Settling the Reward Hypothesis

Michael Bowling, John D Martin, David Abel, Will Dabney
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:3003-3020, 2023.

Abstract

The reward hypothesis posits that, "all of what we mean by goals and purposes can be well thought of as maximization of the expected value of the cumulative sum of a received scalar signal (reward)." We aim to fully settle this hypothesis. This will not conclude with a simple affirmation or refutation, but rather specify completely the implicit requirements on goals and purposes under which the hypothesis holds.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-bowling23a, title = {Settling the Reward Hypothesis}, author = {Bowling, Michael and Martin, John D and Abel, David and Dabney, Will}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {3003--3020}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/bowling23a/bowling23a.pdf}, url = {https://proceedings.mlr.press/v202/bowling23a.html}, abstract = {The reward hypothesis posits that, "all of what we mean by goals and purposes can be well thought of as maximization of the expected value of the cumulative sum of a received scalar signal (reward)." We aim to fully settle this hypothesis. This will not conclude with a simple affirmation or refutation, but rather specify completely the implicit requirements on goals and purposes under which the hypothesis holds.} }
Endnote
%0 Conference Paper %T Settling the Reward Hypothesis %A Michael Bowling %A John D Martin %A David Abel %A Will Dabney %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-bowling23a %I PMLR %P 3003--3020 %U https://proceedings.mlr.press/v202/bowling23a.html %V 202 %X The reward hypothesis posits that, "all of what we mean by goals and purposes can be well thought of as maximization of the expected value of the cumulative sum of a received scalar signal (reward)." We aim to fully settle this hypothesis. This will not conclude with a simple affirmation or refutation, but rather specify completely the implicit requirements on goals and purposes under which the hypothesis holds.
APA
Bowling, M., Martin, J.D., Abel, D. & Dabney, W.. (2023). Settling the Reward Hypothesis. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:3003-3020 Available from https://proceedings.mlr.press/v202/bowling23a.html.

Related Material