more often than those holding a two-digit number (low load), who tended to pick the fruit salad. The interpretation: when System 2 is occupied holding seven digits, it can't override System 1's impulse toward the indulgent choice. Working memory load and self-control draw on overlapping resources. This is a clean, much-cited result, and the direction of the effect — load impairs deliberate override — is well supported across many paradigms.

5. Engaging System 2 reduces certain biases — sometimes. People who score higher on the CRT, or who are prompted to "think carefully," show smaller framing effects, less reliance on stereotypes in syllogistic reasoning, and better calibration on some classic bias tasks. The relationship is real but modest and inconsistent — which is itself one of the most important honest findings, and we'll return to it. Slowing down is not a reliable bias-eraser.

That's the genuine evidence base: effort is physiologically real and metered, it's scarce and serial, the controller is constitutionally lazy, load impairs override, and deliberation helps imperfectly. Notice what's not on that list yet: a validated metabolic fuel that runs out. That's where the honest cut comes.

03

The mechanism — how it actually works

Strip the agent-story away and ask: what is physically happening when "System 2 engages"?

The best current account ties effortful thinking to the central executive of working memory and the lateral prefrontal cortex (PFC), working in concert with the anterior cingulate cortex (ACC) and dorsal attention networks. Here's the chain, marked for confidence.

Working memory is the bottleneck. (Established.) Working memory — most associated with the work of Alan Baddeley and Graham Hitch (1974) — is the small workspace where information is held online and manipulated. Its capacity is famously tiny. George Miller's "magical number seven" (1956) overstated it; later work (Nelson Cowan) puts the real limit closer to about four chunks. Every effortful operation — comparing options, following a rule, suppressing an impulse, doing arithmetic — consumes slots in this tiny workspace. This is why System 2 is serial and load-sensitive: there's barely any room in there. The cake-vs-fruit-salad effect is, mechanistically, "the seven digits ate all the working-memory slots, leaving none to run the override."

The prefrontal cortex implements control. (Established as a region; specifics still researched.) The lateral PFC maintains goals and rules against interference — it holds "the answer might be wrong, check it" active long enough to actually act on it. The ACC is heavily implicated in conflict monitoring: it detects when an automatic response conflicts with the task goal (as in the Stroop task — naming the ink color of the word "RED" printed in blue) and signals that more control is needed. So the loop looks like: ACC flags a conflict → PFC ramps up control → working memory holds the rule → the automatic response gets overridden. When any link is weak — distraction, fatigue, low motivation, alcohol, sleep deprivation — System 1 wins by default.

Thinking costs glucose — but carefully. (Mixed / partially supported.) The brain is metabolically expensive overall — roughly 20% of resting energy use for ~2% of body mass — but here's a crucial correction to a popular claim: the additional energy cost of hard thinking over easy thinking is small. The brain's baseline burn is enormous and fairly constant; a difficult cognitive task adds only a modest increment on top. The seductive story that "you literally run out of brain fuel after a hard meeting" overstates the metabolic differential dramatically. There is something real about prolonged mental work producing fatigue — recent work (e.g., Antonius Wiehler and colleagues, 2022) has proposed that intense, sustained cognitive control may lead to a buildup of glutamate in the lateral PFC, which could be a chemical correlate of mental fatigue and the felt need to disengage. But mark this clearly: it is a hypothesis, not settled fact, based on limited samples, and it is not the same as the simple "glucose runs out" story. The honest position is: mental effort produces real, measurable fatigue; the mechanism of that fatigue is genuinely unresolved.

Why "lazy" is the right word. (Established framing.) Effortful control is metabolically and attentionally costly relative to coasting on automaticity, and evolution built a mind that conserves effort by default. System 2 doesn't engage because something is important in the abstract — it engages when there's a triggering signal: detected conflict, surprise, explicit instruction, high stakes that are emotionally registered, or a deliberate act of will. Absent a trigger, the cheap answer ships. This isn't a character flaw; it's an energy-conservation design that's correct most of the time and catastrophic in exactly the situations (statistical reasoning, novel problems, override of a strong intuition) where the cheap answer is wrong.

04

Sharp examples

Abstractions slide off the mind. Concrete cases stick. Here are the ones that show System 2 — and its laziness — most cleanly.

The bat and the ball. Already given, but sit with why it's perfect. There's no trick, no hidden word, no obscure knowledge. The arithmetic is trivial. The only thing standing between you and the right answer is whether System 2 bothers to check System 1's confident "10 cents." Most people don't. Early studies (Alter and colleagues, 2007) reported that printing the questions in a hard-to-read font improved accuracy, but a large 2015 replication across thousands of participants (Meyer, Frederick and colleagues) found no reliable effect, so treat "disfluency wakes System 2" as unsupported. The error isn't ignorance. It's non-engagement.

The Stroop task. Read the ink color, not the word: a list of color-words printed in mismatched ink. You'll feel the friction directly — your reading System 1 fires "blue" automatically while System 2 strains to say "red." That felt strain is the override happening in real time. The Stroop effect is one of the most replicated phenomena in all of psychology, and it's the cleanest demonstration that automaticity must be actively suppressed by effortful control.

The Müller-Lyer illusion. Two lines of equal length, one with arrowheads pointing in, one out; one looks longer. Now measure them. You know they're equal — and they still look unequal. This is Kahneman's favorite illustration of the systems' independence: System 2's knowledge does not overwrite System 1's perception. You can know the truth and still see the illusion. The lesson generalizes brutally: knowing a bias exists doesn't make you immune to feeling its pull. You have to learn to distrust the cue, not to make the cue go away.

The expert who "just knows." A chess grandmaster glancing at a board and instantly seeing the strong move; a firefighter sensing a floor is about to collapse seconds before it does (Gary Klein's research). This looks like System 1 — fast, intuitive, effortless. And it is. But it's System 1 that was built by years of System 2. The intuition is reliable here because the domain has stable patterns and the expert got thousands of cycles of high-quality feedback. This is the bridge between Kahneman and Anders Ericsson's deliberate practice: effortful, attentive System-2 repetition slowly compiles into fast, accurate System-1 intuition. Skill is the migration of competence from System 2 to System 1.

The exhausted judge, the tired shopper, the late-night argument. Decision quality degrades when System 2 is taxed or depleted — you make the impulsive purchase at the end of a long shopping trip, you snap at your partner at 11 PM, you take the default option on the form because evaluating alternatives is too much. (One famous claim — that parole judges grant fewer favorable rulings as they get hungry before a break — has been contested, with critics arguing case-ordering confounds explain much of it. Mark it disputed.) The general phenomenon — depleted or loaded controllers default to System 1 — is well supported even where specific headline studies are shaky.

05

Where it gets distorted — the honest cut

Now the part that separates a real student from someone who skimmed a TED talk. The popular understanding of System 2 contains several distortions, and one of them is a famous theory that largely failed replication. If you take nothing else from this, take this section.

Distortion 1: "System 2 is the smart, rational you, and the goal is to use it more." This inverts Kahneman's actual message. His point was not "trust System 2" — it was that System 2 is lazy, overconfident, and frequently absent, and that it constructs post-hoc rationalizations for decisions System 1 already made while believing it reasoned its way there. The self that "decides" is largely a narrator explaining choices it didn't author. The goal isn't to live in System 2 (you can't — it's too expensive and too slow). The goal is to recognize the specific situations where System 1 reliably errs and to deliberately invoke System 2 there. Pervasive distrust of your own intuition is as useless as blind trust.

Distortion 2 — THE BIG ONE: ego depletion as System 2's fuel gauge. Here is the claim that self-help swallowed whole and that the science has since largely thrown up.

In the late 1990s, Roy Baumeister and colleagues proposed ego depletion: the idea that self-control draws on a single, limited, depletable resource — like a muscle that fatigues. The signature study (Baumeister, Bratslavsky, Muraven, Tice, 1998): participants who resisted a plate of cookies and ate radishes instead subsequently gave up sooner on an unsolvable puzzle than those who'd been allowed the cookies. The story: exerting self-control in task one depleted the resource, leaving less for task two. This became one of the most influential ideas in social psychology, spawning hundreds of studies and the strong claim — promoted in popular books — that the resource was literally glucose, and that a sugary drink could "refill" your willpower.

It largely did not replicate. In 2016, a Registered Replication Report (Hagger, Chatzisarantis et al., published in Perspectives on Psychological Science) — a large, pre-registered, multi-lab effort with 2,141 participants across 23 labs, designed precisely to test the effect with high power and no room for p-hacking — found an effect size near zero (about d = 0.04). It's worth being precise about what was tested: the RRR did not re-run the original cookies-and-radishes study but a standardized, pre-agreed depletion protocol (a letter-crossing "e-task"), chosen by consensus including depletion researchers, precisely so no one could later claim the wrong paradigm was used. The carefully-coordinated, adversarial-quality replication simply did not reproduce the depletion effect. Subsequent meta-analyses, once corrected for publication bias (the strong tendency for journals to print positive results and bury null ones), suggested the true effect, if any, is far smaller than the original literature claimed — possibly negligible. The "glucose refills willpower" claim fared even worse and is now widely regarded as not credible, partly because, as noted above, the metabolic differential of self-control is too small to be plausibly rescued by a sip of lemonade.

Be precise about what this does and doesn't mean, because this is where intellectual honesty matters most:

  • It does not mean self-control isn't real, or that mental fatigue is fake. People obviously do get tired, lose patience, and default to impulse after a hard day. That phenomenology is real.
  • It does mean the specific theory — a single, glucose-based, muscle-like depletable resource — failed its strongest test. The "willpower is a battery you drain and recharge with sugar" model is not supported.
  • The honest current view leans toward motivation, attention, and shifting priorities rather than fuel depletion. (Work by Carol Dweck and colleagues showed that whether people believe willpower is limited predicts whether they show depletion effects — implicating belief and motivation, not a fixed physiological tank.) The emerging picture is that "running out of willpower" may be the mind reallocating effort away from a task it has decided isn't worth the cost anymore — a strategic shift, not an empty tank.

Why does this matter for a student of the mind? Because ego depletion was the textbook mechanism for System 2's exhaustion, and it became a load-bearing wall in pop psychology — and it crumbled. If you built productivity advice, dieting strategies, or self-narratives on "I'm out of willpower, I need sugar," you built on sand. This is also the single best worked example of the replication crisis the field went through in the 2010s, and Kahneman himself was admirably candid: he publicly acknowledged that a chapter of Thinking, Fast and Slow relying on certain priming studies should not have been written with the confidence it was, given how poorly those findings later replicated. The originator told you the building had structural problems. Listen to him over the people selling certainty.

Distortion 3: "Slow down and you'll be rational." As noted, deliberation's debiasing power is real but modest and conditional. Smart, reflective people fall for biases too — and in some cases their cleverness lets them construct better rationalizations for the conclusion they already wanted (motivated reasoning; see Dan Kahan's work on how numeracy can worsen polarization on politically charged data). System 2 is not a neutral truth-engine. It is frequently a press secretary for System 1, and a more articulate press secretary is not the same as an honest one.

Distortion 4: the literal two-boxes picture. Repeating it because it's the most common misread: there are no two systems sitting in your skull. Many serious dual-process researchers (Jonathan Evans, and especially critics like Arie Kruglanski and Gerd Gigerenzer and proponents of "single-system" or continuous-graded accounts) argue the clean binary is at best a teaching simplification and at worst misleading. Kahneman knew this. The dichotomy is a lens, not an anatomy.

06

The debates and the frontier

This is a live field, not a closed book. Where serious researchers still disagree:

Is the binary even right? (Actively debated.) The sharpest ongoing fight is whether "two systems" is the correct architecture at all. Critics argue cognition is better described as a continuum of more-to-less automatic processing, or as multiple distinct processes that don't cleanly bundle into two. Jonathan Evans and Keith Stanovich have themselves revised the framing repeatedly — Stanovich now distinguishes the algorithmic mind (raw processing power, roughly what IQ tests measure) from the reflective mind (the disposition to actually engage it), arguing that "System 2" conflates capacity with the tendency to use that capacity. This explains a real puzzle: why high-IQ people still fail the CRT. They have the horsepower; they just don't deploy it. That's a reflective-mind failure, not an algorithmic one.

Default-interventionist vs. parallel-competitive. (Open.) Does System 1 always generate the first answer and System 2 only sometimes intervenes (the "default-interventionist" model)? Or do both run in parallel and compete? And — a striking recent line of work from Wim De Neys — do people actually have "intuitive" detection of their own errors? De Neys's research suggests that even people who give the wrong CRT answer often show signs (slower responses, lower confidence, autonomic arousal) that they sensed something was off. This challenges the simple "lazy and oblivious" picture: System 1 may be smarter and more self-monitoring than the original story allowed, sometimes flagging conflict without System 2 fully engaging. The frontier question is how much "logical intuition" System 1 actually has.

What is mental fatigue, really? (Open, hypothesis-stage.) With ego depletion deflated, the field genuinely doesn't have a settled replacement. The glutamate-buildup hypothesis (Wiehler et al.) is intriguing but unconfirmed. The "opportunity cost / motivation reallocation" model (associated with work by Robert Kurzban and others) reframes fatigue as a signal that effort would be better spent elsewhere, not as an empty reservoir. These aren't yet reconciled. If someone tells you confidently why your brain "gets tired," they're ahead of the evidence.

How trainable is the reflective mind? (Open, important.) Can you durably increase the disposition to engage System 2 — to become someone who reliably catches the bat-and-ball error class in real life, not just on a quiz? "Brain training" games largely failed to transfer. But there's more hope for decontextualized cognitive skills and explicit debiasing instruction that teaches you to recognize trigger situations. The jury is still out on how far metacognitive training generalizes.

07

How to apply it — the evidence-based version

Given all the above — including the parts that failed — what should a serious person actually do? The applications below are deliberately conservative: each one survives the honest cut.

1. Build trigger-recognition, not blanket distrust. You can't run System 2 all day, and you don't need to. The leverage is in learning the specific situations where System 1 reliably misfires and pre-committing to engage there: anything involving probabilities, base rates, or statistics; high-stakes, infrequent decisions (hiring, large purchases, medical choices); situations where you feel strong fluency or certainty about a numeric or logical answer (that fluency is exactly when the bat-and-ball trap springs); and any decision made while tired, rushed, emotional, or distracted. The skill is the alarm, not perpetual vigilance.

2. Engineer the environment so the right default wins — don't rely on willpower. This is the practical upshot of ego depletion's collapse: stop budgeting "willpower" like a fuel tank, and start removing the need for it. If System 2 is lazy and unreliable, the answer is to make the good choice the automatic one. Don't keep the cake in the house (no override required). Use checklists (Atul Gawande's work) to force System 2 engagement at critical moments without depending on remembering to "think hard." Pre-decide — set the rule when you're calm so the tired version of you just follows it. Defaults beat intentions, because defaults don't require an effortful override that may never fire.

3. Externalize working memory. System 2's bottleneck is the ~4-slot workspace. So stop holding things in it. Write the problem down. Lay options side by side on paper. Offload the digits so the override has slots to run in. Nearly every "thinking tool" — a decision matrix, a pros/cons list, a spreadsheet, even talking it through aloud — works by moving load out of the cramped internal workspace and onto an external one. This is one of the most reliable cognitive upgrades available, and it's free.

4. Reduce load before consequential thinking, and protect attention. Since load and distraction reliably push you into System 1, the move is to deliberately lower load before decisions that need real deliberation: single-task, kill notifications, do the hard cognitive work when rested rather than at the end of a depleting day. Note this is about attention and motivation management, not "refueling glucose." The mechanism may be debated; the practical direction (don't make important decisions while cognitively swamped) is robust.

5. Convert effortful into automatic through deliberate practice. The endgame isn't to think slowly forever — it's to build better System 1. Reliable expert intuition comes from effortful, attentive, feedback-rich repetition in domains with stable patterns (Ericsson; Kahneman & Klein's joint paper on when intuition can be trusted). Practice the hard thing slowly and with feedback until it becomes fast and accurate. You're not avoiding System 2; you're using it to compile skill down into System 1 so it stops needing System 2 at all.

6. Hold the meta-lesson: distrust your certainty most exactly where it's strongest. The deepest practical takeaway from the whole literature — and from the Müller-Lyer illusion specifically — is that the feeling of being right is generated by System 1 and is not evidence of being right. Fluency, confidence, and "it's obvious" are System 1 signatures, not truth signatures. The discipline of a strong mind is to treat a too-easy, too-confident answer as a flag to check, not a green light.

08

How It Connects

The Law of Cause and Effect — This is the deepest connection, and it cuts in two directions. First, internally: System 2 is the cognitive machinery of cause-and-effect reasoning. Following a causal chain — "if I do X, then Y, which produces Z" — is precisely the slow, serial, rule-governed work that overloads working memory and demands the PFC. System 1 doesn't do causal chains; it does association and correlation ("this feels like that"), which is exactly why intuition is so easily fooled by spurious patterns. To actually trace causes — instead of pattern-matching to a vibe — you must mobilize System 2. Second, and more honestly: the Law of Cause and Effect says nothing happens by chance; every effect has a cause. The replication crisis around ego depletion is that law enforcing itself on the field. Researchers wanted the willpower-battery story to be true and a low-powered, publication-biased literature appeared to confirm it — but the real cause-and-effect structure (no robust depletion effect) was always there underneath the noise, waiting. The high-powered Registered Replication Report just removed the noise and let the actual causal reality show. The law doesn't care what we hoped the effect was. Belief is System 1; tracing the true cause is System 2's hardest and most important job.

Related concepts: System 1 / dual-process theory (the other half of the frame) · the heuristics and biases program (availability, anchoring, representativeness — the catalog of System 1's reliable errors) · working memory and the central executive (the literal bottleneck) · the Cognitive Reflection Test (the lazy controller's measuring stick) · deliberate practice (the migration of competence from System 2 to System 1) · the replication crisis (the scientific immune response that took down ego depletion) · motivated reasoning (System 2 as press secretary, not judge).

Brain structures: lateral prefrontal cortex (executive control, goal maintenance) · anterior cingulate cortex (conflict monitoring — the "something's off, check it" signal) · dorsolateral PFC and the central-executive / frontoparietal control network (working memory) · the autonomic markers of effort (pupil dilation, heart rate, skin conductance).

Thinkers: Keith Stanovich & Richard West (coined System 1/2; the algorithmic-vs-reflective mind) · Daniel Kahneman & Amos Tversky (heuristics and biases; popularized the framing) · Jonathan Evans & Steven Sloman & Seymour Epstein (dual-process foundations) · Shane Frederick (the CRT) · Alan Baddeley & Graham Hitch (working memory) · Roy Baumeister (ego depletion — the famous claim that fell) · Wim De Neys (logical intuition / error-detection in System 1) · Anders Ericsson & Gary Klein (expert intuition and when to trust it).

The Compression

System 2 is not the rational pilot of your mind — it's a lazy, expensive consultant who mostly signs off on whatever System 1 already decided and then takes credit for the reasoning. Your job isn't to use it more; it's to know the exact moments it's the only thing that can save you, and to force it awake there — because it won't wake up on its own.

Your Practice
My Depth
My Notes
← Previous
System 1 Thinking
Next →
Confirmation Bias