The limits of learning: Exploration, generalization, and the development of learning traps
Has this study been replicated?
The atlas records 2 replications and 1 reproduction of this study. Recorded outcomes: 2 successful, 1 mixed. The earliest on record is from 2023.
Replications
- Choice-contingent feedback creates a learning trap: Replication of Rich & Gureckis (2018), Experiment 1, Ostrowski et al. (2023). Outcome recorded: successful.
Our findings represent a successful replication of experiment 1 of Rich & Gureckis (2018). The differences between our replication and the original data were minor, with somewhat lower overall accuracy levels in our sample (as shown in Figure 2). The effect size for the difference in 1D accuracy during the test was similar across the two experiments, with our replication having a slightly lower effect size (Cohen's d original: 0.60; Cohen's d replication: 0.46). We found…
Outcome read from the full text. View paper - SCORE new data replication of Rich & Gureckis (2018), de Leeuw et al. (2025). Outcome recorded: successful.
"We conclude that the result replicated according to SCORE criteria."
Outcome read from OSF report.
Reproductions
- Reproduction (with author data & code): Rich & Gureckis (2018, Journal of Experimental Psychology: General), Parsons et al. (2022). OSF. Outcome recorded: computationally reproducible, robustness not checked.
Computational: "Reproduction outcome: Based on these criteria, the claim did reproduce." (stated for all 5 claims: g44qqk, m5xkkr, myjqqn, b2vooo, m8lkk9)
Outcome read from Computational: per-claim Reproduction outcome statements. View paper
Cite this record
Rich, A. S., & Gureckis, T. M. (2018). The limits of learning: Exploration, generalization, and the development of learning traps. Journal of Experimental Psychology: General, 147(11), 1553–1570. https://doi.org/10.1037/xge0000466
Other studies in the atlas
Failed replications · Successful replications · All browse pages