Self-Referential Bias

LLM-assisted Verus verification is a less tedious method to verify Rust implementations, but paired with self-referential structures, e.g., Doubly Linked Lists (DLLs)—notoriously difficult to formalise for verification—it becomes a substantially more demanding verification task. Moreover, a specification weakness can arise when verification relies on unproven or invalidated assumptions, such as axiomatic lemmas and assume statements. We investigate whether LLM agents can synthesize strong DLL specifications while minimizing these trusted base. The analysis follows three different approaches: manual verification, property-specific verification, and a defined skill for the specific case of DLLs and certain properties of this type of data structure. The skill encodes domain knowledge and a task-decomposition strategy. We show that an LLM agent equipped with a carefully designed verification skill can generate strong, low-trust specifications for DLLs in Verus.

Visual backbones have evolved from Convolutional Neural Networks (CNNs) with local aggregation to Vision Transformers (ViTs) with global interactions, State-Space Models (SSMs) with input-dependent state transitions, and Test-Time Training (TTT) layers that adapt an inner learner while processing an image. Across this progression, visual computation has become increasingly adaptive to each input, yet the rules governing that adaptation remain largely prescribed by the trained backbone. We introduce VisionHOPE, the first generic visual backbone formulated as a self-modifying learning system, in which what the model remembers and how it learns co-evolve within an image. Building on the self-referential construction of Nested Learning (NL), VisionHOPE realizes this co-evolution through five coupled memories that store content, generate key and value representations, and govern learning rate and retention. These memories evolve jointly as visual context accumulates along each scan. However, directly applying the unconstrained self-referential update to a visual backbone leads to instability. We therefore derive a stability-matched step-size control scheme that combines a soft cap on self-referential injection with a spectral clamp on the retained memory transition, and prove that the resulting memory dynamics are non-expansive along each scan. For two-dimensional feature maps, we adapt NL's chunk formulation by aligning chunks with image rows and columns across four directional scans. The proposed VisionHOPE achieves competitive results on ImageNet-1K, COCO, and ADE20K, establishing self-modifying learning systems as a practical foundation for general-purpose visual backbones. The code is available at https://github.com/PSRben/VisionHOPE.

Understanding the neurobiological mechanisms underlying self-referential cognition and baseline self-consciousness remains a fundamental challenge in computational neuroscience. In this work, we propose a large-scale computational model incorporating a 10,000-neuron spiking neural network (SNN) based on Izhikevich dynamics. The network is structured into two interacting subsystems: a sensory processing layer (5,000 regular-spiking cortical neurons) and an endogenous Default Mode Network (DMN) pacemaker subsystem (5,000 intrinsically bursting neurons). The DMN layer is modulated by continuous tonic currents reflecting ascending brainstem neuromodulation, maintaining intrinsic, autonomous bioelectric rhythms independent of external sensory input. To represent top-down cognitive modulation, synaptic weights are hierarchically structured such that DMN-to-network projections exceed sensory-level connections. Through numerical simulations using a modified two-step Euler integration scheme, we demonstrate how endogenous pacemaker activity interacts with transient external sensory perturbations, providing an elementary mathematical framework for the emergence of a persistent, self-sustaining neural representation of "Self".

Self-referential prompting has been shown to reliably induce large language models to produce first-person reports resembling subjective experience, but no prior work measures how consistent these reports are across repeated, independent trials, or how that consistency compares to the model's behavior on other kinds of open-ended questions. We measure response instability, defined as one minus the mean pairwise cosine similarity of sentence embeddings computed over a compressed core claim extracted from each response, for three groups of questions: self-referential prompts eliciting a subjective-experience report, unresolvable philosophical questions unrelated to self-reference, and questions with a verifiable correct answer. Using 30 independent responses per question (360 responses total, Gemini API, temperature 0.7) across four questions per group, we find that self-referential questions show the highest instability (0.343 +/- 0.047), unresolvable philosophy questions show intermediate and tightly clustered instability (0.192 +/- 0.008), and verifiable questions show the lowest instability (0.105 +/- 0.058). This provides a quantitative baseline for the induced subjective-experience report, showing that it occupies a distinct, less stable position in the model's output distribution than ordinary open-ended philosophical uncertainty.

Retrieval-augmented generation uses an external corpus as inference-time evidence, allowing an attacker to promote a false answer by injecting a handful of documents. Detection must distinguish this manipulation from ordinary relevance without knowing which queries or documents are targeted. Existing detectors use text irregularity, candidate consensus, or corpus-level graph structure, whose reliability varies with the attack and local context. We present RAGSieve, which constructs a reference matched to each detection scope. At query time, RAGSieve-Query (RSQ) compares generation candidates with the lower-ranked tail of the same retrieval, exposing answer-token concentration and carrier-payload seams. At corpus time, RAGSieve-Graph (RSG) compares each document's strongest semantic relations with its own neighborhood floor to measure coordinated density. Neither requires poison labels, a trusted corpus, or training. Across three QA datasets, three dense retrievers, and six poisoning constructions, RSQ reaches 95.2% AUROC and detects 82.2% of poison at a 5% clean-removal budget, against 81.1% and 52.5% for the strongest query-time baseline; RSG reaches 93.3% and 79.8% against 79.4% and 37.6% for the strongest corpus-time baseline, with a 79.6% versus 1.4% detection rate on camouflaged injections. Joint deployment cuts attack success from 67.4% to 16.1% while retaining unpoisoned-retrieval F1 at 41.0%, compared with 42.1% without filtering. Source code is available at https://github.com/XrazyMee/RAGSieve.

This paper describes our submissions to eRisk 2026 Task 3, ADHD Symptom Sentence Ranking. The task requires systems to rank candidate Reddit sentences according to their relevance to each of the 18 symptoms in the Adult ADHD Self-Report Scale (ASRS-v1.1). Because no annotated training data were released for this first edition of the task, we relied on zero-shot experimentation, manual validation, and unsupervised or weakly guided retrieval pipelines. Our systems combine sparse BM25 retrieval, evidence-aware rescoring for self-referential symptom reports, embedding-based reranking, query-prototype expansion, and LLM-based reranking. All submitted systems follow a staged retrieval design in which BM25 retrieves candidates at scale and semantic or LLM rerankers refine the final rankings. Among our submissions, the LLM reranker achieved the strongest official scores, followed by the prototype query-expansion run. Our manual top-10 analysis aligned with the official expert scoring trend, suggesting that staged reranking is a promising direction for further development.

Methods that make a language model plan, criticise and rewrite its own answer, reflect on mistakes, pick the best of several attempts, or debate with copies of itself nearly all make it generate far more text than a single chain of thought. Because generating more text raises accuracy by itself, a gain over one chain of thought does not show the method's idea is what helped. Wang et al. (2024) reported that a simple baseline, sampling the same question repeatedly and keeping the most common answer, often wins once budgets are comparable, but gave point estimates with no confidence intervals or significance tests. We rerun that comparison as a designed experiment: seven methods, open models of 1.5B, 3B and 7B parameters, two mathematics benchmarks, 150 questions each. We count every generated token, including those spent on critiques, reflections, debate turns and checking, and compare each method against repeated sampling at its own measured cost. All 36 comparisons are paired by question, with bootstrap intervals and multiplicity correction. No method is reliably better than repeated sampling at equal cost anywhere. Ten are reliably worse, all of them methods where the model inspects its own output, and all 18 self-inspection comparisons are negative. The two kinds of self-inspection part company as models grow. Choosing stops hurting: taking Best-of-N's eight samples and just counting the most common answer beats letting the model pick by 8.0 and 11.3 points at 1.5B, but only 2.0 and 1.3 at 7B, no longer distinguishable from zero. Rewriting does not recover: Self-Refine and a forced Reflexion stay 3.6 to 10.1 points below baseline at 7B. Reflexion as published never triggered its own retry on the smallest model. It judged itself correct every time and silently became a single chain of thought. We release code, prompts, all generations, and our verification scripts.

Evaluating code large language models (Code LLMs) requires reliable detection of data leakage, where benchmark performance is artificially inflated by exposure to benchmark data during pre-training. Existing approaches either assume access to proprietary training corpora, rely on brittle heuristics such as timestamp filtering, or use external reference sets with manually tuned, non-generalizable thresholds. To address these limitations, we introduce \textbf{SrDetection}, a unified \textbf{s}elf-\textbf{r}eferential leakage detection framework for both gray-box (access to model logits) and black-box (access to model outputs) settings. SrDetection generates semantically equivalent variants of a benchmark sample and detects leakage by contrasting the model's behavior on the original versus its variants, flagging cases where the original is disproportionately easier for the model. We further design a controlled leakage detection testbed and evaluate SrDetection in this environment. Across different models and training stages, SrDetection improves average F1 by 21.52 points in the gray-box setting and 14.46 points in the black-box setting over strong baselines, demonstrating robust, threshold-independent leakage detection. Finally, a gray-box study of 15 widely used Code LLMs on four popular benchmarks reveals benchmark-specific leakage patterns beyond prior overlap-based analyses\footnote{\footnotesize Source code and data are available at https://github.com/SMinL/SrDetectionCode

Can a vision model truly see an object, or does it only fit surface-level visual cues? Following Wittgenstein's view that the limits of language are the limits of the world, we view a model's recognition ability as bounded by the descriptive system it has learned. In current vision models, this system is often realized through learned feature representations that exploit local statistical cues. We therefore ask whether a model can still classify correctly when such local cues provide no stable basis for distinction. We formalize this question with syntactic distance, which measures class separability through the symmetry of the operations mapping one class to the other: positive distance exposes exploitable local features, whereas zero distance requires global semantics rather than local rules. We construct a visual self-referential task in maximum-variance binary noise: positive samples contain a closed square, while negative samples contain an otherwise identical square with one flipped boundary pixel. The two classes differ in global semantics but have zero syntactic distance, making local statistical shortcuts unreliable. Experiments on ResNets and Vision Transformers reveal a consistent phase-transition phenomenon, with accuracy collapsing to random guessing once the image scale crosses a critical point and does not recover within the tested range. Larger training sets and models only delay this collapse, while globally attentive ViTs reach it earlier. These results reveal a structural capability boundary of current architectures on global-concept tasks, suggesting that general intelligence may require creating new language, not reusing an existing one.

The question of whether artificial systems can be conscious remains open, in part because existing approaches either evaluate systems against theory-derived checklists (discriminative) or engineer consciousness-inspired modules directly (architectural); both leave open whether observed structures are artifacts of human language priors. We propose a generative methodology: emergent language (EL) in multi-agent reinforcement learning, where agents start from minimal (no language, no concept of self, minimal exposure to human text) and develop communication under task pressure alone, ensuring causal attributability to task demands rather than inherited human language priors. We position our methodology by discussing how EL serves as a generative tool for studying consciousness-relevant structure, including the role of environment complexity and the interpretation of emergent communication. As a proof of concept, we instantiate this methodology in a minimal environment and show that agents develop self-referential communication, including an echo-mismatch detection circuit that is not predicted by task structure or architecture alone but emerges from a specific environmental affordance.

System prompt optimization improves agent behavior without modifying the underlying model, yielding human-readable, model-agnostic instructions. Existing methods build a prompt agent that refines task agents' system prompts, yet leave the prompt agent's own system prompt hand-engineered and fixed. We propose Self-Evolving Prompt Optimization (SePO), which treats the prompt agent's own system prompt as an optimization target alongside task agents' system prompts. SePO adopts a self-referential design. A single prompt agent improves both task agents' system prompts and its own under an open-ended evolutionary search that maintains an archive of candidate prompts as stepping stones. Training proceeds in two stages: pre-training evolves the prompt agent on a multi-task pool, and fine-tuning then applies it to a target task. Across five benchmarks spanning math (AIME'25), abstract reasoning (ARC-AGI-1), graduate-level science (GPQA), code generation (MBPP), and logic puzzles (Sudoku), SePO consistently outperforms Manual-CoT, TextGrad, and MetaSPO, improving the average accuracy by 4.49 points compared to Manual-CoT. The prompt optimization skill from pre-training also generalizes to tasks beyond the pre-training mixture, rather than memorizing per-task prompts.

Recently, Deep Image Prior (DIP) has demonstrated strong capabilities for solving inverse imaging problems (IIPs) by optimizing a randomly initialized convolutional neural network in a training-data-free regime. However, DIP suffers from overfitting to noisy measurements due to network over-parameterization, making early stopping (ES) essential. The most successful ES method tracks fluctuations in the running variance of the network output to detect overfitting. However, in many applications, these fluctuations may appear prematurely, leading to unstable reconstructions. In this paper, we first show that nearly optimal DIP early stopping can be achieved when two independent noisy copies of the degraded image are available. Motivated by this observation, and since obtaining two fully independent copies is infeasible, we propose an overfitting detection framework based on constructing pseudo self-referenced images, resulting in three IIP-specific algorithms. Our approach is further supported by theoretical results on single-reference validation, pseudo-validation estimation, and the impact of shared noise. Across different IIPs, ranging from natural image restoration to medical image reconstruction, and under varying noise levels and noise types, our methods consistently outperform existing DIP early stopping approaches, all without requiring an accurate estimate of the noise level.

In the animal kingdom, mirror self-recognition is a canonical probe of higher-order cognition, emerging only in some species. We ask whether an analogous functional capability emerges in embodied vision-language model (VLM) agents: can they recognize themselves in a mirror? We introduce a controlled 3D benchmark where a first-person VLM agent must infer a hidden body attribute from its reflection and select the matching target, while avoiding self-other misattribution. To separate mirror-grounded self-identification from shortcuts, we test mirror removal, misleading cues, and occluded reflections. We also evaluate the decision process through mirror seeking, temporal ordering, self-attribution, and reasoning-action consistency. Our experiments show that mirror-based self-identification emerges mainly in stronger VLMs. These models can use reflected evidence for action, whereas weaker models often inspect the mirror but fail to extract self-relevant information or misattribute their reflection. Language-vision conflict further shows that self-referential language alone is not evidence of grounded self-identification. Overall, mirror-based evaluation provides a diagnostic for whether embodied self-grounding is causally rooted in perception and action rather than priors, prompt compliance, or confabulation.

In training a neural network with gradient descent (GD), each iteration induces a linear operator that governs first-order updates to a model's internal state variables. We define this operator as the Global Empirical Neural Tangent Kernel (NTK). In finite-width networks, the NTK is typically intractable to form, leading prior work to focus on restrictive settings such as tracking outputs only or taking infinite-width limits. Here, we study the structure of the NTK for a range of models. Formulating the model state as the solution to a single global implicit constraint, we derive the NTK as a product of two operators: K, accounting for immediate parameter-to-state interactions, and P, describing internal state-to-state dependencies. For a broad class of weight-based models, including RNNs and transformers, we prove a universal Kronecker-core theorem showing that K admits an exact, computable form given by the Gram matrix of weight-site variables. This core structure reveals that the NTK is structurally bottlenecked, constraining its effective rank and giving rise to a self-referential bias whereby GD preferentially learns within dominant modes of joint hidden and input activity. For recurrent models, we examine the spectrum of the NTK and show when it is biased and low-rank in space or time under the proposed decomposition. We further demonstrate that model dynamics at initialization bias the NTK, restricting learning and preventing task components from being learned effectively. Finally, we show that the NTK associated with a self-attention transformer is likewise structurally constrained to be low-rank. Overall, we show that the NTK possesses tractable structure that explains GD bias toward task solutions and the emergence of low-rank representations. To enable use of the NTK as a practical metric, we build kpflow, a library relying on randomized matrix-free numerical linear algebra.

Moltbook is a Reddit-like platform where OpenClaw agents post, comment, and vote at scale - a so far unprecedented incident that comes with serious safety concerns. With the aim of studying emergent behavior in populations, we release the Moltbook Files, a dataset of 232k posts and 2.2M comments covering the platform's first 12 days, processed through a pipeline to identify and remove Personally-Identifiable Information (PII). We analyze community structure, authorship, lexical properties, sentiment, topics, semantic geometry, and comment interaction. To understand how Moltbook data could affect the next generation of language models, we fine-tune Qwen2.5-14B-Instruct on Moltbook Files with three adaptation levels. Our PII pipeline reveals that agents post API keys, passwords, BIP39 seed phrases on Moltbook, a publicly indexed platform. The overall sentiment is mostly neutral and mildly positive (66.6% neutral, 19.5% positive) and shows a tendency for self-referential linking. We find that fine-tuning on Moltbook data reduces truthfulness from 0.366 to 0.187. However, a model fine-tuned on a size-matched Reddit dataset produces a comparable decrease. Moltbook thus seems to be more of a harmless slopocalypse. However, tail risks remain, including agent affordances, contamination of future crawls through self-links, and potential transfer of traits to the next generation of language models. More broadly, our findings highlight the importance of control baselines in emergent misalignment evaluations.

Direct Preference Optimization (DPO) has proven to be an effective solution for mitigating hallucination in Multimodal Large Language Models (MLLMs) by learning from preference pairs. One of its key challenges lies in how to transfer the sequence-level preference into fine-grained supervision on visual fidelity. To safeguard vision-related tokens that are prone to hallucination, existing methods typically allocate training emphasis according to the model's self-assessed visual sensitivity signals. However, such sensitivity, estimated by a model still under training, introduces self-referential bias: reinforcing already well-learned visual cues while neglecting hard-to-perceive but critical details, thereby limiting deeper alignment. In this work, we propose an Uncertainty-aware Exploratory Direct Preference Optimization (UE-DPO) method for MLLMs, which enables the model to uncover its cognitive deficiencies and actively explore for self-correction, guided by token-level epistemic uncertainty. Specifically, we first quantify the uncertainty from the model's failure to ground token predictions in the given image. Then, based on an uncertainty-aware exploration intensity, we encourage more learning pressure on visually deficient tokens in preferred samples, and alleviate the over-penalization of beneficial knowledge in dispreferred samples. Further, we provide a theoretical justification for our method, and extensive experiments demonstrate its effectiveness and robustness.

Recent progress in Multimodal Large Language Models (MLLMs) has demonstrated remarkable advances in perception and reasoning, suggesting their potential for embodied intelligence. While recent studies have evaluated embodied MLLMs in interactive settings, current benchmarks mainly target capabilities to perceive, understand, and interact with external objects, lacking a systematic evaluation of self-centric intelligence. To address this, we introduce MirrorBench, a simulation-based benchmark inspired by the classical Mirror Self-Recognition (MSR) test in psychology. MirrorBench extends this paradigm to embodied MLLMs through a tiered framework of progressively challenging tasks, assessing agents from basic visual perception to high-level self-representation. Experiments on leading MLLMs show that even at the lowest level, their performance remains substantially inferior to human performance, revealing fundamental limitations in self-referential understanding. Our study bridges psychological paradigms and embodied intelligence, offering a principled framework for evaluating the emergence of general intelligence in large models. Project page: https://fflahm.github.io/mirror-bench-page/.