Neurons as autonomous agents: A biologically inspired framework for cognitive architectures in artificial intelligence
Bibliographic record
Abstract
Despite impressive recent advances in artificial intelligence (AI), current deep neural networks still lack the adaptability and energy efficiency inherent to biological systems. Here we suggest that this problem may be overcome by taking inspiration from the brain where neurons operate as autonomous agents, each capable of adjusting its synaptic connections and internal states based on local information. Currently, typical artificial neurons are static nodes, which is in striking contrast to the rich, dynamic computations performed by biological neurons. In this review, we propose redesigning artificial neurons as self-regulating, agent-like units, making actions to maximize future energy/reward. Similarly, as single-celled organisms which can autonomously navigate in complex environments in search for food, neurons can also be viewed as autonomous decision-makers, seeking to maximize their own energy resources. Thus, neurons could be operating similarly like reinforcement learning (RL) agents, which make actions to obtain maximum future reward. Here first we review literature illustrating that biological neurons perform complex computations and employ local, predictive learning rules to anticipate future activity to maximize metabolic energy. Next, we provide examples of recent biologically inspired learning algorithms where artificial neurons are empowered with computational flexibility, similarly to autonomous agents. Networks with neurons using such local learning rules can in some examples outperform current AI algorithms. We also discuss how this can improve scalability of current multi-agent systems (MAS) and energy efficiency. Therefore, designing neurons as autonomous agents may provide an important step toward building human-like cognition.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.004 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".