Behavioral International Law: A Methodological Framework

[Tomer Broude is Vice-Dean and Sylvan M. Cohen Chair in Law at the Faculty of Law and Department of International Relations at the Hebrew University of Jerusalem.]

In my previous post, I tried to briefly introduce the merits of “Behavioral International Law”. Experimental research has shown that in many cases human behavior diverges from theoretical assumptions about rationality. Prospect theory, loss aversion, endowment effects, anchoring, hindsight bias, availability bias, conformity effects, framing effects – the list of experimentally proven, systematic diversions from perfect rationality in human behavior is long. The confines of a blog post preclude detailed discussion of any of these biases and heuristics; the literature in cognitive psychology is vast. The important point, pursued by scholars over the last decade or so, is that this knowledge of actual, rather than hypothesized or assumed, human behavior, can have significant implications for legal regulation. Why should this not be the case with respect to public international law?

A number of objections may arise, and I will mention two of them briefly here. The first would be that cognitive psychology and behavioral economics relate primarily to the conduct of individuals as (obviously) unitary actors, while the main subjects of international law are collective entities, primarily states. This presents a type of external validity problem: can the knowledge we have on human behavior, carry over to other actors? In practice, this problem is surmountable. Cognitive psychology also provides insights into the behavior of collectives, such as small decision-making groups (panels of judges) and corporations. Additionally, individual behavior can also be crucial to the operation of international law. Thus, the problem can be addressed through a levels-of-analysis approach. At the individual level, the real rationality of decision-making by military commanders, elite decision-makers such as trade officials and negotiators, domestic officials charged with implementing international human rights, international judges, investors, consumers – the list goes on – can be highly relevant to analysis of particular international legal questions. At the level of small decision-making groups, such as panels of judges or cabinet leadership, an additional set of behavioral dynamics enters into play. And at the state level, one strategy (which I am less sympathetic to), is to do what traditional rational choice has often done: apply assumptions of rationality to states – states are human too – in this case assuming that bounded rationality carries over to the state level. However, building on literature on corporate decision-making and political economy of decision-making in states, and above all, empirical research, it is possible to demonstrate, rather than assume, that certain decision-making biases are evident in state conduct. Indeed, some of the studies I mentioned in the first post do just that.

The second objection would be that the main strength of the claims made by cognitive psychology regarding rationality and decision-making is its grounding in empirical observations derived from experiments made with human subjects, which are difficult and perhaps impossible to replicate meaningfully in the context of international law and international relations. This objection can also be overcome through a spectrum of methodologies. First, experiments and psychological lab work actually are possible – certainly with respect to individuals, but also with respect to small-decision making groups. Second, experimental research is not the only way to gather empirically valid information about the way people make decisions. Indeed, some consider the first-best method to be field research, using observational testing of real behavior. Thus, field-work, including the use of databases (such as on judicial decisions, treaty related conduct or the use of force) and qualitative research, provides a fertile method for testing hypotheses regarding state rationality. A final methodology is theoretical application: the migration of empirical knowledge validated in one field of law and/or behavior, to another. For example, behavioral research has challenged traditional rational choice parameters of deterrence in criminal law, suggesting that criminals are over-optimistic about the chances of being caught (although the empirical grounding is not free of doubt). These insights can be theoretically applied to international criminal law, with respect to the deterrent effect of international criminal tribunals on would be war criminals. The application would, however, remain theoretical, rather than empirically substantiated; it is a methodology that is very useful in generating hypotheses. Theoretical application is perhaps most conveniently applied to states as unitary actors (for a wide-ranging discussion of possible theoretical applications of behavioral economics, e.g., from contract law to treaty design see this excellent recent talk by Anne van Aaken).

This presents an interesting research paradox that poses an interdisciplinary challenge. While international lawyers would be interested in experimental proof that targeting decisions made by military commanders are subject to framing effects (a topic I will expand on in my next post), a cognitive psychologist would be less interested in conducting such research that to her merely extends existing knowledge to a different setting. Thus, it is ideally necessary to generate hypotheses that contribute both to our understanding of international law, and to behavioral science. This is not, however, always possible, and international legal scholars can work with a combination of methodologies to produce research relevant to international law.

The objections discussed above do not, therefore, preclude behavioral research on international law, but rather constitute considerations to be taken into account in optimal research design. To sum up the methodological framework, below is a 3X3 matrix of the possibilities (and I’m grateful to Jeff Dunoff for suggesting this graphic depiction). Within the matrix I’ve listed general assessments of the validity of each rubric, considering also their technical viability.

Thus, experimental research with individuals is both feasible and with strong validity, whereas experimental research with states is essentially impossible. Collection of field-derived data on individual conduct potentially faces other significant problems of sample selection that may impair generalizability (hence, listed as intermediate), whereas field research with observable state behavior has strong validity. Theoretical application is, by definition, of weaker validity than other methodologies. Of course, these are just general observations: the feasibility and validity of each methodological rubric will vary depending on the research question.