Cognitive Cells: A Compositional Framework for Populations of Small Language Models
Abstract
Recent work on large language models and agentic systems raises a basic question that current practice leaves open: how should artificial cognition be decomposed, measured, and composed? We propose studying multi-agent systems from a fixed unit we call a cognitive cell: a small, frozen language model with bounded memory and a message interface. The methodological commitment, the fixed-cell principle, is to hold this unit constant and vary only the population size, the communication topology, the message bandwidth, and the coordination protocol, so that collective behavior becomes a measurable property of a known device rather than an artifact of per-study engineering. We characterize a single cell by a compact datasheet of measurable parameters, and we ask when replicating and connecting cells improves performance: first we measure how one cell behaves alone, then we replicate it and test when voting, communication, and topology help. Instantiating the framework with small frozen models (1.5 and 3 billion parameters), we report a first round of measurements. Adding cells helps only when their errors are not too correlated. A simple correct/incorrect voting model is a useful but conservative null: real open-ended voting can exceed it, because errors are dispersed across many wrong answers rather than concentrated on one. Popular interactive protocols, namely debate, a shared blackboard, and chain revision, do not beat a matched-cost voting baseline in our setting. Finally, a cell's ability to relay several facts, itself a datasheet quantity, predicts whether a population can solve tasks whose evidence exceeds any single cell's memory. We present these as initial measurements within a broader program on scalable artificial cognition, in which multi-agent architectures appear as the special case of cells autonomous enough to be treated as agents.
Related
- Compositional Aeroelastic Operators for Morphing Flexible Multibody Aircraft: A Geometric Framework with Structural Verification
- Measuring Collective Semantic Change in Populations of Language Model Agents
- Large-Language Models as a Cognitive Virus
- A Cognitive-Aware QML-CRL Framework for Detecting Affinity and Romance-Investment Fraud
- GWPTools: An extensible Wolfram Language framework for time-dependent Gaussian wave packet dynamics
BibTeX
@misc{ferreira2026cognitive,
title = {{Cognitive Cells: A Compositional Framework for Populations of Small Language Models}},
author = {Silvan Ferreira},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2608.28606},
eprint = {2608.28606},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.28606},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Cognitive Cells: A Compositional Framework for Populations of Small Language Models
AU - Silvan Ferreira
PY - 2026
DA - 2026/08/01
PB - arXiv
DO - 10.48550/arxiv.2608.28606
UR - https://arxiv.org/abs/2608.28606
AB - Recent work on large language models and agentic systems raises a basic question that current practice leaves open: how should artificial cognition be decomposed, measured, and composed? We propose studying multi-agent systems from a fixed unit we call a cognitive cell: a small, frozen language model with bounded memory and a message interface. The methodological commitment, the fixed-cell principle, is to hold this unit constant and vary only the population size, the communication topology, the message bandwidth, and the coordination protocol, so that collective behavior becomes a measurable property of a known device rather than an artifact of per-study engineering. We characterize a single cell by a compact datasheet of measurable parameters, and we ask when replicating and connecting cells improves performance: first we measure how one cell behaves alone, then we replicate it and test when voting, communication, and topology help. Instantiating the framework with small frozen models (1.5 and 3 billion parameters), we report a first round of measurements. Adding cells helps only when their errors are not too correlated. A simple correct/incorrect voting model is a useful but conservative null: real open-ended voting can exceed it, because errors are dispersed across many wrong answers rather than concentrated on one. Popular interactive protocols, namely debate, a shared blackboard, and chain revision, do not beat a matched-cost voting baseline in our setting. Finally, a cell's ability to relay several facts, itself a datasheet quantity, predicts whether a population can solve tasks whose evidence exceeds any single cell's memory. We present these as initial measurements within a broader program on scalable artificial cognition, in which multi-agent architectures appear as the special case of cells autonomous enough to be treated as agents.
ER -