The hazards of explanation: Overgeneralization in the face of exceptions.
Abstract
Seeking explanations is central to science, education, and everyday thinking, and prompting learners to explain is often beneficial. Nonetheless, in 2 category learning experiments across artifact and social domains, we demonstrate that the very properties of explanation that support learning can impair learning by fostering overgeneralizations. We find that explaining encourages learners to seek broad patterns, hindering learning when patterns involve exceptions. By revealing how effects of explanation depend on the structure of what is being learned, these experiments simultaneously demonstrate the hazards of explaining and provide evidence for why explaining is so often beneficial. For better or for worse, explaining recruits the remarkable human capacity to seek underlying patterns that go beyond individual observations.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Williams et al. (2013) found that when humans are given opportunities to deliberate after receiving feedback, they learn more slowly and make more errors compared to those who do not deliberate.
“In these types of settings, Williams et al. (2013) reveal that when humans are given opportunities to deliberate after receiving feedback, they learn more slowly and make more errors compared to those who do not deliberate.”
From Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse (2024) · cites this paper as Williams et al. (2013)
When learning, people tended to reason about generalizable rules when conducting explanations, increasing the time needed to learn labels for the entire list (Williams et al., 2013).
“When learning, people tended to reason about generalizable rules when conducting explanations, increasing the time needed to learn labels for the entire list (Williams et al., 2013).”
From Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse (2024) · cites this paper as Williams et al. (2013)
Human participants who were prompted to explain the classification of exemplars performed worse because they tended to attempt the former strategy.
“Human participants who were prompted to explain the classification of exemplars performed worse because they tended to attempt the former strategy.”
From Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse (2024) · cites this paper as Williams et al. (2013)
When stimuli are designed such that broad and simple patterns contain exceptions, participants prompted to explain learned more slowly and made more errors.
“But when the stimuli are designed such that these broad and simple patterns contain exceptions, participants who were prompted to explain learned more slowly and made more errors (Williams et al., 2013).”
From Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse (2024) · cites this paper as Williams et al., 2013
When humans try to explain the category membership of exemplars, they tend to hypothesize simple classification rules, which can lead to inefficient learning when data contain arbitrary exceptions to these rules.
“when humans try to explain the category membership of exemplars, they tend to hypothesize simple classification rules, which can lead to inefficient learning when data contain arbitrary exceptions to these rules.”
From Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse (2024) · cites this paper as Williams et al. (2013)
The active form of thought mentally ingrains incorrect patterns that shift when exposed to successive unexpected answers, leading to deceptively incoherent lines of reasoning that hinder tracking correct labels.
“The active form of thought mentally ingrains incorrect patterns that shift when exposed to successive unexpected answers, altogether leading to the creation of many deceptively incoherent lines of reasoning throughout the learning process that hinder the ability to directly keep track of the correct labels even after multiple passes.”
From Mind Your Step (by Step): Chain-of-Thought can Reduce Performance on Tasks where Thinking Makes Humans Worse (2024) · cites this paper as Williams et al. (2013)
Cites
Explanation and Scientific Understanding
Friedman, M. (1974) · The Journal of Philosophy · doi:10.2307/2024924 · Cited by 1,386 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Explanation in artificial intelligence: Insights from the social sciences
Miller, T. (2019) · Artificial Intelligence · doi:10.1016/j.artint.2018.07.007 · Cited by 5,302 · see citing works
Human-centered XAI: Developing design patterns for explanations of clinical decision support systems
Schoonderwoerd, T. A. J., Jorritsma, W., Neerincx, M. A., & van den Bosch, K. (2021) · International Journal of Human-Computer Studies · doi:10.1016/j.ijhcs.2021.102684 · Cited by 210 · see citing works
Generalization Bias in Science
Peters, U., Krauss, A., & Braganza, O. (2022) · Cognitive Science · doi:10.1111/cogs.13188 · Cited by 63 · see citing works
Liu, R., Geng, J., Wu, A. J., Sucholutsky, I., Lombrozo, T., & Griffiths, T. L. (2024) · arXiv · doi:10.48550/arxiv.2410.21333 · Cited by 6 · see citing works
“But when the stimuli are designed such that these broad and simple patterns contain exceptions, participants who were prompted to explain learned more slowly and made more errors (Williams et al., 2013).”
InfluentialSelf-citation
- Authors
- Williams, J. J., Lombrozo, T., & Rehder, B.
- Affiliations
- University of California, Berkeley (US)
- New York University (US)
- Venue
- Journal of Experimental Psychology: General
- Published
- 2013
- Volume
- 142
- Issue
- 4
- Pages
- 1006-1014
- ISSN
- 0096-3445 (ISSN-L)
- 1939-2222
- OpenAlex
- W2110085097
- Cited by
- 88
- References
- 1
- Citations by year
- 2026: 5 · 2025: 3 · 2024: 5 · 2023: 7 · 2022: 6 · 2021: 7 · 2020: 7 · 2019: 8 · 2018: 5 · 2017: 6 · 2016: 11 · 2015: 7 · 2014: 9 · 2013: 1 · 2012: 1
- FWCI
- 11.50
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Journal of Experimental Psychology General (publishedVersion)
- PubMed (publishedVersion)
- Funders
- National Science Foundation (DRL-1056712)
- Topics
- Social Sciences › Psychology › Experimental and Cognitive Psychology › Visual and Cognitive Learning Processes
- Social Sciences › Psychology › Experimental and Cognitive Psychology › Categorization, perception, and language
- Social Sciences › Education › Science Education and Pedagogy
- Keywords
- explanation, overgeneralization, category learning, exception, pattern seeking, social sphere, learning from explanations
- MeSH terms
- Adult, Concept Formation, Generalization, Psychological, Humans, Learning
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
