Nuclear proliferation and the use of force : nuclear coercion and nuclear learning

What is the effect of developing nuclear weapons on a state’s conflict propensity? Extant answers to this question do not model the effects of time and thus leave policy-makers addressing cases such as Iran and North Korea with two contradictory answers – that nuclear proliferation is stabilising and destabilising – that both have extensive empirical support. I argue that nuclear proliferation is dangerous when decision-makers learn that it is safe and safe when they learn it to be dangerous. I develop and test a psychological nuclear learning model that, consistent with recent quantitative research, explains why weak, revisionist/dissatisfied nuclear powers are highly conflict prone but the same experienced nuclear powers are not. In the model, three biases associated with the availability heuristic make weak revisionist new nuclear powers war prone. (1) Illusory correlations cause decision-makers to believe that the immense destructive potential of nuclear weapons causes them to offer similarly large coercive power; (2) self-serving attribution biases cause decision-makers to infer a causal relationship between nuclear compellence threats and subsequent compliance and (3) decision-maker’s nuclear threats, because they are more cognitively accessible than other contextual variables and the operation of these variables in historical cases, cause them to overestimate the probability of successful nuclear coercion. Such new nuclear powers inevitably practice nuclear coercion, which causes a nuclear crisis and fear of imminent nuclear war. I show that this fear moderates the high war propensity of new nuclear powers. It causes (1) pessimistic estimates of the probability of inadvertent nuclear escalation resulting from nuclear coercion, (2) moderation in future nuclear diplomacy and (3) pessimistic risk choices in logically unrelated foreign policies. Recently released Soviet archival data and elite interviews in South Asia, including with former Pakistani President Pervez Musharraf, allow me to test the psychological nuclear learning model against realist, domestic politics and rational learning models.Arts, Faculty ofPolitical Science, Department ofGraduat