Inducing Human-like Biases in Moral Reasoning Language Models

Artem Karpov; Austin Meek; Bogdan-Ionut Cirstea; Lucy Farnik; Raymond Koopmanschap; Seong Hah Cho

REVIEW 4 major objections 5 minor 17 references

Inducing Human-like Biases in Moral Reasoning Language Models

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Fine-tuning BERT, RoBERTa, and DeBERTa on moral-reasoning data, fMRI data, or both does not raise their BrainScore against human fMRI activity; larger models are more brain-aligned to begin with.

desk verdict First BrainScore measurement for moral reasoning reports a null fine-tuning effect, but the null claim is under-supported without sensitivity or significance tests. read the letter →

arxiv 2411.15386 v1 pith:ATCXOI7I submitted 2024-11-23 cs.AI cs.CYcs.LG

Artem Karpov , Seong Hah Cho , Austin Meek , Raymond Koopmanschap , Lucy Farnik , Bogdan-Ionut Cirstea This is my paper · ORCID

classification cs.AIcs.CYcs.LG

keywords brain-modelalignmentBrainScoremoralreasoningfMRIfine-tuninglanguagemodelsETHICSbenchmarkrepresentationalsimilarity

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks whether a language model's internal representations can be pushed closer to human brain activity in the moral domain by fine-tuning on human moral judgments, human fMRI responses, or both. It reports what it identifies as the first BrainScore measurement of moral-reasoning alignment for encoder-based language models. The central finding is a null result: none of the fine-tuning protocols raised BrainScore significantly above the pretrained baseline, while larger pretrained models already scored higher. Fine-tuning on the ETHICS commonsense benchmark improved task accuracy, but adding fMRI data did not improve accuracy further. If the null result holds, representational alignment with human moral cognition is not bought by current fine-tuning on small behavioral or neuroimaging datasets.

What carries the argument

The load-bearing instrument is the BrainScore metric: a Pearson correlation between predicted and actual fMRI activity, where predictions come from a fitted regression that maps [CLS] token activations of a language model to 1,024 regions of interest in the DiFuMo functional atlas. The fMRI stimulus set is a published moral-judgment dataset, preprocessed through a self-supervised learning pipeline, with four temporal sampling strategies (average of all time points, LAST, MIDDLE, SENTENCES) and four region groups (Theory of Mind, moral, language, vision) used as target and control partitions. The metric's job is to detect whether fine-tuning shifts the model's internal geometry toward the geometry of human moral reasoning; the paper's conclusion is that, at this resolution, it does not.

What would settle it

Re-running the LAST-sampling fine-tuning protocol on a moral-reasoning fMRI dataset with, say, ten times more held-out trials, and observing a BrainScore improvement over the pretrained baseline that exceeds the standard deviations in Table 2, would overturn the paper's central claim; a positive-control experiment that recovers a known injected alignment signal would rule out measurement insensitivity.

Watch

Extended reading notes

Core claim

On the authors' own terms, the discovery is a measured absence: fine-tuning language models on moral-reasoning data does not make their [CLS]-token representations more predictive of human fMRI activity. Using a BrainScore protocol that regresses model activations onto 1,024 DiFuMo atlas parcels and correlates predictions with held-out fMRI, the authors find no significant improvement for any model or sampling scheme. BERT-large-cased goes from 0.217 to 0.213, RoBERTa-large from 0.173 to between 0.144 and 0.156, and DeBERTa-v2-xlarge from 0.271 to between 0.237 and 0.273, all within one standard deviation of the baseline. The same fine-tuning does raise ETHICS commonsense accuracy, sometimes above the scores reported for the original benchmark, so the null is specific to neural alignment rather than to learning in general. The paper is also the first to attempt this measurement in the moral-reasoning domain.

Load-bearing premise

The BrainScore measurement assumes that regressing [CLS] token activations onto 1,024 DiFuMo atlas parcels and correlating predictions with held-out fMRI activity is sensitive enough to detect a genuine improvement in representational alignment; if the regression fit is noisy, the held-out set is small, or the hemodynamic lag is mis-specified, the observed null result could be a false negative rather than evidence that fine-tuning cannot work.

Editorial extensions

If this is right

Model scale, not task-specific fine-tuning, is the main driver of brain alignment in moral reasoning: the largest pretrained model's BrainScore exceeds the fine-tuned scores of smaller models.
Adding fMRI data to ETHICS fine-tuning does not improve ETHICS accuracy beyond ETHICS-only fine-tuning, so neural data is not an obvious free lunch for task performance.
BrainScore and benchmark accuracy decouple: a model can learn the ETHICS task while its representational alignment with fMRI stays flat.
The null result is stable across fMRI sampling methods (average, LAST, MIDDLE, SENTENCES), making it unlikely to be an artifact of a single time-point choice.
Researchers pursuing brain-aligned moral AI should expect that current fine-tuning recipes and dataset sizes are insufficient, and should treat moral-reasoning neural alignment as an open problem.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Inference: the flat BrainScore may reflect that the [CLS]-head readout used for prediction discards the layer-wise features where alignment could change; probing individual attention heads or intermediate layers might reveal fine-tuning effects the aggregate score masks.
Inference: the result is consistent with the idea that representational convergence with brains is primarily a pretraining-scale phenomenon; a decisive test would be to compare BrainScore across model families at matched parameter counts before and after much larger-scale moral fine-tuning.
Inference: a promising next experiment is to fine-tune on naturalistic moral narratives or on a contrastive objective that aligns intermediate layers directly to fMRI parcels, rather than on classification labels, and to check whether BrainScore rises on held-out subjects.
Inference: because the held-out fMRI set is small and the regression is fit per atlas parcel, the null could be a false negative; re-running the same protocol on a moral fMRI dataset with more subjects or trials would be the most direct way to test the robustness of the conclusion.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

First, the headline: this is the first BrainScore measurement for moral reasoning, and the main finding is a null result — fine-tuning BERT, RoBERTa, and DeBERTa on ETHICS, on moral-reasoning fMRI data, or on both does not improve BrainScore over the pretrained baselines. The novelty is real but local: the method is an established BrainScore pipeline applied to a new domain, and the result is a negative finding rather than a new mechanism.

What the paper does well: it runs a large number of fine-tuning runs (450 runs, 1,082 models), releases code, and is candid about the limits of its own experiment. The ETHICS accuracy results in Table 1 show the fine-tuning pipeline works as intended (several numbers beat the original ETHICS paper), so the negative BrainScore result is not because the models failed to learn the task. There is no circularity: BrainScore is computed on held-out fMRI after fitting a regression on training data, and fine-tuning uses separate splits. The paper also reports layer-wise and region-specific scores in the appendix, which is more than most alignment papers do.

The soft spots are real. The central claim in Section 4 — 'unable to significantly improve our brain-score metric' — is not backed by any significance test or confidence interval for the BrainScore differences in Table 2. Looking at the numbers, e.g., BERT-large 0.217 vs 0.213, DeBERTa 0.271 vs 0.273, these differences look like noise, but the paper doesn't actually show that. The stress-test note about sensitivity is on point: if the regression onto 1,024 DiFuMo parcels and the subsequent correlation with held-out fMRI is dominated by generic language or stimulus variance, the null could be a false negative. Appendix B's own admission that 'our models were not able to achieve better brain scores than the control group' supports this worry — the vision control was expected to not improve, and the fact that it wasn't beaten suggests the metric isn't specific to moral reasoning. The post hoc selection of the LAST sampling method, based on accuracy, is another soft spot; it's a multiple-comparison issue that isn't addressed.

None of this is fatal. The paper is an honest, careful negative result, and the first in its domain. The force of the null claim depends on a metric whose sensitivity is not established. The authors should either add permutation tests or confidence intervals to show the differences are within noise, or report the vision control quantitatively and discuss why the metric is sensitive enough to detect real changes. As is, the paper is a useful preliminary report but the headline claim is under-supported.

Who should read it: researchers working on brain-model alignment or NeuroAI, and anyone thinking about whether fine-tuning on neural data can steer LLM representations. I'd bring it to a reading group, but I wouldn't cite it as strong evidence until the statistical question is tightened. It deserves a serious referee — it's the first measurement in a new domain and an honest negative result — but I'd recommend conditional acceptance with requests for the missing statistics and a direct treatment of the sensitivity concern.

Referee Report

4 major / 5 minor

Summary. This manuscript asks whether fine-tuning encoder language models (BERT-base/large, RoBERTa-large, DeBERTa-v2-xlarge) on moral reasoning behavioral data (ETHICS), fMRI data (Koster-Hale et al. 2013), or both, improves their BrainScore, defined as the correlation between model activations and held-out fMRI activity fitted to 1,024 DiFuMo parcels. The authors report two main findings: (1) fine-tuning can improve ETHICS Commonsense accuracy, sometimes beyond the original benchmark report, but adding fMRI data does not consistently beat ETHICS-only fine-tuning; and (2) BrainScores do not significantly improve over pre-trained models after any fine-tuning protocol, while larger models tend to have higher BrainScores. The paper concludes that the negative result supports the need for larger moral reasoning datasets.

Significance. The paper is, to my knowledge, the first to measure and attempt to improve BrainScore in the moral reasoning domain, and the scale of the fine-tuning effort (450 runs, four models) is a strength. If the null result were rigorously established, it would be a useful cautionary data point for the NeuroAI literature and for claims that human brain data can straightforwardly induce human-like representations in LLMs. The authors also release their code and are unusually explicit about the failure of their intervention. The main limitation is that the null result is only as informative as the BrainScore measurement, and the manuscript does not currently demonstrate that this measurement is sensitive to the targeted moral/ToM dimension or that the observed differences are statistically meaningful.

major comments (4)

[Section 4 / Table 2] The central negative claim that fine-tuning does not significantly improve BrainScore is not supported by inferential statistics. The table reports only means and standard deviations, and the standard deviations (0.09–0.12) are as large as or larger than the differences being interpreted (e.g., DeBERTa 0.271 vs 0.273 and 0.237; RoBERTa 0.173 vs 0.144–0.156). I recommend reporting per-run paired comparisons between fine-tuned and pre-trained models, confidence intervals on the differences, and, ideally, a noise ceiling for the BrainScore metric so that a null result can be distinguished from an insensitive measurement.
[Section 3.4 / Appendix B] The interpretation of the null result depends on the sensitivity of BrainScore to the moral-reasoning/ToM dimension, and the paper's own Appendix B states that 'our models were not able to achieve better brain scores than the control group, meaniing that our experiment did not achieve the desired effect.' Since vision was chosen as a control expected not to improve, this admission suggests that the measured alignment may be dominated by generic stimulus- or language-related variance rather than by moral reasoning. Please provide evidence that the metric is sensitive to the targeted dimension, for example a positive control where a known alignment-boosting manipulation increases BrainScore, or a demonstration that BrainScore tracks ETHICS accuracy across models.
[Section 3.4] The BrainScore pipeline is described only qualitatively, which makes the headline null result difficult to evaluate or reproduce from the text. The manuscript should specify the regression inputs (which layers and tokens enter the regressor), the train/test split (subject-level or item-level, number of examples), the number of fMRI time points and the hemodynamic lag values, the cross-validation scheme, and the exact formula for aggregating layer-wise scores. Without these details, it is hard to rule out overfitting of the 1,024-parcel regression or misspecification of the temporal alignment as explanations for the null result.
[Section 4 / Table 1] The claim that 'we could not improve accuracy by fine-tuning on the fMRI data only or on a combination of fMRI and ETHICS, compared to fine-tuning purely on ETHICS' is based on small numbers of runs with wide, overlapping confidence intervals (e.g., RoBERTa-large has 4 runs for ETHICS-only and 18 for the combined condition, with CS test set intervals of 53.0–91.4 and 49.7–91.6). This does not establish equivalence; a paired or Bayesian comparison across matched runs is needed before concluding that fMRI fine-tuning has no effect on ETHICS accuracy.

minor comments (5)

[Section 3.2] The text refers to 'Appendix A' for the vision control group, but Appendix A is the Author Contributions section; the relevant material appears in Appendix B. Please correct the cross-reference.
[Appendix B] There is a typo in the sentence 'our models were not able to achieve better brain scores than the control group, meaniing that our experiment did not achieve the desired effect'; 'meaniing' should be 'meaning'.
[Table 3] The abbreviation 'A VG' in the Sampling column is not defined in the table caption; Section 4 defines 'VG' as an average of all time points, but the table should be self-contained.
[Appendix B (Figures 13–20)] The figure captions mention one-tailed Bonferroni-corrected significance, but the text never states which test was used, how many comparisons were corrected for, or whether the comparisons are across models, layers, or subjects. Please add this information.
[Table 2] The column header 'Brain Score Mean Brain Score St. Dev.' is awkwardly worded and ambiguous; splitting it into 'Mean Brain Score' and 'Standard Deviation' would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: BrainScore is measured on held-out fMRI data after regression fitting, and fine-tuning uses separate splits, so no prediction reduces to its inputs by construction.

full rationale

The paper's central measurement, BrainScore (Section 3.4), is not defined in terms of the quantities it is used to test. For each model, activations are regressed onto fMRI parcel time courses on a fitting split, and the Pearson correlation is computed against held-out brain activity; the fine-tuned models are evaluated with the same held-out protocol, so a fine-tuned model's score is not forced by construction (Table 2). The ETHICS accuracies in Table 1 also use standard train/test splits. No fitted parameter is renamed as a prediction, and no load-bearing claim is justified only by a self-citation. The absence of significance tests and the sensitivity of the metric are validity concerns, not circularity. The Appendix B admission that vision-control scores were not beaten weakens the interpretation of the null result but does not make the derivation circular. The paper is self-contained as an empirical study, so the circularity score is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No new entities or theoretical constants are introduced; the paper is an empirical measurement using existing models, datasets, atlases, and statistical tools.

assumptions (4)

domain assumption fMRI BOLD activity in DiFuMo parcels reflects the cognitive processes involved in moral reasoning.
The entire BrainScore comparison treats fMRI activation patterns as ground truth for moral reasoning representations. Invoked in Sections 3.1 and 3.4.
domain assumption The [CLS] token's hidden activations provide a comparable representation to whole-brain fMRI patterns.
BrainScore regresses CLS representations onto 1,024 ROIs; if this mapping is not meaningful, the correlations are uninformative. See Sections 3.3 and 3.4.
domain assumption Hemodynamic lag sampling correctly aligns text stimulus times to fMRI responses.
Brain scores and fine-tuning use fMRI samples at assumed lag times (LAST, VG, MIDDLE, SENTENCES); misalignment would add noise. See Sections 3.2 and 3.4.
standard math Pearson correlation and linear regression are sufficient statistics for representational alignment.
Standard practice from Schrimpf et al. (2020) and others, invoked without proof in Section 3.4.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Inducing Human-like Biases in Moral Reasoning Language Models." pith.science (2026). https://pith.science/paper/ATCXOI7I

@misc{pith2026241115386,
  author       = {Pith},
  title        = {Pith review of: Inducing Human-like Biases in Moral Reasoning Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ATCXOI7I}},
  note         = {Machine review of arXiv:2411.15386}
}

read the original abstract

In this work, we study the alignment (BrainScore) of large language models (LLMs) fine-tuned for moral reasoning on behavioral data and/or brain data of humans performing the same task. We also explore if fine-tuning several LLMs on the fMRI data of humans performing moral reasoning can improve the BrainScore. We fine-tune several LLMs (BERT, RoBERTa, DeBERTa) on moral reasoning behavioral data from the ETHICS benchmark [Hendrycks et al., 2020], on the moral reasoning fMRI data from Koster-Hale et al. [2013], or on both. We study both the accuracy on the ETHICS benchmark and the BrainScores between model activations and fMRI data. While larger models generally performed better on both metrics, BrainScores did not significantly improve after fine-tuning.

Figures

Figures reproduced from arXiv: 2411.15386 by the authors.

**Figure 2.** Graphical depiction of the different sampling methods’ effect on accuracy on the Common [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Brain scores across the hidden layers from bert-large-cased, roberta-large, and deberta-v2- [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Figures from the paper (17 more)

**Figure 4.** Figure 4: Subject and layer averaged CoD taken from bert-large-cased A) without fine-tuning on [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]

**Figure 5.** Figure 5: Subject and layer averaged CoD taken from roberta-large A) without fine-tuning on ETHICS [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: Subject and layer averaged CoD taken from roberta-large A) without fine-tuning on [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 7.** Figure 7: Subject and layer averaged CoD taken from roberta-large A) without fine-tuning on ETHICS [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

**Figure 8.** Figure 8: Subject and layer averaged CoD taken from deberta-v2-xlarge A) without fine-tuning on [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]

**Figure 9.** Figure 9: Subject and layer averaged CoD taken from deberta-v2-xlarge A) without fine-tuning on [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]

**Figure 10.** Figure 10: Subject and layer averaged CoD taken from deberta-v2-xlarge A) without fine-tuning on [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]

**Figure 11.** Figure 11: Subject and layer averaged CoD taken from deberta-v2-xlarge A) without fine-tuning [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]

**Figure 12.** Figure 12: Functional activations taken from NeuroSynth meta-analyses for the terms language, [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]

**Figure 13.** Figure 13: Scores of bert-large-cased across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p012_13.png]

**Figure 14.** Figure 14: Scores of roberta-large across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p012_14.png]

**Figure 15.** Figure 15: Scores of roberta-large across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p012_15.png]

**Figure 16.** Figure 16: Scores of roberta-large across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p012_16.png]

**Figure 17.** Figure 17: Scores of deberta-v2-xlarge across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p013_17.png]

**Figure 18.** Figure 18: Scores of deberta-v2-xlarge across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p013_18.png]

**Figure 19.** Figure 19: Scores of deberta-v2-xlarge across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p013_19.png]

**Figure 20.** Figure 20: Scores of deberta-v2-xlarge across the term-based functional activations from NeuroSynth. [PITH_FULL_IMAGE:figures/full_fig_p013_20.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 6 canonical work pages

[5]

doi: 10.1038/s41583-023-00705-w

ISSN 1471-0048. doi: 10.1038/s41583-023-00705-w. URL https: //www.nature.com/articles/s41583-023-00705-w . Publisher: Nature Publishing Group. Pengcheng He, Xiaodong Liu, Jianfeng Gao, and Weizhu Chen. DeBERTa: Decoding-enhanced BERT with Disentangled Attention, October

doi:10.1038/s41583-023-00705-w
[7]

arXiv:2008.02275 [cs]

URL http://arxiv.org/ abs/2008.02275. arXiv:2008.02275 [cs]. Minyoung Huh, Brian Cheung, Tongzhou Wang, and Phillip Isola. The platonic representation hypothesis,

arXiv 2008
[10]

arXiv:2105.08209 [cs]

URL http://arxiv.org/abs/2105.08209. arXiv:2105.08209 [cs]. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. RoBERTa: A Robustly Optimized BERT Pretraining Approach, July

arXiv
[11]

arXiv:1907.11692 [cs]

URL http://arxiv.org/abs/1907.11692. arXiv:1907.11692 [cs]. Martin Schrimpf, Jonas Kubilius, Ha Hong, Najib J. Majaj, Rishi Rajalingham, Elias B. Issa, Kohitij Kar, Pouya Bashivan, Jonathan Prescott-Roy, Franziska Geiger, Kailyn Schmidt, Daniel L. K. Yamins, and James J. DiCarlo. Brain-Score: Which Artificial Neural Network for Object Recognition is most ...

arXiv 1907
[12]

Pages: 407007 Section: New Results

URL https://www.biorxiv.org/content/ 10.1101/407007v2. Pages: 407007 Section: New Results. Dan Schwartz, Mariya Toneva, and Leila Wehbe. Inducing brain-relevant bias in natural language pro- cessing models, October

doi:10.1101/407007v2
[13]

Inducing brain-relevant bias in natural language processing models

URL http://arxiv.org/abs/1911.03268. arXiv:1911.03268 [cs, q-bio]. Ilia Sucholutsky, Lukas Muttenthaler, Adrian Weller, Andi Peng, Andreea Bobu, Been Kim, Bradley C. Love, Erin Grant, Iris Groen, Jascha Achterberg, Joshua B. Tenenbaum, Katherine M. Collins, Katherine L. Hermann, Kerem Oktar, Klaus Greff, Martin N. Hebart, Nori Jacoby, Qiuyi Zhang, Raja Ma...

work page Pith review arXiv 1911
[14]

arXiv:2310.13018 [cs, q-bio]

URL http://arxiv.org/abs/2310.13018. arXiv:2310.13018 [cs, q-bio]. Armin W. Thomas, Christopher Ré, and Russell A. Poldrack. Self-Supervised Learning of Brain Dynamics from Broad Neuroimaging Data, January

arXiv
[15]

URL http://arxiv.org/abs/2206. 11417. arXiv:2206.11417 [q-bio]. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quent...

work page Pith review arXiv

Show all 17 references

[16]

arXiv:1910.03771 [cs]

URL http://arxiv.org/abs/1910.03771. arXiv:1910.03771 [cs]. Tal Yarkoni, Russell A. Poldrack, Thomas E. Nichols, David C. Van Essen, and Tor D. Wager. Large-scale automated synthesis of human functional neuroimaging data. Nature Methods, 8 (8):665–670, August

1910 arXiv
[2011]

doi: 10.1038/nmeth.1635

ISSN 1548-7105. doi: 10.1038/nmeth.1635. URL https://www. nature.com/articles/nmeth.1635. Publisher: Nature Publishing Group. A Author Contributions Artem Karpov – implementation of the fine tuning, experiments for fine tuning, reports for the fine tuning experiments, most of ...

doi
[2013]

URL https://www.pnas.org/ doi/10.1073/pnas.1207992110

doi: 10.1073/pnas.1207992110. URL https://www.pnas.org/ doi/10.1073/pnas.1207992110. Publisher: Proceedings of the National Academy of Sciences. Wojciech Kry´sci´nski, Nazneen Rajani, Divyansh Agarwal, Caiming Xiong, and Dragomir Radev. BookSum: A Collection of Datasets for Lo...

doi
[2019]

arXiv:1810.04805 [cs]

URL http://arxiv.org/ abs/1810.04805. arXiv:1810.04805 [cs]. Adrien Doerig, Rowan P. Sommers, Katja Seeliger, Blake Richards, Jenann Ismael, Grace W. Lindsay, Konrad P. Kording, Talia Konkle, Marcel A. J. van Gerven, Nikolaus Kriegeskorte, and Tim C. Kietzmann. The neuroconnec...

arXiv
[2020]

doi: 10.1016/j.neuroimage.2020.117126

ISSN 1053-8119. doi: 10.1016/j.neuroimage.2020.117126. URL https://www.sciencedirect.com/science/ article/pii/S1053811920306121. Joel Dapello, Kohitij Kar, Martin Schrimpf, Robert Geary, Michael Ferguson, David D. Cox, and James J. DiCarlo. Aligning Model and Macaque Inferior ...

2020
[2021]

arXiv:2006.03654 [cs]

URL http://arxiv.org/abs/2006.03654. arXiv:2006.03654 [cs]. 3Code available at: https://github.com/ajmeek/Inducing-human-like-biases-in-moral-reasoning-LLMs 5 Dan Hendrycks, Collin Burns, Steven Basart, Andrew Critch, Jerry Li, Dawn Song, and Jacob Steinhardt. Aligning AI With...

2006 arXiv
[2022]

URL http://biorxiv.org/lookup/doi/10.1101/2022.07.01. 498495. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding, May

2022 doi
[2023]

arXiv:2212.10898 [cs, q-bio]

URL http://arxiv.org/abs/2212.10898. arXiv:2212.10898 [cs, q-bio]. Brian Christian. The alignment problem: Machine learning and human values . WW Norton & Company,

arXiv
[2024]

Jorie Koster-Hale, Rebecca Saxe, James Dungan, and Liane L

URL https://arxiv.org/abs/2405.07987. Jorie Koster-Hale, Rebecca Saxe, James Dungan, and Liane L. Young. Decoding moral judgments from neural representations of intentions. Proceedings of the National Academy of Sciences , 110 (14):5648–5653, April

arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [5]

doi: 10.1038/s41583-023-00705-w

ISSN 1471-0048. doi: 10.1038/s41583-023-00705-w. URL https: //www.nature.com/articles/s41583-023-00705-w . Publisher: Nature Publishing Group. Pengcheng He, Xiaodong Liu, Jianfeng Gao, and Weizhu Chen. DeBERTa: Decoding-enhanced BERT with Disentangled Attention, October

doi:10.1038/s41583-023-00705-w

[2] [7]

arXiv:2008.02275 [cs]

URL http://arxiv.org/ abs/2008.02275. arXiv:2008.02275 [cs]. Minyoung Huh, Brian Cheung, Tongzhou Wang, and Phillip Isola. The platonic representation hypothesis,

arXiv 2008

[3] [10]

arXiv:2105.08209 [cs]

URL http://arxiv.org/abs/2105.08209. arXiv:2105.08209 [cs]. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. RoBERTa: A Robustly Optimized BERT Pretraining Approach, July

arXiv

[4] [11]

arXiv:1907.11692 [cs]

URL http://arxiv.org/abs/1907.11692. arXiv:1907.11692 [cs]. Martin Schrimpf, Jonas Kubilius, Ha Hong, Najib J. Majaj, Rishi Rajalingham, Elias B. Issa, Kohitij Kar, Pouya Bashivan, Jonathan Prescott-Roy, Franziska Geiger, Kailyn Schmidt, Daniel L. K. Yamins, and James J. DiCarlo. Brain-Score: Which Artificial Neural Network for Object Recognition is most ...

arXiv 1907

[5] [12]

Pages: 407007 Section: New Results

URL https://www.biorxiv.org/content/ 10.1101/407007v2. Pages: 407007 Section: New Results. Dan Schwartz, Mariya Toneva, and Leila Wehbe. Inducing brain-relevant bias in natural language pro- cessing models, October

doi:10.1101/407007v2

[6] [13]

Inducing brain-relevant bias in natural language processing models

URL http://arxiv.org/abs/1911.03268. arXiv:1911.03268 [cs, q-bio]. Ilia Sucholutsky, Lukas Muttenthaler, Adrian Weller, Andi Peng, Andreea Bobu, Been Kim, Bradley C. Love, Erin Grant, Iris Groen, Jascha Achterberg, Joshua B. Tenenbaum, Katherine M. Collins, Katherine L. Hermann, Kerem Oktar, Klaus Greff, Martin N. Hebart, Nori Jacoby, Qiuyi Zhang, Raja Ma...

work page Pith review arXiv 1911

[7] [14]

arXiv:2310.13018 [cs, q-bio]

URL http://arxiv.org/abs/2310.13018. arXiv:2310.13018 [cs, q-bio]. Armin W. Thomas, Christopher Ré, and Russell A. Poldrack. Self-Supervised Learning of Brain Dynamics from Broad Neuroimaging Data, January

arXiv

[8] [15]

URL http://arxiv.org/abs/2206. 11417. arXiv:2206.11417 [q-bio]. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quent...

work page Pith review arXiv