Why Six Dimensions? On the Overlapping Nature of Reasoning Errors
When we first began building the taxonomy, the obvious approach was a single flat list. Logical fallacies, propaganda techniques, cognitive biases — one giant catalog, alphabetically sorted. Simple. Easy to search. And, it turned out, deeply misleading about how these things actually operate.
The overlap problem
Consider appeal to fear. It's a logical fallacy — the argument structure is invalid because threat of harm doesn't establish truth. But it's also a manipulation technique, because skilled communicators use it deliberately to bypass rational evaluation. And it exploits a cognitive bias — the availability heuristic makes vivid threats feel more probable than statistics suggest.
So where does "appeal to fear" belong? In the logical fallacies list? The propaganda techniques list? The cognitive biases list? The honest answer: all three. A flat taxonomy forces an artificial choice that distorts the concept.
What the dimensions capture
The six dimensions aren't categories you assign an aspect to — they're lenses you apply to analyze a communication event:
- Logical Fallacies ask: is the argument structure valid? Does the conclusion actually follow from the premises?
- Manipulation & Propaganda ask: is this technique deliberately deployed to produce belief without evidence?
- Cognitive Biases ask: which systematic error in human cognition is being triggered or exploited?
- Statistical & Methodological Errors ask: are the numbers, studies, or evidence being used correctly?
- Argumentation Schemes ask: what is the underlying structural pattern of this argument, and is it being used appropriately?
- Discourse Mechanics ask: how is the conversational context — framing, tone, turn-taking — shaping what conclusions seem available?
A single piece of political rhetoric might activate all six simultaneously. The dimensions help you see which is the dominant mechanism in a particular case — and which others are contributing.
Why this matters for detection
A skilled debater who knows only the logical fallacies list will catch invalid argument structures — but will miss the emotional manipulation that made the invalid structure feel convincing. A media critic who focuses only on propaganda techniques will identify the intentional moves — but might not notice the statistical sleight of hand buried in the supporting evidence.
Real-world persuasion is multi-layered. The taxonomy needs to be multi-layered too.
The connections between aspects
Beyond the six dimensions, each aspect in the taxonomy links to related aspects — sometimes within the same dimension, often across dimensions. Scapegoating relates to hasty generalization (logical), emotional appeal (manipulation), and outgroup homogeneity bias (cognitive). Following these links reveals the full structure of a manipulation strategy.
The Knowledge Graph visualization makes these cross-dimension connections visible in a way a list never could.
The practical implication
When TellDear analyzes a text, it doesn't ask "which category does this belong to?" It asks "which aspects are active here, across all dimensions?" A result that flags a cognitive bias and a propaganda technique and a statistical error in the same passage isn't indicating confusion — it's indicating that the text is doing several things at once. That's usually the point.