Skip to content

Considering The Opposite

Principle · All principles
Evidence · 13 claims (6 for, 7 mixed) · 23 studies (7 causal, 6 quant-synthesis, 5 review, 2 theoretical, 1 associational, 1 qualitative, 1 design), q1–q4 · 5 of 23 report an effect size · 7 claims rest on one study

Conditional relationship

This page owns one move at one moment: after a learner has committed to a judgment (a diagnosis, a prediction, an interpretation of a source, a verdict on a design or an argument) on a task where the evidence could point more than one way, prompting them to generate specific reasons why the opposite, or a named alternative, could be right, and then to decide again, is expected to reduce confirmation bias, anchoring and overconfidence in that judgment more than a general instruction to "be objective". The knowledge type is mainly principle (causal explanations, diagnoses, interpretations) and metacognitive-strategy (checking one's own judgment); the outcomes are the quality of the revised judgment, assessment-accuracy of the learner's confidence, and, where the alternative was wrong, conceptual-understanding of why. Whether the habit transfers to judgments made without the prompt, or lasts, is a separate question that nothing here answers.

No claim in this wiki tests the relationship directly. The one claim about the strategy, A systematic review of bias-mitigation training names considering the opposite as a previously validated way to reduce judgment biases, without testing it itself [+M], rests on a systematic review (12 studies, 17 experiments) that names "consider the opposite" as an established debiasing method and cites primary experiments on hindsight bias and anchoring, but does not test it in its own data; the primary experiments (Lord, Lepper & Preston 1984; Mussweiler, Strack & Pfeiffer 2000, in Key Sources) are not recorded as evidence in the wiki. The rest of the evidence below is carried over from neighbouring relationships: comparing contrasting cases, conflict-based instruction for misconceptions, refutation text and argumentation. Those test instruction in which learners meet or produce an opposing case, not the self-generated reversal at the moment of judgment.

The relationship is conditional on four things: that the learner has actually formed a position (the move has nothing to act on otherwise); that the learner knows enough of the domain to produce a plausible alternative, which a novice often cannot, so the alternative may have to be supplied; that evidence bearing on both sides is available to them; and that the question is not a personal conviction the learner is unlikely to examine on request (Critical disposition is least spontaneous when confronting one's most basic prejudices or deeply held convictions [~W], a theoretical argument only).

How this sits beside converted pages. Cognitive Disequilibrium holds the general model of confronting a learner's explanation with evidence it cannot account for and then supporting the resolution; this page is the self-directed, judgment-stage version of that move, and does not repeat its resolution policy. Evaluating Sources holds the trust-a-source judgment, where checking what others say about a source is a specific way of looking for the disconfirming case. The Debate and Structured Academic Controversy patterns build arguing the other side into a group sequence (SAC's position reversal resembles this move); use them when the opposite is to be argued by a group rather than considered by one learner. Reflection and Self-Monitoring hold the wider habits this prompt is one instance of.

Default design, while the relationship is untested

The earlier page's guidance, kept as a concrete default. Every step is a design proposal; its evidence status is stated beside it.

  1. Commit first, in writing, with a confidence rating. Before any prompt, each learner records their judgment and how sure they are (for example 50–100%). The commitment gives the prompt something to act on and the rating gives a before-and-after observation. Untested proposal; from the earlier page ("the moment of commitment is exactly when bias is strongest").
  2. Ask for reasons for the opposite, not for open-mindedness. The prompt is specific: "Give two or three reasons the other diagnosis (prediction, reading) could be right", or "Suppose you turn out to be wrong: what is the most likely reason?" Do not substitute "be objective" or "consider other views". Basis: A systematic review of bias-mitigation training names considering the opposite as a previously validated way to reduce judgment biases, without testing it itself [+M] (second-hand: a review naming the method; it contrasts it with generic warnings, which it reports fail because people believe they are already objective).
  3. Supply the alternative when the learner cannot produce one. If the learner's alternatives are absent, implausible or straw men, give the alternative (a second case, a rival explanation, a statement of the common misconception) and ask them to make the best case for it, then compare it side by side with their own. Basis, carried from neighbouring relationships: comparing cases side by side [+M] (meta-analysis of 57 experiments, d = .50, against studying cases singly or in sequence; larger when learners look for similarities and when the principle comes after the comparison) and refutation text and conflict interventions [+M] (science and reading-education syntheses, read from abstracts). Neither tests a learner-generated opposite.
  4. Check the alternative against evidence, then decide again. Give the learner the evidence that discriminates the two (a test result, a source, an observation, the outcome of the prediction) and ask for a final judgment and a second confidence rating, with one sentence on what changed or why it did not. Untested as a step here; from the earlier page ("time and structure for genuine evaluation of the alternative, not a token gesture"). Conflict alone may not be enough: in one experiment staged contradictions helped only learners who reported being confused (conflict-based instruction [~M]), so the resolution step is the part not to cut.
  5. Treat a changed mind as a good outcome. Say so before the first use, and in feedback credit the quality of the reasons on both sides, not whether the first answer survived. Untested proposal; from the earlier page.
  6. Bound it. One alternative and a fixed time per judgment (about 3–5 minutes for a single judgment in class; one line in a fast-moving task), then commit. Use it on judgments that are uncertain or consequential, not on every step, so it does not become second-guessing. Untested proposal; the earlier page warned of decision paralysis.
  7. Plan the arc across sessions, and fade the prompt. A default, untested: sessions 1–2: the teacher models the move aloud on one judgment, and the alternative is supplied for learners to argue for; sessions 3–5: learners generate the alternative themselves from a prompt card, and a partner or the teacher checks that it is plausible; sessions 6 and after: the card is removed, and the teacher records whether learners state an alternative unprompted before committing. A learner moves on from supplied to self-generated alternatives once they give at least two domain-relevant reasons for the alternative on two consecutive cases, and from prompted to unprompted once they do so without the card on two consecutive cases. These thresholds are untested proposals. How to withdraw the support is modelled in Scaffolding and Fading.

When a condition is not met: if no position has been formed, ask for one first; if there is only one defensible answer reached by a set method, this move has nothing to work on, so contrast two worked methods instead (row below); if no discriminating evidence is available, stop at step 3 and record the judgment as open rather than resolved.

Fitting the design to a situation

The three facts that most change the decision:

  • Whether the learner can produce a plausible alternative. A novice in the domain often cannot, and a supplied alternative or contrasting case does the work; an advanced learner or professional needs only the prompt. If unstated, ask: when this learner is wrong on a task like this, what is the usual alternative, and could they name it themselves?
  • What the judgment is about. An open diagnostic, predictive or interpretive judgment suits the move; a procedure with one correct method does not; a personal or identity-linked conviction needs a different, more private design. If unstated, ask: is there a defensible alternative answer here, and would asking for it touch something the learner holds personally?
  • What counts as success. A better final judgment, better-calibrated confidence, a written counterargument, and a changed explanation of a phenomenon are different targets with different checks. If unstated, ask: after the activity, what should the learner be able to do or decide, and how will you see it?
If the brief says… Then change… Basis
Learners are novice in the domain, or their alternatives are implausible Supply the alternative (a rival explanation, a second case, the common misconception) and ask learners to make its best case, side by side with their own; drop to self-generated alternatives only when they meet the threshold in the arc above Comparing cases side by side improves learning [+M] (carried from case comparison; not a test of this move); expertise reversal [~M] (review of cognitive-load studies; the reverse case)
Learners are advanced or practising professionals Prompt only, with no supplied alternative; use the red-team version (one person argues against the leading proposal) on consequential decisions Instructional guidance that helps novices can become redundant or counterproductive as expertise grows. [~M] (cognitive-load studies, carried to this prompt); the red-team form is untested
Learners are child or adolescent, in science, holding an everyday misconception Predict–observe–explain: commit to a prediction, ask "how would someone who thinks the opposite explain it?", show the phenomenon, then discuss which account fits; use a refutation text that names the misconception and answers it Conflict-based instruction improves science conceptual learning [+M] (meta-analyses of science-education studies; staged contradiction helped only confused learners in one small experiment); refutation text [+M] (no developmental pattern found in the review)
The goal is argument writing (complex-skill) Add a required counterargument-and-rebuttal paragraph: the strongest objection to the learner's thesis, then the reply; score it with the brief's rubric Argumentation improves reasoning [+M] (online science argumentation meta-analysis, 46 studies; one multiyear dialogue intervention with young adolescents); prompted peer feedback raised addressing the opposing view [+W] (one study, 12 groups)
The goal is calibrated confidence Rate confidence before and after the opposite step, then score against the outcome; show each learner their own record across several judgments Prior Knowledge Needed For Accurate Self Assessment [~M] (lowest performers grossly overconfident in one undergraduate course; it does not test this remedy)
The task is a procedure with one correct method Do not use the move; contrast two worked methods side by side and ask which suits which problem Comparing cases side by side [+M] (seventh-grade algebra: method comparison raised procedural knowledge and flexibility, not conceptual knowledge)
Adults in work settings (workplace-clinical), e.g. diagnosis A diagnostic time-out before the decision is final: list the findings that do not fit, name the most dangerous alternative, say what test would separate them A systematic review of bias-mitigation training names considering the opposite as a previously validated way to reduce judgment biases, without testing it itself [+M], carried to clinical judgment, where it is untested in the wiki
The decision must be made quickly, or stakes are high and time is short One alternative, a fixed short time, then commit; record what was considered; never make it open-ended untested; from the earlier page's warning about paralysis
online-self-paced, or phone only A required field before submission: "the strongest reason you could be wrong"; then show the expert's alternative and the deciding evidence untested proposal
Some learners remote and some in the room Do the commit and the alternative in writing in one shared space (chat or a shared document) so remote learners contribute on equal terms; pair across modes for the "best case for the other side" step untested proposal
One teacher for many learners, or scarce facilitator time (a volunteer, a busy shift) A printed or on-screen prompt card does the prompting; partners swap and write the case against each other's judgment; the facilitator reads three to five at random and gives whole-group feedback on the commonest weak alternative untested proposal
A single session only, or a fixed short time budget (say 10 minutes) One judgment: 2 minutes to commit with a confidence rating, 4 minutes for the opposite case, 2 minutes to see the evidence and decide again, 2 minutes on what changed; claim nothing about a habit untested proposal
The question touches learners' convictions, identity or private experience Begin on low-stakes questions; let the learner keep their reasons for the opposite private and choose what to share; never require a learner to argue against their own identity in public Critical disposition is least spontaneous when confronting one's most basic prejudices [~W] (theoretical argument only)
Learners have low confidence or low literacy Do it orally or with a sentence starter ("Someone who disagrees would say … because …"); frame the alternative as another person's view, not the learner's error untested proposal
Disagreement arises among peers in group work Do not rely on the conflict itself; after the discussion, each learner writes alone what they now think and what changed Peer conflicts worked only for students prepared to reflect [~M] (seven dyads, qualitative)
Learners use an AI tool or decision-support output Have the learner judge first, then compare with the tool's answer and argue against whichever answer they now favour Conflict with a decision-support system's output induces deeper reflection than confirmation [+W] (reported second-hand in a review, no detail)
The brief names its own routine (a claim–evidence–reasoning frame, a diagnostic checklist, a critique protocol) Keep it; add one line to it: "the strongest alternative, and why you reject or accept it". Do not replace the routine untested proposal
The brief gives its own categories, criteria or rubric Judge the quality of the alternative with those criteria, and add no test or quiz the brief did not ask for untested proposal
The output is a product or performance for a real audience (a published essay, a client design review) Keep the opposite-consideration in drafting notes; in the final product address the strongest objection only where that audience expects it untested proposal
The brief wants a whole task kept whole (a full case, a simulation) Put the prompt at the commit point inside the task, not as a separate exercise before or after it untested proposal
The brief reports an error learners actually made Use that error as the alternative: learners make the best case for it, then show what evidence rules it out untested proposal; resembles refutation text [+M]

Eleven of the 21 rows link a claim (one only as a resemblance), and none of those tests the prompt itself in that setting: each carries a neighbouring finding (case comparison, conflict-based science instruction, argumentation, calibration) or a second-hand report, and says so. The other ten are untested proposals.

Observation, state and explanation

What can be observed: the committed judgment and its confidence rating; the alternative the learner wrote (how many reasons, whether they are plausible and use domain evidence); the final judgment and confidence; and what the learner said changed. "Debiased", "open-minded" and "has the habit" are inferred states. A changed answer is not by itself a better one, and an unchanged answer is not by itself bias.

Response Plausible explanations Observation that could change the interpretation
Writes a weak or straw-man alternative Lacks the domain knowledge to build one; protecting the first answer; does not see why the task matters Supply a strong alternative and ask for its best case; if this produces good reasons, knowledge was the gap. Ask what would have to be true for the alternative to be right.
Changes the answer after every prompt Low confidence or wish to please; the prompt read as a signal that the first answer was wrong; genuine uncertainty Use the prompt on a judgment that was right too, and compare; ask for confidence before and after.
Never changes the answer, high confidence throughout Answers were right; anchoring; the question touches a conviction Score the first judgments against outcomes; offer a low-stakes question of the same form.
Confidence falls but accuracy does not change Calibration improved; the prompt induced general doubt Track calibration across several judgments, not one.
Uses the move with the card, not without it Not yet a habit; the unprompted task does not look like one that calls for it Give an unprompted judgment of the same kind, then one of a different kind, and record whether an alternative appears.

These rows are proposals, untested as diagnostics. A probe is itself a prompt and changes what it measures, so record each one.

Evidence and qualifications

No claim in the wiki tests a considering-the-opposite prompt against a generic warning or no prompt, in any learning setting. The claims below test the claim's second-hand basis and the neighbouring relationships it borrows from.

Limits:

The learners, tasks and outcomes differ (undergraduates on negotiation cases, school science classes, online argumentation, course exams), so do not rank them by effect size or add them up into a verdict for this move.

The evidence by question

Further evidence, not yet read against this model

Claims that bear on the topic but are not part of the model above.

Objective and learner-valued goal

The designer's objective is usually a better judgment under uncertainty, with confidence that matches accuracy, on a stated kind of task. Elicit separately what the learner wants: to be right, to look competent, to finish, to win the argument, or to get the real decision right at work. These can pull against the move. A learner who values looking competent may treat a prompt to consider the opposite as an accusation of error and defend the first answer; one who values winning may produce only straw-man alternatives. Record what the learner says they want, and use it: a trainee clinician who wants to avoid missing a serious condition will take "name the most dangerous alternative" more seriously than "consider other views".

For example, a nursing student may want to pass the practical assessment, while the designer's objective is diagnostic reasoning that holds up on the ward. Practice on assessment-style cases serves the first; an unprompted case on placement, where the student is watched for whether they name an alternative before acting, tests the second. Record both, so that a good assessment score is not read as the habit.

What would revise this model?

The expectation should weaken if primary studies, once recorded, show that the prompt does not change judgments in learning tasks beyond what a generic warning does, or that its effect disappears once the prompt is removed. If supplied alternatives work as well as self-generated ones for every learner, the model should prefer the supplied form, which is cheaper. If confidence falls without accuracy rising across several judgments, the move is producing doubt, not debiasing. If the habit never appears in unprompted judgments, the move is a task feature, not a capability, and the arc above should be dropped. Do not protect the model by calling every failure "a token gesture" or "too little domain knowledge" after the fact.

A better judgment on the prompted task, calibrated confidence, a written counterargument, conceptual change and an unprompted habit in the setting the learner values are separate claims. The present evidence establishes none of them for this move directly, and sets no dose, no threshold and no number of sessions.

  • Purposeful Reflection — considering the opposite is a specific, high-leverage reflection prompt applied at the moment of judgment
  • Explaining Their Thinking — self-explanation and counter-consideration both force generative processing of one's own reasoning
  • Pairing Non-Examples With Examples — non-examples externalize the "opposite" case, supporting discrimination between correct and incorrect applications
  • Cognitive Disequilibrium — deliberately engineered contradiction is the motivational engine this technique exploits

Examples

Illustrative

"Consider the opposite" prompts in clinical education — Medical and nursing educators ask trainees who have committed to a diagnosis to list three reasons the patient might have a different condition before finalizing the chart. This mirrors the original Lord et al. (1984) manipulation and is a standard element of diagnostic-error curricula.

Devil's advocate / red-team structures — In design critiques, debate formats, and intelligence analysis (e.g., the structured analytic techniques taught in the U.S. intelligence community's Tradecraft 101 tradition), one participant or team is formally assigned to argue against the leading proposal, institutionalizing the opposite-consideration step.

Prediction journals with confidence ratings — Learners record a prediction with a confidence level, then write the strongest case against it before checking the outcome. Used in statistics and research-methods courses to build calibrated judgment.

Misconception confrontations in science class — After students commit to an answer on a concept inventory item (e.g., "heavier objects fall faster"), the teacher asks them to explain how someone who believed the opposite would justify it, then confronts both accounts with evidence — a pattern central to physics education research curricula such as Tutorials in Introductory Physics (University of Washington).

Counterargument drafting in writing instruction — Before finalizing a persuasive essay, students must draft the strongest objection to their own thesis and respond to it, a routine embedded in many argumentation frameworks (e.g., the Claim–Evidence–Reasoning structures used in NGSS-aligned science writing).

Key Sources

Page metadata
Field Value
type principle
id considering-the-opposite
title Considering The Opposite
description For a learner who has committed to a judgment where the evidence could point either way, a prompt to generate specific reasons the opposite could be right, then to decide again, is expected to reduce confirmation, anchoring and overconfidence biases more than a general warning to be objective; no claim in this wiki tests it directly, and the one claim on it rests on a review that names the method without testing it.
status review
canonical True
generated by claude/unspecified on 2026-10-15
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.