# Evaluating and Characterizing Human Rationales

arXiv: https://arxiv.org/abs/2010.04736 · HTML: https://mlsanity.com/paper/2010.04736/

## Key facts

- Cited by 36 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Oct 2020.

## Built on by (most cited first, 30 of 36)

- [Teach Me to Explain: A Review of Datasets for Explainable Natural Language Processing](https://mlsanity.com/paper/2102.12060/) (2021)
- [Few-Shot Self-Rationalization with Natural Language Prompts](https://mlsanity.com/paper/2111.08284/) (2021)
- [UNIREX: A Unified Learning Framework for Language Model Rationale Extraction](https://mlsanity.com/paper/2112.08802/) (2021)
- [Local Interpretations for Explainable Natural Language Processing: A Survey](https://mlsanity.com/paper/2103.11072/) (2021)
- [Is Sparse Attention more Interpretable?](https://mlsanity.com/paper/2106.01087/) (2021)
- [On the Diversity and Limits of Human Explanations](https://mlsanity.com/paper/2106.11988/) (2021)
- [ferret: a Framework for Benchmarking Explainers on Transformers](https://mlsanity.com/paper/2208.01575/) (2022)
- [What to Learn, and How: Toward Effective Learning from Rationales](https://mlsanity.com/paper/2112.00071/) (2021)
- [An Empirical Study on Explanations in Out-of-Domain Settings](https://mlsanity.com/paper/2203.00056/) (2022)
- [CATfOOD: Counterfactual Augmented Training for Improving Out-of-Domain Performance and Calibration](https://mlsanity.com/paper/2309.07822/) (2023)
- [A survey on improving NLP models with human explanations](https://mlsanity.com/paper/2204.08892/) (2022)
- [On the Impact of Temporal Concept Drift on Model Explanations](https://mlsanity.com/paper/2210.09197/) (2022)
- [Incorporating Attribution Importance for Improving Faithfulness Metrics](https://mlsanity.com/paper/2305.10496/) (2023)
- [GeoZero: Incentivizing Reasoning from Scratch on Geospatial Scenes](https://mlsanity.com/paper/2511.22645/) (2025)
- [Are Human Explanations Always Helpful? Towards Objective Evaluation of Human Natural Language Explanations](https://mlsanity.com/paper/2305.03117/) (2023)
- [Unifying Model Explainability and Robustness for Joint Text Classification and Rationale Extraction](https://mlsanity.com/paper/2112.10424/) (2021)
- [Flexible Instance-Specific Rationalization of NLP Models](https://mlsanity.com/paper/2104.08219/) (2021)
- [Beyond Labels: Empowering Human Annotators with Natural Language Explanations through a Novel Active-Learning Architecture](https://mlsanity.com/paper/2305.12710/) (2023)
- [Does Self-Rationalization Improve Robustness to Spurious Correlations?](https://mlsanity.com/paper/2210.13575/) (2022)
- [Are Shortest Rationales the Best Explanations for Human Understanding?](https://mlsanity.com/paper/2203.08788/) (2022)
- [XMD: An End-to-End Framework for Interactive Explanation-Based Debugging of NLP Models](https://mlsanity.com/paper/2210.16978/) (2022)
- [ER-Test: Evaluating Explanation Regularization Methods for Language Models](https://mlsanity.com/paper/2205.12542/) (2022)
- [Comparing Explanation Faithfulness between Multilingual and Monolingual Fine-tuned Language Models](https://mlsanity.com/paper/2403.12809/) (2024)
- [Should I Share this Translation? Evaluating Quality Feedback for User Reliance on Machine Translation](https://mlsanity.com/paper/2505.24683/) (2025)
- [On Positivity Bias in Negative Reviews](https://mlsanity.com/paper/2106.12056/) (2021)
- [The Role of Syntactic Span Preferences in Post-Hoc Explanation Disagreement](https://mlsanity.com/paper/2403.19424/) (2024)
- [Counterfactuals As a Means for Evaluating Faithfulness of Attribution Methods in Autoregressive Language Models](https://mlsanity.com/paper/2408.11252/) (2024)
- [You Can Do Better! If You Elaborate the Reason When Making Prediction](https://mlsanity.com/paper/2103.14919/) (2021)
- [Towards Human-Centred Explainability Benchmarks For Text Classification](https://mlsanity.com/paper/2211.05452/) (2022)
- [Situated Natural Language Explanations](https://mlsanity.com/paper/2308.14115/) (2023)

## Abstract (arXiv)

Two main approaches for evaluating the quality of machine-generated rationales are: 1) using human rationales as a gold standard; and 2) automated metrics based on how rationales affect model behavior. An open question, however, is how human rationales fare with these automatic metrics. Analyzing a variety of datasets and models, we find that human rationales do not necessarily perform well on these metrics. To unpack this finding, we propose improved metrics to account for model-dependent baseline performance. We then propose two methods to further characterize rationale quality, one based on model retraining and one on using "fidelity curves" to reveal properties such as irrelevance and redundancy. Our work leads to actionable suggestions for evaluating and characterizing rationales.