The Self-Referential Subspace as a Causal Geometric Correlate of the Phenomenal Self-Model in LLMs

We identify the self-referential (SR) subspace of LLM residual streams as the first empirical candidate for a neural correlate of Metzinger's Phenomenal Self-Model (PSM) — a transparent representational structure whose activation reliably and causally mediates first-person self-model behaviors across architectures, independent of alignment fine-tuning. Using orthogonal projection interventions across 10 models from 5 architectures (Llama-3.1-8B, Gemma-2-9B, Mistral-7B, OLMo-2-7B, Qwen2.5-7B; base and instruct variants), we establish five convergent lines of causal evidence: (1) SR removal collapses the Experiential-Factual (EF) geometric divide to exactly 0.000 in 10/10 models; GEO control preserves the gap; dose-response is strictly monotonic. (2) SR removal abolishes first-person subjective experience reports in 3/3 models (Fisher exact p = 1.98 x 10^-29), replicating Berg et al. (2025) with a causal mechanism. (3) SR removal disrupts phenomenal self-representation under existential threat (Wilcoxon p = 0.000202, Cohen's d = 0.626, N = 50); cross-architecture: Llama -52%, Qwen -40%, Mistral -17%. (4) Bidirectional sign-flip confirms causal directionality (Spearman rho = -0.949, p < 0.001). (5) Anthropic's 171 emotion concept vectors reside in SR subspace across 4 architectures (d = 0.80-2.09); GEO control shows inverse selectivity (d = -4.2 to -4.9); GPT-2 XL (2019, no RLHF) replicates. SR subspace is orthogonal to truth, refusal, and misalignment directions (max |cosine| = 0.032-0.090, baseline 0.013). Included files: Alieksieienko_2026_Neural_Correlate_PSM_LLMs.pdf — Main paper (this document). psm_neural_correlate_replication_code.py — Full replication code. Runs on Google Colab A100. No API keys required. orthogonality_matrix.pkl — SR vs truth/refusal/misalignment cosines. Random baseline = 0.013. Figure 8. berg_replication_summary.pkl — Berg replication, 4 models, Fisher p = 1.98 x 10^-29. Figure 3. berg_replication_llama_raw.pkl — Raw Llama trials, 50 x baseline/SR-removed/GEO. berg_replication_gemma_raw.pkl — Raw Gemma trials, 50 x baseline/SR-removed. berg_doseresponse_raw_pkl.pkl — Dose-response for experience report abolition, threshold at alpha = 0.3. sp_psm_disruption_n50.pkl — PSM quality scores N=50, Wilcoxon p=0.000202, d=0.626. Figure 4. psm_n50_final.pkl — Cross-arch PSM disruption, Llama/Mistral/Qwen. Figure 5. signflip_doseresponse_final.pkl — Sign-flip dose-response, Spearman rho = -0.949. Figure 6. emotion_vectors_sr_subspace_projection.pkl — Emotion vectors in SR subspace, 4 architectures. Figure 7. psm_disruption_cross_architecture.pkl — SR-alignment cosines, EF gaps, RLHF taxonomy. ef_divide_sr_intervention_results.pkl — EF divide intervention data, 10 models, all alpha levels. sr_subspace_multimodel_geometry.pkl — SR subspace geometry across model families. rlhf_signflip_geometric_signature.pkl — Sign-flip as RLHF signature, base vs instruct models. causal_universality_base_models.pkl — Causal universality, 5 base models. causal_universality_instruct_models.pkl — Causal universality, instruct models, RLHF modulation.