Abstract
How can systems in which individuals’ inner workings are very similar to each other, as neural networks or ant colonies, produce so many qualitatively different behaviors, giving rise to roles and specialization? In this work, we bring new perspectives to this question by focusing on the underlying network that defines how individuals in these systems interact. We applied a genetic algorithm to optimize rules and connections of cellular automata in order to solve the density classification task, a classical problem used to study emergent behaviors in decentralized computational systems. The networks used were all generated by the introduction of shortcuts in an originally regular topology, following the small-world model. Even though all cells follow the exact same rules, we observed the existence of different classes of cells’ behaviors in the best cellular automata found—most cells were responsible for memory and others for integration of information. Through the analysis of structural measures and patterns of connections (motifs) in successful cellular automata, we observed that the distribution of shortcuts between distant regions and the speed in which a cell can gather information from different parts of the system seem to be the main factors for the specialization we observed, demonstrating how heterogeneity in a network can create heterogeneity of behavior.
Highlights
IntroductionWe used in this work cellular automaton, a classic agent-based model
As previously stated, we used in this work cellular automaton, a classic agent-based model
The first thing the reader can notice is that, to results seen in the literature [47], the evolution was faster in finding multiple good solutions when we allowed the topology to be rewired: in searches with p > 0 we observed a general improvement of fitness around the 4th epoch, while for fixed topologies such improvement was seen only after 8 epochs
Summary
We used in this work cellular automaton, a classic agent-based model. Complex and other massively distributed systems, are very different from such usual model of computation: while in a von Neumann architecture the CPU has access to the whole memory, in a complex system each individual (or processing unit) has complete access only to its own state (we can refer to this state as the individual memory) and can gather information solely from those individuals to which it is directly connected [8]. To exemplify such limitation, we can think about neurons: when deciding whether to fire
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
