Back to Kant: Reinterpreting the Democratic Peace as a Macrohistorical Learning Process.

After more than a decade of intense scholarly debate, the claim that democratic states hardly ever fight each other remains contested. Although most analysts appear to support the democratic peace hypothesis, a small but determined minority of realist scholars does not accept it. Despite this fundamental disagreement, both camps agree that Immanuel Kant laid the intellectual foundation of the hypothesis. in the late eighteenth century.(1) Whether aiming at corroboration or refutation, most contemporary scholars appear to believe that they are operationalizing and testing some version of the Kantian thesis. Yet, although some analysts have come closer to Kant's original conjecture by embracing more of its analytical dimensions, neither side of the debate succeeds in fully capturing the dynamic and dialectical logic of the process.

Does it really matter whether Kant has been misunderstood? I argue that it does. Rather than engage in exegesis for its own sake, I maintain that only partial representations are responsible for many of the empirical and theoretical disputes haunting the current debate. By squeezing Kant's fundamentally dynamic argument into a Procrustean bed of static regression equations, today's researchers typically expect the "Kantian effect" to be time-invariant.

It is worth revisiting Kant's original peace plan. Instead of viewing the democratic peace argument as merely a "second-image" claim about the pacific effect of a particular regime type on dyadic relations, I propose a Kantian reinterpretation of the democratic peace hypothesis as a dynamic and dialectical learning process. It is dynamic rather than stationary in that states alter their behavior as a consequence of taking past experience into account. Thus, the effect of democracy changes over time. The process is dialectical because catastrophic reversals, such as world wars, drive home the point that there is little choice but to eliminate violence in interstate relations.

Once recast in these terms, the democratic peace argument can be tested in a way that reflects at least some of the conjecture's subtlety. Indeed, Kant ([1784] 1970a, 50) anticipated that, one day, it would become feasible to evaluate the empirical validity of his postulates:

 
   The real test is whether experience can discover anything to indicate a 
   purposeful natural process of this kind. In my opinion, it can discover a 
   little; for this cycle of events seems to take so long a time to complete, 
   that the small part of it traversed by mankind up till now does not allow 
   us to determine with certainty the shape of the whole cycle, and the 
   relation of its parts to the whole. 

Writing in the late eighteenth century, Kant was in a less privileged position than social scientists are today. Yet, most of his predictions have had an almost uncanny tendency to be borne out by history, despite the scarcity of information on which he based his theorizing.

In this article, I focus primarily on the dynamic dimension of the famous peace plan and merely correct for exogenously given dialectical disturbances. The empirical part draws on data from standard quantitative sources that have so far been analyzed in exclusively static terms. The evidence strongly supports the core of the Kantian learning hypothesis. Since the first half of the nineteenth century, democratic states appear to be much better learners in their mutual relations than when faced with other states, or than nondemocratic states in their own interactions. There is no support, however, for the (possibly Kantian) view that learning can only take place among democracies. Indeed, some learning appears to spill over into other relations.