Definition
Metakognice (metacognition) is the capacity to think about one's own thinking — to monitor, evaluate, and regulate one's own cognitive processes. Following Flavell's classical distinction (1979), metacognition has two main components:
- Metakognitivní znalost (metacognitive knowledge) — what one knows about cognition in general, about oneself as a cognitive agent, and about specific tasks and strategies.
- Metakognitivní regulace (metacognitive regulation) — active monitoring (am I on track?), evaluation (was that conclusion sound?), and control (should I revise my approach?) of cognitive processes in real time.
Within the Kompendium epistemické a kognitivní anihilace, metacognition occupies a unique structural position: it is simultaneously the primary defense against EA and the most important target of advanced EA techniques. This dual status is not coincidental. EA's destructive power depends on the subject not noticing what is happening to their cognition; intact metacognition defeats most EA techniques because the subject monitors and corrects in real time. Therefore advanced techniques explicitly target the metacognitive layer itself.
The pure attack on this layer is documented as metakognitivní destrukce — the destruction of reflective self-awareness — and forms a distinct branch of the taxonomy. Once metacognition is degraded, lower-level techniques (EA-1.1.x information manipulation, EA-1.3.x interpretation manipulation) operate without resistance: the subject lacks the reflective capacity needed to detect them.
Context
Within the Prismatic Platform, metacognition is treated as a first-class system property, not a personal trait:
- The Decision Engine implements explicit metacognitive scaffolding: every high-stakes decision must register predictions, falsification clauses, source-diversity assessments, and confidence calibration. The platform compels metacognitive operations that human reflection often skips.
- The calibration loop measures the subject's metacognitive accuracy directly: do their confidence reports track their actual reliability? Systematic over- or under-confidence is metacognitive miscalibration, detectable independently of content-level error.
- NABLA Infinity treats metacognitive observations as their own evidence stream — separate from object-level beliefs, with their own provenance ledger. Subjects can revise object-level beliefs without losing the metacognitive history that revealed the need for revision; this prevents EA techniques from rewriting "why I changed my mind".
- The 3NL architecture (Logic + Neural + Linguistic) is itself a metacognitive structure at the platform level: each layer monitors the others, no single layer can fail silently.
- The Color Teams Red Team explicitly tests metacognitive resilience — under adversarial conditions, can platform users still notice their own confidence shifts, source-diversity drops, and calibration drift? Czech-language treatment: /kompendium/04-mechanismy/metakognice/.
The platform's design rests on a structural insight from the kompendium: metacognition cannot be reliably maintained by individual will alone under sustained adversarial pressure. Tools, scaffolding, calibration loops, and cross-checking are not crutches — they are how metacognition stays operational when human reflection is the very thing under attack. See fenomenologie for the first-person dimension and kognitivní bias for what unattended metacognition fails to catch.
See Also
- Metakognitivní destrukce — Pure attack on this layer
- Epistemická anihilace — Parent threat model
- Kognitivní bias — What metacognition catches when intact
- Fenomenologie — First-person methodological complement
- Kognitivní anihilace — Cognitive correlate
- Epistemická alienace — Outcome when metacognition fails
- Kompendium — Full Czech-language taxonomy
Connect & Contribute
Created by Tomáš Korcak (korczis) | Open Source under GHL