@inproceedings{yang-etal-2025-option,
title = "Option Symbol Matters: Investigating and Mitigating Multiple-Choice Option Symbol Bias of Large Language Models",
author = "Yang, Zhen and
Jian, Ping and
Li, Chengzhi",
editor = "Chiruzzo, Luis and
Ritter, Alan and
Wang, Lu",
booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
month = apr,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.naacl-long.95/",
pages = "1902--1917",
ISBN = "979-8-89176-189-6",
abstract = "Multiple-Choice Question Answering (MCQA) is a widely used task in the evaluation of Large Language Models (LLMs). In this work, we reveal that current LLMs' performance in MCQA could be heavily influenced by the choice of option symbol sets, due to the option symbol bias. That is, when altering only the option symbols (e.g., A/B/C/D$\rightarrow$i/ii/iii/iv), the results could vary sharply, leading to a margin of approximately 10{\%} in accuracy. To uncover the mechanisms behind this, we investigate the internal components of LLMs from a causal perspective. By measuring the causal effects, we identify a small subset of attention heads responsible for the symbol bias. Subsequently, we interpret these key components in a human-understandable way, showing that attention heads with higher causal effects are more likely to focus on only option symbols, while those with lower causal effects tend to distribute their attention across the content of questions and options. It also motivates us to pursue debiasing based on the causal effects. Specifically, to mitigate such bias, we propose a tuning-free, causal effect driven debiasing method which intervenes the activations of identified components according to their causal effects, with stronger interventions corresponding to higher causal effects. Experimental results demonstrate that the proposed method not only alleviates aforementioned bias, but also improves the MCQA performance of LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yang-etal-2025-option">
<titleInfo>
<title>Option Symbol Matters: Investigating and Mitigating Multiple-Choice Option Symbol Bias of Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ping</namePart>
<namePart type="family">Jian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chengzhi</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-04</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Luis</namePart>
<namePart type="family">Chiruzzo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Ritter</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Albuquerque, New Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-189-6</identifier>
</relatedItem>
<abstract>Multiple-Choice Question Answering (MCQA) is a widely used task in the evaluation of Large Language Models (LLMs). In this work, we reveal that current LLMs’ performance in MCQA could be heavily influenced by the choice of option symbol sets, due to the option symbol bias. That is, when altering only the option symbols (e.g., A/B/C/D\rightarrowi/ii/iii/iv), the results could vary sharply, leading to a margin of approximately 10% in accuracy. To uncover the mechanisms behind this, we investigate the internal components of LLMs from a causal perspective. By measuring the causal effects, we identify a small subset of attention heads responsible for the symbol bias. Subsequently, we interpret these key components in a human-understandable way, showing that attention heads with higher causal effects are more likely to focus on only option symbols, while those with lower causal effects tend to distribute their attention across the content of questions and options. It also motivates us to pursue debiasing based on the causal effects. Specifically, to mitigate such bias, we propose a tuning-free, causal effect driven debiasing method which intervenes the activations of identified components according to their causal effects, with stronger interventions corresponding to higher causal effects. Experimental results demonstrate that the proposed method not only alleviates aforementioned bias, but also improves the MCQA performance of LLMs.</abstract>
<identifier type="citekey">yang-etal-2025-option</identifier>
<location>
<url>https://aclanthology.org/2025.naacl-long.95/</url>
</location>
<part>
<date>2025-04</date>
<extent unit="page">
<start>1902</start>
<end>1917</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Option Symbol Matters: Investigating and Mitigating Multiple-Choice Option Symbol Bias of Large Language Models
%A Yang, Zhen
%A Jian, Ping
%A Li, Chengzhi
%Y Chiruzzo, Luis
%Y Ritter, Alan
%Y Wang, Lu
%S Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
%D 2025
%8 April
%I Association for Computational Linguistics
%C Albuquerque, New Mexico
%@ 979-8-89176-189-6
%F yang-etal-2025-option
%X Multiple-Choice Question Answering (MCQA) is a widely used task in the evaluation of Large Language Models (LLMs). In this work, we reveal that current LLMs’ performance in MCQA could be heavily influenced by the choice of option symbol sets, due to the option symbol bias. That is, when altering only the option symbols (e.g., A/B/C/D\rightarrowi/ii/iii/iv), the results could vary sharply, leading to a margin of approximately 10% in accuracy. To uncover the mechanisms behind this, we investigate the internal components of LLMs from a causal perspective. By measuring the causal effects, we identify a small subset of attention heads responsible for the symbol bias. Subsequently, we interpret these key components in a human-understandable way, showing that attention heads with higher causal effects are more likely to focus on only option symbols, while those with lower causal effects tend to distribute their attention across the content of questions and options. It also motivates us to pursue debiasing based on the causal effects. Specifically, to mitigate such bias, we propose a tuning-free, causal effect driven debiasing method which intervenes the activations of identified components according to their causal effects, with stronger interventions corresponding to higher causal effects. Experimental results demonstrate that the proposed method not only alleviates aforementioned bias, but also improves the MCQA performance of LLMs.
%U https://aclanthology.org/2025.naacl-long.95/
%P 1902-1917
Markdown (Informal)
[Option Symbol Matters: Investigating and Mitigating Multiple-Choice Option Symbol Bias of Large Language Models](https://aclanthology.org/2025.naacl-long.95/) (Yang et al., NAACL 2025)
ACL