The Alien Space of Science: Sampling Coherent but Cognitively Unavailable Research Directions
Abstract
Scientific discovery is constrained not only by what is true, but by what is cognitively available to researchers. Many directions are coherent in light of the literature yet unlikely to be proposed because no existing community combines the relevant concepts, methods, and intuitions. Modern language models inherit this bias, recombining high-density regions of the literature when prompted for novel ideas. We target the complementary region, the alien space of science: directions that are scientifically plausible but unlikely under the distribution of existing researchers. Our method decomposes papers into conceptual units and clusters them into a vocabulary of idea atoms. A coherence model scores whether atom combinations form viable research directions, while an availability model scores whether any existing author community is positioned to produce them. We rank combinations to maximize coherence while minimizing availability. On 16,068 peer-reviewed LLM papers from NeurIPS, ICLR, ICML, and major NLP venues, the sampler produces more diverse research portfolios than frontier LLM baselines, including batch-prompted variants that see their entire output, while matching or exceeding their per-idea quality under blind human, LLM, and downstream experimental evaluation. In temporal validation, all components use only pre-2025 data. Selecting less than 1% of the candidate space recovers 2025 combinations at 35.9 times chance; weighting unavailability increases recovery of strictly novel combinations to 3.24 times chance. By separating scientific plausibility from community availability, our framework points toward AI ideation that complements rather than merely accelerates human science, expanding exploration into coherent directions the community may overlook.
NeurIPS 2026 Main Track. Paper page on Lacuna: lacuna.tiptreesystems.com/work/the-alien-space-of-science
Related
- Nonuniform iterative phasing framework and sampling requirements for 3D dynamical inversion from coherent surface scattering imaging
- Coherent and robust Bayesian inference under informative sampling via sample-level weight models
- Determinant Quantum-Quantum Monte Carlo: Coherent Auxiliary-Field Sampling
- Current state and future directions of nursing and allied health research in implantable left ventricular assist devices
- Burkitt lymphoma: current understanding, unmet needs, and research directions∗
BibTeX
@misc{artiles2026alien,
title = {{The Alien Space of Science: Sampling Coherent but Cognitively Unavailable Research Directions}},
author = {Alejandro H. Artiles and Martin Weiss and Levin Brinkmann and Iyad Rahwan and Bernhard Schölkopf and Christopher Pal and Hugo Larochelle and Anirudh Goyal and Nasim Rahaman},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2603.01092},
url = {https://zenodo.org/records/23575388},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - The Alien Space of Science: Sampling Coherent but Cognitively Unavailable Research Directions
AU - Alejandro H. Artiles
AU - Martin Weiss
AU - Levin Brinkmann
AU - Iyad Rahwan
AU - Bernhard Schölkopf
AU - Christopher Pal
AU - Hugo Larochelle
AU - Anirudh Goyal
AU - Nasim Rahaman
PY - 2026
DA - 2026/10/21
PB - Zenodo
DO - 10.48550/arxiv.2603.01092
UR - https://zenodo.org/records/23575388
AB - Scientific discovery is constrained not only by what is true, but by what is cognitively available to researchers. Many directions are coherent in light of the literature yet unlikely to be proposed because no existing community combines the relevant concepts, methods, and intuitions. Modern language models inherit this bias, recombining high-density regions of the literature when prompted for novel ideas. We target the complementary region, the alien space of science: directions that are scientifically plausible but unlikely under the distribution of existing researchers. Our method decomposes papers into conceptual units and clusters them into a vocabulary of idea atoms. A coherence model scores whether atom combinations form viable research directions, while an availability model scores whether any existing author community is positioned to produce them. We rank combinations to maximize coherence while minimizing availability. On 16,068 peer-reviewed LLM papers from NeurIPS, ICLR, ICML, and major NLP venues, the sampler produces more diverse research portfolios than frontier LLM baselines, including batch-prompted variants that see their entire output, while matching or exceeding their per-idea quality under blind human, LLM, and downstream experimental evaluation. In temporal validation, all components use only pre-2025 data. Selecting less than 1% of the candidate space recovers 2025 combinations at 35.9 times chance; weighting unavailability increases recovery of strictly novel combinations to 3.24 times chance. By separating scientific plausibility from community availability, our framework points toward AI ideation that complements rather than merely accelerates human science, expanding exploration into coherent directions the community may overlook. NeurIPS 2026 Main Track. Paper page on Lacuna: lacuna.tiptreesystems.com/work/the-alien-space-of-science
ER -