The evolutionary origins of modularity

By Jeff Clune, Jean-Baptiste Mouret, Hod Lipson. 2012

624 citations recorded by the source corpus.

A central biological question is how natural organisms are so evolvable (capable of quickly adapting to new environments). A key driver of evolvability is the widespread modularity of biological networks—their organization as functional, sparsely connected subunits—but there is no consensus regarding why modularity itself evolved. Although most hypotheses assume indirect selection for evolvability, here we demonstrate that the ubiquitous, direct selection pressure to reduce the cost of connections between network nodes causes the emergence of modular networks. Computational evolution experiments with selection pressures to maximize network performance and minimize connection costs yield networks that are significantly more modular and more evolvable than control experiments that only select for performance. These results will catalyse research in numerous disciplines, such as neuroscience and genetics, and enhance our ability to harness evolution for engineering purposes.

Subfields: Artificial Intelligence, Genetics, Sociology and Political Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Genetics

In Sociology and Political Science

Citations in the corpus

Weight Agnostic Neural Networks (2019, methodology): “We apply the connection cost technique from [15] shown to produce networks that are more simple, modular, and evolvable.”
Deep learning in neural networks: An overview (2014, background): “One may penalize the task-specific total length of connections (e.g., Legenstein and Maass, 2002; Schmidhuber, 2012, 2013b; Clune et al., 2013) and communication costs of SLIM NNs implemented on the 3dimensional brain-like multi-processor hardware to be expected in the future.”
The Surprising Creativity of Digital Evolution: A Collection of Anecdotes from the Evolutionary Computation and Artificial Life Research Communities (2018, neutral): “Yet both philosophical arguments [13, 18, 31, 14, 34] and high-profile publications [35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50] suggest that digital evolution can be a useful tool to aid and complement the study of biological evolution.”
Modular Networks: Learning to Decompose Neural Computation (2018, background): “Furthermore, the modular structure of biological neural connections [22] is hypothesized to optimize energy cost [7], improve adaption to changing environments and mitigate catastrophic forgetting [21].”
Diffusion-based neuromodulation can eliminate catastrophic forgetting in simple neural networks (2017, methodology): “Two methods to promote structural modularity during the evolution of an ANN include the CCT mentioned above, and constantly switching between different test problems that have the same subgoals [38].”
Modular Brain Networks (2015, background): “Clune et al. (2013) have argued that the adaptive advantages of modularity, for example, in yielding a more flexible system design, may not be sufficiently strong or direct to explain the modularity currently observed in biological organisms.”
The modular and integrative functional architecture of the human brain (2015, background): “However, adding a selective pressure to minimize wiring costs, or varying the goals of the network, leads to the evolution of modular networks that quickly adapt to new environments (75, 76).”
Neural Modularity Helps Organisms Evolve to Learn New Skills without Forgetting Old Skills (2015, methodology): “[23] in this different context of networks with within-life learning.”
Illuminating search spaces by mapping elites (2015, methodology): “Networks have to answer whether an object is present on both the left and right sides of the retina 32,44 .”
Unshackling evolution: evolving soft robots with multiple materials and a powerful generative encoding (2013, methodology): “[7] J. Clune and H. Lipson.”
Constraints Evolve: Context Dependency of Gene Effects Allows Evolution of Pleiotropy (2015, background): “Multiple groups have modeled the evolution of the structure of GP maps, primarily in the context of the evolution of modularity (Clune et al. 2013, Kashtan & Alon 2005, Le Nagard et al. 2011).”
Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Masks (2020, background): “At ﬁrst glance, explicitly modular networks (Clune et al., 2013; Andreas et al., 2016; Kirsch et al., 2018; Chang et al., 2019; Bahdanau et al., 2019) could provide a solution for the discovered problems.”
A review of modularization techniques in artificial neural networks (2019, methodology): “2014) is biologically inspired as evidence suggests that the evolution of the brain, guided by the minimization of wiring length, besides improving information transfer efficiency, produces modular structure (Clune et al. 2013; Bullmore and Sporns 2009).”
Communication dynamics in complex brain networks (2017, background): “…found that a hierarchical modular organization emerges as an optimal cost-efficient solution to diverse problems such as facilitating the coexistence of functional integration and segregation 13 , and enabling evolvability and adaptability 180 , while minimizing wiring and energetic cost 1,4,93 .”