Instructed to Bias: Instruction-Tuned Language Models Exhibit Emergent Cognitive Bias

Itay Itzhak, Gabriel Stanovsky, Nir Rosenfeld, Yonatan Belinkov


Abstract
Recent studies show that instruction tuning (IT) and reinforcement learning from human feedback (RLHF) improve the abilities of large language models (LMs) dramatically. While these tuning methods can help align models with human objectives and generate high-quality text, not much is known about their potential adverse effects. In this work, we investigate the effect of IT and RLHF on decision making and reasoning in LMs, focusing on three cognitive biases—the decoy effect, the certainty effect, and the belief bias—all of which are known to influence human decision-making and reasoning. Our findings highlight the presence of these biases in various models from the GPT-3, Mistral, and T5 families. Notably, we find a stronger presence of biases in models that have undergone instruction tuning, such as Flan-T5, Mistral-Instruct, GPT3.5, and GPT4. Our work constitutes a step toward comprehending cognitive biases in instruction-tuned LMs, which is crucial for the development of more reliable and unbiased language models.1
Anthology ID:
2024.tacl-1.43
Volume:
Transactions of the Association for Computational Linguistics, Volume 12
Month:
Year:
2024
Address:
Cambridge, MA
Venue:
TACL
SIG:
Publisher:
MIT Press
Note:
Pages:
771–785
Language:
URL:
https://aclanthology.org/2024.tacl-1.43
DOI:
10.1162/tacl_a_00673
Bibkey:
Cite (ACL):
Itay Itzhak, Gabriel Stanovsky, Nir Rosenfeld, and Yonatan Belinkov. 2024. Instructed to Bias: Instruction-Tuned Language Models Exhibit Emergent Cognitive Bias. Transactions of the Association for Computational Linguistics, 12:771–785.
Cite (Informal):
Instructed to Bias: Instruction-Tuned Language Models Exhibit Emergent Cognitive Bias (Itzhak et al., TACL 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.tacl-1.43.pdf