Many systems in nature, engineering, and human society are 'adapted complex systems', i.e. they consist of many somewhat autonomous interacting parts but are optimized (or should be) to function well as a group. Social insect colonies are one example of this (evolved to maximize colony fitness through individual worker actions), but so are multicellular organisms (evolved to maximize organism fitness through individual cell actions), cluster computers (multiple 'nodes' contributing to a calculation/process), power grids, and companies or organizations (with group-level goals and individual human contributors). While much of our work is thus inspired by the desire to understand social insects, who are beautiful and wonderful in their own way, we also want to contribute to the broad understanding of fundamental principles that govern all of life and much of the human experience: understanding how complex system behavior emerges from individual behavior, and why particular strategies of organization work better than others and in which environments.
Insights on why particular collective strategies evolved in social insects can also be put to much more immediate use. As mentioned above, several human-constructed systems, in computing, engineering, and organizations, can also be viewed as 'adapted complex systems'. These systems, from computer networks to companies, also employ task allocation algorithms, they have information flow among agents, and they solve search problems. Thus, we believe that both a general theoretical understanding of such topics as 'task allocation in complex systems' as well as specific knowledge about the set of mechanisms that social insects employ on task allocation will be helpful to these disciplines, and to improving human-made complex systems.
Using biological processes to inform the design of human-made systems is called bioinspiration or biomimetics. Bioinspiration has a long history in engineering and particularly computing (e.g. see this and this). However, we believe that fundamental advances in the way in which bioinspiration works are possible, particularly with regard to collective or complex system behavior, for these reasons:
- Engineers who are designing 'bioinspired' algorithms are often fascinated with biology, but have no biology training. Thus, the 'inspiration' is often derived from somewhat random encounters with biological tidbits. We, as biologists, strive to work directly with computer scientists to allow them to access deeper biological insights and more awareness of the diversity of biological solutions. A plethora of solutions have evolved in biological organisms; a more complete understanding of this solution set also allows us to pick the best material to be inspired.
- In particular, it is not widely known that the field within biology that studies evolutionary optimization of behavior is called 'Behavioral Ecology'. As behavioral ecologists, we are trained not to tell 'just-so stories' about why we think a particular algorithm works well; we are able to rigorously test such ideas. As a result, we can gain understanding how different environments or constraints promote the evolution of particular strategies; this means being able to not just gain some novel ideas for engineering, but to actually understand which biological processes are good examples of the kinds of problems and solutions sought for a particular application. In concrete terms, we can only make use of the optimization process that evolution has already provided if we know which biological phenomenon is in fact a solution to the kind of problem, with the kind of constraints, that we want to solve.
- Bioinspiration is particularly relevant in the context of complex systems because people are not very good at designing individual-level rules for best system-level outcomes. Despite our high intelligence and creativity, our mind essentially evolved to think of group tasks as best solved by hierarchies or linear processes. Thus, it is actually very difficult for us intuitively understand what group-level outcomes will result from certain individual-level behaviors, and therefore, it is also difficult for us to engineer desirable individual-level algorithms. Several authors have thus pointed out the usefulness of learning from social insects (e.g. here, here, here, and here).
We have received funding from NSF in the Biological and Computing Shared Principles initiative in an attempt to generalize biological insights for computing, and to use the techniques already developed in computing to advance biology, for the question of optimal task allocation mechanisms.
We are also working with a small company (Novateur) to develop novel strategies for active cyberdefense.
We co-organize an annual workshop on 'Biological Distributed Algorithms' to which biologists are invited and which is associated with the computer science conference PODC (Principles of Distributed Computing), to facilitate collaborations and information exchange between biologists and computer scientists.
We regularly engage with public media and the public directly to spread insights about social organization (directly resulting in a paper in the Journal of Bioeconomics and reports in Salt; see also our page 'in the news'; other articles which do not directly reference our work on this topic are here, and here). All these articles contain ideas or examples of using social-insect-derived strategies, such as dealing with randomness, lack of hierarchical control, or flexible task allocation, to improve performance of human organizations or companies.