Hostname: page-component-7dd5485656-g8tfn Total loading time: 0 Render date: 2025-10-31T16:56:39.377Z Has data issue: false hasContentIssue false
  • English
  • Français

Meta-learning: Data, architecture, and both

Published online by Cambridge University Press:  23 September 2024

Marcel Binz Open the ORCID record for Marcel Binz [Opens in a new window] ,
Ishita Dasgupta ,
Akshay Jagadish ,
Matthew Botvinick ,
Jane X. Wang  and
Eric Schulz
Marcel Binz*
Affiliation:
Max Planck Institute for Biological Cybernetics, Tübingen, Germany Helmholtz Institute for Human-Centered AI, Munich, Germany marcel.binz@helmholtz-munich.de akshay.jagadish@helmholtz-munich.de eric.schulz@helmholtz-munich.de
Ishita Dasgupta
Affiliation:
Google DeepMind, London, UK idg@google.com botvinick@google.com wangjane@google.com
Akshay Jagadish
Affiliation:
Max Planck Institute for Biological Cybernetics, Tübingen, Germany Helmholtz Institute for Human-Centered AI, Munich, Germany marcel.binz@helmholtz-munich.de akshay.jagadish@helmholtz-munich.de eric.schulz@helmholtz-munich.de
Matthew Botvinick
Affiliation:
Google DeepMind, London, UK idg@google.com botvinick@google.com wangjane@google.com
Jane X. Wang
Affiliation:
Google DeepMind, London, UK idg@google.com botvinick@google.com wangjane@google.com
Eric Schulz
Affiliation:
Max Planck Institute for Biological Cybernetics, Tübingen, Germany Helmholtz Institute for Human-Centered AI, Munich, Germany marcel.binz@helmholtz-munich.de akshay.jagadish@helmholtz-munich.de eric.schulz@helmholtz-munich.de
*
*Corresponding author.
Get access Rights & Permissions [Opens in a new window]

Abstract

We are encouraged by the many positive commentaries on our target article. In this response, we recapitulate some of the points raised and identify synergies between them. We have arranged our response based on the tension between data and architecture that arises in the meta-learning framework. We additionally provide a short discussion that touches upon connections to foundation models.

Information

Type
Authors' Response
Information
Behavioral and Brain Sciences , Volume 47 , 2024 , e170
Check for updates
Copyright
Copyright © The Author(s), 2024. Published by Cambridge University Press

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Article purchase

Temporarily unavailable

References

Anderson, J. R. (2013). The adaptive character of thought. Psychology Press.CrossRefGoogle Scholar
Binz, M., & Schulz, E. (2022). Modeling human exploration through resource-rational reinforcement learning. Advances in Neural Information Processing Systems, 35, 31755–31768.Google Scholar
Binz, M., & Schulz, E. (2023). Turning large language models into cognitive models. arXiv preprint arXiv:2306.03917.Google Scholar
Binz, M., Gershman, S. J., Schulz, E., & Endres, D. (2022). Heuristics from bounded meta-learned inference. Psychological Review, 129(5), 1042.CrossRefGoogle ScholarPubMed
Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., … Amodei, D. (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems, 33, 1877–1901.Google Scholar
Buschoff, L. M. S., Schulz, E., & Binz, M. (2023). The acquisition of physical knowledge in generative neural networks. In International Conference on Machine Learning (pp. 30321–30341). PMLR.Google Scholar
Chan, S., Santoro, A., Lampinen, A., Wang, J., Singh, A., Richemond, P., … Hill, F. (2022). Data distributional properties drive emergent in-context learning in transformers. Advances in Neural Information Processing Systems, 35, 18878–18891.Google Scholar
Conwell, C., Prince, J. S., Kay, K. N., Alvarez, G. A., & Konkle, T. (2022). What can 1.8 billion regressions tell us about the pressures shaping high-level visual representation in brains and machines?. BioRxiv, 2022-03.Google Scholar
Dasgupta, I., Schulz, E., Tenenbaum, J. B., & Gershman, S. J. (2020). A theory of learning to infer. Psychological Review, 127(3), 412.CrossRefGoogle ScholarPubMed
De Raedt, L., Manhaeve, R., Dumancic, S., Demeester, T., & Kimmig, A. (2019). Neuro-symbolic= neural+ logical+ probabilistic. In NeSy’19@ IJCAI, the 14th International Workshop on Neural-Symbolic Learning and Reasoning.Google Scholar
Griffiths, T. L., Kemp, C., Tenenbaum, J. B.. (2008). Bayesian models of cognition. In Sun, R. (Ed.), The Cambridge handbook of computational psychology (pp. 59–100). Cambridge University Press.Google Scholar
Harrison, P., Marjieh, R., Adolfi, F., van Rijn, P., Anglada-Tort, M., Tchernichovski, O., … Jacoby, N. (2020). Gibbs sampling with people. Advances in Neural Information Processing Systems, 33, 10659–10671.Google Scholar
Hussain, Z., Binz, M., Mata, R., & Wulff, D. U. (2023). A tutorial on open-source large language models for behavioral science. PsyArXiv preprint.CrossRefGoogle Scholar
Jagadish, A. K., Binz, M., Saanum, T., Wang, J. X., & Schulz, E. (2023). Zero-shot compositional reinforcement learning in humans.CrossRefGoogle Scholar
Jagadish, A. K., Coda-Forno, J., Thalmann, M., Schulz, E., & Binz, M. (2024). Ecologically rational meta-learned inference explains human category learning. arXiv preprint arXiv:2402.01821.Google Scholar
Kumar, S., Correa, C. G., Dasgupta, I., Marjieh, R., Hu, M. Y., Hawkins, R., … Griffiths, T. (2022). Using natural language and program abstractions to instill human inductive biases in machines. Advances in Neural Information Processing Systems, 35, 167–180.Google Scholar
Lake, B. M., & Baroni, M. (2023). Human-like systematic generalization through a meta-learning neural network. Nature, 623(7985), 115–121.CrossRefGoogle ScholarPubMed
Marcus, G. F. (1998). Rethinking eliminative connectionism. Cognitive Psychology, 37(3), 243–282.CrossRefGoogle ScholarPubMed
Muttenthaler, L., Dippel, J., Linhardt, L., Vandermeulen, R. A., & Kornblith, S. (2022). Human alignment of neural network representations. arXiv preprint arXiv:2211.01201.Google Scholar
Ortega, P. A. (2020). Twitter. Retrieved from https://twitter.com/adaptiveagents/status/1334609817432940550?lang=bnGoogle Scholar
Ortega, P. A., Wang, J. X., Rowland, M., Genewein, T., Kurth-Nelson, Z., Pascanu, R., … Legg, S. (2019). Meta-learning of sequential strategies. arXiv preprint arXiv:1905.03030.Google Scholar
Palminteri, S., & Lebreton, M. (2022). The computational roots of positivity and confirmation biases in reinforcement learning. Trends in Cognitive Sciences, 26(7), 607–621.CrossRefGoogle ScholarPubMed
Saanum, T., Éltető, N., Dayan, P., Binz, M., & Schulz, E. (2023). Reinforcement learning with simple sequence priors. Advances in Neural Information Processing Systems, 36, 61985–62005.Google Scholar
Schubert, J. A., Jagadish, A. K., Binz, M., & Schulz, E. (2024). In-context learning agents are asymmetric belief updaters. arXiv preprint arXiv:2402.03969.Google Scholar
SIMA Team, Raad, M. A., Ahuja, A., Barros, C., Besse, F., Bolt, A., … Young, N. (2024). Scaling instructable agents across many simulated worlds. Technical Report.Google Scholar
Sutton, R. (2019). The bitter lesson. Incomplete Ideas (blog), 13(1), 38.Google Scholar
Team, A. A., Bauer, J., Baumli, K., Baveja, S., Behbahani, F., Bhoopchand, A., … Zhang, L. (2023). Human-timescale adaptation in an open-ended task space. arXiv preprint arXiv:2301.07608.Google Scholar
Tian, Y., Sun, C., Poole, B., Krishnan, D., Schmid, C., & Isola, P. (2020). What makes for good views for contrastive learning? Advances in Neural Information Processing Systems, 33, 6827–6839.Google Scholar
Wang, G., Xie, Y., Jiang, Y., Mandlekar, A., Xiao, C., Zhu, Y., … Anandkumar, A. (2023). Voyager: An open-ended embodied agent with large language models. arXiv preprint arXiv:2305.16291.Google Scholar
Wang, J. X., Kurth-Nelson, Z., Tirumala, D., Soyer, H., Leibo, J. Z., Munos, R., … Botvinick, M. (2016). Learning to reinforcement learn. arXiv preprint arXiv:1611.05763.Google Scholar
Yax, N., Anlló, H., & Palminteri, S. (2023). Studying and improving reasoning in humans and machines. arXiv preprint arXiv:2309.12485.Google Scholar