What Is Society?

And related concepts: interest group, polity, ultrasociety

In May I participated in a working group organized by Mark Moffett in New Haven. The workshop brought together an interdisciplinary group that included psychologists, archaeologists, sociologists, and evolutionary scientists. The goal was to develop a definition of “society” that would apply not only to modern nations but also “to diverse hunter-gatherer and tribal groups as well as to certain groups among other species for which comparison to humans could be instructive,” as Mark wrote to me in an email.

Mark has recently published a target article in Behavioral and Brain Sciences, What is a society? Building an interdisciplinary perspective and why that’s important (“target articles” are those for which the journal invites commentaries on, which are published in the same issue as the target article). In this article, and during the May workshop, Mark argued that we need a definition that describes societies in terms of shared group identification rather than social interactions. This goes against the usual approach in social sciences that in fact emphasizes interaction. For example, the Wikipedia article on “Society” defines it as “a group of individuals involved in persistent social interaction.”

Mark, instead, proposes that defining features of a society are that it has:

(1) a mechanism for group identification, by which the members distinguish those who belong from those who do not;

(2) the possibility for this membership to persist through the generations; and

(3) primary control over a shared physical space.

My approach is different. While I agree that definitions are important in science, they are, in a sense, secondary to research questions. Or definitions will be affected by what we want to know about the entities in question.

One of the “big questions” that has motivated my research over the past 25 years is formulated this way (this is from the description on the back cover of my forthcoming book The Great Holocene Transformation):

During the Holocene (the last 10,000 years), human societies have been transformed utterly: from small groups of nomadic foragers to our current interconnected world of large-scale societies organized as states. Population numbers, agricultural productivity, technological development, political and social complexity have all seen spectacular growth. This shift—the Great Holocene Transformation—deserves to be seen as a “Major Evolutionary Transition,” as momentous as the appearance of multicellular life, or the emergence of complex cognition in humans during the Paleolithic era. Why and how did this transformation take place? Past thinkers and modern social scientists have developed myriad theories, and new ones continue to be proposed. Yet we still don’t have a widely accepted answer to the puzzle.

The key dimension of this transformation (and in fact, any major evolutionary transition) is the dramatic increase in the scale at which humans cooperate—from hundreds or a few thousands 10,000 years ago to millions, or even more than a billion, today. A new kind of entity appeared a few thousand years ago that I called “ultrasociety” (for more, see my book Ultrasociety).

And this, finally, brings me to my answer on how we should define “society.” I start with the standard definition that emphasizes interaction, but immediately shift to making specific the content of interaction: relationships between individuals and their actions that enable and sustain cooperation.

Cooperation is a controversial concept because of the tension between group-level benefits (such as production of public goods) and individual-level costs. As a result, cooperation can easily unravel because of the temptation by individuals to free-ride on the contributions of others. How evolution overcomes this free-riding problem is explained in Ultrasociety (and the theory is then extensively tested and empirically supported in The Great Holocene Transformation).

My definition of society, then, is a collection of individuals that cooperate to achieve some common goal. But because the word society has a lot of historical “baggage”—can mean so many different things—in practice I prefer the term interest group. And that is also the term that we use in Seshat DB for various entities that we collect data on. You can see why definitions are important. What is it that experienced a major transformation during the Holocene? When we want to test rival theories about this transformation, what are the historical entities for which we collect data in Seshat?

The bulk of data in Seshat currently focuses on a specific kind of interest groups—polities. A polity is defined as an independent political unit. The term covers a wide range of societies, from small, independent villages to massive states and empires.

The key glue that holds together a polity is cooperation. This doesn’t mean that polities cannot be oppressive and despotic. But some degree of cooperation among some segments of population (most importantly, the elites) is necessary—when it disappears, the polity collapses.

Other features of polities, on which both the standard and Mark’s definition of society focus are important, but secondary. These are features that are often present because they enhance the capacity of polity for internal cooperation. Thus, the first part of Mark’s definition above (“a mechanism for group identification, by which the members distinguish those who belong from those who do not”) is a key characteristic of groups that enables cooperation within them. And so on. Because cooperation is by its nature fragile, it requires a lot of secondary characteristics to strengthen the capacities of interest groups to sustain it.

This is especially important for ultrasocieties—large-scale (e.g., more than one million members) complex human societies—because the larger the scale, the more difficult it is to establish and sustain cooperation.

A final note. While polity has been the primary focus of Seshat until recently, we are now expanding data collection to other kinds—settlements, political parties, religious cults, and so on (and there are many other kinds, such as corporations, although currently we have no plans to collect data on those). There is a lot more conceptual development needed to define, for example, a city (a kind of settlement). We need to know what a city is to collect data for cities located in China, Mesopotamia, and Peru. In November I am running a workshop that is devoted to this issue.

Eventually, we will need a flexible and general definition of a settlement, similar to what we’ve done for a polity. Take a look at how we dealt with such difficult issues as to defining when a polity starts and when it ends; the degree of centralization; and relationships with super- and sub-polities. It all required defining a variety of Seshat general variables capturing such nuance and different layers of “politiness” that cannot be forced into a Procrustean, binary distinction. Is Holy Roman Empire a polity? When did it stop being a polity? What was the geographic space that it controlled?

To conclude, good definitions become important but not at the beginning of a research program, but in one of its phases. We start with a general question, then delineate a range of theories attempting to answer it, translate theories into explicit mathematical or computational models, extract predictions, and determine how we can test them with data. This is the stage where we need general yet flexible definitions that enable us to collect data that will be used to test theories.