Isaiah Berlin on the Scientific Method in History

I

It is, in my experience, beneficial for one’s scientific creativity and methodological nous to be, as far as time and personal interest allow, conversant with the goals, theories, methods, evidence, and philosophies of fields of inquiry distinct from one’s own. I find that awareness of both the theories and methods of other fields of inquiry enables one to think more deeply about how one approaches one’s own research, provides inspiration for the borrowing of methodological apporaches and theoretical inspiration through analogy, and enables one to consider problems in one’s home field with a new perspective. That said, the better reason for reading work from other fields is that it is generally interesting and enriching. I therefore read with both interest and pleasure Isaiah Berlin’s The Concept of Scientific History, collected in Vintage Books’ The Proper Study of Mankind, an anthology of some of Berlin’s writing which I’ve begun to slowly pick my way through.

Berlin argues that historical enquiry differs from scientific enquiry in that where the (natural) sciences aim for the discovery of general laws, history aims for to discover particular phenomena or facts. He offers the contrast of two examples to make his case. In the first, we consider the case where a man did not see the sun rise in the morning. In this case we do not throw out our existing theories of celestial mechanics, but instead attempt to construct an ad hoc hypothesis which preserves our present system of physics. We might suggest, for instance, that the man was asleep, or that he was lying or perhaps playing some kind of prank. It would certainly not be good scientific practice to take particularly seriously the evidence of the man’s testimony as set against the whole of the body of hard-won evidence in the field of physics.

His second example is if, say, some historian refused to accept some piece of evidence that Napoleon wore a three-cornered hat on the grounds of a general theory which suggested French generals did not wear three-cornered hats on the battlefield, we would likely regard this historian and their argument as suspect. We would not, in other words, seek to construct some ad hoc hypothesis: we would be suspicious of the historian’s motivations, perhaps regard them as being overly dogmatic or wedded to some particular theory and thus prone to attempting to manipulate the available facts and evidence. History in this account is thus concerned with particulars, where science is concerned with general laws.

The methodological education of a historian, in Berlin’s account, consists not of the scientific method but of learning general maxims such as not to pay too much attention to heroic indivudals, to give attention to the lives of ordinary people, to give due attention to social and economic factors, to recall the existence of dull, slow, and imperceptible effects. Maxims of this nature do not follow the rules of deductive or inductive inference, or the rules of specialised techniques. Instead, Berlin argues that they are distilled from the accumulated experience of a practitioner and resemble a skill rather than factual knowledge.

II

I first encountered Berlin during my undergraduate degree via his most famous essay, Two Concepts of Liberty, which also happens to be collected in the same anthology. In this essay, Berlin articulates and defends the notion of a separation between negative and positive liberty. The former is the absence of obstacles, or constraints: the latter is the possibility of acting. Berlin believed that while liberty is rarely controversial as a value, these interpretations reflect competing political traditions. He was concerned by the positive definition of liberty, which he feared ¹. This essay is, to be sure, a lodestone in political thought, but I have personally been more influenced by his concept of value pluralism, which is touched on in the essay but better expressed in the first essay of the collection, The Pursuit of the Ideal.

Value pluralism is, in short, a recognition that it is rational for humans to hold different values to one another, and for these values to clash. In The Pursuit of the Ideal Berlin uses the example of one person who believes in always telling the truth, and another who believes it is sometimes permissible to lie because of the harm truth-telling can cause. More than merely clash however, even if there are some values which are universally seen as good - equality and liberty being two such values - there is not reason why these values cannot contradict one another, and no reason why your or my interpretation of these values should be the same. Value pluralism is the basic acceptance of the plurality of human values and therefore the inevitability of political conflict. In Two Concepts of Liberty, one of the points Berlin makes is that many totalitarian movements sought to do away with politics entirely: from the perspective of a value pluralist, such hopes represent the kernel of totalitarianism.

There are, of course, contradictions and tensions within value pluralism. One is that value pluralism itself, or at the very least the preference for the tolerance of differing values, is itself only one value among many. Another is Popper’s familiar paradox of tolerance². It is important to stress at the same time that value pluralism is not a relativistic position which sees moral value as arbitrary, but rather it is a rejection of the idea that all values can be reduced to a single, irreducible universal value³. Univerisal values might exist, but insofar as they do they cannot necessarily be reconciled to one another. You might wish to maximise security even at the expense of privacy, I might wish to maximise privacy, even at the expense of security. Trade-offs must be made. Berlin was not a systematic philosopher and never gave a complete account of value pluralism, but nor did he need to: the basic power of his argument is that it forces us to confront the basic fact of human difference and its implication⁴.

Like other Cold War liberals, Berlin was deeply shaped by his engagement with Marxism and the general 20th century experience of both Soviet and Fascist totalitarianism. He rooted his critique of Marxism and other totalitarianisms along with his value pluralism in a sceptical social and historical epistemology, wherein he rejected both the notion of teleology in history and the possibility absolute moral certainty⁵. This concern with a limited epistemology underpins Berlin’s philosophy of history, and his objection to the application of the scientific method to history.

III

In The Concept of Scientific History, Berlin goes on to argue that

“The natural sciences largely consist of logically linked laws about the behaviour of objects in the world. In certain cases these generalisations can be represented in the form of an ideal model … The electron, the chromosome, the state of perfect competition, the Oedipus complex, the ideal democracy are all such models; they are useful to the degree to which the actual behaviour of real entities in the world can be represented with lesser or greater precision in terms of their deviation from the frictionless behaviour of the perfect model.”

Berlin’s concern here is with the role of abstraction in scientific model-building. For Berlin historical analysis involves an entirely different kind of knowledge to science. Scientific discplines aim at a sort of “thin” experience, consisting of deliberately isolated strands of experience. Science, in other words, depends to a large degree on abstraction and generalisation away from many components of reality. History, by constrast, aims at “thick” experience: it amagamates evidence from across disciplines and aims to represent human experience in its full complexity. It depends on the selection and arrangement of facts, on interpretation, on the ability to understand the structure of himan experience. For Berlin, then, history cannot be reduced to general formulae or models.

For those of us in the social sciences, rather than the humanities, Berlin’s arguments here are familiar. In virtually all of the social sciences there exists some methodenstreit between, on the one hand, schools of thought which favour qualitative analysis, and on the other schools of thought which favour quantitative analysis. I am not concerned here with the clash between empirical and normative and critical approaches: but rather a clash between different kinds of empiricist. I do not think such clashes are particularly salient in the social sciences, or at least in my discipline of political science, not least because most scholars are reasonably pragmatic, and many I think suspect that the difference is overstated. Quantitative modelling after all depends strongly on qualitative knowledge, and the results of quantitative work can enter qualitative reasoning.

By now, there exist a number of historical scholars who have adopted the methods of the quantitative social sciences, which is to say their work involves the collection of data and the building of statistical models with that data. The Seshat databank⁶ is perhaps the best example of this methodological movement. It represents a collection of a range of historical variables covering, for instance, the use of certain warfare technologies, the presence or absence of human sacrifice, the presence of certain forms of social technology such as roads, and so on. It has been gathered through the effort of a large number of historians, with each coding in the dataset being closely linked to existing scholarly work⁷. As a result, it has become possible to use common statistical techniques such as PCA on the dataset make inferences about the dimensionality of social complexity⁸.

All the usual objections within social scientific practice might apply: perhaps the method is inappropriate, perhaps a different set of indicators should be used, and so on and so forth. But plainly, these scholars have at the very least displayed the possibility of applying the quantitative methods of social science to historical analysis. To do so, requires, as Berlin argued, a degree of abstraction away from specifics and towards more general variables. Whether doing so uncovers ‘laws’ depends partly on how ‘laws’ are defined. It cannot uncover laws in the determinsitic sense of, say, the law of gravity. But empirical laws in the sense of empirically discoverable regularities in human societies seems a more plausible account for this kind of analysis.

What makes application of the scientific method possible in the social domain is a willingness to abstract and quantify, to make inductive inferences through statistical modelling, to accept that one’s findings represent average effects or trends, and to accept that one’s findings will often fail to generalise over time and space (or the question of generalisation over time and space might even be nonsensical), and one will almost never uncover some deterministic law of human societies. One cannot recover a teleology through the quantification of history, but one can make interesting observations about general patterns or trends across many societies at a scale that would not otherwise be possible in qualitative analysis. It depends, in the end, on the goals of one’s analysis. I suspect Berlin would not have accepted this work as historical work, because it is necessarily “thin” and not “thick”. but I think it is clear that we learn things about history from this kind of work.

Footnotes