Path dependency in formation of academic disciplines

The topic of the historicity of academic disciplines has come up numerous times in this forum. It is a conviction of mine that disciplines demonstrate a great deal of path dependency over time in their evolution. We can think of a discipline as being constituted at a time by some or all of these elements:

  • a definition of important questions for research
  • a definition of appropriate methods of research and analysis
  • a model of explanation in the field
  • some key examples of what theories and hypotheses ought to look like
  • institutions for supporting, organizing, and directing research efforts
  • institutions for validating and disseminating research findings
  • institutions for training young researchers in the key elements of the discipline

This sounds a lot like Kuhn’s idea of a paradigm, Lakatos’s idea of a research community, or the definitions of scientific enterprise offered by historians and sociologists of science and researchers in the tradition of STS studies (link). An academic discipline is an assemblage of ideas, networks of individuals, institutions, and locations (libraries, laboratories, research institutes).

If this is a reasonable approximation to the social reality of an academic discipline, what does it suggest about contingency and path-dependency in the development of the discipline? For one thing, it suggests multiple sources of contingency both internal to the intellectual enterprise and external to it. Internally, a discipline like philosophy or a sub-discipline like the philosophy of mind is driven in part by a somewhat logical process of attack on existing problems — what Kuhn referred to as “normal science”, and partly by large, compelling breakthroughs by individuals or small groups (for example, the Vienna Circle). Externally, it is straightforward to identify political and institutional influences that shape the research agenda at various times in various disciplines — the preference for positivism in sociology that was advanced by considerations of the Cold War, for example. And within the institutional setting of the disciplines there are contingencies as well — for example, a strong editor of a leading journal or research laboratory can set the agenda for theory and methodology in a discipline for a generation. (Andrew Abbott describes this kind of influence in Department and Discipline: Chicago Sociology at One Hundred.)

Almost every element in this list is itself visibly dependent on historical circumstances in multiple ways. Take the issue of defining the important questions for research. There are political and governmental influences on the definition of research problems — witness the influence of the Cold War on the development of the social sciences, the role that is played by governmental funding agencies like the NSF or NIH, and the occasional intrusion of political pressure into scientific fields like environmental science and sociology.

Within the community of individuals currently pursuing the discipline or proto-discipline there is a range of levels of talent and innovation, on the one hand, and prestige and influence, on the other. (The two categories don’t necessarily correlate perfectly.) One charismatic individual or local group (Wittgenstein, say) may exert influence over the direction of a sub-field through charisma and the power of his or her ideas. Another may exert influence over the strategic placement he or she occupies in the institutions of influence — major graduate schools or prominent journals, for example. And in each case, the discipline moves to a new phase with new questions and ideas.

Bourdieu’s theoretical construct of the field (link) is very relevant to these forms of influence on the development of a given academic discipline. By locating various individuals within the network of institutions, scholars, and funding sources it is possible to attempt to piece together the ways in which their own research agendas unfolded (responding to incentives created by their field) and the influence they exerted on other scholars. Neil Gross’s sociological biography of Richard Rorty illustrates this kind of analysis (link), as does much of George Steinmetz’s research on the development of sociology as a discipline in France, Germany, and the US.

What all of this seems to support is the idea that the academic disciplines are in fact highly contingent in their development, and that there is no reason to expect convergence around a single “best” version of the discipline. The history of disciplines should better be understood in analogy to the brachiation and differentiation associated with the evolution of species and sub-species over time — lots of contingency, with a consequent specialization of the intermediate results to the demands of a particular point in time. This implies that a discipline like sociology or political science could have developed very differently, with substantially different ideas about research questions and methods. And this seems to be true for similar reasons in the humanities as well as the natural sciences and mathematics. Finally, this suggests that there is no end-point — no “universal sociology,” no “final philosophy,” no “complete mathematics.” Instead, every discipline in its search for knowledge and new ideas is charting new intellectual space.

How the calendar matters

It is interesting to consider how the timing of a routine social event can have a major effect on outcomes. Malcolm Gladwell observes that the most talented Canadian hockey players in the NHL are disproportionately likely to have birthdays in the months of January or February in his recent book, Outliers: The Story of Success. Observers of the current US presidential election may speculate that, if the financial crisis of September had occurred in May, the outcome of the election might have been different. The generation of Americans born around 1915 are much like those born around 1945 — except for the searing experience their generation had of the great depression.

The lesson to be drawn here might seem to be the obvious and trivial one — context matters in human affairs. Because youth hockey leagues define the age of a player based on his age on December 31, the January children have a major advantage in size and physical development over the November children. And this advantage creates a small headstart that amplifies over time. The fact that the financial crisis of 2008 created a major disadvantage for the McCain ticket less than 60 days ahead of the election made it very difficult for the candidate to recover in the polls. The cohort experience of poverty and insecurity made the 1920 generation much more risk averse than the 1950 generation. So context and the timing of contextual events matters.

But perhaps the importance of the calendar goes deeper than this. In an earlier posting I discussed Victor Lieberman’s discovery of an unexpected synchronicity of political change at the far ends of Eurasia, over the course of a millenium. We tried to understand this pattern in terms of hypothetical social mechanisms that might have produced these parallels. But what is striking about the example is not simply the fact that there must have been underlying causal mechanisms; it is that the result is a weakly synchronized system of events — that is, a system of events with a regular temporal association — that might never have been noticed.

What this suggests to me is that the social sciences can profitably give more attention to the temporal features of social phenomena — the simultaneous experiences a group of people would have had in virtue of being part of the same age cohort, the temporal parallels that might exist between the rise of a mass ideology and the sales of particular books, the accidents of simultaneity that have major repercussions decades later. Causal analysis implicitly imposes a temporal structure on events (causes precede effects). But often the research goal is to strip away the particular timing and temporal context, and to treat causal structures purely abstractly. And this means deliberately taking causal pairs out of their particular temporal contexts and comparing them with temporally disconnected alternative examples.

Andrew Abbott takes up some aspects of these issues in “Conceptions of Time and Events in Social Science Methods” in Time Matters: On Theory and Method. And William Sewell’s critique of some forms of causal reasoning in comparative historical sociology in “Three Temporalities” in Logics of History: Social Theory and Social Transformation is highly relevant as well.

Turning points

Are there turning points in history? How would we know if we’re in the midst of one? Does the current financial crisis represent a turning point in the development of the US economy? Did the election of Ronald Reagan represent a turning point in American politics and government?

Often what is announced as a turning point eventually seems like a change without a difference — an example of rearranging the deck chairs on the Titanic, of changing drivers but not direction. Nguyen van Thieu takes office in Vietnam in 1967 and a new era is announced; but then the same old policies persist and Vietnam slides ever further towards Communist victory.

A turning point might be defined as an event, action, or choice, that profoundly alters the direction of a whole series of subsequent events. The New Deal is perhaps a candidate in the development of the political-social culture of the United States — a new set of policies, laws, and strategies that set the United States on a new direction and that substantially constrained later choices by government. The notion of a turning point conveys the situation of contingency — up until T things might have continued within the existing pattern P, but after T things shifted to P‘. And it conveys the idea of path dependency as well — now that the turning point has occurred and P’ is embodied, it is much more difficult to return to P. So a turning point results from some contingent event that occurs within a system at a particular time and substantially inflects the future dynamics of development of the system. The idea turns on the background assumption that there are mechanisms or forces that sustain the development of the system, and that contingent events can “push” the system onto a different course for a while.

What sorts of things can have turning points? Can an individual have one? What about a family or a marriage? How about a business or a university? And how about a nation or a civilization? We might say that anything that has a recognizable and somewhat stable pattern of development can display a turning point. So each of these orders of human affairs can do so. An individual may be influenced by a traumatic event or a charismatic person and may change his ways; from that point forward he may behave differently — more honestly, more cautiously, more compassionately. The event was a turning point on his development. A “velvet revolution” may be on a course that gives great importance to non-violent tactics. Then something happens — a violent repression by the state, the emergence of a new clique of leaders more open to violence. The velvet revolution undergoes a turning point and becomes more violent in its strategies.

Schematically, the idea of a turning point involves an ontology something like this: system properties in a state of persistence > singular event > new system properties in a state of persistence.

So how could we know that we’re at a turning point? The answer seems to be: we can’t. Only the larger course of history can indicate whether contemporary changes will be large and persistent, or cosmetic and evanescent.

The idea of a “turning point” is perhaps one of the analytical categories that we use to characterize and analyze the sweep of history. It is a narrative device that highlights persistence, contingency, and direction. And, it would appear, we’ve got to wait until the Owl of Minerva spreads its wings before we can say with confidence when they occur.

%d bloggers like this: