Article Summary: “Taxonomic Definitions in Social Science” Geoffrey M. Hodgson

Summary of Hodgson, Geoffrey M. (2018) ‘Taxonomic Definitions in Social Science, with Firms, Markets and Institutions as Case Studies’, Journal of Institutional Economics, published online. Forthcoming in issue 15(2), April 2019. DOI: 10.1017/S1744137418000334. 

By Geoffrey M. Hodgson, Loughborough University London, UK. 

This paper is open access and is available HERE.


Definitions are crucial for social science. This article explains the nature of a taxonomic definitions, with particular attention to their use in institutional research.

Although there will always be some ambiguity and fuzziness, there is a failure in some parts of social science to develop adequate common understandings of terms. Some scholars have argued that definitional tasks should be abandoned. There is little recognition that several different types of definition have been identified in philosophy. Yet without workable definitions, science can make little progress. 

It is sometimes suggested that definitions should express everything vital about the phenomenon defined. Such a task would be impossible, and is refuted by examples of effective taxonomic definitions, such as ‘a mammal is a clade of animal where the females suckles their young’. This parsimonious but effective definition omits numerous vital features of mammals. 

A common mistake in the social sciences is to overload taxonomic definitions with elements that are important but not vital to classify types. Taxonomic definitions identify the minimum number of properties that are sufficient to demarcate one group of entities from all other entities. Their role is to demarcate, not to express in detail the nature of that kind.

Science is a social process, wherein communication and understanding are vital. Scientists need to interact with one another and establish common understandings, including of the terms being used. This transmission of meaning is vital for scientific advance.

For example, investigation of the (true or false) claim that ‘capitalism generates inequality’ requires an adequate common understanding of what is meant by ‘capitalism’ and ‘inequality’. If there were some dispute over whether or not to include (say) prehistoric trade, ancient Rome, or the former Soviet Union under the category ‘capitalist’, then attempts to critique or justify the inequality claim would be confounded by major differences over what entities were being analysed, thus frustrating the search for possible generic mechanisms that might explain capitalist inequality (Hodgson 2015a).

To get anywhere, participants in the dialogue need to establish some minimal common understanding of a term such as ‘capitalism’, so that inquirers can then identify possible causal mechanisms and place them under some kind of empirical or other scrutiny. This is why taxonomic definitions are important.

Definitions in the social sciences are likely to have fuzzy boundaries. But this is not an excuse for abandoning the tasks of definition and the vital communication of meaning. Scientists must first establish an agreed rough understanding of the phenomena they are investigating. Then they try to focus on the problem, using a taxonomic definition as a means of demarcation. They sometimes change this definition. But definitions matter at every stage. 

Different kinds of definition

Although there is relatively little published on definitions in modern philosophy, the nature and role of definitions was a major topic in classical Greek philosophical analysis. 

Aristotle (1901, book 2, chs 8-10) distinguished between nominal and real definitions. A real definition involves an explanation of the origin and nature of a type of entity: it is an attempt to outline its essence and function. By contrast, a nominal definition elaborates the meaning of the name that we use. 

Clearly, a nominal definition is a necessary condition for attempting a real definition. To attempt to investigate the nature of an object or object denoted by X, we must first adopt a view of what class is denoted by X. But the use of a nominaldefinition does not imply a rejection of philosophical realism – the view that a (social) world exists, which is independent of our mind, beliefs or senses. Nominal and real definitions are different kinds of definition, each compatible with various philosophical outlooks. 

The term taxonomic definition is introduced here to denote a mode of nominal definition that addresses not a single entity but a populationof entities with some common properties. Populations are commonplace in the social sciences. There are populations of structures, cultures, ideologies, organizations, families, economies, states, individuals and so on. Taxonomic definitions in social science concern populations of social phenomena that exhibit some degree of commonality and some degree of diversity. As in biology, social scientists are forever attempting to classify phenomena in the face of endless diversity and change.

Richard Robinson (1950, p. 19) distinguished between ‘lexical and stipulative definition. Here by “lexical definition” I mean reporting the customary or dictionary meaning of a word, and by “stipulative definition” I mean establishing or announcing one’s own meaning for a word.’ A lexical definition must simply convey existing meaning, while a nominal definition may involve some stipulation (and some departure from customary or dictionary usages) to establish a more precise meaning of a term.

Thislexical-stipulativedistinction refers to the degree to which we may rely on the baggage of meaning that a term may carry in pre-existing discourse. When we invent an entirely new word or phrase to identify a class of entities, we are being entirely stipulative. Alternatively, a stipulative definition may use an already-existing term but stipulate a new meaning. Many definitions in mathematics have a strongly stipulative character. 

But outside mathematics, there are problems with purely stipulative definitions. Words typically connote prior understandings. It is difficult to get rid of all this baggage of meaning. The persuasive task becomes more difficult if we ask every reader to expunge all preconceptions and pre-associations, and adopt a number of terms with strikingly novel meanings. 

Hence, as Robinson (1950 p. 80) argued: ‘The supreme rule of [definitional] stipulation is surely to stipulate as little as possible.’ Similarly, Aristotle wrote in his Topica (II, 2): ‘we ought to use our terms to mean the same things as most people mean by them’. Thomas Robert Malthus (1827 p. 4) argued likewise: with terms ‘in the common conversation of educated persons, we should define and apply them, so as to agree with the sense in which they are understood in this ordinary use of them.’ Likewise, Alfred Marshall (1920, p. 43) argued that the terms used in economics should ‘conform … to the familiar terms of everyday life, and so far as possible [economists] must use them as they are commonly used.’

A definitional tool-kit

Biology establishes taxonomic definitions with reference to evolutionary descent from a common ancestor. The existence of a shared common ancestor is the necessary criterion for establishing a taxonomic class. The ancestral lineage is traced from an approximate point where a distinctive change or mutation occurs, so that later descendants are identifiably different from earlier ancestors. Some distinctive features appear and then persist for among descendants, despite the ongoing generation of variation and tree-like divergence.

By contrast, in the social sphere, hybridization and the replication of hybrids is commonplace. For example, the English language is a fusion of multiple language types, with Latin, Germanic and other roots. Some prominent legal systems involve combinations of common and civil law. Institutions that develop in one country or setting are frequently copied and installed in another. 

Social evolution widely involves reticulate(with divergence and much recombination) rather than hierarchic(involving divergence but with minimal recombination) patterns of descent. Consequently, most species of social phenomena cannot be conceptualized in terms of a single, earliest, common ancestor, from which all others in tree-like formation descend. 

Hence the social scientist must look to other grounds for taxonomic definitions. The old philosophical contest between nominal and realist ontologies becomes relevant. Nominalism denies the existence of universals. But the realist-nominalist ontological dilemma is different from the real-nominal dichotomy concerning types of definition. Aristotle addressed both real and nominal definitions, but from the perspective of a realist ontology that accepted universals.

By contrast, from a nominalist ontological perspective, all taxonomic definitions must be regarded as categorizations of convenience that we impose upon collections of entities. The warrant for any particular taxonomic classification is that it proves useful for analysis. 

A realist ontological approach does not dispense with such difficulties, but it offers an additional grounding for taxonomic proposals. As Plato reported of Socrates in Phaedrus,the approach is to ‘carve’ reality ‘where the joint is’. This raises the contested Aristotelean concept of essence. Aristotle distinguished between essentialand accidental(or incidental) properties of a kind.The essential properties of an entity are those properties of the object that it must have if it is to be an object of its kind. By contrast, an accidental property is one that may be found in the population of that type, but is not universal for that type. 

As Aristotle explained in his Metaphysics, the essence of what an entity iscannot be adequately defined in terms of what an entity does, including by the patterns that it generates. If we make this confusion, then we wrongly imply that when the entity interrupts its characteristic activity, then it ceases to be such an entity. Sometimes birds fly. But what helps define a bird is the (existing or ancestral) capacityto fly, not flying itself. If a bird were defined as a flying animal, then any bird sitting on a branch or pecking on the ground would cease to be a bird. 

If a firm were defined as an organization producing goods or services, then when the workers were on holiday the firm would cease to exist. It would be better to define a firm as an organization with the capacityto produce goods or services. The substitution of transient behaviours for dispositions or capacities is one of the major methodological mistakes in the social sciences. 

From an Aristotelian perspective, the task is to identify a minimumnumber of essential properties that can substantiate a taxonomic definition. These particular essential properties are chosen for their taxonomic effectiveness, not for (say) descriptive adequacy or causal relevance. Taxonomic definitions are not meant to be adequate accounts of the essenceTheir role is to demarcate, not to describe adequately or to analyse. An ontological realist would argue, because of reticulaterather than hierarchicevolutionary descent, taxonomic definition in the social sciences should refer to essences and choose key elements from those properties.

From a realist ontological viewpoint, the usefulness and persuasiveness of a taxonomic definition depends in part on whether it points to a subset of distinguishing properties of a common essence among a population of entities. They may have similar causal origins that are worthy of investigation, or they have correlated properties that support generalisations, even if they do not share a common ancestry. By contrast, ontological nominalists appeal primarily to analytical utility and the demonstrated capacity for scientific progress. In both cases the proof of the pudding is in the eating.

From whatever perspective, taxonomic definitions should be parsimonious. The efficiency of communication is improved by judicious parsimony, which can help to form a scientific consensus upon the basic meaning of a term. Parsimony impels us to establish a minimal list of definitional properties with maximal effectiveness. A taxonomic definition of a kind cannot include everything that is vital for the existence of that kind. Such an impractical definition would consist of an unending list of everything required for the entities to function. Furthermore, the over-extension of the list of definitional properties can spur disagreement and impair communication. A definition ‘should be empty of assertional content beyond its ability to explain meaning’ (Belnap 1993, p. 122). 

We must seek some precision with words, to help us dissect a messy reality with sharp concepts of the mind. Taxonomic definitions point to specific identifying features: they are not just descriptive lists of attributes. The definer has also to persuade others that the identification of a particular class of entities (even if well understood) would point to useful, further theoretical and empirical exploration. 

After establishing a taxonomic definition, analysts then have the big job of understanding the origin, nature, structure, composition, survival, operations and functions of this type of entity. This larger task is likely to be advanced further using a division of labour in a scientific community. The taxonomic definition is a necessary preliminary step for science – to ensure that multiple analysts are talking about the same thing. Although taxonomic definitions are affected and updated by ongoing analysis, definition and analysis are different tasks. 

Hence a taxonomic definition of (say) capitalism is not a theoretical analysis of capitalism. Also, a taxonomic definition of capitalism is not a list of characteristics that are vital for capitalism to exist. A taxonomic definition may seem simple, but it is not a simplifying assumption. It does not give us license to ignore in analysis important properties that do not appear in the definition. 

The kind of definition we require involves a minimal list of characteristics that can differentiate (say) capitalism from other social formations. Its primary purpose is to ensure that researchers can communicate their meaning effectively when using a category such as capitalism. It is vital to understand what a researcher means by the term when he or she uses it. This preliminary understanding is necessary before the researchers try to explain the emergence or functioning of the kind of entity involved. 

Although formal models may be heuristically useful, they may over-infuse the process with elements of stipulation that are more suited to mathematics than taxonomy, and they may divert attention from the vital empirical material. Taxonomic definitions that derive principally from mathematical models are often suspect.

Case studies: firms, markets and instititions

Three case studies illustrate different sorts of problems in this area. They highlight different degrees of consensus on workable taxonomic definitions. But these problems are not confined to these cases. Unfortunately, they are widespread.

The first case study highlights a failure to establish a shared taxonomic definition of a firm, alongside influential but misguided warnings that such a definition should not even be attempted. We may speculate that one reason why innovative research into ‘the theory of the firm’ has slowed down dramatically may be this failure to establish a shared taxonomic definition of the central object of analysis. This is an example of the failure of taxonomic definition, and of some of the possible consequences. 

The second case involves a range of taxonomic alternatives, in this instance over the meaning of the term market. A case is made for a narrow definition, where trade is a broader category and markets are a special case, involving organized and recurrent exchanges. Nevertheless, it is possible to make progress with a broader definition alongside the narrower one, as long as the chosen meaning is made clear and explicit.

The third case involves the definition of the term institution. Unlike many other terms in the social sciences, there is a near-consensus on a taxonomic definition in this case. But some important disagreements remain (Epstein 2015; Hindriks and Guala 2015; Hodgson 2015a, 2016b; Guala 2016; Miller 2010). Nevertheless, there has been sufficient taxonomic consensus over this object of analysis to enable considerable advance. This in turn may open up a vista of more extensive and well-grounded empirical enquiry, spanning multiple academic disciplines.


Aristotle (1901) Aristotle’s Posterior Analytics, translated E. S. Bouchier (Oxford: Blackwell). 

Aristotle (1956) Metaphysics, edited and translated by John Warrington with an introduction by W. David Ross (London: Dent).

Belnap, Nuel D. (1993) ‘On Rigorous Definitions’, Philosophical Studies, 72(2/3), December, pp. 115-46. 

Epstein, Brian (2015) The Ant Trap: Rebuilding the Foundations of the Social Sciences (Oxford and New York: Oxford University Press). 

Hindriks, Frank and Guala, Francesco (2015) ‘Institutions, Rules, and Equilibria: A Unified Theory’, Journal of Institutional Economics, 11(3), September, pp. 459-480.

Guala, Francesco (2016) Understanding Institutions: The Science and Philosophy of Living Together (Princeton: Princeton University Press). 

Hodgson, Geoffrey M. (2015a) Conceptualizing Capitalism: Institutions, Evolution, Future (Chicago: University of Chicago Press). 

Hodgson, Geoffrey M. (2015b) ‘On Defining Institutions: Rules versusEquilibria’, Journal of Institutional Economics, 11(3), September 2015, pp. 499-505. 

Malthus, Thomas Robert (1827) Definitions in Political Economy(London: John Murray).

Marshall, Alfred (1920) Principles of Economics: An Introductory Volume, 8th edn. (London: Macmillan).

Miller, Seumas (2010) The Moral Foundations of Social Institutions: A Philosophical Study(Cambridge and New York: Cambridge University Press). 

Robinson, Richard (1950) Definition (Oxford: Clarendon Press). 

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.