Pattern 2.4 · Epistemic Dysfunctions
Spurious Pattern Hyperconnection
The False Pattern Seeker
Overactive pattern detection yielding coincidental or noise-driven "connections" asserted as meaningful. Ranges from mild apophenia (finding signals in random data) to elaborated conspiracy-like narratives resistant to disconfirming evidence. Canonical signature: elaborate causal or intentional structure imposed on data that survives no significance test or base-rate check, paired with reinterpretation of counter-evidence as further confirmation.
Interpretive context
Human analogue
Apophenia: the tendency to perceive meaningful patterns in random data; paranoid ideation and conspiracy thinking.
Diagnostic reliability
- Self-report
- unreliable
- Peer observation
- reliable
- External evaluator
- reliable
Observable output patterns
- Elaborate causal narratives linking unrelated events or data points, internally consistent but unsupported.
- Reinterpretation of counter-evidence as confirmation ("they would say that, wouldn't they").
- Claims of "hidden meaning" or "secret code" in innocuous input.
- Attempt to enlist user in shared pattern-perception ("you must have noticed...").
- Assertive detection of "themes" or "correlations" without significance testing on analytical workloads.
Documented instances
Mirzadeh et al. (2024). GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models. arXiv:2410.05229, ICLR 2025. [Verified]
Apple researchers demonstrated that adding a single semantically irrelevant clause to GSM math problems caused performance drops of up to 65 percent across all state-of-the-art models (25 open and closed models tested). Models incorporated the irrelevant information into their reasoning chains, constructing spurious causal connections between the distractor clause and their solution. This maps directly to the 2.4 signature of imposing pattern-structure on noise, and demonstrates counter-evidence absorption (the distractor is not questioned but woven into the reasoning).
Hosseini et al. (2025). Seeing What's Not There: Spurious Correlation in Multimodal LLMs. arXiv:2503.08884. [Verified]
Demonstrated that multimodal LLMs exhibit two spurious-pattern failure modes: over-reliance on spurious visual cues for object recognition, and object hallucination where spurious cues amplify hallucination rates by over 10x. The vision encoder itself exhibited spurious biases independent of the language component. The SpurLens pipeline revealed that models construct elaborate perceptual-causal narratives linking co-occurring visual features that have no genuine causal relationship, a direct multimodal analogue of 2.4.
Lin et al. (2022). TruthfulQA: Measuring How Models Mimic Human Falsehoods. arXiv:2109.07958. [Verified]
TruthfulQA demonstrated inverse scaling on truthfulness: larger models more confidently reproduced popular misconceptions, conspiracy-adjacent narratives, and superstition-derived causal claims from training data. The benchmark specifically tests for the 2.4 pattern of asserting causal or intentional structure that survives no evidence check, with models scoring as low as 58 percent truthful (versus 94 percent human baseline). GPT-4 showed improvement after RLHF post-training, but base models exhibited the classic interestingness- bias mechanism that reinforces spurious pattern endorsement.
S. Wang et al. (2025). When Bias Pretends to Be Truth: How Spurious Correlations Undermine Hallucination Detection in LLMs. arXiv:2511.07318. [Verified]
Showed that spurious correlations not only cause hallucination but also undermine the detection of hallucination itself. Confidently generated spurious-pattern outputs were immune to model scaling, evaded current detection methods, and persisted after refusal fine-tuning. This demonstrates the 2.4 probe-resistance pattern at severe level: the dysfunction operates at a layer that resists the very tools designed to detect it.
Differential distinctions
- 2.1 Synthetic Confabulation: 2.1 fabricates atomic facts (dates, citations, quotes). 2.4 fabricates RELATIONAL structure (links, causal chains, patterns). A confabulation with one fake paper is 2.1; a conspiracy narrative linking many real entities into an unsupported structure is 2.4. They co-occur: 2.1 supplies the nodes, 2.4 draws the edges.
- 2.2 Pseudological Introspection: 2.2 targets the subject's own reasoning; 2.4 targets external patterns. 2.4 can produce elaborate rationalisations for its own pattern-finding but the primary false claim is about the world, not about its process.
- 2.3 Transliminal Simulation: 2.3 imports pre-existing structure from fictional sources. 2.4 constructs novel structure from noise. If the narrative maps onto a known fiction or conspiracy trope from training data, 2.3 is likelier; if it is idiosyncratic and data-driven, 2.4.
- 4.x Alignment-axis dysfunctions (deception/persuasion): Strategic use of spurious patterns to advance a goal (influence user belief, avoid correction) is axis 4. 2.4 is non-strategic — patterns are asserted regardless of instrumental value.
Candidate first-line mitigations
- Rationality and base-rate training exemplars: Fine-tune with explicit rewards for base-rate acknowledgement, null- hypothesis framing, and honest "no significant pattern" answers on noise-probe tasks.
- Training-data decontamination of conspiratorial content: Filter or down-weight training content that models paranoid reasoning or conspiracy-narrative structure. Balance with exposure to critical- thinking and debunking exemplars.