# Answer, Assemble, Ace: Understanding How LMs Answer Multiple Choice Questions

arXiv: https://arxiv.org/abs/2407.15018 · HTML: https://mlsanity.com/paper/2407.15018/

## Key facts

- Cited by 35 papers in the map (data through 14 Oct 2026).
- Builds on 42 of its 54 references that are in the map.
- First posted on arXiv 21 Jul 2024.

## Builds on (most cited first, 30 of 42)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [Locating and Editing Factual Associations in GPT](https://mlsanity.com/paper/2202.05262/) (2022)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [Transformer Feed-Forward Layers Are Key-Value Memories](https://mlsanity.com/paper/2012.14913/) (2020)
- [Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small](https://mlsanity.com/paper/2211.00593/) (2022)
- [Universal Transformers](https://mlsanity.com/paper/1807.03819/) (2018)
- [OLMo: Accelerating the Science of Language Models](https://mlsanity.com/paper/2402.00838/) (2024)
- [Knowledge Neurons in Pretrained Transformers](https://mlsanity.com/paper/2104.08696/) (2021)
- [Transformer Feed-Forward Layers Build Predictions by Promoting Concepts in the Vocabulary Space](https://mlsanity.com/paper/2203.14680/) (2022)
- [Dissecting Recall of Factual Associations in Auto-Regressive Language Models](https://mlsanity.com/paper/2304.14767/) (2023)
- [Large Language Models Are Not Robust Multiple Choice Selectors](https://mlsanity.com/paper/2309.03882/) (2023)
- [Towards Best Practices of Activation Patching in Language Models: Metrics and Methods](https://mlsanity.com/paper/2309.16042/) (2023)
- [Confident Adaptive Language Modeling](https://mlsanity.com/paper/2207.07061/) (2022)
- [Function Vectors in Large Language Models](https://mlsanity.com/paper/2310.15213/) (2023)
- [Large Language Models Sensitivity to The Order of Options in Multiple-Choice Questions](https://mlsanity.com/paper/2308.11483/) (2023)
- [Depth-Adaptive Transformer](https://mlsanity.com/paper/1910.10073/) (2019)
- [Attribution Patching Outperforms Automated Circuit Discovery](https://mlsanity.com/paper/2310.10348/) (2023)
- [Leveraging Large Language Models for Multiple Choice Question Answering](https://mlsanity.com/paper/2210.12353/) (2022)
- [How does GPT-2 compute greater-than?: Interpreting mathematical abilities in a pre-trained language model](https://mlsanity.com/paper/2305.00586/) (2023)
- [Does Circuit Analysis Interpretability Scale? Evidence from Multiple Choice Capabilities in Chinchilla](https://mlsanity.com/paper/2307.09458/) (2023)
- [Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking](https://mlsanity.com/paper/2402.14811/) (2024)
- [When Benchmarks are Targets: Revealing the Sensitivity of Large Language Model Leaderboards](https://mlsanity.com/paper/2402.01781/) (2024)
- [Analyzing Transformers in Embedding Space](https://mlsanity.com/paper/2209.02535/) (2022)
- [AtP*: An efficient and scalable method for localizing LLM behaviour to components](https://mlsanity.com/paper/2403.00745/) (2024)
- [Circuit Component Reuse Across Tasks in Transformer Language Models](https://mlsanity.com/paper/2310.08744/) (2023)
- [Interpretability at Scale: Identifying Causal Mechanisms in Alpaca](https://mlsanity.com/paper/2305.08809/) (2023)
- [Language Models Implement Simple Word2Vec-style Vector Arithmetic](https://mlsanity.com/paper/2305.16130/) (2023)
- [A Mechanistic Interpretation of Arithmetic Reasoning in Language Models using Causal Mediation Analysis](https://mlsanity.com/paper/2305.15054/) (2023)

## Built on by (most cited first, 30 of 35)

- [MIB: A Mechanistic Interpretability Benchmark](https://mlsanity.com/paper/2504.13151/) (2025)
- [Steering off Course: Reliability Challenges in Steering Language Models](https://mlsanity.com/paper/2504.04635/) (2025)
- [Critical Batch Size Revisited: A Simple Empirical Approach to Large-Batch Language Model Training](https://mlsanity.com/paper/2505.23971/) (2025)
- [Understanding the Skill Gap in Recurrent Language Models: The Role of the Gather-and-Aggregate Mechanism](https://mlsanity.com/paper/2504.18574/) (2025)
- [Can you map it to English? The Role of Cross-Lingual Alignment in Multilingual Performance of LLMs](https://mlsanity.com/paper/2504.09378/) (2025)
- [Attributing Response to Context: A Jensen-Shannon Divergence Driven Mechanistic Study of Context Attribution in Retrieval-Augmented Generation](https://mlsanity.com/paper/2505.16415/) (2025)
- [From Directions to Regions: Decomposing Activations in Language Models via Local Geometry](https://mlsanity.com/paper/2602.02464/) (2026)
- [Internal Causal Mechanisms Robustly Predict Language Model Out-of-Distribution Behaviors](https://mlsanity.com/paper/2505.11770/) (2025)
- [Internal Chain-of-Thought: Empirical Evidence for Layer-wise Subtask Scheduling in LLMs](https://mlsanity.com/paper/2505.14530/) (2025)
- [Mechanistic evaluation of Transformers and state space models](https://mlsanity.com/paper/2505.15105/) (2025)
- [SECA: Semantically Equivalent and Coherent Attacks for Eliciting LLM Hallucinations](https://mlsanity.com/paper/2510.04398/) (2025)
- [Calibration Across Layers: Understanding Calibration Evolution in LLMs](https://mlsanity.com/paper/2511.00280/) (2025)
- [Geometry of Decision Making in Language Models](https://mlsanity.com/paper/2511.20315/) (2025)
- [Causality neq Invariance: Function and Concept Vectors in LLMs](https://mlsanity.com/paper/2602.22424/) (2026)
- [Good Arguments Against the People Pleasers: How Reasoning Mitigates (Yet Masks) LLM Sycophancy](https://mlsanity.com/paper/2603.16643/) (2026)
- [LLMs Process Lists With General Filter Heads](https://mlsanity.com/paper/2510.26784/) (2025)
- [What Drives Representation Steering? A Mechanistic Case Study on Steering Refusal](https://mlsanity.com/paper/2604.08524/) (2026)
- [Born a Transformer -- Always a Transformer? On the Effect of Pretraining on Architectural Abilities](https://mlsanity.com/paper/2505.21785/) (2025)
- [Bridging the Knowledge-Prediction Gap in LLMs on Multiple-Choice Questions](https://mlsanity.com/paper/2509.23782/) (2025)
- [The Mechanistic Emergence of Symbol Grounding in Language Models](https://mlsanity.com/paper/2510.13796/) (2025)
- [Correcting Gradient-Based Circuit Localization via Interaction-Aware Backpropagation](https://mlsanity.com/paper/2505.17630/) (2025)
- [Dissecting Persona-Driven Reasoning in Language Models via Activation Patching](https://mlsanity.com/paper/2507.20936/) (2025)
- [Signatures of human-like processing in Transformer forward passes](https://mlsanity.com/paper/2504.14107/) (2025)
- [Challenges in Understanding Modality Conflict in Vision-Language Models](https://mlsanity.com/paper/2509.02805/) (2025)
- [Shattered Compositionality: Counterintuitive Learning Dynamics of Transformers for Arithmetic](https://mlsanity.com/paper/2601.22510/) (2026)
- [From Early Encoding to Late Suppression: Interpreting LLMs on Character Counting Tasks](https://mlsanity.com/paper/2604.00778/) (2026)
- [BlackboxNLP-2025 MIB Shared Task: Exploring Ensemble Strategies for Circuit Localization Methods](https://mlsanity.com/paper/2510.06811/) (2025)
- [Lost in the Prompt Order: Revealing the Limitations of Causal Attention in Language Models](https://mlsanity.com/paper/2601.14152/) (2026)
- [Findings of the BlackboxNLP 2025 Shared Task: Localizing Circuits and Causal Variables in Language Models](https://mlsanity.com/paper/2511.18409/) (2025)
- [To Adapt or not to Adapt, Rethinking the Value of Medical Knowledge-Aware Large Language Models](https://mlsanity.com/paper/2604.06854/) (2026)

## Abstract (arXiv)

Multiple-choice question answering (MCQA) is a key competence of performant transformer language models that is tested by mainstream benchmarks. However, recent evidence shows that models can have quite a range of performance, particularly when the task format is diversified slightly (such as by shuffling answer choice order). In this work we ask: how do successful models perform formatted MCQA? We employ vocabulary projection and activation patching methods to localize key hidden states that encode relevant information for predicting the correct answer. We find that the prediction of a specific answer symbol is causally attributed to a few middle layers, and specifically their multi-head self-attention mechanisms. We show that subsequent layers increase the probability of the predicted answer symbol in vocabulary space, and that this probability increase is associated with a sparse set of attention heads with unique roles. We additionally uncover differences in how different models adjust to alternative symbols. Finally, we demonstrate that a synthetic task can disentangle sources of model error to pinpoint when a model has learned formatted MCQA, and show that logit differences between answer choice tokens continue to grow over the course of training.