478 Explanation Kaufmann, W. (1968). Existentialism from Dostoevsky be correlated. For example, a falling barometer to Sartre. Cleveland, OH: Meridien Books. may correlate highly with the coming of rain, but Kruks, S. (1990). Situation and human existence: the barometer’s falling has no causal connection to Freedom, subjectivity, and society. London: Unwin the rain. Merely fulfilling the D-N model, thus, Hyman. cannot be sufficient for explanation. Likewise, rea- Merleau-Ponty, M. (1962). Phenomenology of perception sons may count as causes, but reasons do not pro- (C. Smith, Trans.). London: Routledge. vide mechanism and typically rationalize actions Poster, M. (1975). Existential Marxism in postwar rather than explain them in the sense of making the France. Princeton, NJ: Princeton University Press. actions a logical consequence of some antecedently Sartre, J.-P. (1966). Being and nothingness (H. E. Barnes, specified conditions. Trans.). New York: Washington Square Press. A key element involved in criticisms of logical Warnock, M. (1970). Existentialism. Oxford, UK: positivism in the social sciences concerns the place Oxford University Press. of the D-N model and its heirs as defining proper scientific explanation. While some criticisms can be understood as philosophically valid, they can also reflect an animus toward the presumed EXPLANATION absence of any normative concerns underlying this model. There may also be doubts regarding the Three key issues drive discussion of explanation existence in the social sciences of the sorts of laws in the social sciences: First, must explanation be or robust generalizations on which application of of a certain form to be regarded as scientific? the model relies. The question that lingers, how- Second, how does explanation contrast with the ever, is how to evaluate purported explanations, notion of understanding? Third, how can expla- absent some such models. nations be evaluated? The answers interrelate in important ways. Explanation Versus Interpretation The notion of explanation has at its root philo- sophical usage a connection to the provision of In the philosophy of the social sciences, the term causes, and in particular of causal mechanisms. explanation is typically used as a contrast to that This leads by the usual logic of causal or mechanis- of interpretation. The contrast involves many dif- tic explanation to a demand for laws and predic- ferences. The most fundamental concerns whether tions. For, the thought goes, a law is just a or not the goal or purpose of social inquiry should regularity in nature. Hence, any event that can be be to reconstruct actions (i.e., purposive behavior) explained (the explanandum) by adverting to the in terms of an agent’s (individual actor’s) perspec- relevant law that identifies its cause should also be tive. An alternative view maintains that the pur- predictable, given knowledge of the presence of the pose of social inquiry should be to develop antecedent conditions (the explanatia). This is the explanations, that is, accounts that have robust essence of the deductive-nomological (D-N) model predictive value. Explanation, from this point of of explanation. A great virtue of this model (long view, is unconstrained by any need to reflect or associated with logical positivism but surviving its reconstruct the view of agents with respect to the demise) involves the fact that it provides a purely events to be explained. So, for example, economic formal or syntactic criterion by which to adjudge explanations of choice of marital partner or eco- whether an explanation exists. The D-N model logical explanations of cultural traditions employ also provides a straightforward criterion for evalu- underlying regularities to account for observed ating the soundness and not just the formal ade- behaviors but diverge from any account that might quacy of an explanation. be offered of those whose behavior is so However, at this point, matters become conten- explained. tious, for well-known counterexamples to this The aims of interpretation and explanation model exist. The counterexamples demonstrate appear prima facie incompatible. An interpreta- that law-like connections may well have no causal tive account focuses on reasons specific to an connection to the phenomenon to which they may agent, a time, and a situation. The expected result Explanation 479 would be an account that did not generalize but suffices to prove that a statement about some applied only to the people and the time studied. observed regularity expresses a law. These historicist or ethnographic approaches to However, observational data can disprove a the study of human behavior must be distin- supposed law of nature by providing just one guished from case studies because the latter but counterinstance. Thus, some hold that the purpose not the former are taken to provide a basis for of testing should be to seek disconfirming instances generalizations. of laws. Laws that survive repeated attempts to While we owe the term positivism to Auguste disconfirm them may then have claim to some Comte, logical positivism names the philosophical degree of verisimilitude. But whatever one’s view movement (also referred to as the Vienna Circle) of the evaluation of proffered explanations by that emerged in the 1920s and 1930s through the empirical test, the orthodoxy holds that proper writings of associated philosophers, scientists, and explanation requires laws and empirical informa- social scientists. Following the dispersion of these tion that allows for a logical connection between thinkers to England and the United States due to the explanans and the explanandum. Only by vir- the rise of fascism throughout central Europe, tue of the logical connection can one determine logical positivism became the dominant account of whether the observed instances follow from the what defined scientific explanation. Within the statements said to explain them, and only by vir- social sciences, the term positivism enjoys wide use tue of the logical connection can one assess as an epithet attached to all views that insist that whether the generalization appealed to in the the purpose of social science explanation requires explanation has been confirmed or survived dis- being able to apply generalizations to account for confirmation, and so has a real use for purposes of observed cases. Ideally, generalizations used to explanation. All models of scientific explanation, explain would also allow social scientists to make in short, take explanation to consist in a relation predictions. In the terminology of logical positiv- between some theory and instances that the theory ism, a statement of that which is to be explained is is a theory of. termed the explanandum (pl. explananda); explan- Absent any logical connection between explan- ans is the class of statements (pl. explanantia) that ans and explanandum, however, no rational assess- account for the explanandum. ment of an offered explanation appears possible. Because a primary goal of logical positivists This insistence on the syntactic formulation of was to help distinguish scientific from pseudo- explanation, one that provides a logical connec- scientific explanations, two key criteria were tion between explanans and explanandum, remains developed that were thought to capture the rele- an important yet largely unappreciated feature of vant differences. The first offered a syntactic for- positivist accounts of explanations. This syntactic mulation for an explanation. This requirement standard allowed a clear formal criterion by which held that good explanations had a specific form, to identify a candidate explanation. and this form allowed for the logical derivation of A recurrent criticism of positivist models of the explanandum from the explanans. This format explanation concerns the principled absence of any also required the use of an appropriate law-like or need to incorporate the perspective of those whose statistical generalization. actions are to be explained. In addition, the syntac- The second requirement concerns the need for tic structure of positivist explanation is viewed as empirical testability. Logical positivists themselves also precluding any normative evaluations of the differed with regard to the so-called logic of con- actions. Scientific explanations are typically held firmation for explanations. The basic split con- to be purely descriptive. Absent an overt inclusion cerned whether testing of hypotheses should stress of a normative premise, no normative conclusion confirmation or disconfirmation. The essence of can be derived. the controversy involved the fact that no law can For example, in explaining why people might ever be fully confirmed by observed positive cases; be co-opted to be mass murders in the case of the law makes claims about cases seen and unseen, the Holocaust, the use of generalizations that a as well as past, present, and future. Thus, no finite large majority of people simply tend to obey amount of observed evidence, however great, authority might be employed, together with other 480 Explanation information. However, this might lead to complaints experience to apparently agree with what theory that perpetrators are in fact thoughtless, that is, not implies can be accommodated in an ad hoc fashion portrayed as making a choice about how to act. To by the explanatory framework. the extent that perpetrators did not make choices, Moreover, it is important to note that even the then they might appear to be less morally culpable