attck-research

Evidence review: How Cognitive Bias Shapes Digital Decisions

Article: How Cognitive Bias Shapes Digital Decisions
Reviewed: November 2026 · Method version 2 · How these reviews work

How this review works

Each factual claim in the article was checked against the source it cites, then challenged by reviewers whose job was to find the strongest evidence against it: replication attempts, meta-analyses, later papers that cite the original work, and research from other fields that study the same mechanism. A separate reviewer weighed both sides, and a script re-checked every quoted source. Where a claim was overstated or misattributed, the article was corrected. The research below supports the article as it now reads.

Background research

Introduction

In reality, research on judgment and decision-making suggests that much of what shapes a decision happens quickly and automatically, with little deliberate thought.

Kahneman’s account of fast, automatic System 1 judgment is the basis for this sentence; classic work by Nisbett and Wilson (1977) also showed that people often cannot report what influenced their choices.

The Science of Speed

System 1: Fast, automatic, and effortless. This is the brain on “autopilot.” System 2: Slow, effortful, and deliberate. This is the brain working through a hard problem.

Kahneman describes System 1 as fast, automatic, effortless and associative, and System 2 as slow, serial and effortful (Nobel lecture, 2002).

Online, most users scan pages rather than read them word by word, and they slow down to read closely only when the content matters enough to them.

In Nielsen’s usability studies, 79% of test users scanned new pages and only 16% read word by word.

If an interface makes a task feel hard, users are less willing to start it or see it through.

In Song and Schwarz’s experiments, instructions printed in a hard-to-read font made the same task seem harder and people less willing to do it.

The Science of Speed (pull quote)

Gigerenzer and Gaissmaier define heuristics as efficient cognitive processes, conscious or unconscious, that ignore part of the information.

Gigerenzer and Gaissmaier (2011) define heuristics this way in the abstract of their Annual Review of Psychology paper.

The Anchoring Effect

One of the best-documented biases in judgment is Anchoring: a number people see early, even an irrelevant one, pulls their later estimates toward it. In studies of what people will pay, the effect is real but more moderate than early experiments suggested.

A meta-analysis of 53 studies of willingness to pay and accept found a moderate anchoring effect (Li, Maniadis and Sedikides 2021).

The prices have not changed, but the order can shift what feels reasonable. Studies of price order find that listing higher prices first nudges people toward pricier options, mainly when they read price as a sign of quality.

Suk, Lee and Lichtenstein (2012) found that presenting prices from high to low led consumers to choose higher-priced options.

The Default Effect

People tend to stick with whatever option is already selected. This is known as the Default Effect. A pre-selected option usually keeps far more people than it would attract otherwise, although the size of the effect varies a lot by setting, and defaults work less well when people already have strong preferences.

A meta-analysis of default studies (Jachimowicz et al. 2019) finds that most show positive effects, and that defaults work through effort, implied endorsement and a sense of ownership.

The most famous example of this comes from a study by Johnson and Goldstein (2003) on organ donation. In countries with an “opt-in” policy, about 15% of people were registered as donors on average. In countries with an “opt-out” policy, consent rates were 86% or higher. The default changed who was registered far more than it changed actual donations, though: later studies find a much smaller difference in donations, and some find none.

Johnson and Goldstein’s figures describe effective consent; their own estimate of the effect on actual donation was a rise from 14.1 to 16.4 donors per million (Science, 2003).

If most of your users choose “Express Shipping,” do not make them select it every time. Make it the default. This saves most users a step, and people tend to read a default as the recommended choice.

McKenzie, Liersch and Finkelstein (2006) showed that people treat a policy default as an implicit recommendation from whoever set it.

Loss Aversion

In many studies, people weigh a loss about twice as heavily as a gain of the same size. This is Loss Aversion. The size of the effect varies with the situation, and with how familiar people are with what is at stake.

A 2024 meta-analysis of loss aversion estimates puts the average coefficient at about 1.96, close to Kahneman’s estimate of 2 to 2.5.

Case Study: The Economist

Behavioral economist Dan Ariely famously illustrated the Decoy Effect (or Asymmetric Dominance), first documented by Joel Huber and colleagues in 1982, using a subscription offer from The Economist.

Asymmetric dominance was first reported by Huber, Payne and Puto (1982); Ariely’s Economist example in Predictably Irrational (2008) made it widely known.

In Ariely’s reading, the print-only option was a decoy: almost no one would want it, but it made the web-and-print bundle look like a much better deal. The Economist never explained the offer, and the ad was later withdrawn.

Ariely presents the print-only tier as a decoy that made the bundle look more attractive (Predictably Irrational, 2008; TED, 2008). He has said he asked The Economist what they were thinking and received no explanation.

The Choice Factory

Richard Shotton’s The Choice Factory gives each bias a short chapter, follows a shopper through an ordinary day, and shows how to use each bias, often with his own small experiments.

Harriman House describes the book as 25 short chapters, each addressing one cognitive bias.

What changed in the article

These edits were made on November 12, 2026 after the review.

Introduction: supported with caveats

Claim reviewed: Research on judgment and decision-making suggests much of what shapes a purchasing decision happens below the threshold of conscious thought.

Now reads: In reality, research on judgment and decision-making suggests that much of what shapes a decision happens quickly and automatically, with little deliberate thought.

Why: Kahneman’s book does describe fast, automatic judgment that people often cannot report, so the general idea has a source. But the book is about judgment in general, not purchasing, and a major critical review argues that unconscious influences on decisions have been overstated, partly because some priming studies did not hold up. ‘Below the threshold of conscious thought’ should become ‘quickly and automatically’.

Introduction: not supported

Claim reviewed: Purchasing decisions are not steered by logic but by cognitive biases, mental shortcuts that evolved to help the brain process a flood of information without burning through its energy reserves.

Now reads: These decisions lean heavily on cognitive biases and heuristics, mental shortcuts that let people decide quickly when time, attention and information are limited.

Why: No source is given, and both halves of the sentence run ahead of the evidence. Whole-brain energy use barely changes during hard thinking, and the evolutionary literature explains biases by the asymmetric cost of errors, not by energy saving; the closest thing to support is a 2022 study suggesting that long cognitive work builds up metabolic by-products, which is not the same as draining ‘energy reserves’. The article’s own source also argues that ignoring information often makes judgments more accurate, which cuts against ‘not steered by logic’. Now cites Gigerenzer & Gaissmaier, Heuristic Decision Making, Annual Review of Psychology 62 (2011).

Introduction: supported with caveats

Claim reviewed: When an interface presents every option with equal weight, users often freeze or default to the lowest-effort option.

Now reads: When an interface presents many similar options with no guidance, some users put off the decision or take the easiest option. This is most likely when the options are hard to compare and users are unsure what they want.

Why: The sentence has no source, but its core mechanism is well documented: when options are hard to compare, people are more likely to put off the decision or take the default, and larger sets make this worse. What the evidence does not support is a general rule. The average ‘choice overload’ effect across experiments is close to zero; it shows up when options are complex, preferences are unclear and the set gives no guidance, which is the situation the article describes, so the sentence should state those conditions. Now cites Chernev, Böckenholt & Goodman, Choice overload: A conceptual review and meta-analysis (2015).

The Science of Speed: misattributed

Claim reviewed: Dual-process theory was popularized by Nobel laureate Daniel Kahneman and psychologist Amos Tversky, who identified two distinct modes of thinking, System 1 and System 2.

Now reads: To understand why users behave “irrationally,” we can look to dual-process theory, popularized by Nobel laureate Daniel Kahneman, who built on decades of research with psychologist Amos Tversky. Kahneman describes two modes of thinking, using the labels System 1 and System 2 coined by psychologists Keith Stanovich and Richard West:

Why: Kahneman did popularize System 1 and System 2 in the cited book and did win the Nobel. But in his own Nobel lecture and biography he credits the two-system labels to Sloman and to Stanovich and West, and he records that Tversky learned in early 1996 that he had months to live, so Tversky did not ‘identify’ these two modes. Kahneman and Tversky’s joint work was on heuristics, biases and prospect theory. Now cites Kahneman, Maps of Bounded Rationality (Nobel lecture, 2002).

The Science of Speed: supported with caveats

Claim reviewed: System 1 is fast, automatic, emotional and subconscious; System 2 is slow, effortful, logical and calculating.

Now reads: System 1: Fast, automatic, and effortless. This is the brain on “autopilot.” System 2: Slow, effortful, and deliberate. This is the brain working through a hard problem.

Why: This matches the popular description in Kahneman’s book. Researchers who study dual processes say speed and effort are the defining features, while ‘emotional’ and ‘logical’ are loose associations: fast thinking can be logical and slow thinking can be emotional. Some theorists also question whether there are two distinct systems at all. Now cites Kahneman, Maps of Bounded Rationality (Nobel lecture, 2002).

The Science of Speed: supported with caveats

Claim reviewed: In digital contexts users operate almost exclusively in System 1: they do not read pages, they scan them.

Now reads: Online, most users scan pages rather than read them word by word, and they slow down to read closely only when the content matters enough to them.

Why: The scanning part is right: Nielsen’s usability studies found most users scan new pages rather than read word by word. But 79% is not ‘almost exclusively’, and NN/g’s later work says people read closely when the content matters to them. Linking scanning to System 1 is the article’s own interpretation, not something the studies measured. Now cites Nielsen, How Users Read on the Web, NN/g (1997).

The Science of Speed: not supported

Claim reviewed: Because System 2 thinking is metabolically expensive, the brain actively avoids it.

Now reads: Because System 2 thinking feels effortful, people tend to avoid it when an easier path is available.

Why: People do avoid demanding mental work when an easier path exists, and a 2024 meta-analysis of 170 studies finds effort is reliably experienced as unpleasant. The stated reason, that System 2 thinking is ‘metabolically expensive’, is not supported: whole-brain energy use does not measurably rise during hard thinking, the glucose and ego-depletion experiments behind the idea have failed in large preregistered tests, and current reviews say the source of effort costs is still unknown. One 2022 study points to metabolic by-products building up after hours of work, which is the strongest case for a metabolic cost but does not mean the brain avoids System 2 to save energy. Now cites Kool, McGuire, Rosen & Botvinick, Decision making and the avoidance of cognitive demand (2010).

The Science of Speed: supported with caveats

Claim reviewed: If an interface forces a user to think too hard, they experience cognitive strain and often abandon the task.

Now reads: If an interface makes a task feel hard, users are less willing to start it or see it through.

Why: Experiments support the core idea: when a task looks or feels harder, people are less willing to do it, and they prefer less demanding options. No study here measured interface abandonment caused by ‘cognitive strain’, so ‘often abandon’ is an extrapolation. A softer wording with a citation fixes it. Now cites Song & Schwarz, If It’s Hard to Read, It’s Hard to Do (2008).

The Anchoring Effect: supported with caveats

Claim reviewed: Anchoring is one of the most powerful biases in pricing; the brain relies heavily on the first piece of information it receives (the anchor) when making subsequent judgments.

Now reads: One of the best-documented biases in judgment is Anchoring: a number people see early, even an irrelevant one, pulls their later estimates toward it. In studies of what people will pay, the effect is real but more moderate than early experiments suggested.

Why: Anchoring is one of the best-replicated findings in judgment research, and Kahneman documents it. Two things overreach: the anchor need not be the ‘first’ piece of information (any salient number works, even an irrelevant one), and a meta-analysis of price studies finds a moderate effect, smaller than early studies reported. ‘One of the most powerful biases in pricing strategy’ has no source. Now cites Li, Maniadis & Sedikides, Anchoring in Economics: A Meta-Analysis (2021).

The Anchoring Effect: supported with caveats

Claim reviewed: On a SaaS pricing page, showing the $299 Enterprise plan first makes the $99 plan feel like a bargain, while showing the $29 Basic plan first makes the $99 plan feel expensive; perception of value shifts entirely based on the anchor.

Now reads: The prices have not changed, but the order can shift what feels reasonable. Studies of price order find that listing higher prices first nudges people toward pricier options, mainly when they read price as a sign of quality.

Why: The pricing-page example is hypothetical, but research on price order backs its direction: shoppers who see higher prices first tend to choose pricier options. That effect depends on people reading price as a sign of quality, so ‘shifted entirely’ is too strong. Soften the last sentence and add a source. Now cites Suk, Lee & Lichtenstein, The Influence of Price Presentation Order on Consumer Choice (2012).

The Default Effect: supported with caveats

Claim reviewed: Humans have a profound biological tendency to follow the path of least resistance; when a choice is pre-selected, the majority of users will accept it rather than expend the energy to change it.

Now reads: People tend to stick with whatever option is already selected. This is known as the Default Effect. A pre-selected option usually keeps far more people than it would attract otherwise, although the size of the effect varies a lot by setting, and defaults work less well when people already have strong preferences.

Why: Defaults are among the strongest tools in choice design, and Nudge builds on them. But ‘profound biological tendency’ has no source, and effort is only one reason defaults work; people also read them as recommendations or feel they already own the option. A meta-analysis finds most default studies show an effect, but some show none and two show the opposite, so ‘the majority of users will accept it’ is not guaranteed. Now cites Jachimowicz, Duncan, Weber & Johnson, When and why defaults influence decisions, Behavioural Public Policy 3(2) (2019).

The Default Effect: supported with caveats

Claim reviewed: Johnson and Goldstein (2003): countries with opt-in organ donation policies typically see consent rates around 15%, while opt-out countries see consent rates exceeding 90%.

Now reads: The most famous example of this comes from a study by Johnson and Goldstein (2003) on organ donation. In countries with an “opt-in” policy, about 15% of people were registered as donors on average. In countries with an “opt-out” policy, consent rates were 86% or higher. The default changed who was registered far more than it changed actual donations, though: later studies find a much smaller difference in donations, and some find none.

Why: The figures match the paper’s chart of effective consent rates: opt-in countries range from about 4% to 28% (around 15% on average) and opt-out countries from about 86% to nearly 100%. The catch is that ‘consent’ here means not having opted out, not actual donation. The same paper estimates a much smaller gain in real donations, and several later cross-country studies find no significant difference in donation rates between opt-out and opt-in countries. Now cites Dallacker et al., Opt-out defaults do not increase organ donation rates, Public Health 236 (2024).

The Default Effect: supported with caveats

Claim reviewed: Making the majority-preferred option the default reduces friction and signals to the user that it is the recommended or normal course of action.

Now reads: If most of your users choose “Express Shipping,” do not make them select it every time. Make it the default. This saves most users a step, and people tend to read a default as the recommended choice.

Why: No source is cited, but the ‘recommendation’ half is well supported: experiments show people read a default as what the choice architect recommends, and the article’s own source names this mechanism. The 80% figure is a made-up example and ‘reduces friction’ is common sense rather than a measured result. Add a citation; the wording can stay. Now cites McKenzie, Liersch & Finkelstein, Recommendations Implicit in Policy Defaults (2006).

Loss Aversion: supported with caveats

Claim reviewed: The pain of losing something is roughly twice as powerful as the pleasure of gaining something of equal value.

Now reads: In many studies, people weigh a loss about twice as heavily as a gain of the same size. This is Loss Aversion. The size of the effect varies with the situation, and with how familiar people are with what is at stake.

Why: The ‘roughly twice’ figure has strong support: Kahneman’s own estimate is 2 to 2.5, and a 2024 meta-analysis puts the average at about 1.96. But the size varies with the situation, some experiments make it vanish or reverse, and one review argues losses are not generally more powerful than gains. Say ‘in many studies’ and note the variation. Now cites Brown, Imai, Vieider & Camerer, Meta-analysis of Empirical Estimates of Loss Aversion (2024).

Loss Aversion: contested

Claim reviewed: A message that highlights what the user stands to lose is often more compelling than one highlighting what they stand to gain; a loss frame (‘Stop overpaying $50/year’) triggers a stronger behavioral response than a gain frame (‘Save $50’).

Now reads: In copywriting and interface design, framing matters, but the direction is not fixed. Loss-framed incentives and some loss-framed sales messages have outperformed gain frames in field experiments, while reviews of health and product messages find gain frames doing as well or slightly better. Treat “Save $50” versus “Stop overpaying $50/year” as something to test with your own users, not a rule.

Why: This is a different finding from loss aversion, and the evidence points in different directions by domain. Loss framing has won in field experiments where a real loss is at stake, such as bonuses workers or teachers can forfeit and a 1995 credit-card mailing. But meta-analyses of health messages find no general loss advantage, with gain frames slightly ahead for prevention, and a 2022 meta-analysis of product-attribute framing found gain frames did better. No study shows that ‘Stop overpaying’ beats ‘Save $50’, so the article should present the direction as something to test, not a rule. Now cites Gallagher & Updegraff, Health message framing effects on attitudes, intentions, and behavior: a meta-analytic review (2012).

Case Study: The Economist: supported with caveats

Claim reviewed: Dan Ariely demonstrated the Decoy Effect (asymmetric dominance) using a subscription offer from The Economist with three tiers: Web Only $59, Print Only $125, Web + Print $125.

Now reads: Behavioral economist Dan Ariely famously illustrated the Decoy Effect (or Asymmetric Dominance), first documented by Joel Huber and colleagues in 1982, using a subscription offer from The Economist.

Why: The tiers and prices match Ariely’s example as widely reported, and it is the right source for it. But Ariely did not discover or first demonstrate the decoy effect; Huber, Payne and Puto documented asymmetric dominance in 1982, and Ariely popularized it with this example. Change ‘demonstrated’ to ‘illustrated’ and credit the original work. Now cites Huber, Payne & Puto, Adding Asymmetrically Dominated Alternatives, Journal of Consumer Research 9(1) (1982).

Case Study: The Economist: not supported

Claim reviewed: When the decoy was present, the vast majority of subscribers chose the expensive bundle; when Ariely removed the decoy, users flocked to the cheaper Web Only plan.

Now reads: When Ariely gave these three options to 100 MIT students, 84% chose the bundle and none chose print only. When he gave a version without the print-only option to another 100 students, most chose the cheaper web-only plan. This was a classroom exercise with hypothetical choices, and later field studies find the decoy effect is small or absent in many real shopping situations, including a large test on a flight-booking site.

Why: Ariely’s own account confirms the numbers came from two groups of 100 MIT students answering a hypothetical question, not from Economist subscribers, so ‘the vast majority of subscribers’ misstates the source; the direction of his result is as reported. Separately, the newest field evidence is sobering for a design team: a 140,000-session experiment on a flight-booking site found no general attraction effect, and a 3.6 million-transaction wine study found one of about one percentage point. The effect is real in the lab and defended by its originators, but it is small or absent in many realistic settings. Now cites Ariely, Are we in control of our own decisions? (TED talk transcript, 2008).

Case Study: The Economist: supported with caveats

Claim reviewed: The Print Only option was a decoy whose sole purpose was to make the bundle look like a deal; the design team manipulated the perceived value of the premium product.

Now reads: In Ariely’s reading, the print-only option was a decoy: almost no one would want it, but it made the web-and-print bundle look like a much better deal. The Economist never explained the offer, and the ad was later withdrawn.

Why: Calling the print-only option a decoy is Ariely’s interpretation, and his student experiment supports it. But Ariely himself says he called The Economist to find out what they were thinking, got passed around, and the ad disappeared with no explanation, so nobody knows what the publisher intended. ‘Its sole purpose’ and ‘the design team manipulated’ state a guess as fact; attribute the reading to Ariely. Now cites Ariely, Are we in control of our own decisions? (TED talk transcript, 2008).

Conclusion: contested

Claim reviewed: Cognitive biases are not glitches but features that allow us to navigate a complex world efficiently.

Now reads: The mental shortcuts behind cognitive biases are not simply “glitches” in the human operating system. They usually let us decide quickly and well enough in a complex world, even though they also produce predictable errors.

Why: This is one side of a live debate, and the side the article takes has more backing than pass 4 showed: evolutionary biologists argue biases can improve decisions by steering people away from the costlier error, resource-rational models reproduce anchoring as a sensible use of limited time, and Gigerenzer’s review finds little evidence that biases cost people health, wealth or happiness. Against that, the heuristics-and-biases tradition that the article’s main source belongs to defines biases as the systematic errors heuristics produce, philosophers object that an adaptive rationale for a shortcut does not make the resulting judgment correct, and a medical systematic review links common biases to diagnostic errors. Say the shortcuts are useful and the biases are their predictable side effects. Now cites Gigerenzer & Gaissmaier, Heuristic Decision Making, Annual Review of Psychology 62 (2011).