Glossary

An argument map displays the logical skeleton of a text: Claims (main conclusions), Premises (supporting statements), Counters (opposing points), and Rhetorical elements (persuasion techniques). Edges show how parts relate: Supports (reinforces), Rebuts (contradicts), Qualifies (limits scope), or Annotates (adds context). This visualization helps identify weak points, hidden assumptions, and the overall strength of an argument at a glance.

A question attributed to the Roman consul Lucius Cassius. It doesn't prove a claim is wrong just because someone benefits from it — but it reveals potential conflicts of interest and hidden motivations. When a pharmaceutical company funds a study showing their drug works, or when a politician makes claims that conveniently justify their policy, asking "cui bono?" is the first step toward understanding the full picture.

If your argument can't defeat the best version of the opposing view, it can't defeat the opposing view at all. Steelmanning forces intellectual honesty: before criticizing, you reconstruct the other side's position in its most charitable, most defensible form. Only then do you respond. It's harder than it sounds — our natural instinct is to attack the weakest version (straw man).

Argumentation schemes are templates for acceptable reasoning. Unlike formal logic, they capture how people actually argue: "Expert X says Y, so Y is probably true" (appeal to expert opinion), or "A is similar to B in relevant ways, so what's true of A is true of B" (argument from analogy). Douglas Walton cataloged 96 such schemes, each with critical questions that reveal when the pattern breaks down. A scheme isn't automatically fallacious — it becomes one when the critical questions expose a flaw. TellDear's d5 dimension covers 32 of the most important schemes.

Developed by Frans van Eemeren and Rob Grootendorst at the University of Amsterdam, pragma-dialectics defines 10 rules for an ideal "critical discussion" — and every violation of these rules corresponds to a specific fallacy. For example, Rule 1 ("parties must not prevent each other from advancing standpoints") is violated by ad hominem attacks. Rule 8 ("arguments must be logically valid") catches formal fallacies. This framework connects fallacy theory directly to discourse norms.

Every argument has at least one premise and a conclusion. "All humans are mortal" is a premise. "Socrates is human" is another premise. "Therefore Socrates is mortal" is the conclusion. In real-world arguments, premises are often unstated (hidden assumptions) — and that's where many reasoning errors hide.

Stephen Toulmin proposed this model in 1958 as an alternative to formal logic for analyzing everyday arguments. Claim: what you're arguing. Data: the evidence supporting it. Warrant: the logical bridge connecting data to claim (often unstated). Backing: support for the warrant itself. Qualifier: the degree of certainty ("probably", "in most cases"). Rebuttal: conditions under which the claim doesn't hold. Most real-world arguments leave the warrant and qualifier implicit — Toulmin analysis makes them explicit.

Cognitive biases are not random mistakes. They are predictable, systematic errors built into how our brains process information. Evolution gave us fast heuristics that work well enough most of the time — but they fail in predictable ways. The anchoring effect makes us cling to the first number we hear. Confirmation bias makes us seek evidence that supports what we already believe. The Dunning-Kruger effect makes beginners overestimate their competence. There are over 180 documented cognitive biases. TellDear catalogs 115 of the most relevant ones in dimension d3.

Critical thinking isn't about being negative or skeptical of everything — it's about being appropriately skeptical. It means asking: What's the evidence? Is the reasoning valid? Are there alternative explanations? Who benefits from this claim? TellDear exists to make these skills more accessible by breaking them down into 452 concrete, learnable patterns across 6 dimensions.

TellDear organizes its 452 reasoning aspects into 6 dimensions: Logical Fallacies (d1, red) — formal and informal errors in reasoning; Manipulation & Propaganda (d2, orange) — deliberate persuasion techniques; Cognitive Biases (d3, blue) — systematic thinking errors; Statistical Errors (d4, purple) — misuse of data and numbers; Argumentation Schemes (d5, green) — common but defeasible argument patterns; Discourse Mechanics (d6, pink) — structural features of debate and dialogue. Each dimension has its own color throughout the platform.

Fallacies come in two flavors. Formal fallacies have an invalid logical structure — the conclusion doesn't follow from the premises regardless of content (e.g., affirming the consequent: "if it rains the street is wet; the street is wet; therefore it rained"). Informal fallacies have a valid-looking structure but rely on irrelevant premises, ambiguity, or psychological tricks (e.g., ad hominem: attacking the person instead of the argument). TellDear covers 95 fallacies in dimension d1.

"All cats are animals. My dog is an animal. Therefore my dog is a cat." Both premises are true, but the conclusion is nonsense. That's a logical fallacy — the structure is broken. Some fallacies are this obvious; most are not. They hide behind eloquent language, emotional appeals, or sheer complexity. TellDear's dimension d1 covers 95 logical fallacies, both formal (structural errors) and informal (content-based errors).

Media literacy goes beyond just reading and understanding text. It means asking: Who created this? What's their purpose? What techniques are they using? What's left out? In an age of algorithmic feeds, deepfakes, and AI-generated content, these questions matter more than ever. TellDear is, at its core, a media literacy tool — it helps you see the machinery behind the message.

FOL (First-Order Logic) uses quantifiers (∀ = for all, ∃ = there exists), connectives (∧ = and, ∨ = or, ⇒ = implies, ¬ = not), and predicates to capture the essential form of a reasoning pattern. For example, the Ad Hominem fallacy: Person(x) ∧ HasFlaw(x) ⇒ Invalid(Claim(x)). These patterns allow automated verification of logical validity.

Formal verification uses an SMT (Satisfiability Modulo Theories) solver — specifically Z3 — to mathematically check whether an argument's logical structure is valid. Each reasoning pattern is translated into First-Order Logic and tested: Can the premises be true while the conclusion is false? If yes, it's formally invalid. If no, it's formally valid. Many real-world patterns (analogies, heuristics) cannot be fully captured in formal logic — these are marked as not formally decidable, which doesn't mean they're wrong.

Quick (~30s, ~$0.02) is a free-form scan that detects obvious patterns without using the full taxonomy. Standard (~2 min, ~$0.10) uses all 452 aspects and returns the top 15 findings with verification steps. Thorough (~5 min, ~$0.25) performs an exhaustive analysis across every aspect and provides detailed findings with examples and counter-strategies. Higher depth means more tokens, longer duration, and higher cost — but also more comprehensive results.

Each of TellDear's 452 aspects represents a distinct reasoning pattern — from formal logical fallacies (ad hominem, begging the question) to cognitive biases (anchoring, confirmation bias) to propaganda techniques (loaded language, false dilemma). Every aspect has a definition, real-world examples, verification steps for AI detection, a formal logic pattern, and links to related aspects.

Relations connect aspects in the knowledge graph: is_a (classification), sub_type_of (specialization), triggers (one pattern often leads to another), correlates_with (patterns that frequently co-occur). Understanding these connections helps recognize how reasoning errors cluster and reinforce each other — for instance, how confirmation bias often triggers cherry picking, which in turn enables hasty generalization.

The Deep Dive section provides in-depth information about each aspect: a real-world example showing the pattern in action, an explanation of why it works psychologically, practical advice on how to counter it, alternative names, and links to related aspects.

The knowledge graph maps relationships between all 452 reasoning aspects. Nodes represent aspects, colored by dimension. Edges show relationships like 'is a subtype of', 'triggers', or 'correlates with'. Available in 2D and 3D views. It helps discover how reasoning patterns interact — how a cognitive bias might trigger a logical fallacy, or how propaganda exploits known biases.

Like biological taxonomy classifies organisms into kingdoms, phyla, and species, TellDear's taxonomy classifies reasoning patterns into dimensions (6 major types), categories (e.g., formal vs. informal fallacies), and individual aspects (452 specific patterns). This hierarchy helps navigate the complex landscape of reasoning errors and makes systematic analysis possible.

Each of the 452 aspects has verification steps — simple yes/no questions designed to systematically detect whether a pattern appears in a text. For ad hominem: "Does the argument attack a person rather than their claim?" For false dichotomy: "Are only two options presented when more exist?" This ensures consistent, reproducible analysis.