Post-training influences introspective capability expression

Different post-training strategies substantially influence introspection task performance; 'helpful-only' variants show higher false positives but some achieve strong net performance.

Source paper

extracted_from

Emergent Introspective Awareness in Large Language Models

(2026) · Lindsey, Jack

Neighborhood — ranked by edge-count

Findings (1)

finding

Post-training is key to eliciting introspective awareness
supports
Base pretrained models show high false positive rates and achieve no net task performance on concept injection detection; post-training essential for introspection.

Communities (4)

community

Mechanistic interpretability & model evaluation
members_of
Spans attention head decomposition, benchmark awareness, and genomic pathogenicity prediction via neural models.
Mechanistic introspection in language models
members_of
Empirical investigation of how LMs access and report internal states across layers, using concept injection and thought detection on Claude models.
LLM functional introspective awareness
members_of
Empirical probing of language models' ability to detect and report their own internal concept representations
Post-training emergence of model introspection
members_of
How instruction tuning and RLHF elicit latent introspective capabilities in language models beyond base pretraining.

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Post-training strategies can strongly influence performance on introspective tasksclaim0.880
Assertion about the role of post-training in eliciting introspection.
Post-training is key to eliciting strong introspective awareness; base pretrained models do not show above-chance detectionclaim0.873
Finding that base models have high false positives and no net positive performance.
Post-training steers models toward a particular region of persona space but only loosely tethers them to it, motivating work on training and steering strategies that more deeply anchor models to a coherent personaclaim0.798
Central interpretive claim and motivation for future work
Introspective capabilities may continue to develop with further improvements to model capabilitiesclaim0.796
Forward-looking statement about future models.
Are there examples of models recognizing their introspective capability and then suppressing it?question0.784
Cube Flipper's question prompted by the idea that supernormal capabilities might be hidden.
Introspective awareness correlates with overall model capabilityclaim0.781
Most capable models (Opus 4, 4.1) show greatest introspective awareness; trend suggests introspection aided by improvements in model intelligence.
Introspective signals appear in middle layers but are suppressed by later post-training-shaped layers.finding0.780
Mechanistic finding by Lindsey (2026) explaining how contemplative prompt may work: enables mid-layer introspection to reach output.
Post-training elicits rather than installs personas; the persona direction is already laid down in pretrainingclaim0.771
Authors' interpretive endorsement of PSM view, backed by transfer experiments