Evolving Symbiosis, from Barricelli's Legacy to Collective Intelligence: a simulated and conceptual approach

This report from the ALICE 2026 workshop details the SymBa group's efforts to revive and extend Nils Aall Barricelli's 1953 research on symbiogenesis by replicating his 1D numerical organisms, developing 2D extensions and DNA-norm experiments, and exploring the implications of symbiogenesis for the origins of life, open-ended evolution, and collective intelligence in artificial systems.

The Gist

Imagine a digital sandbox where numbers aren't just counting things, but are actually living creatures.

This paper is a report from a group of scientists (dubbed "SymBa") who spent a week in Copenhagen playing with a very old, very cool idea from 1953. They wanted to answer a big question: How does life (or intelligence) actually get started?

Here is the story of their experiment, explained simply.

1. The Grandfather of the Idea: Nils Barricelli

Back in the 1950s, a pioneer named Nils Barricelli asked: "If you have a bunch of simple things that can copy themselves and make mistakes (mutate), will they eventually become complex life?"

He tried it with a computer that had less memory than a modern calculator. He created a world of 1D lines of numbers.

• The Rules: A number (say, a "4") tries to jump 4 spaces forward to copy itself.
• The Twist: If two numbers land on the same spot, they don't just crash. They interact based on a "rulebook" (called a Norm). Sometimes they merge, sometimes they fight, sometimes they create a new number.

Barricelli found that these numbers didn't just sit there. They formed Symbioorganisms. These were groups of numbers that stuck together, copied themselves as a team, and even developed "parasites" (numbers that lived off others) and "self-repairing" patterns.

The Analogy: Imagine a game of musical chairs where, instead of sitting down, the players grab hands and form a chain. If the chain is strong enough, it survives the music stopping. Barricelli found that these chains could evolve and get smarter.

2. The Big Question: Is "Teamwork" the Secret Sauce?

The group asked: Is "Symbiogenesis" (the act of two separate things merging to become one new thing) the missing key to life?

In biology, we know this happened. Our cells have "mitochondria" (the power plants) which used to be free-floating bacteria that decided to move inside our cells and stay there forever. They became one team.

• The Paper's Theory: Maybe life didn't start with one perfect organism. Maybe it started with a bunch of messy, independent parts that realized, "Hey, if we hold hands, we can do way more than if we run alone."

3. What Did This Group Do? (The Experiment)

Since Barricelli's old computer was too slow to prove everything, this team used modern computers to replay and upgrade his game.

A. The 1D and 2D Worlds

They recreated Barricelli's 1D line of numbers. It worked! They saw the same "symbioorganisms" emerge.
Then, they asked: "What if we give them a 2D grid (like a chessboard) instead of a line?"
They built this from scratch (Barricelli never got to finish it). They found that in 2D, these number-creatures formed waves, spirals, and complex patterns that moved and grew, just like cells in a petri dish.

B. The "DNA" Upgrade

Barricelli once joked that maybe these rules should look more like DNA. So, the team built a new version of the game using A, C, G, and T (the letters of DNA) instead of random numbers.

• The Rules: They made rules for how these letters stick together (A with T, C with G).
• The Result: When they let these DNA-letters interact, they didn't just float around. They started forming long, stable chains that could copy themselves and split apart.
• The Metaphor: Imagine a soup of loose Lego bricks. In the "old" version, they just bumped into each other. In the "DNA" version, the bricks have magnets. They snap together to build a tower, and then the tower splits in half to build two new towers. This is how life replicates!

4. Why Does This Matter for AI and Us?

The paper argues that this isn't just about numbers; it's about Intelligence.

• Collective Intelligence: Just like our cells work together to make you think, maybe AI needs to stop trying to build one giant "brain" and start building teams of small, simple agents that learn to cooperate.
• The "Interpreter" Problem: In our current AI, the rules are fixed by humans. But in nature, the "rules" (biology) and the "players" (genes) evolve together. The paper suggests that for AI to truly become alive, the "software" (the code) and the "hardware" (the machine reading the code) need to evolve together, like a symbiotic relationship.

5. The Takeaway

The paper concludes with a fun, honest admission: "It's complicated!! But fun :)"

They didn't solve the mystery of life. Instead, they took an old, dusty idea from 70 years ago, dusted it off, and showed that cooperation is a powerful engine for creating complexity.

The Final Metaphor:
Think of the universe as a giant, chaotic dance floor.

• Old View: Everyone dances alone, trying to be the best soloist.
• Barricelli's View (and this paper): The magic happens when two dancers realize that if they hold hands and spin together, they can create a move neither could do alone. That "holding hands" is Symbiogenesis, and it's how we went from single cells to humans, and maybe, how we will go from simple code to true artificial life.

Technical Summary

Here is a detailed technical summary of the manuscript "Evolving Symbiosis, from Barricelli's Legacy to Collective Intelligence: a simulated and conceptual approach."

1. Problem Statement

The paper addresses a fundamental gap in understanding the origins of life, open-ended evolution, and collective intelligence. While standard evolutionary theory relies on replication and mutation, the authors argue these mechanisms are insufficient to explain the emergence of complex, multi-gene organisms and major evolutionary transitions (e.g., the origin of eukaryotes).

The core problem is the lack of rigorous computational investigation into symbiogenesis—the creation of new entities through mutually beneficial (or otherwise) relationships between pre-existing entities. The authors aim to:

• Revive and rigorously analyze the pioneering work of Nils Aall Barricelli (1950s), who proposed "numerical organisms" in 1D cellular automata (CA) that evolve via symbiosis.
• Determine if symbiogenesis is a necessary mechanism for open-ended evolution and collective intelligence.
• Extend Barricelli's abstract numerical models to modern substrates, including 2D spaces and DNA-inspired interaction norms, to bridge the gap between abstract computation and biochemical reality.

2. Methodology

The study employs a multi-faceted approach combining historical replication, novel simulation, and theoretical analysis:

• Replication of Barricelli's 1D System:

  • The team re-implemented Barricelli's original 1D cellular automaton where "genes" are integers in a discrete space.
  • Mechanism: Numbers replicate by shifting to a new position based on their value ($a + n$). If a collision occurs (two numbers land on the same cell), specific collision norms (mutation rules) determine the outcome (e.g., annihilation, summation, or exclusion).
  • Symbiosis: Defined as a mechanism where interacting numbers trigger secondary replications, allowing multi-gene structures ("symbioorganisms") to emerge spontaneously.

• Extension to 2D Cellular Automata:

  • The authors generalized the 1D system to a 2D grid. Instead of scalar shifts, numbers represent 2D vectors (direction and magnitude).
  • Replication and collision norms were adapted to handle spatial interactions in two dimensions, testing if complex dynamics persist in higher-dimensional substrates.

• DNA-Norm Expansion (Minimal Biochemical Simulation):

  • To move beyond abstract numbers, the authors introduced a minimal DNA-norm system.
  • Substrate: A pool of monomers (A, C, G, T) and polymers (single/double strands).
  • Three Core Norms:
    1. Elongation: Monomers add to the 3' end of a strand.
    2. Complementary Association: Strands anneal via base-pairing; overlapping fragments merge to form longer double strands.
    3. Split: Double strands dissociate into two single strands (reproduction).
  • Experimental Conditions: The team compared a baseline (Elongation only) against a full DNA-norm condition (Elongation + Association + Split) in both well-mixed "soup" environments and spatially explicit 1D CAs.

• Boolean CA Symbiogenesis:

  • A preliminary experiment gating a standard Boolean CA rule (Rule 110) with symbiogenetic norms to test if symbiosis can emerge in binary substrates.

• Metrics and Analysis:

  • Information Theory: Calculation of Shannon entropy and Mutual Information to quantify system complexity.
  • Motif Repetition: Measuring the fraction of repeated $k$-mers (sequence motifs) to detect the emergence of stable, heritable structures.
  • Robustness Testing: Simulating perturbations (collisions with foreign numbers) to measure symbioorganism survival rates.

3. Key Contributions

• Historical Replication & Validation: Successfully reproduced Barricelli's 1D numerical experiments, confirming that symbiotic interactions (collisions leading to secondary replication) are sufficient to generate self-replicating, multi-gene structures without explicit design.
• 2D Generalization: Provided the first implementation of Barricelli's system in 2D, demonstrating that complex, self-propagating wave patterns and symbioorganisms emerge in higher-dimensional spaces.
• DNA-Norm Implementation: Developed and simulated a minimal set of DNA-inspired rules (elongation, annealing, splitting) that bridge the gap between abstract numerical automata and molecular biology.
• Theoretical Framework: Proposed a conceptual link between symbiogenesis and collective intelligence, arguing that symbiotic integration is a mechanism for scaling problem-solving capabilities and creating new levels of individuality (from cells to holobionts).
• Open-Endedness Hypothesis: Posited that symbiogenesis is a critical driver for open-ended evolution, allowing systems to combine independently evolved modules into higher-order functional units.

4. Results

• 1D & 2D Numerical Systems:
  • Sparse initial conditions rapidly organize into dense fields of repeating, self-propagating patterns.
  • Under "0 norm" (annihilation on collision), strong selection pressure favors spatial strategies that minimize conflict.
  • 2D simulations exhibit wave propagation and persistent structures similar to, but more complex than, the 1D case.
• DNA-Norm Experiments:
  • Motif Repetition: In the "soup" environment, the full DNA-norm condition (Condition B) significantly outperformed the elongation-only baseline (Condition A) in generating repeated longer motifs ($k=6, 8$). This confirms that complementary association and splitting are essential for the reuse and duplication of long sequence segments.
  • Spatial Structure: In the 1D CA embedding, Condition A produced only short-lived, sparse streaks. Condition B generated contiguous, cone-shaped domains of dominant motifs that expanded and persisted. This demonstrates that local interactions (annealing/splitting) can sustain heritable spatial lineages and diversity.
• Robustness: Preliminary analysis suggests symbioorganisms exhibit resilience to damage, with survival rates increasing when multiple organisms cooperate, validating Barricelli's original claims.
• Boolean CA: Early results show that gating Rule 110 with symbiogenetic norms can produce persistent, duplicating patterns, suggesting symbiosis can emerge even in binary substrates.

5. Significance and Implications

• For Artificial Life (ALife): The work revitalizes Barricelli's legacy, proving that his "numerical organisms" are a valid substrate for studying the origins of life. It suggests that symbiosis is not just a biological curiosity but a fundamental computational principle for generating complexity.
• For Artificial Intelligence (AI):
  • Collective Intelligence: The paper reframes collective intelligence as a process of symbiotic integration, where agents merge to solve problems they cannot tackle individually.
  • Evolvability: Symbiogenesis offers a pathway for systems to rapidly acquire new capabilities by combining existing functional modules, potentially solving the "evolvability" bottleneck in evolutionary algorithms.
  • Co-evolving Interpreters: The authors propose a critical future direction: moving beyond fixed interaction rules (interpreters) to systems where the interpreter itself evolves alongside the genome. This mirrors biological reality where the cellular machinery (ribosomes, regulatory networks) co-evolves with genetic code.
• Theoretical Biology: The findings support the "Combinatorial Functional Niching" hypothesis, suggesting that life evolves by integrating orthogonal capabilities to explore new functional niches, a process driven by symbiotic interactions rather than just competition.

Conclusion:
This report serves as a foundational step toward a "renaissance" of symbiotic approaches in AI and ALife. By combining rigorous simulation of historical models with novel DNA-inspired norms, the authors provide evidence that symbiogenesis is a robust mechanism for open-ended evolution, collective intelligence, and the emergence of complex, self-repairing systems. The work calls for future research into co-evolving interpreters and the application of these principles to neural networks and multi-agent systems.