Elamur Research⌕
☰
home / papers / 2026 / Cognitive Effects and Biases in Large Language Models

Cognitive Effects and Biases in Large Language Models

Paper · 2026 · Markus Schedl, Ralph Hertwig, Antonela Tommasel et al.

Summary

This entry is a tutorial abstract (EACL 2026, Volume 6), not a research paper, so it reports no data, experiments or results. It addresses an interdisciplinary problem. NLP researchers increasingly probe large language models for cognitive phenomena such as anchoring, order or position effects, and confirmation bias, often by borrowing psychology experiments. The organizers argue that these adaptations rely on specific experimental paradigms with their own assumptions about human performance. They also argue that NLP work tends to treat every cognitive effect as a "bias," ignoring that some of these phenomena may be adaptive. The tutorial is co-organized by NLP researchers and a decision scientist. It first explains how psychology defines, validates and disputes such effects. It then surveys LLM bias studies and compares their methods with psychological practice. In a hands-on session, participants try to detect one effect, positional bias, in multilingual LLMs and run into the pitfalls of doing so. It closes with open research questions. Its value lies in shared concepts and methods across the two fields, not in a new resource or benchmark.

Key points

  • Tutorial T2 at EACL 2026 (Proceedings Vol. 6: Tutorial Abstracts, pp. 4–6) took place on 28 March 2026, 14:00–17:30, in Rabat. Materials are at cpjku.github.io/cobis llm eacl2026.
  • Three of the organizers, Markus Schedl, Antonela Tommasel and Shahed Masoudian, are from Johannes Kepler University Linz. The fourth, Ralph Hertwig, works in Berlin at the Max Planck Institute for Human Development's Center of Adaptive Rationality. Tommasel is also affiliated with UNCPBA/CONICET, Argentina.
  • The central conceptual claim is that cognitive effects should be separated from biases, since some effects may serve a useful function.
  • The practical part is an exercise on detecting positional bias in multilingual LLMs.
  • The abstract reports no empirical findings, datasets, models or metrics.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksSocial Sciences · 1,954 worksSafety Research · 626 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,894 works

Cite

Cite as Markus Schedl, Ralph Hertwig, Antonela Tommasel and Shahed Masoudian. 2026. Cognitive Effects and Biases in Large Language Models. Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 6: Tutorial Abstracts).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.eacl-tutorials.2

@inproceedings{schedl2026cognitive,
  title = {{Cognitive Effects and Biases in Large Language Models}},
  author = {Markus Schedl, Ralph Hertwig, Antonela Tommasel and Shahed Masoudian},
  booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 6: Tutorial Abstracts)},
  year = {2026},
  doi = {10.18653/v1/2026.eacl-tutorials.2},
  url = {https://doi.org/10.18653/v1/2026.eacl-tutorials.2}
}

Updated

This page in Markdown → https://md.research.elamur.ai/free/en/papers/schedl-2026-cognitive-effects-biases-large-language/