Solve A Biological Mystery, Boost Artificial Intelligence
    By News Staff | February 3rd 2013 01:00 PM | 7 comments | Print | E-mail | Track Comments

    If a numerical model simulating 25,000 generations of evolution within computers is to be believed, Cornell University engineering and robotics researchers may have discovered why biological networks tend to be organized as modules – a finding that will lead to a deeper understanding of the evolution of complexity.

    And they say it will help evolve artificial intelligence. Robot brains with the grace and cunning of animals? What could go wrong?

    From brains to gene regulatory networks, many biological entities are organized into modules – dense clusters of interconnected parts within a complex network. For decades biologists have wanted to know why humans, bacteria and other organisms evolved in a modular fashion. Like engineers, nature builds things modularly by building and combining distinct parts, but that does not explain how such modularity evolved in the first place. Biologists from Richard Dawkins to Günter P. Wagner to the late Stephen Jay Gould identified the question of modularity as central to the debate over "the evolution of complexity."

    For years, the prevailing assumption was simply that modules evolved because entities that were modular could respond to change more quickly, and therefore had an adaptive advantage over their non-modular competitors. But that may not be enough to explain the origin of the phenomena.

    The team discovered that evolution produces modules not because they produce more adaptable designs, but because modular designs have fewer and shorter network connections, which are costly to build and maintain. As it turned out, it was enough to include a "cost of wiring" to make evolution favor modular architectures. 

    To test the theory, the researchers simulated the evolution of networks with and without a cost for network connections.

    "Once you add a cost for network connections, modules immediately appear. Without a cost, modules never form. The effect is quite dramatic," says  Jeff Clune, assistant professor of computer science at the University of Wyoming .

    The results may help explain the near-universal presence of modularity in biological networks as diverse as neural networks – such as animal brains – and vascular networks, gene regulatory networks, protein-protein interaction networks, metabolic networks and even human-constructed networks such as the Internet.

    "Being able to evolve modularity will let us create more complex, sophisticated computational brains," says Clune.

    "We've had various attempts to try to crack the modularity question in lots of different ways. This one by far is the simplest and most elegant,"  says Hod Lipson, Cornell associate professor of mechanical and aerospace engineering.

     Published in Proceedings of the Royal Society 


    Actually, this "cost of wiring" is not only simple and elegant, is is also quit reasonable, when one thinks about practically any "real world" cases.  The only possible exceptions are those "virtual" networks such as interaction/regulation networks, and the like.

    Perhaps more thought needs to be applied as to why there "should" be any "cost of wiring" for such "virtual" networks, or what analogous "cost" may play a similar role.


    Thor Russell
    The more "wiring" the more potential for overfitting etc in computational networks.However for something like the internet, more wiring seems to mean more robustness, without cost (except physical material cost etc)
    I remember seeing this result reported quite a few months ago in I think.
    Thor Russell

    Of course, the "nice" thing about this result is that they get the "benefit" of the "right thing" for cases such as computational networks, quite "naturally", by simply imposing the quite natural/realistic cost of physical wiring.

    The "trouble", as I see it, is that they get the observed phenomena for "virtual" networks, that don't seem to have an apparent "cost of wiring", without even questioning, as far as I can tell from this article, what the equivalent is to "cost of wiring".

    As for whether this is a first publication of this result:  Proceedings publications are often preceded by other publication of results.



    There is a section of the paper that addresses your question:

    "The concept of connection costs is straightforward in networks with phy- sical connections (e.g. neural networks), but costs and physical limits on the number of possible connections may also tend to limit interactions in other types of networks such as genetic and metabolic pathways. For example, adding more connections in a signalling pathway might delay the time that it takes to output a critical response; adding regulation of a gene via more transcription factors may be difficult or impossible after a certain number of prox- imal DNA binding sites are occupied, and increases the time and material required for genome replication and regulation; and adding more protein – protein interactions to a system may become increasingly difficult as more of the remaining surface area is taken up by other binding interactions. Future work is needed to investigate these and other hypoth- eses regarding costs in cellular networks. "

    From: Clune J, Mouret J-B, Lipson H (2013) The evolutionary origins of modularity. Proceedings of the Royal Society B. 280: 20122863.

    PS. The MIT Technology Review article is here:

    Thank you, ScienceIsAwesome.  That answers my concern quite nicely.  :)


    Back in the day, I tried to get one of the designers of IC interconnect routing software to support 2 signal attributes, data, and control, for just such a purpose. But I don't think he applied it, at least not then.
    Never is a long time.
    What a horribly misleading title (also that goes for

    So now any research in computer science and/or evolution can make grandiose claims to "boost artificial intelligence?" Nothing is mentioned in the original article about improving any kind of AI.

    What they do speculate on is quite interesting yet not mentioned in the hypesludge:

    Knowing that selection to reduce connection costs produces
    modular networks will substantially advance fields
    that harness evolution for engineering, because a longstanding
    challenge therein has been evolving modular designs