Most decisions about a page are made by the fast mind.

Type 1 processing is fast, intuitive and autonomous thinking that runs with little effort and places little load on working memory. Daniel Kahneman popularised it as System 1. A snippet is judged by it in a fraction of a second.

Type 2 processing is slow, deliberate and reflective thinking that requires effort and draws on working memory. Kahneman called it System 2. Reading a page closely and comparing offers engages it.

Bounded rationality, a concept by Herbert Simon, describes decision-making limited by time, information and cognitive capacity. People satisfice and look for good enough options. Dual-process theory explains many of the shortcuts this produces.

Working memory is the limited resource that deliberate thinking depends on. Type 1 processes run largely without it. When working memory is loaded, the fast answer wins.

Predictive processing describes the brain as constantly predicting input and updating on errors, a view developed by Andy Clark, Karl Friston and others. Much of it resembles fast, automatic processing. It offers one framework that could unite both types.

Symbolic classical cognition is rule-based manipulation of explicit representations, the kind of thinking classical cognitive science modelled. It corresponds to the deliberate side of dual-process theory. Logic and calculation belong here.

Heuristic processes select relevant information quickly and without deliberation, as Jonathan Evans described in his heuristic-analytic theory. They deliver the first, default response. They decide which parts of a page are considered at all.

Analytic processes evaluate the information that heuristic processes have selected, with deliberate effort. They can confirm or correct the first response. They only work on what the heuristic stage let through.

The elaboration likelihood model, developed by Richard Petty and John Cacioppo, describes two routes of persuasion that depend on a person's motivation and ability to think. It is a dual-process model of attitude change. It explains why the same page persuades some readers by its arguments and others by its surface.

The central route to persuasion involves careful scrutiny of arguments. It requires motivation and ability. Attitudes formed this way last longer and resist counter-arguments.

The peripheral route relies on cues such as attractiveness, apparent expertise or the number of arguments, without evaluating their strength. It operates when motivation or ability is low. Most people skimming a result page are on this route.

The reflective-impulsive model by Fritz Strack and Roland Deutsch describes behaviour as the product of two parallel systems. One works through knowledge and goals, the other through associations. Both are active at once.

The reflective system guides behaviour through explicit knowledge, reasoning and goals. It is slow and needs capacity. It can decide against an impulse, if it has the resources.

The impulsive system triggers behaviour through learned associations and motivational orientations. It works continuously and fast. A familiar brand name in the results activates it immediately.

The default-interventionist model holds that fast processes produce a default response that slow processes may monitor and override. Override needs effort and working memory. Without intervention, the default stands.

Cognitive decoupling, a concept emphasised by Keith Stanovich, is the ability to keep a representation apart from reality in order to think hypothetically. It is a defining feature of Type 2 thinking. It lets a reader imagine what an offer would mean before accepting it.

Fluid intelligence, the capacity to reason with new problems, correlates with Type 2 processing and working memory. It predicts how well people can override intuitive errors. Capacity is necessary and it is not the whole story.

Prospect theory, published by Daniel Kahneman and Amos Tversky in 1979, describes how people evaluate gains and losses relative to a reference point, with losses weighing more. Many of its effects arise from fast evaluation. It linked psychology with economics.

What triggers the switch, and what intuition does well. The usual treatment contrasts fast and slow thinking. It leaves out how conflict between the two is detected, the feeling of correctness that decides whether anyone checks, the finer architecture of the mind, the tests that measure reflection, the brain systems involved, and the view that fast heuristics are often smart. For search, that missing part decides whether a reader trusts the first impression or looks again.

Logical intuition is the fast, implicit sensitivity to logical principles, as Jonathan De Neys and colleagues have argued. Intuition can be right as well as biased. Fast judgements are not automatically errors.

The algorithmic mind, in Keith Stanovich's tripartite model, is the level of computational capacity that executes deliberate thinking. It corresponds to intelligence as tests measure it. Having the capacity does not guarantee using it.

The reflective mind is Stanovich's level of dispositions and goals that decides whether deliberate thinking is started at all. It covers traits such as open-mindedness and the tendency to check. It separates the willingness to think from the ability to think.

The autonomous mind is Stanovich's term for the set of fast processes that run on their own, from innate responses to over-learned skills. They are triggered by stimuli without control. Much reading behaviour on the web runs here.

The mindware gap is the absence of the knowledge, rules or strategies a person would need to reason correctly, such as basic probability. Effort cannot fix what is missing. Some errors are gaps in knowledge, beyond laziness.

Cognitive miserliness is the tendency to default to the least effortful way of thinking, a term from Susan Fiske and Shelley Taylor. It preserves resources. Pages that demand effort before they deliver lose readers to it.

Conflict monitoring is the detection of a clash between competing responses, a function associated with the anterior cingulate cortex. Without detection there is no correction. Many reasoning errors are failures to notice a conflict.

The feeling of correctness, or feeling of rightness, is the confidence that accompanies an intuitive answer, studied by Valerie Thompson. A strong feeling reduces the chance of checking. A page that feels right is rarely examined.

Metacognitive monitoring is the ongoing assessment of one's own thinking, including how sure one is and whether to think more. It decides when slow thinking is switched on. It is the gatekeeper between the two modes.

The tripartite model of mind, proposed by Keith Stanovich, divides cognition into the autonomous, the algorithmic and the reflective mind. It refines the two-type picture. It explains why intelligent people still make intuitive errors.

The Cognitive Reflection Test, published by Shane Frederick in 2005, consists of short problems with a tempting wrong answer, such as the bat and ball problem. It measures the tendency to override the first response. Many educated people fail it.

If people checked the first answer that comes to mind before writing it down, students at selective universities would solve all three items. Outside MIT, fewer than a third did.

The parallel-competitive model holds that fast and slow processes run at the same time and compete for control, a view associated with Steven Sloman. It differs from the idea of a default that is later corrected. Both modes can pull in opposite directions at once.

The belief bias task presents syllogisms whose logical validity conflicts with the believability of their conclusions, as Jonathan Evans, Julie Barston and Paul Pollard showed in 1983. People accept believable conclusions even when the logic fails. Belief beats logic by default.

If logic alone decided, each row would read the same from left to right. The participants had been told to answer yes only when a conclusion followed necessarily.

The Wason selection task, devised by Peter Wason in 1966, asks which cards must be turned to test a conditional rule. Most people choose the wrong cards. It is one of the most replicated demonstrations of how reasoning goes astray.

A System 1 override is the suppression of an intuitive response in favour of a deliberate one. It requires detecting the conflict, having the right knowledge and spending the effort. All three can fail.

The somatic marker hypothesis, proposed by Antonio Damasio, holds that bodily signals associated with emotions guide decisions, often quickly and below awareness. Patients with damage to the ventromedial prefrontal cortex decide poorly despite intact reasoning. Feeling is part of deciding.

Dual-system neuroarchitecture refers to attempts to map fast and slow processing onto brain networks. The mapping is debated. It connects the psychological theory to measurable brain activity.

The ventromedial prefrontal cortex integrates value and emotion in decisions. Damage to it disrupts real-life choices while test intelligence remains. It is a key region for fast, value-based judgement.

The anterior cingulate cortex is associated with detecting conflict and errors. It signals when an intuitive response may be wrong. It is a candidate for the trigger of slow thinking.

Fuzzy-trace theory, developed by Valerie Reyna and Charles Brainerd, holds that people store both verbatim details and gist meanings, and that reasoning tends to rely on gist. With experience, people reason more intuitively and often better. It offers an alternative to fast-versus-slow.

A verbatim memory trace is a precise representation of exact details such as numbers and wording. It fades faster than gist. Exact figures on a page are remembered less well than the meaning they carried.

A gist memory trace is a representation of the bottom-line meaning. People rely on it when they decide. What a reader takes away from a page is its gist.

The ratio bias is the preference for a larger number of chances over a better probability, as in choosing a bowl with several winning beans among 100 over a bowl with 1 in 10, although the odds are worse. Seymour Epstein and Veronika Denes-Raj documented it. Large counts feel better than good ratios.

The affect heuristic is the use of a quick feeling of good or bad to judge risks and benefits, as Paul Slovic and colleagues described. When something feels good, its risks seem low. First impressions of a page steer every later judgement.

Hot executive function is cognitive control in emotionally charged situations involving reward or threat, a distinction made by Philip Zelazo and colleagues. Control is harder when emotions run high. Worried searchers decide under hot conditions.

Cool executive function is cognitive control in emotionally neutral situations, such as abstract problem solving. It is what most laboratory tasks measure. Real searches are often less cool.

Need for cognition is a stable personal tendency to enjoy and engage in effortful thinking, measured by a scale from John Cacioppo and Richard Petty. People high in it take the central route more often. Audiences differ in how much thinking they want.

Actuarial judgement uses statistical rules to make predictions, as opposed to clinical intuition. Paul Meehl showed in 1954 that such rules often outperform expert judgement. Formal checks can catch what intuition misses.

Nudge theory, set out by Richard Thaler and Cass Sunstein in 2008, holds that small changes in how choices are presented can steer behaviour predictably without restricting options. Nudges work mostly through fast processes. Every page nudges, intended or not.

Choice architecture is the design of the environment in which people choose, including defaults, order and framing. It shapes fast decisions. A result page is a choice architecture, and so is every page after the click.

If people held a settled view on organ donation, the default would barely move the numbers. Wherever doing nothing means consent, nearly everyone consents.

Fast and frugal heuristics are simple decision rules that use little information and time, studied by Gerd Gigerenzer and his research group. They often perform as well as complex models. Fast thinking can be smart.

Ecological rationality is the match between a decision strategy and the structure of the environment in which it is used. A heuristic is good or bad relative to its environment. The same shortcut works in one setting and fails in another.

The recognition heuristic infers that a recognised option has the higher value when only one of two options is recognised, as Daniel Goldstein and Gerd Gigerenzer showed. Recognition alone can be a good cue. In search, a known name wins the click.

The take-the-best heuristic compares options cue by cue in order of validity and decides on the first cue that discriminates. It ignores the rest. Readers compare results the same way, stopping at the first decisive difference.

A result is chosen and a page is judged mostly by fast processes: recognition, a feeling of correctness, the gist. Slow checking happens when a conflict is noticed and the reader has reason and capacity to think. Pages win the first decision with clarity and familiarity, and they win the second with arguments that survive scrutiny.

Make the first impression accurate, because it is rarely revised. Give fast readers the gist and deliberate readers the evidence, and signal clearly where the two differ. A short question that asks what the reader needs is easy to answer fast, which is why Supervised Search works with both modes.