mlsanityData to 14 Oct 2026

PsyMemBehav: Psychology-Informed and Memory-Grounded Two-Level Identity-to-Action Reasoning for Social-Media Behavior Simulation

arXiv 2610.14570 ↗First posted 10 Oct 2026cs.AI cs.SIcited by 0

On the map: LLM reasoning → Reasoning benchmarks and evaluation also fits: Reasoning for agents, planning and code

Abstract (arXiv)

Large language models (LLMs) support social simulation, but faithfully reproducing real users requires grounding both structured platform decisions and language in sparse, noisy behavioral evidence. We introduce PsyMemBehav, which maps psychology-informed profiles and retrieved memories to target-first platform behavior through two-level identity-to-action reasoning. To support this approach with comprehensive behavioral traces, we construct a dataset PsyMemBehav-bench from 5.57M observed behaviors of 5,862 real users (approximately 950 per user) across Reddit, Bluesky, and Twitter. Each recorded engagement is modeled as target selection followed by Comment, Like, or Repost and optional response generation. PsyMemBehav combines 41-dimensional user descriptors, top-5 BM25 memories, reverse-CoT supervision, and rubric-guided DPO over complete reasoning-action trajectories. Experiments show that PsyMemBehav 27B ranks first in 11 of 12 reported individual-level metrics, improving average response quality over GPT-5.4 from 54.28 to 59.68 (9.95% relative), and obtains the lowest group-level MAE (0.081) and MSE (0.009) among simulated systems. These results support evidence-grounded user modeling for observed-thread continuation; the benchmark remains conditional on recorded engagement and does not model no-action decisions.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 19 of its 60 references that are in the map. First posted on arXiv 10 Oct 2026.

Builds on

19 of 60 references.

References · 19

  1. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  2. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
  3. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 3,050
  4. 2023G-Eval: NLG Evaluation using GPT-4 with Better Human Alignmentcited by 2,276
  5. 2026DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligencecited by 1,040
  6. 2023RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedbackcited by 625
  7. 2023Prometheus: Inducing Fine-grained Evaluation Capability in Language Modelscited by 575
  8. 2023Character-LLM: A Trainable Agent for Role-Playingcited by 339
  9. 2023Large Language Models Empowered Agent-based Modeling and Simulation: A Survey and Perspectivescited by 257
  10. 2016Ex Machina: Personal Attacks Seen at Scalecited by 210
  11. 2024CharacterEval: A Chinese Benchmark for Role-Playing Conversational Agent Evaluationcited by 143
  12. 2025MOSAIC: Modeling Social AI for Content Dissemination and Regulation in Multi-Agent Simulationscited by 27
7 more
  1. 2024BASES: Large-scale Web Search User Simulation with Large Language Model based Agentscited by 25
  2. 2025GA-S3: Comprehensive Social Network Simulation with Group Agentscited by 19
  3. 2025Can LLMs Simulate Social Media Engagement? A Study on Action-Guided Response Generationcited by 9
  4. 2026AdaMARP: An Adaptive Multi-Agent Interaction Framework for General Immersive Role-Playingcited by 9
  5. 2025PsyMem: Fine-grained psychological alignment and Explicit Memory Control for Advanced Role-Playing LLMscited by 5
  6. 2026RippleMem: From Isolated Retrieval to Associative Recollection for Long-Term Agent Memorycited by 3
  7. 2026DREAM: LLM-based Dynamic Role-playing via Event-Aware Memory Graphcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025textttBluePrint: A Social Media User Dataset for LLM Persona Evaluation and Training
  2. 2025SVBench: Evaluation of Video Generation Models on Social Reasoning
  3. 2025TWICE: Modeling the Temporal Evolution of Personalized User Behavior via Event-Driven Agents
  4. 2026Benchmarking Psychological Dynamics in Generative Agents
  5. 2026Recon: Reconstruction-Guided Reasoning Synthesis for User Modeling
  6. 2026Persona-Based Simulation of Human Opinion at Population Scale
  7. 2026MiroBench: Benchmarking Realism in Agentic Simulation of Real-world Discussions
  8. 2026Socio-Foundation: A Model for Generalizable Individual Behavior Simulation via Hierarchical Capability Distillation