Behavioural psychology library
Search in the language of the product problem. Each ingredient explains what may be happening, where the evidence applies, and what to protect when you design with it.
Browse all ingredients
You do not need to know the academic term. Search by a behaviour, product outcome or interface problem, then narrow the results if it helps.
Decision Making
A preference for doing something when action is not clearly better than waiting or doing nothing.
+Recognising possible action bias can help a team compare the likely outcomes of acting, waiting and doing nothing instead of assuming that movement is always progress.
−Designing to “use” action bias can push people into premature, unnecessary or hard-to-reverse decisions. Speed and conversion can conceal confusion, regret, avoidable cost or harm.
Perception & Cognition
A readily available positive or negative feeling can become a shortcut for judging an option’s risk, benefit or overall value.
+Recognising affect as an input can help teams investigate why an option feels reassuring, threatening or appealing and ensure that valid concerns are represented alongside factual risk and benefit information.
−Pleasant imagery, friendly tone or a positive moment can make a product seem safer or more beneficial without changing its terms. Fear, urgency or distress can likewise inflate perceived danger. Using those shifts to secure consent, payment or retention can obscure rather than support an informed decision.
Decision Making
When outcome probabilities are unclear, some people prefer a comparable option with known probabilities—but not in every context.
+Recognising ambiguity can help teams show what is known, what remains uncertain, where an estimate came from and when it will be updated. This supports informed comparison without pretending that an imprecise forecast is a known risk.
−A product can exploit ambiguity by hiding uncertainty around one option, presenting false precision for a preferred option or making a familiar choice look safer without better evidence. Removing every unknown can also suppress exploration, and a preference for the known option may be sensible rather than biased.
Decision Making
An earlier number can pull a later estimate towards it, especially when the answer is uncertain.
+Helps teams spot when prices, estimates, targets or ranges may become unintended reference points—and replace them with relevant, explainable comparisons.
−Irrelevant or inflated numbers can distort perceived price, effort, risk or value, especially when people have limited knowledge. Using them to make an option look better can be manipulative.
Usability & Interaction
A person may consider how they would feel if a chosen outcome proved worse than a foregone alternative, but anticipated regret can favour action, inaction, risk or safety depending on feedback and context.
+As a decision-support audit, anticipated regret draws attention to whether people will learn the result of forgone options, whether a decision can be revised and whether the interface exaggerates blame. Clear comparisons, cooling-off and recovery paths can improve the decision without predicting which option someone should choose.
−“You’ll regret missing this” messages can manufacture fear, urgency and self-blame. Avoiding a decision does not reliably prevent regret, and the least regrettable option is not necessarily the safest or best-supported option. High-stakes guidance must not substitute emotion framing for probabilities, consequences or professional advice.
Usability & Interaction
Attention can be allocated differently to threat or personally relevant cues, particularly in anxiety—but the pattern depends on the person, task and timing.
+The evidence can help teams ask whether warnings, health information or other threat-related content affect anxious or high-risk audiences differently. This encourages proportionate risk communication, accessible support and tests of what important neutral information people may miss.
−Using fear, anger or distressing imagery to capture attention can exploit vulnerability and may reduce comprehension. A click, heatmap or glance does not establish attentional bias, and an average laboratory reaction-time difference does not predict one person's behaviour in an interface.
Social & Emotional Influence
Perceived legitimate expertise or institutional authority can receive more weight than the supporting reasons warrant.
+Recognising authority cues helps teams make recommendations, certifications and expert claims attributable and checkable, so people can judge whether the messenger has relevant expertise for the decision.
−Borrowed prestige, unexplained “recommended” labels and vague “science-backed” claims can suppress questions or create false assurance. Authority may be irrelevant, conflicted, outdated or wrong, and deference is especially risky when a choice is consequential or hard to reverse.
Usability & Interaction
When estimating how common or likely something is, people may substitute how readily relevant examples come to mind.
+Recognising the shortcut can help teams support uncertain judgments with representative records, denominators, time windows and comparable alternatives instead of relying on whichever example feels easiest to recall.
−Making selected incidents vivid, recent or prominent can manufacture accessibility without improving the evidence. In safety, health, money, consent or eligibility decisions, this can inflate or suppress perceived risk and make an anecdote feel more representative than it is.
Usability & Interaction
A correction makes belief in the false claim stronger than it would have been without the correction—a possible but uncommon result.
+The concept can remind teams to measure whether a correction improves belief accuracy for the intended audience instead of assuming that publishing a notice worked. It also encourages a clear factual claim, comparison group, timing and follow-up.
−Overstating backfire can leave harmful misinformation uncorrected or encourage products to personalise facts around ideology. Corrections can still be incomplete, short-lived or ineffective, but those outcomes require better evidence and explanation—not silence, ridicule or a claim that people are irrational.
Social & Emotional Influence
Knowing that many other people chose or support an option can change demand or judgement, even when popularity is not evidence that the option fits this person.
+As an audit concept, the effect helps teams test whether genuine popularity information helps people discover commonly useful options. Showing a current count, denominator, time period and relevant comparison group allows users to judge what the signal means instead of treating a vague “trending” badge as quality evidence.
−Popularity can amplify an arbitrary early advantage, suppress minority needs and create a feedback loop in ranking systems. Fabricated counts, selective reference groups and stale labels are deceptive. Even a real majority is not evidence of quality, safety or personal fit, and one observed popular choice cannot establish why someone selected it.
Social & Emotional Influence
Broad, high-base-rate statements can feel uniquely accurate when they are presented as a personal assessment.
+Recognising the effect helps teams audit personality quizzes, AI-generated insights and health, career or financial recommendations for false specificity. It encourages products to show what information genuinely changed an output and how uncertain or widely applicable each statement is.
−Exploiting broad flattering feedback can create unwarranted trust, disclosure, purchases or consequential decisions. Satisfaction, recognition or engagement does not validate an assessment, while dismissing every broadly useful statement as Barnum can also overlook genuine evidence and shared needs.
Usability & Interaction
A vivid case-specific clue can receive too much weight, or the relevant starting frequency too little, when people estimate how likely an outcome is.
+Recognising base-rate neglect helps teams explain alerts, forecasts and risk estimates without letting a striking case or isolated accuracy figure dominate. A relevant reference class, shared denominator and natural-frequency display can make true positives, false positives and uncertainty easier to inspect.
−A base rate can mislead when its reference class, time period or data quality does not match the decision. Historical rates may hide data drift or unequal treatment, while presenting an average as a verdict about one person can erase important evidence. Teams can also manipulate judgement by choosing a convenient denominator.
