#33 Elster & Landemore on the philosophy of social science
Boundaries, obscurantism, causality, rationality, self-interest …
Elster & Landemore wrote:
This chapter on the philosophy of social sciences addresses problems and themes in the social sciences, where the latter are understood in the specific sense of sciences that have (or that should have) the following minimal characteristics: their object of study is human behavior, and they follow a certain number of methodological principles, including: (1) a marked effort towards analytical clarity; (2) the investigation of causal explanations through the formulation of causal laws or at least causal mechanisms; and (3) a subscription to a form of methodological individualism, if an amended one, which puts at the heart of social science the notion of choice…
The Boundaries of the Social Sciences:
… The disciplines traditionally classified within the “humanities”— philosophy (at least as a history of ideas), ancient languages, literature, literary criticism, and certain branches of law— are excluded a priori by … criteria of object and method. Although these disciplines have as their object of study the human being in all of its complexity, they are only incidentally interested in the question of choice and do not directly pursue the investigation of causal laws. On the other hand, economics, sociology, political science, certain branches of law (as in the recent discipline of “law and economics,” more developed in the Anglo- Saxon world than in continental Europe), history, information and communication sciences, psychology and anthropology, are included within the field of social sciences, insofar as they satisfy, at least minimally, the criteria formulated above. We insist in particular on the artificial and dated character of the distinction between history and social science. Even if causal laws and causal mechanisms are not always explicitly mentioned in qualitative approaches, they do underlie the selection and the description of the reported events and facts. Conversely, quantitative approaches strive to demonstrate statistical regularities that serve as causal explanations (unfortunately often incomplete in our view)…
Soft Obscurantism:
“Soft” obscurantism is closer to literary criticism or even to literature itself (about which the authors offer no value judgment, except to say that it does not constitute social science) than to the qualitative empirical research that literary criticism often claims to be. Despite their creativity and possible power of suggestion, postmodernism, postcolonial studies, subaltern studies, and Kleinian or Lacanian deconstructionism, among others, fit into this category because, and insofar as, they represent sectarianisms based more on common linguistic reflexes than on an investigative principle of universal, inclusive rationality and a search for truth. To the extent that it is impossible to falsify that which does not expose itself to falsification, we refer back to [literature cited] for a definitive proof by absurdity.
Hard Obscurantism:
Hard obscurantism characterizes one part of the research conducted in the “quantitative” social sciences to the extent that measuring exercises, data analysis and modeling have no more than a thematic relation to the reality of the human phenomena that they are meant to explain or predict. This critique targets, in part, rational choice theory, including theories of individual decision-making, game theory and social choice, as well as certain practices of the statistical branches of economics and political science … As far as statistical analysis is concerned, it is worth emphasizing the fact that it is closer to a profession or a technique, in which precision and value increase with practice and experience, than a science properly understood. In this sense, the quantification of social phenomena, and the simple act of identifying a regularity in a mass of data, are not proof of good social science. There are many things at play, indeed, in the selection of variables and the interpretation of data. Moreover, it is necessary to denounce the common practices of adjusting curves to the data (“curve fitting”), or of selectively choosing data to fit them to curves (“data snooping”). The measuring rod against which practitioners of statistical analysis in social science should assess their results is prediction, or “postdiction,” which consists of predicting one part of the observations obtained on the basis of the analysis performed on another part of the same observations. One way (perhaps impractical) to prevent post hoc curve or data manipulation would be to force researchers to present their datasets and hypotheses to the journals in which they intend to publish their results, say, two years before the submission of those results…
Explanation in the Social Sciences:
The principal mission of the social sciences is that of explaining social phenomena. In this regard, we maintain that every explanation is causal, in the sense that to explain a phenomenon (an explanandum) is to specify an antecedent phenomenon that caused it. According to Hume’s billiard-ball model, one can thus say than a certain event A is caused by a certain event B, in the same way that a billiard ball is pushed by another ball … As a first, rough approximation, one can say that social scientists aim to produce causal explanations on the model of the causal explanations found in the natural sciences. They aim to identify causal laws from which the explanandum can be logically deduced. According to this deductive-nomological (or hypothetico-deductive) model, social scientists choose a theory, that is, a set of mutually interrelated causal propositions, then specify a hypothesis that applies the theory to a given question, and finally show that the explanandum follows logically from the hypothesis. A causal explanation thus understood will be more or less accepted depending on how well, relative to rival explanations, it can accommodate already observed facts as logical consequences, as well as help predict “new facts”, that is fact observed after the causal explanation has been formulated…
Why It Is More Costly to Give Up on the Rationality Hypothesis:
The rationality and self-interest hypotheses are logically independent from one another. Thus, the rationality hypothesis does not imply the self-interest hypothesis, and vice versa. The hypothesis of self-interested or egoistic motivation may be combined with the rationality hypothesis to lead to a particular case of rationality, even an important case, but there is no methodological reason to privilege it. Inversely, self-interested behavior may be irrational in that the agent does not apply the most adequate means for pursuing his egoistic desires. Nonetheless, there is a certain asymmetry between the two hypotheses, to the extent that rationality is also a norm that human beings seek out over its opposite, irrationality, whereas self-interest is a purely contingent motivation from an empirical point of view, since we do not always have reason to privilege personal interest. The rationality norm constitutes a permanent counterweight to irrational tendencies, which is not the case with self-interest. For explicative purposes, it is therefore more useful to preserve the idea of maximizing a utility function, even if the maximized object includes the wellbeing of others, than it is to preserve the idea that the object of the attempted action is individual interest…
Do Selfless Actions Exist?
Even if the self-interest hypothesis is less necessary, to economic analysis in particular, than the rationality hypothesis, a problem arises from the fact that, contrary to the irrationality hypothesis, which is well-established for certain actions, the self-interest hypothesis is not necessarily plausible, nor, in particular, easy to verify. Selflessness refers to motivation detached from all personal interest. One example of this might be triply anonymous donations to charitable works, in which neither the identity of the recipients, nor of the organizer of the charity, nor of the public is known. An example of a triply anonymous donation is given by the case of a person who deposits a 100-dollar bill in the donation box of an empty church. The problem is that it is always possible to say, even in this example, that the charitable action is motivated not by real selflessness but by the desire to please God and save one’s soul, or the desire to win internal plaudits from one’s own conscience. The question posed, then, is based on Kant’s model of the existence of a good intention: does there exist, in this depraved world, one single authentically selfless action? One can imagine an even more convincing paradigm than the case of the triply anonymous donor, that of the anonymous, atheistic kamikaze (unless he experiences — a possibility, according to Kant, we are unable to exclude — a feeling of self- satisfaction at the instant preceding his death that would ruin the hypothesis of complete disinterest). We propose that there exist three forms of authentic selflessness: selflessness in fact, selflessness by choice, and selflessness through negligence. The first corresponds, broadly speaking, to the disinterestedness of the judge, who has no personal stake in the question of which he is an a priori impartial arbitrator. The second corresponds to the selflessness of the altruist, who consciously chooses to pursue the interests of others above his own. Finally, the third corresponds to that of the revenge-seeking individual whose passion carries him back to self-interest…
[Conclusion] … it seems to us that the social sciences would be better off, for now, cultivating a certain epistemological modesty. This modesty would in part imply abandoning, at least temporarily, the search for general laws and instead concentrating on the clarification of a certain number of fundamental hypotheses, like those of rationality or self-interest, and on the collection of mechanisms.
Jon Elster and Hélène Landemore, ‘Philosophy of Social Sciences’, in The Philosophy of Science: A Companion, edited by Anouk Barberousse, Denis Bonnay, and Mikaël Cozic, Oxford 2018 [pp. 510, 511-512, 512-513, 516, 535, 536-537]