@inproceedings{kiegeland-etal-2024-reverse,
title = "Reverse-Engineering the Reader",
author = "Kiegeland, Samuel and
Wilcox, Ethan and
Amini, Afra and
Reich, David and
Cotterell, Ryan",
editor = "Al-Onaizan, Yaser and
Bansal, Mohit and
Chen, Yun-Nung",
booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.emnlp-main.526",
pages = "9367--9389",
abstract = "Numerous previous studies have sought to determine to what extent language models, pretrained on natural language text, can serve as useful models of human cognition.In this paper, we are interested in the opposite question: whether we can directly optimize a language model to be a useful cognitive model by aligning it to human psychometric data.To achieve this, we introduce a novel alignment technique in which we fine-tune a language model to implicitly optimize the parameters of a linear regressor that directly predicts humans{'} reading times of in-context linguistic units, e.g., phonemes, morphemes, or words, using surprisal estimates derived from the language model. Using words as a test case, we evaluate our technique across multiple model sizes and datasets and find that it improves language models{'} psychometric predictive power.However, we find an inverse relationship between psychometric power and a model{'}s performance on downstream NLP tasks as well as its perplexity on held-out test data.While this latter trend has been observed before (Oh et al., 2022; Shain et al., 2024), we are the first to induce it by manipulating a model{'}s alignment to psychometric data.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kiegeland-etal-2024-reverse">
<titleInfo>
<title>Reverse-Engineering the Reader</title>
</titleInfo>
<name type="personal">
<namePart type="given">Samuel</namePart>
<namePart type="family">Kiegeland</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ethan</namePart>
<namePart type="family">Wilcox</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Afra</namePart>
<namePart type="family">Amini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Reich</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ryan</namePart>
<namePart type="family">Cotterell</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yaser</namePart>
<namePart type="family">Al-Onaizan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohit</namePart>
<namePart type="family">Bansal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yun-Nung</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Miami, Florida, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Numerous previous studies have sought to determine to what extent language models, pretrained on natural language text, can serve as useful models of human cognition.In this paper, we are interested in the opposite question: whether we can directly optimize a language model to be a useful cognitive model by aligning it to human psychometric data.To achieve this, we introduce a novel alignment technique in which we fine-tune a language model to implicitly optimize the parameters of a linear regressor that directly predicts humans’ reading times of in-context linguistic units, e.g., phonemes, morphemes, or words, using surprisal estimates derived from the language model. Using words as a test case, we evaluate our technique across multiple model sizes and datasets and find that it improves language models’ psychometric predictive power.However, we find an inverse relationship between psychometric power and a model’s performance on downstream NLP tasks as well as its perplexity on held-out test data.While this latter trend has been observed before (Oh et al., 2022; Shain et al., 2024), we are the first to induce it by manipulating a model’s alignment to psychometric data.</abstract>
<identifier type="citekey">kiegeland-etal-2024-reverse</identifier>
<location>
<url>https://aclanthology.org/2024.emnlp-main.526</url>
</location>
<part>
<date>2024-11</date>
<extent unit="page">
<start>9367</start>
<end>9389</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Reverse-Engineering the Reader
%A Kiegeland, Samuel
%A Wilcox, Ethan
%A Amini, Afra
%A Reich, David
%A Cotterell, Ryan
%Y Al-Onaizan, Yaser
%Y Bansal, Mohit
%Y Chen, Yun-Nung
%S Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
%D 2024
%8 November
%I Association for Computational Linguistics
%C Miami, Florida, USA
%F kiegeland-etal-2024-reverse
%X Numerous previous studies have sought to determine to what extent language models, pretrained on natural language text, can serve as useful models of human cognition.In this paper, we are interested in the opposite question: whether we can directly optimize a language model to be a useful cognitive model by aligning it to human psychometric data.To achieve this, we introduce a novel alignment technique in which we fine-tune a language model to implicitly optimize the parameters of a linear regressor that directly predicts humans’ reading times of in-context linguistic units, e.g., phonemes, morphemes, or words, using surprisal estimates derived from the language model. Using words as a test case, we evaluate our technique across multiple model sizes and datasets and find that it improves language models’ psychometric predictive power.However, we find an inverse relationship between psychometric power and a model’s performance on downstream NLP tasks as well as its perplexity on held-out test data.While this latter trend has been observed before (Oh et al., 2022; Shain et al., 2024), we are the first to induce it by manipulating a model’s alignment to psychometric data.
%U https://aclanthology.org/2024.emnlp-main.526
%P 9367-9389
Markdown (Informal)
[Reverse-Engineering the Reader](https://aclanthology.org/2024.emnlp-main.526) (Kiegeland et al., EMNLP 2024)
ACL
- Samuel Kiegeland, Ethan Wilcox, Afra Amini, David Reich, and Ryan Cotterell. 2024. Reverse-Engineering the Reader. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 9367–9389, Miami, Florida, USA. Association for Computational Linguistics.