Intelligence Without a Brain—Perspectives from the Aneural Slime Mold as a Computing Model for Artificial Intelligence


Physarum polycephalum (slime mold) has demonstrated numerous sophisticated behaviors that qualify as "intelligent" despite being a single-celled organism without a nervous system. Key experiments and capabilities include:

Pathfinding and Network Optimization

  • Maze-solving: In landmark experiments, plasmodia placed in mazes with food sources at opposite ends explored the entire structure, then retracted all unnecessary protoplasmic tubes, leaving only the shortest path connecting the food sources.
  • Multi-source optimization: When presented with multiple food sources, Physarum forms networks that follow mathematical principles, optimizing not just for shortest paths but also for robustness and fault tolerance (maintaining alternative routes in case of disconnection).
  • These abilities inspired new algorithms for solving computational problems like the traveling salesman problem and minimum spanning tree calculations.

Decision-Making and Nutritional Intelligence

  • Risk-assessment: When choosing between low-quality food in darkness (safe) versus higher-quality food in light (normally avoided), Physarum weighs the trade-offs—selecting the safer option unless the nutritional difference is substantial enough to justify the risk.
  • Nutritional balancing: Given multiple food sources with suboptimal nutrient ratios, Physarum doesn't select a single option but strategically spreads across several sources so that the combined intake achieves its preferred carbohydrate-to-protein ratio—demonstrating integration of multiple information sources.
  • Context-dependent choices: In three-option experiments with balanced nutritional values, introducing a third undesirable option (low-value food in darkness) shifted preferences away from the lit option in well-fed plasmodia—displaying the same "irrational" context-dependent decision-making seen in humans and animals.
  • Variable decision speeds: Individual plasmodia show consistent personality differences in decision-making speed, with some consistently making faster choices than others.

Learning and Memory

  • Anticipation: After experiencing aversive stimuli at regular intervals just three times, plasmodia began slowing their movement at the expected time—even when the stimulus was absent—demonstrating temporal anticipation. This memory could be renewed with a single reinforcement after fading.
  • Habituation: Plasmodia crossing quinine-laced bridges initially avoided the repellent (narrow tubes, long crossing times), but after nine repetitions behaved as if crossing normal agar—showing decreased response to repeated non-harmful stimuli. This memory was:
    • Specific (habituation to quinine didn't transfer to caffeine)
    • Temporary (spontaneously returned after 48 hours without exposure)
  • Memory transfer: When a habituated plasmodium fused with a non-habituated one, cytoplasmic exchange transferred the habituation memory—demonstrating non-neural information transfer.
  • External spatial memory: Physarum deposits extracellular slime as it moves and avoids areas marked with this slime, effectively using the environment as an externalized memory system to avoid re-exploring areas—unless completely surrounded by its own slime, when it overrides this avoidance to prevent self-trapping.

Complex Problem Solving

  • Two-armed bandit problem: Physarum successfully solved this decision-making challenge (choosing between two options with variable rewards) by sampling both paths and adjusting behavior based on outcomes—performing comparably to vertebrate animals even in unpredictable environments.

These experiments collectively demonstrate that Physarum polycephalum exhibits adaptive behavior, information integration, learning, memory formation and transfer, risk assessment, and optimization capabilities—all without neurons or a centralized brain—challenging traditional definitions of intelligence and offering insights for bio-inspired computing architectures.


Last modified on 13-Feb-26

/ EMMIND - Electromagnetic Mind