The23 proposed SSCM does coverthe case of non-zero variance, but currently the identifiability proof is only shown in a24 specific case. Inour simulations under non-zero variance settings, we neverobserved that the procedure converged25 to wrong solutions, suggesting that the non-zero-variance case is also identifiable. For the fMRI and cellular data, the null hypothesis was rejected at significance level 0.01. Regarding causal28 structure variation, for fMRI data, it is well-known that neural connectivities may change across different external29 stimuliorintrinsicstates. Forcellular32 data, causal structure may be different across conditions/interventions.(0)Theyare different.

We develop a method for generating causal post-hoc explanations of black-box classifiers based on a learned low-dimensional representation of the data. The explanation is causal in the sense that changing learned latent factors produces a change in the classifier output statistics. To construct these explanations, we design a learning framework that leverages a generative model and information-theoretic measures of causal influence. Our objective function encourages both the generative model to faithfully represent the data distribution and the latent factors to have a large causal influence on the classifier output. Our method learns both global and local explanations, is compatible with any classifier that admits class probabilities and a gradient, and does not require labeled attributes or knowledge of causal structure. Using carefully controlled test cases, we provide intuition that illuminates the function of our causal objective. We then demonstrate the practical utility of our method on image recognition tasks.

Deep graph learning has advanced Alzheimer's (AD) disease classification from MRI, but most models remain correlational, confounding demographic and genetic factors with disease specific features. We present Causal-GCN, an interventional graph convolutional framework that integrates do-calculus-based back-door adjustment to identify brain regions exerting stable causal influence on AD progression. Each subject's MRI is represented as a structural connectome where nodes denote cortical and subcortical regions and edges encode anatomical connectivity. Confounders such as age, sec, and APOE4 genotype are summarized via principal components and included in the causal adjustment set. After training, interventions on individual regions are simulated by serving their incoming edges and altering node features to estimate average causal effects on disease probability. Applied to 484 subjects from the ADNI cohort, Causal-GCN achieves performance comparable to baseline GNNs while providing interpretable causal effect rankings that highlight posterior, cingulate, and insular hubs consistent with established AD neuropathology.

Understanding causality in event sequences with thousands of sparse event types is critical in domains such as healthcare, cybersecurity, or vehicle diagnostics, yet current methods fail to scale. We present OSCAR, a one-shot causal autoregressive method that infers per-sequence Markov Boundaries using two pretrained Transformers as density estimators. This enables efficient, parallel causal discovery without costly global CI testing. On a real-world automotive dataset with 29,100 events and 474 labels, OSCAR recovers interpretable causal structures in minutes, while classical methods fail to scale, enabling practical scientific diagnostics at production scale.

Large Language Models (LLMs) trained with reinforcement learning and verifiable rewards have achieved strong results on complex reasoning tasks. Recent work extends this paradigm to a multi-agent setting, where a meta-thinking agent proposes plans and monitors progress while a reasoning agent executes subtasks through sequential conversational turns. Despite promising performance, we identify a critical limitation: lazy agent behavior, in which one agent dominates while the other contributes little, undermining collaboration and collapsing the setup to an ineffective single agent. In this paper, we first provide a theoretical analysis showing why lazy behavior naturally arises in multi-agent reasoning. We then introduce a stable and efficient method for measuring causal influence, helping mitigate this issue. Finally, as collaboration intensifies, the reasoning agent risks getting lost in multi-turn interactions and trapped by previous noisy responses. To counter this, we propose a verifiable reward mechanism that encourages deliberation by allowing the reasoning agent to discard noisy outputs, consolidate instructions, and restart its reasoning process when necessary. Extensive experiments demonstrate that our framework alleviates lazy agent behavior and unlocks the full potential of multi-agent framework for complex reasoning tasks. Techniques such as chain-of-thought prompting (Wei et al., 2022; Kojima et al., 2022) and structured methods like Tree-of-Thoughts and Graph-of-Thoughts (Y ao et al., 2023; Besta et al., 2024) expand the space for deliberation. More recently, multi-agent frameworks enable LLMs with specialized roles to collaborate via planning, delegation, and debate, echoing human team dynamics (Li et al., 2023; Wu et al., 2024a; Chen et al., 2023; Du et al., 2023; Y uan & Xie). To support multi-agent and multi-turn reinforcement learning, multi-turn Group Relative Preference Optimization (GRPO) (Wan et al., 2025; Shi et al., 2025; Wei et al., 2025) and its variants (Guo et al., 2025b; Zhang et al., 2025c; Ning et al., 2025; Xue et al., 2025) compute advantages and importance ratios at the turn level, enabling finer-grained optimization and more precise credit assignment. Building on this foundation, ReMA (Wan et al., 2025) introduces a multi-agent LLM reasoning framework with two specialized roles: a meta-thinking agent, which decomposes tasks, sets intermediate goals, and adapts based on feedback, and a reasoning agent, which performs step-by-step 1 The agents alternate sequentially, but since only a final outcome reward is available, ReMA computes a group advantage following GRPO (Shao et al., 2024) and uniformly assigns this trajectory-level signal to every turn in the rollout.

Mechanistic interpretability seeks to uncover how internal components of neural networks give rise to predictions. A persistent challenge, however, is disentangling two often conflated notions: decodability--the recoverability of information from hidden states--and causality--the extent to which those states functionally influence outputs. In this work, we investigate their relationship in vision transformers (ViTs) fine-tuned for object counting. Using activation patching, we test the causal role of spatial and CLS tokens by transplanting activations across clean-corrupted image pairs. In parallel, we train linear probes to assess the decodability of count information at different depths. Our results reveal systematic mismatches: middle-layer object tokens exert strong causal influence despite being weakly decodable, whereas final-layer object tokens support accurate decoding yet are functionally inert. Similarly, the CLS token becomes decodable in mid-layers but only acquires causal power in the final layers. These findings highlight that decodability and causality reflect complementary dimensions of representation--what information is present versus what is used--and that their divergence can expose hidden computational circuits.

We develop a method for generating causal post-hoc explanations of black-box classifiers based on a learned low-dimensional representation of the data. The explanation is causal in the sense that changing learned latent factors produces a change in the classifier output statistics. To construct these explanations, we design a learning framework that leverages a generative model and information-theoretic measures of causal influence.