# Do Large Language Models Perform Latent Multi-Hop Reasoning without Exploiting Shortcuts?

arXiv: https://arxiv.org/abs/2411.16679 · HTML: https://mlsanity.com/paper/2411.16679/

## Key facts

- Cited by 28 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge- and tool-augmented reasoning, one of 17 directions in LLM reasoning.
- Builds on 47 of its 81 references that are in the map.
- First posted on arXiv 25 Nov 2024.

## Placed in

- LLM reasoning / [Knowledge- and tool-augmented reasoning](https://mlsanity.com/lr/knowledge-and-tool-augmented-reasoning/)

## Builds on (most cited first, 30 of 47)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context](https://mlsanity.com/paper/2403.05530/) (2024)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [Locating and Editing Factual Associations in GPT](https://mlsanity.com/paper/2202.05262/) (2022)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [BLOOM: A 176B-Parameter Open-Access Multilingual Language Model](https://mlsanity.com/paper/2211.05100/) (2022)
- [Mixtral of Experts](https://mlsanity.com/paper/2401.04088/) (2024)
- [MuSiQue: Multihop Questions via Single-hop Question Composition](https://mlsanity.com/paper/2108.00573/) (2021)
- [Gemma: Open Models Based on Gemini Research and Technology](https://mlsanity.com/paper/2403.08295/) (2024)
- [Yi: Open Foundation Models by 01.AI](https://mlsanity.com/paper/2403.04652/) (2024)
- [Dissecting Recall of Factual Associations in Auto-Regressive Language Models](https://mlsanity.com/paper/2304.14767/) (2023)
- [Evaluating the Ripple Effects of Knowledge Editing in Language Models](https://mlsanity.com/paper/2307.12976/) (2023)
- [Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Models](https://mlsanity.com/paper/2401.06102/) (2024)
- [Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?](https://mlsanity.com/paper/2405.05904/) (2024)
- [Active Example Selection for In-Context Learning](https://mlsanity.com/paper/2211.04486/) (2022)
- [Causal Mediation Analysis for Interpreting Neural NLP: The Case of Gender Bias](https://mlsanity.com/paper/2004.12265/) (2020)
- [A Careful Examination of Large Language Model Performance on Grade School Arithmetic](https://mlsanity.com/paper/2405.00332/) (2024)
- [Do Large Language Models Latently Perform Multi-Hop Reasoning?](https://mlsanity.com/paper/2402.16837/) (2024)
- [The pitfalls of next-token prediction](https://mlsanity.com/paper/2403.06963/) (2024)
- [Taken out of context: On measuring situational awareness in LLMs](https://mlsanity.com/paper/2309.00667/) (2023)
- [Physics of Language Models: Part 3.2, Knowledge Manipulation](https://mlsanity.com/paper/2309.14402/) (2023)
- [What's In My Big Data?](https://mlsanity.com/paper/2310.20707/) (2023)
- [On the Paradox of Learning to Reason from Data](https://mlsanity.com/paper/2205.11502/) (2022)
- [How Do Large Language Models Acquire Factual Knowledge During Pretraining?](https://mlsanity.com/paper/2406.11813/) (2024)
- [Compositional Questions Do Not Necessitate Multi-hop Reasoning](https://mlsanity.com/paper/1906.02900/) (2019)
- [Unfamiliar Finetuning Examples Control How Language Models Hallucinate](https://mlsanity.com/paper/2403.05612/) (2024)
- [Testing the General Deductive Reasoning Capacity of Large Language Models Using OOD Examples](https://mlsanity.com/paper/2305.15269/) (2023)

## Built on by (most cited first, 28 of 28)

- [Towards Holistic Evaluation of Large Audio-Language Models: A Comprehensive Survey](https://mlsanity.com/paper/2505.15957/) (2025)
- [Roll the dice & look before you leap: Going beyond the creative limits of next-token prediction](https://mlsanity.com/paper/2504.15266/) (2025)
- [Do Latent Tokens Think? A Causal and Adversarial Analysis of Chain-of-Continuous-Thought](https://mlsanity.com/paper/2512.21711/) (2025)
- [Large Language Model Reasoning Failures](https://mlsanity.com/paper/2602.06176/) (2026)
- [How do Transformers Learn Implicit Reasoning?](https://mlsanity.com/paper/2505.23653/) (2025)
- [Implicit Reasoning in Transformers is Reasoning through Shortcuts](https://mlsanity.com/paper/2503.07604/) (2025)
- [Think Fast: Estimating No-CoT Task-Completion Time Horizons of Frontier AI Models](https://mlsanity.com/paper/2606.07157/) (2026)
- [Language models can learn implicit multi-hop reasoning, but only if they have lots of training data](https://mlsanity.com/paper/2505.17923/) (2025)
- [CaKE: Circuit-aware Editing Enables Generalizable Knowledge Learners](https://mlsanity.com/paper/2503.16356/) (2025)
- [Internal Chain-of-Thought: Empirical Evidence for Layer-wise Subtask Scheduling in LLMs](https://mlsanity.com/paper/2505.14530/) (2025)
- [Latent Reasoning with Supervised Thinking States](https://mlsanity.com/paper/2602.08332/) (2026)
- [How Do Language Models Compose Functions?](https://mlsanity.com/paper/2510.01685/) (2025)
- [Can LLMs Reconcile Knowledge Conflicts in Counterfactual Reasoning](https://mlsanity.com/paper/2506.15732/) (2025)
- [Benefits and Limitations of Communication in Multi-Agent Reasoning](https://mlsanity.com/paper/2510.13903/) (2025)
- [The Illusion of Superposition? A Principled Analysis of Latent Thinking in Language Models](https://mlsanity.com/paper/2604.06374/) (2026)
- [Are Transformers Able to Reason by Connecting Separated Knowledge in Training Data?](https://mlsanity.com/paper/2501.15857/) (2025)
- [Layer-Order Inversion: Rethinking Latent Multi-Hop Reasoning in Large Language Models](https://mlsanity.com/paper/2601.03542/) (2026)
- [Programming by Backprop: An Instruction is Worth 100 Examples When Finetuning LLMs](https://mlsanity.com/paper/2506.18777/) (2025)
- [Investigating Language Model Capabilities to Represent and Process Formal Knowledge: A Preliminary Study to Assist Ontology Engineering](https://mlsanity.com/paper/2509.10249/) (2025)
- [Reading Between the Dots: Decoding Hidden Computation across Filler Tokens](https://mlsanity.com/paper/2607.03502/) (2026)
- [Bilinear representation mitigates reversal curse and enables consistent model editing](https://mlsanity.com/paper/2509.21993/) (2025)
- [The Unlearning Mirage: A Dynamic Framework for Evaluating LLM Unlearning](https://mlsanity.com/paper/2603.11266/) (2026)
- [Composition Collapse: Stable Factual Knowledge Does Not Imply Compositional Reasoning](https://mlsanity.com/paper/2605.26789/) (2026)
- [NeuroFaith: Evaluating LLM Self-Explanation Faithfulness via Internal Representation Alignment](https://mlsanity.com/paper/2506.09277/) (2025)
- [Out-of-Context Abduction: LLMs Make Inferences About Procedural Data Leveraging Declarative Facts in Earlier Training Data](https://mlsanity.com/paper/2508.00741/) (2025)
- [Not All Needles Are Found: How Fact Distribution and Don't Make It Up Prompts Shape Retrieval, Reasoning, and Hallucination in Long-Context LLMs](https://mlsanity.com/paper/2601.02023/) (2026)
- [Multi-Hop Knowledge Composition is Bound by Pretraining Exposure](https://mlsanity.com/paper/2606.09338/) (2026)
- [Silent Dissent: LLM Agents That Yield to the Majority Still Represent Their Original Premise](https://mlsanity.com/paper/2610.02702/) (2026)

## Abstract (arXiv)

We evaluate how well Large Language Models (LLMs) latently recall and compose facts to answer multi-hop queries like "In the year Scarlett Johansson was born, the Summer Olympics were hosted in the country of". One major challenge in such evaluation is that LLMs may have developed shortcuts by encountering the head entity "Scarlett Johansson" and the answer entity "United States" in the same training sequences or merely guess the answer based on frequency-based priors. To prevent shortcuts, we exclude test queries where the head and answer entities might have co-appeared during training. Through careful selection of relations and facts and systematic removal of cases where models might guess answers or exploit partial matches, we construct an evaluation dataset SOCRATES (ShOrtCut-fRee lATent rEaSoning). We observe that LLMs demonstrate promising latent multi-hop reasoning abilities without exploiting shortcuts, but only for certain types of queries. For queries requiring latent recall of countries as the intermediate answer, the best models achieve 80% latent composability, but this drops to just 5% for the recall of years. Comparisons with Chain-of-Thought highlight a significant gap between the ability of models to reason latently versus explicitly. Analysis reveals that latent representations of the intermediate answer are constructed more often in queries with higher latent composability, and shows the emergence of latent multi-hop reasoning during pretraining.