NLP's Clever Hans Moment has Arrived
The DOI matches its OpenAlex record.
Abstract
Large, pretrained language models have led to a flurry of new state-of-the-art results being reported in many areas of natural language processing. However, recent work has also shown that such models tend to solve language tasks by relying on superficial cues found in benchmark datasets, instead of acquiring the capabilities envisioned by the task designers. In this short opinion piece, I review a report by Niven & Kao (2019) of this so-called Clever Hans effect on an argument reasoning task and discuss possible solutions for its prevention.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cited by
Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data
Bender, E. M., & Koller, A. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.463 · Cited by 1,082
“One approach to providing grounding is to train distributional models on corpora augmented with perceptual data, such as photos (Hossain et al., 2019) or other modalities (Kiela and Clark, 2015; Kiela et al., 2015).”
BackgroundHutchinson, B., Smart, A., Hanna, A., Denton, R., Greer, C., Kjartansson, O., Barnes, P., & Mitchell, M. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445918 · Cited by 279
On the genealogy of machine learning datasets: A critical history of ImageNet
Denton, E., Hanna, A., Amironesei, R., Smart, A., & Nicole, H. (2021) · Big Data & Society · doi:10.1177/20539517211035955 · Cited by 221
“2021) and publications that focus solely on dataset creation tend to be devalued within traditional peer-review processes (Heinzerling, 2019).”
BackgroundBringing the People Back In: Contesting Benchmark Machine Learning Datasets
Denton, R., Hanna, A., Amironesei, R., Smart, A., Nicole, H., & Scheuerman, M. K. (2020) · arXiv · doi:10.48550/arxiv.2007.07399 · Cited by 100
“…curriculum (e.g. Goodfellow et al. (2016)), the under-specification or data decisions in publications accompanying new datasets (Geiger et al., 2020; Scheuerman et al., 2020), and the routine undervaluing of the work that goes into the construction of datasets (Heinzerling, 2019; Jo & Gebru, 2020).”
BackgroundData and its (dis)contents: A survey of dataset development and use in machine learning research
Paullada, A., Raji, I. D., Bender, E. M., Denton, E., & Hanna, A. (2021) · Patterns · doi:10.1016/j.patter.2021.100336 · Cited by 548
Metadata
- Authors
- Heinzerling, B.
- Venue
- The Journal of Cognitive Science
- Published
- 2020-01-01
- Volume
- 21
- Issue
- 1
- Pages
- 159-167
- OpenAlex
- W3026320104
- Cited by
- 19
- References
- 0
- FWCI
- 1.89
- Type
- journal article
- Language
- en
- Open access
- closed
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: NLP's Clever Hans Moment has Arrived. https://app.cochat.ai/papers/10.17791/jcs.2020.21.1.161
- The paper
- Heinzerling, B. (2020). NLP's Clever Hans Moment has Arrived. The Journal of Cognitive Science. https://doi.org/10.17791/jcs.2020.21.1.161
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.