Hedonic Neurons: A Mechanistic Mapping of Latent Coalitions in Transformer MLPs

• URL: http://arxiv.org/abs/2509.23684v1
• Date: Sun, 28 Sep 2025 06:55:24 GMT
• Title: Hedonic Neurons: A Mechanistic Mapping of Latent Coalitions in Transformer MLPs
• Abstract summary: We introduce a mechanistic interpretability framework based on coalitional game theory.<n>We extract stable coalitions of neurons whose joint ablation has non-additive effects.<n>We then track their transitions across layers as persistence, splitting, merging, or disappearance.
• Score: 18.335932098245895
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Fine-tuned Large Language Models (LLMs) encode rich task-specific features, but the form of these representations, especially within MLP layers, remains unclear. Empirical inspection of LoRA updates shows that new features concentrate in mid-layer MLPs, yet the scale of these layers obscures meaningful structure. Prior probing suggests that statistical priors may strengthen, split, or vanish across depth, motivating the need to study how neurons work together rather than in isolation. We introduce a mechanistic interpretability framework based on coalitional game theory, where neurons mimic agents in a hedonic game whose preferences capture their synergistic contributions to layer-local computations. Using top-responsive utilities and the PAC-Top-Cover algorithm, we extract stable coalitions of neurons: groups whose joint ablation has non-additive effects. We then track their transitions across layers as persistence, splitting, merging, or disappearance. Applied to LLaMA, Mistral, and Pythia rerankers fine-tuned on scalar IR tasks, our method finds coalitions with consistently higher synergy than clustering baselines. By revealing how neurons cooperate to encode features, hedonic coalitions uncover higher-order structure beyond disentanglement and yield computational units that are functionally important, interpretable, and predictive across domains.

Related papers

• Revenge of Monosemanticity: Specialized Neurons Improve Data Efficiency in MLPs [17.43268776924599]
  We show that multilayer perceptrons (MLPs) naturally develop monosemantic specialized neurons.<n>Rather than learning a single global low-dimensional representation, relevants learn a collection of local low-dimensional representations that can collectively span a high-dimensional space.
  arXiv  Detail & Related papers  (2026-08-25T02:54:58Z)
• Sparse Inter-Layer Dependencies of Transformer FFN Neurons [32.00544060412007]
  We study whether the activation of an FFN neuron can be explained by a sparse set of preceding neuron activations and attention outputs.<n>We introduce a training-free attribution method that estimates the relative influence of upstream neurons and attention outputs on a target neuron's activation.
  arXiv  Detail & Related papers  (2026-07-13T15:04:50Z)
• Beyond the Aggregation Dilemma: Prior-Retaining Decoupled Learning for Multimodal Graphs [69.56407708794845]
  MultimodalAttributed Graph Learning (MAGL) integrates node attributes with structural topology via graph aggregation.<n>As pretrained encoders evolve into Large Foundation Models (LFMs), the landscape of MAGL fundamentally shifts.<n>Under high-confidence LFM priors, mandatory aggregation introduces topological intrinsic noise that overwhelms discriminative signals.<n>We propose SUPRA (Shared-Unique Prior-Retaining Architecture), a decoupled dual-pathway paradigm.<n> SUPRA processes modality-specific features through topology-agnostics while capturing structural synergy via a lightweight shared GNN, with auxiliary deep supervision counteracting gradient starvation.
  arXiv  Detail & Related papers  (2026-05-23T17:42:25Z)
• Rare Events, Real Signals: Functional Ensembles as Units of Computation in Deep Spiking Networks [0.0]
  We investigate how internal representations emerge across hierarchical processing systems by introducing a neuroscience-inspired framework for analyzing deep spiking neural networks (SNN)<n>We form the first-order functionally-connected (1FC) group of a neuron based on its statistically significant pairwise correlations with neurons from the previous layer of a trained SNN architecture.<n>Our analysis shows that several principles of functional connectivity previously observed in biological cortex are preserved in spiking ResNet architectures.
  arXiv  Detail & Related papers  (2026-05-21T15:44:32Z)
• Structural Graph Probing of Vision-Language Models [9.364818672785757]
  Vision-language models (VLMs) achieve strong multimodal performance.<n>How computation is organized across populations of neurons remains poorly understood.<n>We studyVLMs through the lens of neural topology, representing each layer as a within-layer correlation graph.
  arXiv  Detail & Related papers  (2026-03-28T01:14:40Z)
• Do LLMs and VLMs Share Neurons for Inference? Evidence and Mechanisms of Cross-Modal Transfer [65.72553715508691]
  We show that large vision-language models (LVLMs) lag behind strong text-only large language models (LLMs) on tasks that require multi-step inference and compositional decision-making.<n>We propose Shared Neuron Low-Rank Fusion (SNRF), a parameter-efficient framework that transfers mature inference circuitry from LLMs to LVLMs.<n>Our results demonstrate that shared neurons form an interpretable bridge between LLMs and LVLMs, enabling low-cost transfer of inference ability into multimodal models.
  arXiv  Detail & Related papers  (2026-02-22T06:04:05Z)
• Association-sensory spatiotemporal hierarchy and functional gradient-regularised recurrent neural network with implications for schizophrenia [1.7711010795772193]
  This study characterises the AS hierarchy of patients with schizophrenia in a comparison with controls.<n>We found that schizophrenia compresses the AS hierarchy indicating reduced functional differentiation.<n>By modelling neural timescale with the Ornstein-Uhlenbeck process, we observed that the most specialised, locally cohesive regions at the gradient extremes exhibit dynamics with a longer time constant.
  arXiv  Detail & Related papers  (2025-11-04T16:45:55Z)
• Neuronal Group Communication for Efficient Neural representation [85.36421257648294]
  This paper addresses the question of how to build large neural systems that learn efficient, modular, and interpretable representations.<n>We propose Neuronal Group Communication (NGC), a theory-driven framework that reimagines a neural network as a dynamical system of interacting neuronal groups.<n>NGC treats weights as transient interactions between embedding-like neuronal states, with neural computation unfolding through iterative communication among groups of neurons.
  arXiv  Detail & Related papers  (2025-10-19T14:23:35Z)
• Distributed Specialization: Rare-Token Neurons in Large Language Models [8.13000021263958]
  Large language models (LLMs) struggle with representing and generating rare tokens despite their importance in specialized domains.<n>We investigate whether LLMs develop internal specialization mechanisms through discrete modular architectures or distributed parameter-level differentiation domains.
  arXiv  Detail & Related papers  (2025-09-25T13:49:38Z)
• CodeBrain: Towards Decoupled Interpretability and Multi-Scale Architecture for EEG Foundation Model [52.466542039411515]
  EEG foundation models (EFMs) have emerged to address the scalability issues of task-specific models.<n>We present CodeBrain, a two-stage EFM designed to fill this gap.<n>In the first stage, we introduce the TFDual-Tokenizer, which decouples heterogeneous temporal and frequency EEG signals into discrete tokens.<n>In the second stage, we propose the multi-scale EEGSSM architecture, which combines structured global convolution with sliding window attention.
  arXiv  Detail & Related papers  (2025-06-10T17:20:39Z)
• BrainNetMLP: An Efficient and Effective Baseline for Functional Brain Network Classification [2.969929079464237]
  We propose a pure deep learning architecture, named BrainNetMLP, for functional brain network classification.<n>We evaluate our proposed BrainNetMLP on two public and popular brain network classification datasets.
  arXiv  Detail & Related papers  (2025-05-14T10:55:51Z)
• NeurFlow: Interpreting Neural Networks through Neuron Groups and Functional Interactions [16.00223741620103]
  We propose a novel framework that transitions the focus from analyzing individual neurons to investigating groups of neurons.<n>Our automated framework, NeurFlow, first identifies core neurons and clusters them into groups based on shared functional relationships.
  arXiv  Detail & Related papers  (2025-02-22T06:01:03Z)
• P-SpikeSSM: Harnessing Probabilistic Spiking State Space Models for Long-Range Dependency Tasks [1.9775291915550175]
  Spiking neural networks (SNNs) are posited as a computationally efficient and biologically plausible alternative to conventional neural architectures.<n>We develop a scalable probabilistic spiking learning framework for long-range dependency tasks.<n>Our models attain state-of-the-art performance among SNN models across diverse long-range dependency tasks.
  arXiv  Detail & Related papers  (2024-06-05T04:23:11Z)
• Fully Spiking Actor Network with Intra-layer Connections for Reinforcement Learning [51.386945803485084]
  We focus on the task where the agent needs to learn multi-dimensional deterministic policies to control. Most existing spike-based RL methods take the firing rate as the output of SNNs, and convert it to represent continuous action space (i.e., the deterministic policy) through a fully-connected layer. To develop a fully spiking actor network without any floating-point matrix operations, we draw inspiration from the non-spiking interneurons found in insects.
  arXiv  Detail & Related papers  (2024-01-09T07:31:34Z)
• Language Knowledge-Assisted Representation Learning for Skeleton-Based Action Recognition [71.35205097460124]
  How humans understand and recognize the actions of others is a complex neuroscientific problem. LA-GCN proposes a graph convolution network using large-scale language models (LLM) knowledge assistance.
  arXiv  Detail & Related papers  (2023-05-21T08:29:16Z)
• And/or trade-off in artificial neurons: impact on adversarial robustness [91.3755431537592]
  Presence of sufficient number of OR-like neurons in a network can lead to classification brittleness and increased vulnerability to adversarial attacks. We define AND-like neurons and propose measures to increase their proportion in the network. Experimental results on the MNIST dataset suggest that our approach holds promise as a direction for further exploration.
  arXiv  Detail & Related papers  (2021-02-15T08:19:05Z)
• How Neural Networks Extrapolate: From Feedforward to Graph Neural Networks [80.55378250013496]
  We study how neural networks trained by gradient descent extrapolate what they learn outside the support of the training distribution. Graph Neural Networks (GNNs) have shown some success in more complex tasks.
  arXiv  Detail & Related papers  (2020-09-24T17:48:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.