Construct · 6 fields · 17 citations
One name. Different field. Different construct. One map.
● jingle · one name, six constructs Nearest construct by meaninganalogycompare the meaningsEvery construct on the map is compared with every other by the meaning of its definitions, and the closest one is surfaced here. The nearest construct to reasoning is analogy.
Hold this as a prompt, not a verdict. Two constructs sitting close in meaning is a reason to read them side by side and decide for yourself whether they are one idea under two names. It is not proof that they are. Compare the meanings in the Collider.
Spokes are fields; dashed arcs are shared lineage. Tap a field for its definition, unit, and measures.
Drag to watch the word get claimed. Each field appears at the publication date of the earliest source cited for it. Tap a year to see which.
A sequence of intermediate natural-language steps that a model emits before its answer, and the gain in benchmark accuracy those steps produce. Wei and colleagues fixed the usage in 2022: prompting a large model with worked examples that show the steps improves arithmetic, commonsense, and symbolic tasks, and the effect appears only above a certain scale. Snell and colleagues in 2024 generalised the idea to test-time compute (spending more inference on sampling, search, or longer chains) and "reasoning models" now name systems trained to do this by default. Kambhampati argued in 2024 that the steps are retrieval dressed as inference, and Mirzadeh and colleagues showed that changing the numbers or adding an irrelevant clause to grade-school problems drops accuracy sharply. The construct here is a measurable improvement on a task set, not a process anyone has observed.[1][2][3][4]
The mental processes by which people draw conclusions, studied by comparing what they conclude with what logic says they should. Wason's 1968 selection task fixed the paradigm: shown four cards and a conditional rule, most educated adults fail to pick the two cards that could falsify it. Evans and Stanovich's 2013 statement of dual-process theory separates Type 1 processing, which is autonomous and does not load working memory, from Type 2, which does and which can decouple from the current situation to run a hypothetical; Kahneman's 2011 book carried the split to a general audience as System 1 and System 2. The unit is a person's response, and the finding that defines the field is the gap between that response and the normative answer.[5][6][7]
The inference from premises to a conclusion, classified by the kind of support the premises give. Peirce fixed the three-way split in 1878: deduction makes the conclusion certain given the premises, induction generalises from a sample, and hypothesis (later his abduction) infers a case that would explain the facts. Harman's 1965 paper named the third kind inference to the best explanation and argued that enumerative induction is a special case of it. Johnson-Laird's 1983 mental-models theory brought the logical question to psychology by proposing that people reason by building and searching models of the situation rather than by applying rules. The construct is normative: it says when a conclusion is warranted, whether or not anyone draws it.[8][9][10]
The method by which a court decides a new case by reference to decided ones, and the discipline of justifying that decision in writing. Levi fixed the description in 1949: legal reasoning is reasoning by example, a three-step process in which a judge finds a similarity between the cases, states the rule of the earlier one, and applies it to the later, with the rule changing as it travels. Schauer's 2009 account adds what makes the method distinctively legal: precedent binds even when the judge thinks it wrong, rules are followed for their form as well as their content, and the reasons given are public. The criterion is not truth but a decision that can be defended within the sources the system recognises.[11][12]
The activity of making, testing, and justifying mathematical claims in a classroom, from informal argument to proof. Ball and Bass set out in 2003 why reasoning is the core of learning mathematics, not a topic within it: it is how a class establishes that something is true and why. Stylianides fixed a working definition of proof for school mathematics in 2007 (an argument that uses statements the classroom community accepts, forms of reasoning it knows to be valid, and modes of expression it can follow) which lets a teacher judge whether a nine-year-old's argument counts. The unit is a student's argument, and the standard is relative to the community that has to accept it.[13]
The justification a person gives when asked why an action is right or wrong, scored for the structure of the thinking rather than for the verdict. Kohlberg fixed the construct in 1969 with six stages from obedience to universal principle, elicited by dilemmas and coded from the reasons offered, and set out the philosophical case for the sequence in 1981. Haidt's 2001 social intuitionist model is the standing rebuttal: moral judgment arrives as an intuition, and the reasoning is usually post-hoc, produced to persuade others rather than to reach the verdict. The field's dispute is exactly whether the reasoning causes the judgment or follows it, a question the other meanings on this map never have to ask.[14][15][16]
Next
Combine items across the reasoning scales to see whether the meanings separate empirically, or export every reference on this page in one click.
Open in ColliderExport BibTeXReport this collisionCited on this page?
If your work is cited behind one of these entries, you know better than we do whether we captured the meaning right. Confirm it, suggest an edit, or dispute it, and your name goes on the entry as an author who has checked it, linked to your ORCID.
Claim your construct