This website is currently undergoing live development!
Christopher X J. Jensen
Associate Professor, Pratt Institute

Emergence

Research > Interests > Emergence

emergence-0200pxMost of the very successful science of the twentieth century was reductionist. The scientific method was to reduce observed phenomena to the behavior of the smallest possible unit. This approach has worked very well in chemistry and in physics. It also worked relatively well in biology when it came to understanding individual organisms: by reducing our understanding of physiology down to the level of organs, tissues, cells, and even individual biomolecules, we now have a very strong understanding of the anatomy of a diverse number of organisms.

Ecology and evolution have also had their successes by taking a reductionist approach. In ecology, a lot of understanding has been generated by studying the behavior of particular groups organisms, by considering only single populations, and by observing only small patches of ecosystems. In evolution, much of the reductionist revolution has been about genetics: the ability to show how natural selection acts on particular genes gave us our first glimpse of evolution in action.

I do not dispute the importance of the reductionist approach, but I do worry that its exclusive and extreme use places severe limits on our ability to understand living systems. Reductionist approaches work under special conditions: if smaller phenomena scale up to produce the larger-scale phenomena that we seek to explain, reductionism is the proper tool to generate understanding. But what if a simple understanding of the constituent parts of a system is not sufficient to explain the observed behavior of the larger system? We call this “the whole is not the sum of its parts” property emergence, and systems with emergent properties are all around us.

Emergent systems share some common characteristics. They are composed of smaller parts that interact with each other in a networked, nonhierarchical fashion. It is the nature of these complex networked interactions that produces emergence. Emergence is often seen by newcomers to the concept as somehow “magical” because very interesting and sometimes complex system-wide behaviors can emerge from very simple individual behaviors. It is important to note that there is nothing at all magical about emergence: emergent systems are simply more rich with possible dynamics than simple systems which scale up linearly from the behavior of their constituent parts. One way of understanding the reductionist approach is to see it as a selector of “low hanging scientific problems”: we have come to understand those phenomena that are easily explained by scaling up smaller-scale behaviors.

The problem is that extreme reductionism biases us towards only explaining those phenomena without emergent dynamics. Many, many important phenomena can only be explained by understanding how emergent behaviors are generated by the smaller-scale behaviors of that system’s constituents. The examples in ecology are obvious: one cannot explain an entire ecosystem simply by studying all the individuals, all the populations, or even all the communities that make up that ecosystem. Not only would such an extreme reductionist approach be impossible, it would also be a waste of effort: to really understand an ecosystem, we need to understand how the properties of that ecosystem emerge from the interactions that go on within that ecosystem. Evolution has similar problems: we can study how very particular environmental changes affect very particular traits and even perhaps the genes that generate those traits, but what if we want to understand the overall trajectory of the evolutionary process within a larger ecosystem? And what if the trait that we discover is being selected for is actually governed by a complex network of interacting genes? These interesting questions – the real questions that when answered will make ecology and evolution socially-valuable sources of knowledge – can only be answered through understanding emergence.

Emergence is also very important if we want to try to understand human social systems. Humans are probably the most complex-networked social species on the planet. All of the important systems that have made the human species successful emerge from countless minute interactions that take place between individual people. If we want to understand the consequences of various economic and social policies, we need to understand how the emergent properties of our social systems come about. And just in case you were under the impression that emergence really only matters for systems at the scale above individual organisms, do not forget that we lack basic understanding of how the brain works. Why is this organ so hard to understand when others have been so thoroughly explained through a reductionist approach? You guessed it: it is because the brain is composed of networked cells whose complex interactions produce the emergent phenomena of consciousness and bodily control (and a whole lot more).

Looking at emergent phenomena requires that one reject extreme and exclusive reductionism, not reductionism in general. In some sense the question that emergent phenomena force us to ask is “to what level of complexity should I reduce my phenomenon of interest?”. If I am looking for the impact of a predator population on a population of prey, I certainly do not need to investigate down to the atomic level, but I might need to consider the behavior of individuals or groups of individuals. Choosing that correct level of emergence is the key to reducing a given phenomena to its underlying mechanisms.

Emergence is a huge field with influence in many disciplines, so in part I try to be a good student of the latest science investigating emergence. But I also seek to contribute to the field in my own small ways. The fieldTest simulator allows us to understand the emergence social properties of individual organisms who seek to maximize the availability of local resources. Beyond this particular agent-based model, I plan to use models of this type to understand emergent phenomena in the areas of ecology and evolution.


You can see all of my blog posts related to emergence here.