# Looking Is Not Picking: An Attention-Segment Account of Tool-Selection Failures in LLM Agents arXiv: https://arxiv.org/abs/2606.16364 · HTML: https://mlsanity.com/paper/2606.16364/ ## Key facts - Cited by 3 papers in the map (data through 14 Oct 2026). - Placed in Tool learning and function calling, one of 16 directions in LLM agents and tool use. - Builds on 27 of its 39 references that are in the map. - First posted on arXiv 15 Jun 2026. ## Placed in - LLM agents and tool use / [Tool learning and function calling](https://mlsanity.com/agents/tool-learning-and-function-calling/) ## Builds on (most cited first, 27 of 27) - [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017) - [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024) - [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022) - [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023) - [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023) - [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023) - [ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs](https://mlsanity.com/paper/2307.16789/) (2023) - [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024) - [τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains](https://mlsanity.com/paper/2406.12045/) (2024) - [Attention is not Explanation](https://mlsanity.com/paper/1902.10186/) (2019) - [Quantifying Attention Flow in Transformers](https://mlsanity.com/paper/2005.00928/) (2020) - [Yi: Open Foundation Models by 01.AI](https://mlsanity.com/paper/2403.04652/) (2024) - [Attention is not not Explanation](https://mlsanity.com/paper/1908.04626/) (2019) - [AttnLRP: Attention-Aware Layer-Wise Relevance Propagation for Transformers](https://mlsanity.com/paper/2402.05602/) (2024) - [Seal-Tools: Self-Instruct Tool Learning Dataset for Agent Tuning and Detailed Benchmark](https://mlsanity.com/paper/2405.08355/) (2024) - [Which Attention Heads Matter for In-Context Learning?](https://mlsanity.com/paper/2502.14010/) (2025) - [Detecting High-Stakes Interactions with Activation Probes](https://mlsanity.com/paper/2506.10805/) (2025) - [Attention in Large Language Models Yields Efficient Zero-Shot Re-Rankers](https://mlsanity.com/paper/2410.02642/) (2024) - [Fortify the Shortest Stave in Attention: Enhancing Context Awareness of Large Language Models for Effective Tool Use](https://mlsanity.com/paper/2312.04455/) (2023) - [Spotlight Your Instructions: Instruction-following with Dynamic Attention Steering](https://mlsanity.com/paper/2505.12025/) (2025) - [Mixture of In-Context Experts Enhance LLMs' Long Context Awareness](https://mlsanity.com/paper/2406.19598/) (2024) - [Accurate Failure Prediction in Agents Does Not Imply Effective Failure Prevention](https://mlsanity.com/paper/2602.03338/) (2026) - [MindGuard: Intrinsic Decision Inspection for Securing LLM Agents Against Metadata Poisoning](https://mlsanity.com/paper/2508.20412/) (2025) - [Tool Calling is Linearly Readable and Steerable in Language Models](https://mlsanity.com/paper/2605.07990/) (2026) - [Model-Adaptive Tool Necessity Reveals the Knowing-Doing Gap in LLM Tool Use](https://mlsanity.com/paper/2605.14038/) (2026) - [Beyond the Black Box: Interpretability of Agentic AI Tool Use](https://mlsanity.com/paper/2605.06890/) (2026) - [Spectral Guardrails for Agents in the Wild: Detecting Tool Use Hallucinations via Attention Topology](https://mlsanity.com/paper/2602.08082/) (2026) ## Built on by (most cited first, 3 of 3) - [SkillGate: Training In-Policy Skill Selection in Long-Horizon Agents](https://mlsanity.com/paper/2608.18852/) (2026) - [When Retrieval Metrics Mislead: Measuring Policy Signal in Long-Horizon Tool-Use Agents](https://mlsanity.com/paper/2606.23937/) (2026) - [The Default Trap: Rethinking Plan Evaluation in Tool-Using LLM Agents](https://mlsanity.com/paper/2609.36829/) (2026) ## Abstract (arXiv) LLM agents mis-call tools, and the natural guess is that the model failed to see the right tool in a crowded harness. We show the opposite through a lens concurrent work sets aside -- the model's attention to labeled tool-definition segments. On real BFCL failures, by per-candidate attention argmax the model attends most to the correct tool 80% of the time (vs. 21% chance), and the gold is the under-attended segment on only 10%: it looks at the right tool and still picks wrong. This directly refutes the intuitive "crowded-harness / lost-in-the-middle" explanation: the failure is at the decision readout, not the harness, and we pin it there three ways. (1) Input vs. readout: repairing the prompt (reordering or duplicating the gold tool) recovers <=23% of failures, while readout-side interventions recover 59-91%. (2) Representation-invariance: two gold-pointed interventions in different representations -- an additive attention-logit bias and a residual-stream steering vector -- recover largely the same failures (per-task Jaccard 0.865 pooled, 0.79-0.91 per model), so the bottleneck is localized to the readout independent of which representation is poked. (3) A training-free, gold-free selector: per-segment attention closes most of the gold-free-vs-oracle gap on BFCL (+11.9 pts pooled function-name selection vs. +17.9-pt oracle headroom) and adds +14.9 pts on Seal-Tools; every model positive (exact McNemar p<=8e-4 each). Scopes differ: the causal attention-bias dose-response is bidirectional and monotonic on 10 mask-honoring models (3-32B), the full 0.5-32B span carrying only the correlational diagnostic; the deployable selector is evaluated on 5 single-turn models and does not yet transfer to a multi-turn loop.