Abstract
While Herbert Spencer is widely recognized as one of the first figures to speak about differentiation as an important sociological concept, accounts of the discipline’s development often give the impression that his ideas about this process have little in common with those of later thinkers, or were quickly superseded by more sophisticated efforts. Countering this view, the following article argues that the theories of differentiation articulated by Talcott Parsons and his followers in the middle of the 20th century were highly derivative of, and in some respects entirely identical to, those proposed by Spencer. By examining some illustrative parallels between Spencerian theory and its ‘structural-functionalist’ counterpart, I demonstrate that Parsons repeats Spencer in several key respects, and show that the similarities in question are a logical consequence of the metatheoretical commitments shared by both figures. The analysis suggests that these commitments will lead to common features, and generate common problems, in any account of differentiation that begins from their assumptions.
Keywords
Introduction
Herbert Spencer, the ‘forgotten theoretical giant’ (Turner, 1981), has all but lost his status as a thinker worthy of note. Though he played an important role in the formation of sociology as a discipline, and awakened a widespread interest in the nascent field (Carneiro, 1974), it has become a tradition (or perhaps a cliché) to begin any discussion of Spencer’s work with the famous lines with which Parsons (1949: 3), borrowing from Brinton, chose to begin The Structure of Social Action: ‘Who now reads Spencer? Spencer is dead.’ In line with this view, contemporary sociologists tend to treat Spencerian theory as little more than a historical curiosity: interesting for its time, but quickly superseded as a ‘total structure’ of thought (ibid.: 3). For many, Spencer failed to institutionalize a theoretical tradition, and, in comparison to other classical authors (still well liked and still well read), seems to have had little influence on the development of subsequent thought.
Despite some recent attempts to combat this perception, the idea that Spencerian theory is, as Offer (2019: 1) puts it, ‘alien, quaint and a spent force’ remains quite prominent – especially among theorists of differentiation. Though Spencer is generally recognized as one of the first figures to identify differentiation as an important social process, historical accounts of the differentiation concept regularly downplay the significance of his ideas. While these narratives may mention Spencer as an early progenitor of differentiation thinking, his approach is usually presented in a cursory manner, and often with an assumption that it has little relationship to those of later theorists. Placed in a realm of ‘proto-sociology’, Spencer is too often glossed over entirely, or has his contributions credited to a range of more fashionable figures.
Examples of this tendency are easily found. For example, Parsons (1961c: 220) claims that the notion of differentiation, while being ‘central to Spencer’s thought’, is ‘in its more modern phase above all associated with Durkheim’. Alexander (1992: 181), writing 30 years later, concurs, arguing that the ‘modern theory of social change as differentiation began with Durkheim’ (who ‘put Spencer's earlier theory in a new form and started a research program that extends to the present day’). Steintrager (2016: 310), ignoring Spencer entirely, states that ‘functional differentiation as a sociological concept has its roots in the work of Talcott Parsons and before that in Max Weber and Émile Durkheim’. Finally, Alvehus and Hallonsten (2022: 11), while acknowledging that Spencer provides an ‘early sociological conceptualization’ of differentiation, argue that it was Weber, not Spencer, who was the primary point of reference in the development of ‘functionalist differentiation theory’.
The following article provides a counterpoint to these assertions, arguing that Spencer himself anticipated, if not originated, many of the elements that came to characterize subsequent theories of differentiation. Through a detailed comparison between the work of Spencer – the first theorist to develop a ‘sociological’ concept of differentiation – and Parsons – the chief advocate of this concept post-World War II – I show that differentiation theory was, in its mid-century form, far more Spencerian than most commentators have typically cared to acknowledge. I thus intend to trace a point of irony: that Parsons, despite emerging with an emphatic claim to have ‘evolved beyond Spencer’ (Parsons, 1949: 3), ultimately produced an account of differentiation that contains the same set of analogies, assumptions, and conceptual problems. I thus demonstrate that there are elements of Spencer’s work that are more contemporary, and elements of Parsons’ that are more classic, than is generally recognized, and show that we are, in neglecting Spencer, likely not only to repeat his major errors, but to miss a range of insights that continue to be relevant for current theoretical efforts.
Differentiation in Spencer
Let us begin by examining Spencer’s positions. Spencer (1898: Vol. 1, 434) defines differentiation as the process by which a social aggregate ‘passes from an indefinite incoherent homogeneity to a definite coherent heterogeneity’. Two elements of this definition should be noted. First, differentiation creates heterogeneity, and for Spencer this heterogeneity is conceptualized along functional lines. As he notes, differentiation transforms an entity with ‘alike’ parts to one ‘full of structural and functional unlikeness’ (ibid.: Vol. 1, 596). Second, with this change comes an increase in the coherence and definitiveness of the social aggregate. While ‘social organization’ is, in Spencer’s (ibid.: Vol. 1, 596) view, ‘at first vague’, each individual having within themselves all the necessary requisites of independent existence, the differentiation process creates bonds of mutual dependence (or what Durkheim will later, repeating this exact argument, refer to as organic solidarity). For Spencer (ibid.: Vol. 1, 452), bonds of this kind make a society, transforming a group of associated individuals into a definite ‘living whole’.
While these principles are easily summarized, we must, to make instructive comparisons with Parsons’ approach, consider the specific claims with which Spencer fills out his account. Five such claims will be examined: (a) that differentiation is an interdisciplinary concept, (b) that differentiated parts perform distinct societal functions, (c) that differentiation is a process of decomposition, (d) that differentiation is a criterion of evolutionary progress, and (e) that differentiation is fundamentally linked with integration.
Differentiation is an interdisciplinary concept
Although Spencer developed a novel theory of social differentiation, his understanding of this process relies on a host of well-established theories in other fields of inquiry. For example, his thought that early societies were undifferentiated aggregates, in which gradually emerged multilayered distinctions between specialized components, is a simple generalization of Von Baer’s laws of embryonic development, according to which specialized forms always arise from more general ones (Taylor, 2007: 60). Likewise, his idea that differentiation is a process of functional localization, in which responsibility for certain functional exigencies becomes concentrated in specific units, follows the logic of a key doctrine in 19th-century biology: the notion (itself generalized from economic theory) of a physiological division of labour (see Spencer, 1898: Vol. 1, 452).
Generalizations of this kind reflect the general fashion for analogical reasoning in 19th-century thought, but in Spencer’s case they also offer an early example of what has come to be known, following von Bertalanffy (1968: 15), as ‘general systems theory’: an interdisciplinary effort to ‘investigate the isomorphy of concepts, laws, and models in various fields’. Motivated by the idea of a unified scientific vocabulary, Spencer (1898: Vol. 1, 592) posits that entities of a vastly different material character can share ‘fundamental principles of organization’, such that one can not only describe them using the same abstract concepts (speaking of ‘systems’, with ‘structures’, that ‘function’, and can ‘differentiate’), but can analyse them using the same set of abstract propositions (Elwick, 2014).
Spencer’s famous – and often misunderstood – analogy between organisms and societies is merely one example of this kind of thinking. As Offer (2019) has explored in detail, comparisons between societies and organisms are, for Spencer, an example of ‘remote analogical reasoning’, in which likenesses are often distant, and in which similarities need to be ‘inductively confirmed’ through empirical study (Spencer, 1855: 141; see also Spencer, 1898: Vol. 1, 592). Indeed, any reader of Spencer cannot but be struck by the care he takes to establish points of disanalogy between these two types of entities (e.g. 1855: 144; 1898: Vol. 1, 457–9). At the same time, Spencer clearly believes that societies and organisms are marked by a common pattern of organization (a ‘parallelism of principle in the arrangement of components’ [Spencer, 1898: Vol. 1, 448]), and that one can use analogies between them as a ‘scaffolding to help in building up a coherent body of sociological inductions’ (ibid.: Vol. 1, 592–3; see also Offer, 2015: 339–42). They are, for example, both aggregates of elemental units, and are both subject to growth. Moreover, in both this growth is accompanied by a progressive differentiation of structures and functions (Spencer, 1898: Vol. 1, 449ff.). As a result, Spencer believes that one can, in both cases, speak about differentiation and integration: a ‘sub-division of functions and mutual dependence of parts’ (ibid.: Vol. 1, 460). This is, in fact, the only real analogy that he wishes to make. As he notes, ‘There are no analogies between the body politic and a living body, save those necessitated by that mutual dependence of parts which they display in common’ (ibid.: Vol. 1, 592; see also Spencer, 1891: Vol. 3, 411).
Differentiation is defined with reference to functions
As we can see, Spencer’s approach to differentiation is based on a distinctly mereological ontology, which is interested in how the various parts of an aggregate work together ‘in service of the whole’ (Spencer, 1898: Vol. 1, 598). This ontology is, as Turner and Maryanski (1988) note, central to functionalist analysis, whose characteristic arguments link the action of component parts with some quality, capacity, or condition of a wider container entity.
For Spencer, the method of functional analysis is, like the notion of differentiation, an interdisciplinary tool. Indeed, Spencer goes so far as to claim that the ‘proper working’ of all systems can be expressed via the same general concept: that of equilibrium. This concept describes a state in which the forces acting on a system are balanced by forces emanating from the system (Spencer, 1855: 374). Spencer’s basic idea is that systemic entities – including both organisms and societies – are characterized by a distinctive pattern of internal organization (societies, for example, show a ‘general persistence of the arrangements among [individuals]’ within a given territory [Spencer, 1898: Vol. 1, 448]). As a result, to exist as systems, these entities must contain mechanisms that counteract the influence of environmental forces that threaten, disrupt, or otherwise disturb their internal condition (death, Spencer [1873: 339] notes, always results from an ‘inability to withstand some environing action’). The forces produced by such mechanisms are considered functions; in fact, Spencer (1864: 154) defines this term as ‘the statical and the dynamical distributions of force which an organism [or a system] opposes to the forces brought to bear on it’.
The notion of force is thus central to Spencer’s functionalism, and it is by distinguishing different aspects of force that he begins to build an account of differentiation. As articulated in First Principles, Spencer (1870: 154) believes that one can make a foundational distinction between the accumulation, expenditure, and transfer of force. From this comes the idea that all systems must, to maintain equilibrium, contain mechanisms to handle these three tasks. In the case of organisms, for example, Spencer (1864: 155, 306) speaks of mechanisms of alimentation and aeration (addressing accumulation), of nervous and muscular action (addressing expenditure), and of circulation (addressing transfer). In the case of societies, he speaks of three distinct functional subsystems, these being the ‘sustaining’ system (accumulation), the ‘regulative’ system (expenditure), and the ‘distributive’ system (transfer).
While this particular conceptualization is not always maintained, and while The Principles of Sociology contains a range of other ‘functional’ distinctions (such as a distinction between ‘regulative’ and ‘operative’ parts [Spencer, 1898: Vol. 1, 472], between agencies oriented to the ‘regulation’, ‘sustentation’, ‘defence’, and ‘augmentation’ of life [ibid.: Vol. 3, 179], and between six groups of functional institutions – the domestic, ceremonial, political, ecclesiastical, professional, and industrial [ibid.: Vol. 1]) – Spencer (ibid.: Vol. 3, 179) clearly believes that the ‘maintenance of the life of a society’ ultimately depends on the performance of certain analytically distinguishable functions, and that certain macrosocial entities operate in service of specific functional ends (professional institutions, for example, are said to serve the augmentation of life). 1 These macrosocial entities – whether subsystems or institutions – are explicitly conceptualized as sociological equivalents to systems of organs (see ibid.: Vol. 1, Ch. 6), in that they are collections of lower-level structural elements that collectively address a single functional exigency.
From this follows an important point. Because the exigencies of which Spencer speaks must – as universal requirements of social existence – impinge on all societies, one will find, in every existing society, systems of parts that collectively work to meet these functional ends. If, however, the society has a low level of differentiation, in which we see ‘repetitions of one element’, or in which ‘each portion of the community performs the same duties with every other portion’ (Spencer, 1851: 453), these systems will overlap, and be largely indistinct. Conversely, a society with a higher level of differentiation will be marked by specialized, functionally distinct units. When such units emerge, the systems in which they participate grow more separate on the empirical level, becoming ‘more coherent and definite’ in form (Spencer, 1898: Vol. 1, 588).
Differentiation is a process of decomposition
Central to this account is the idea that differentiation refers to the localization or concentration of functions, rather than their diversification or multiplication. This follows from Spencer’s key reference analogies: the notion of the division of labour in manufacture, as theorized by political economists like Smith, Mandeville, Hutcheson, and Ferguson, and the corresponding notion of the physiological division of labour, most notably advanced by Milne-Edwards. Consider Smith’s famous example of the former: the manufacture of pins. Comparing different forms of manufacture, Smith (1880[1776]: 6–7) notes that pin making depends on some ‘eighteen distinct operations’, which are in some manufactories ‘performed by distinct hands’, in others assigned to roles that perform ‘two or three distinct operations’, and, in their pre-industrial form, all performed by one individual. While the first case shows a higher division of labour, this process has not changed the tasks performed; instead, it has simply redistributed these tasks to more specialized individuals.
Similarly, Milne-Edwards’ (1856) account of the physiological division of labour argues that all organisms, regardless of their particular morphology, have the same fundamental capacities, and that more ‘advanced’ organisms merely possess more specialized systems with which to generate these capacities. And just as Smith posited that the division of labour in manufacture allows for a higher level of production, so too did Milne-Edwards (ibid.: 11–12) argue that the division of labour in physiology, by concentrating certain functions in various ‘parts of the animal economy’, helps organisms meet certain universal exigencies (such as nutrition, relation, and reproduction) in a more efficient manner (D’Hombres, 2012: 6).
These ideas had a lasting influence on Spencer (1873: 335–6), who speaks of the ‘luminous conception’ of M. Milne-Edwards, and who claims that the ‘recognition of the advantages which an individual organism gains when parts of it, originally alike and having like activities, divide these activities among them’ is a ‘truth which Biology borrows from Sociology and returns with vast interest’. Following this template, Spencer thus adopts what has come to be known as a decomposition paradigm, in which differentiation is conceptualized as the division of an antecedent unity (Schimank, 2005). For example, Spencer claims that ‘all institutions, at first confusingly intermingled, slowly separate’ (1898: Vol. 1, 597), that functionally distinct structures arise from an ‘indefinite germ in which [they are] mingled with other structures’ (ibid.: Vol. 3, 526), ‘that all social structures result from specializations of a relatively homogeneous mass’ (ibid.: Vol. 3, 181), and that modern professional roles must, regardless of their ostensible novelty, have differentiated out of some ‘pre-existing social tissue’ (ibid.: Vol. 3, 181): a multifunctional role constellation in which one could find incipient forms of all professional labour. 2
Differentiation is a criterion of evolutionary progress
Why argue that all societies, regardless of their particular features, contain the same kinds of functional ‘work’, differing only in the specificity of the parts that perform this work? As D’Hombres (2012) has explored in detail, this approach serves a clear theoretical purpose: it facilitates the use of the differentiation construct as a comparative variable (see also Taylor, 2024). One sees this clearly in Smith’s example of the production of pins, whose demonstrative success relies on a comparison between forms of manufacture that differ only in their respective divisions of labour, rather than in their end products or in the tasks that they perform. The only variable that changes across these hypothetical firms is their respective division of labour, and this helps the example demonstrate a clear link between specialization and productive capacity.
The decomposition paradigm applies this logic at the level of societies. By conceptualizing societies as functional totalities with the same set of requisites or capacities, one creates a common yardstick for comparison, and avoids the problem of ‘comparing incomparables’ (Goldschmidt, 1966: 32; Turner and Maryanski, 1988: 119). In other words, by depicting functions as a stable point of reference, the decomposition paradigm ensures that differentiation can be treated as a universal comparative construct, applicable to, and identical in meaning across, all societies. One can thus assess the degree to which the differentiation process has advanced in different societies, and can, by virtue of comparative control, develop generalized propositions about its relationships with other relevant traits.
Spencer is clear that the development of such propositions is his ultimate aim; indeed he explicitly designates his own project as general sociology – its purpose being to discover ‘facts of structure and function displayed by societies in general, dissociated, so far as may be, from special facts due to special circumstances’ (1898: Vol. 1, 36–7). In line with his interdisciplinary commitments, many such facts are generalized from other domains of inquiry: none more significant than that ‘all embracing truth’, drawn from political economy, but also described as a ‘fundamental principle’ of biology, that specialization helps to perfect productive forces (Spencer, 1873: 335). Applied to sociology, this becomes the claim that a more differentiated social system can meet its exigencies in a more efficient manner, and thus maintain its equilibrium in a wider variety of external conditions (Spencer, 1855: 376–9).
For Spencer, as for many naturalists of the 19th century, this claim was of special significance. As D’Hombres (2012) has noted, pre-Darwinian biology, operating with a progressivist notion of evolution, persistently struggled with a major philosophical question, that of the criteria by which to judge certain organisms as more ‘evolved’ than others. For figures like Milne-Edwards, functional localization – already linked, in economics, to the improvement, advancement, and ‘universal opulence’ of social organization (Smith, 1880[1776]: 12) – provided an ideal candidate; indeed, Milne-Edwards explicitly cites ‘the principle developed by modern economists’ to argue that the functional differentiation of anatomical elements is the means by which nature achieves the ‘perfectioning of all beings’ (Milne-Edwards, 1826, cited in D’Hombres, 2012: 6).
These ideas clearly influenced Spencer, who regularly speaks of ‘more-evolved’ and ‘less-evolved’ societies (e.g. 1898: Vol. 1, 97, 597), and consistently argues that an ‘increasing division of labour … characterises social evolution everywhere’ (1904: 297). Differentiation is thus, for Spencer, equivalent to evolution – not in the sense that it is an inevitable direction of change, but in that it is one possible response to a loss of equilibrium (1898: Vol. 1, 95), and the only response that Spencer (following Milne-Edwards) treats as an evolutionary advance. In other words, Spencer holds that states of disequilibrium – such as those brought on by ‘[the] pressure of population and consequent difficulty in satisfying wants’ (ibid.: Vol. 3, 368) – prompt the emergence of compensatory change. While such changes need not ‘[constitute] a step in evolution’, and while ‘usually neither advance nor recession results’ (ibid.: Vol. 1, 96), such pressures may, he believes, occasion a shift to a ‘higher’ social structure, which will not only be characterized by greater differentiation, but also by what Spencer (ibid.: Vol. 1, 471) refers to as the ‘primary trait of evolution’: the deeper integration of differentiated parts.
Differentiation is accompanied by integration
In accordance with his mereological ontology, and his view that differentiation increases the definiteness and coherence of a social aggregate, Spencer argues that processes of differentiation are regularly accompanied by processes of integration, which consolidate differentiated parts into a single, functional whole (Spencer, 1898: Vol. 1; Taylor, 2007: 62). Although such processes are numerous, they generally address one of two requirements: first, the need for mechanisms that facilitate cooperation between differentiated parts, and second, the need for mechanisms that motivate these parts to cooperate in the first place.
In regard to the former, Spencer believes that societies have developed a range of structural innovations that allow differentiated units to combine their efforts, make use of each other’s outputs, or coordinate their behaviour. For example, the development of a monetary mechanism (which Spencer [1891: Vol. 1, 293], displaying his penchant for imaginative analogies, likens to the emergence of ‘blood-discs’ in ‘higher individual organisms’) allows for the growth of diverse professions, providing a greater basis for cooperative exchange than a system of barter (Spencer, 1898: Vol. 3, 404–5). Similarly, the progressive differentiation of the regulative system creates a need for appliances by which to exercise control: most notably ‘a media of communication’ through which a political centre may exert its influence over an array of subordinate parts (ibid.: Vol. 1, 533–7).
Cooperation must, however, also be motivated, and here Spencer speaks of a range of mechanisms that unify social components by binding them to a wider aggregate. Anticipating Durkheim’s well-known distinction between mechanical and organic solidarity, Spencer argues that unifying mechanisms will differ depending on the size and complexity of a society. Relatively small, undifferentiated societies are, he believes, mostly held together by ideational sources of cohesion: an ‘invisible framework’ of norms and customs that assures consensus and facilitates group identity (1898: Vol. 2, 322–3). While these sources continue to operate in more differentiated societies (which may be united, Spencer [ibid.: Vol. 2, 284] suggests, by ‘pecuniary interests, family bonds, and love of country’), the growth of the social aggregate – typically associated with territorial diversity, cultural heterogeneity, and organizational complexity – will, he argues, weaken the force of shared sentiment. As such, societies of this kind are primarily integrated by organizational sources of cohesion, or by ties of interdependence between differentiated parts (ibid.: Vol. 1, 593–4).
As in Durkheim’s famous account of the function of the division of labour, Spencer thus argues that differentiation is itself an integrative mechanism. Though functionally differentiated parts thus assume activities of ‘unlike kinds’, these activities are in Spencer’s (1898: Vol. 1, 462) view, interrelated in such a way as to ‘make one another possible’. As a result, differentiation creates ties of ‘mutual dependence’ (ibid.: Vol. 1, 706). By ‘increasing [the] inability in each part to perform the functions of other parts’ (ibid.: Vol. 1, 594), differentiation processes make it harder for these parts to separate from the aggregate and survive. This, Spencer believes, ‘conduces to consolidation of the group as an organic whole’, and ultimately creates a more ‘compacted’, ‘coherent’, and more definitely ‘marked off’ society (ibid.: Vol. 1, 706, 260).
Differentiation in Parsons
Having considered the key features of Spencer’s approach, let us now look for evidence of corresponding elements in the work of Parsons. For Parsons, differentiation is a process in which a unit, subsystem, or category of units having a single, relatively well-defined place in the society divides into units or systems (usually two) which differ in both structure and functional significance for the wider system. (Parsons, 1966: 22)
To aid in our comparison, I will consider the five areas discussed above.
Differentiation is an interdisciplinary concept
Like Spencer, Parsons begins from the ontological assumption that societies are systems, comparable, in terms of their behaviours, structures, and internal processes, with entities studied in other fields of inquiry. It is, however, generally not to Spencer that Parsons attributes this idea, but to the work of those connected with the broad enterprise of ‘general systems science’ – an approach that had, by the time of Parsons’ middle period writings, developed into a fully fledged theoretical paradigm (see Parsons, 1970). Spencer therefore does not generally feature on Parsons’ list of influences, with Parsons (ibid.) himself claiming that he instead drew inspiration from more contemporary figures (including Lawrence Henderson, Walter Cannon, Alfred Emerson, A. N. Whitehead, Vilfredo Pareto, and Joseph Schumpeter).
There is, however, a notable line of influence that connects Parsonian systems theory to its Spencerian counterpart. As Parsons (1970: 830) notes, Henderson, one of his primary influences, was a ‘great admirer of Claude Bernard’, and Bernard’s notion of the milieu intérieur was deeply indebted to Spencer’s concept of equilibration (Normandin, 2007: 502). Although Parsons tends to minimize this connection (at one point claiming that the ‘conception of a relatively stable internal environment’, ‘developed for physiology’ by Bernard, was developed for sociology by Durkheim (!) [Parsons and Platt, 1973: 15, 22]), his understanding of the system concept is clearly coloured by its Spencerian heritage. For example, Parsons shares Spencer’s belief that systems are sets of ‘determinate relationships’ among ‘parts or variables’ (Parsons and Shils, 1951: 107), and continues to see equilibrium as a central concept. As a consequence, just as Spencer argued that societies must, to preserve their distinctiveness in relation to an environment, contain internal forces that counteract external perturbations, Parsons writes that every social system ‘tends to maintain a relative stability and distinctiveness of pattern and behaviour as an entity by contrast with its – social or other – environment’ (1954: 143), and that systems ‘must have a tendency to self-maintenance … generally expressed in the concept of equilibrium’ (Parsons and Shils, 1951: 107).
Despite this, there is one major difference between the two theorists. While Spencer speaks of social systems as ‘discrete’ rather than ‘concrete’ (in that their elements are not, like those of organisms, bound together in a physically coherent body), he still sees social aggregates as entities: ‘living wholes’ (1898: Vol. 1, 460) that have a ‘certain concreteness’ implied by the general persistence of arrangements within a population and territory (ibid.: Vol. 1, 448). Parsons, by contrast, repeatedly supports Whitehead’s idea that the system concept is entirely analytic, in that it describes relationships between concrete units, rather than referring to a concrete unity (Parsons, 1970: 839). At the same time, Parsons frequently fails to apply this principle when discussing societies. He often speaks, for example, of ‘concrete social systems’ (Parsons, 1961b: 38; 1951: 39), and often uses terminology – such as that of functionality and self-maintenance – which implies that these systems are actually existing entities. Indeed, Parsons (1977: 230; 1970: 831) consistently speaks of social systems as ‘living systems’, which ‘[maintain] an element of distinctiveness in the face of variations in environmental conditions’ and are, for that reason ‘analogous to [organisms]’ (Parsons, 1954: 143). Despite his philosophical manoeuvres, Parsons thus ends up repeating one of Spencer’s most famous and fundamental arguments, in that he continues to believe that organisms and societies share certain organizational principles, and continues to hold that certain concepts (such as function, equilibrium, evolution, and differentiation) are applicable to both kinds of phenomena.
Differentiation is defined with reference to functions
While differentiation could take many forms, Parsons, like Spencer, uses this term with reference to functional distinctions, at one point claiming (without any clear justification) that the ‘broad lines along which societal differentiation is likely to proceed … must be functional’ (1966: 24). As in Spencer, functions are conceptualized in a mereological fashion: differentiation, Parsons (ibid.: 24) claims, ‘involves the development of subsystems specialized about more specific functions in the operation of the system as a whole’. Equilibrium also continues to be a central concept, with Parsons arguing that the function term designates processes that mediate between the exigencies imposed by the ‘relative constancy or “givenness” of a structure, and those imposed by the ‘givenness of the environing situation external to a system’ (Parsons, 1961b: 36). In his view, a component unit acts as a functional mechanism when it contributes to the preservation of system structure, or works to maintain the equilibrium of its containing system (Parsons, 1977: 101).
Parsons also follows Spencer in arguing that one can – with the aid of foundational theoretical principles – deduce an exhaustive catalogue of functional requisites. While Spencer presents a threefold typology, derived from his theory of force, Parsons presents us with the well-known ‘four-function’ or ‘AGIL’ scheme, justified with reference to two distinct ‘axes’ of action orientations. The first, the internal-external axis, postulates that action can be oriented to either a system (internal) or its environment (external). The second, the instrumental-consummatory axis, postulates that action can be oriented to either the gathering (instrumental) or the utilization (consummatory) of resources (Parsons and Platt, 1973: 10–12).
By cross-tabulating these two axes, Parsons posits that social systems have four fundamental requirements: (A)daptation (external x instrumental), the need to gather resources for the attainment of system goals; (G)oal attainment (external x consummatory), the need to ‘reduce the discrepancy between the needs of the system … and the conditions in the environing systems that bear upon … [these] needs’; (I)ntegration (internal x consummatory), the need for mechanisms that allow differentiated components to operate in a complementary or coordinated manner; and pattern maintenance and (L)atent-tension management (internal x instrumental), the need to stabilize the system’s internal structure and provide sufficient motivation for units to act in accordance with structural patterns (Parsons, 1961b: 38–9).
Although this scheme obviously differs from Spencer’s, there are some notable similarities. For example, the distinction between adaptation systems (which procure resources) and goal-attainment systems (which deploy them) repeats Spencer’s idea that differentiation will initially proceed by creating a distinction between ‘operative’ and ‘regulative’ classes: the former concerned with producing activities, the latter concerned with the execution and direction of ‘external actions’ (Spencer, 1898: Vol. 1, 492). Likewise, Spencer also recognizes the importance of the ‘internal-external’ axis described by Parsons, noting that in ‘systems of organs’ (both ‘individual and social’), one sees a primary differentiation between ‘inner and outer parts’: ‘the one having to deal with environing actions and agents, the other having to use internally-placed materials’ (ibid.: Vol. 1, 519).
A further, and more significant similarity, is that both theorists believe that their functional catalogues are not just applicable to societies, but to other kinds of systems as well. Thus, just as Spencer argues that his threefold division of force – derived from the world of physics – is equally applicable to both organic and ‘super-organic’ (i.e. social) entities, so too does Parsons claim that the AGIL paradigm – initially developed in the analysis of social systems – can be applied at other levels of reference. For example, Parsons believes that the AGIL needs of social systems are met by the operation of four ‘functional subsystems’: the economy (A), the polity (G), the societal community (I), and the fiduciary system (L). Because, however, the AGIL paradigm is intended to apply to all systems of action, Parsons believes that these four subsystems will, as systems, also have their own set of AGIL needs, and thus their own set of AGIL subsystems. The pattern thus repeats itself: each AGIL subsystem contains four AGIL subsubsystems (designated as AA, AG, AI, etc.), which have a further set of AGIL subsubsubsystems, and so on ad infinitum. 3
In Parsons’ model, this hierarchical scheme also applies at levels above that of social systems, which, while having their own requisites met by four internal subsystems, are themselves subsystems (specifically ‘I’ subsystems) for so-called ‘systems of action’ (also comprised of ‘behavioural’ [A], ‘personality’ [G], and ‘cultural’ [L] systems). In fact, Parsons goes so far, in one of his final works, as to claim that these ‘action systems’ – alongside ‘physio-chemical’, ‘organic’, and ‘telic’ systems – collectively constitute the I, A, G, and L dimensions of what he terms ‘the human condition’ (Parsons, 1978). In such accounts, the AGIL paradigm is stretched to ambitious extremes, leading to a grand vision of unified science that is highly reminiscent of Spencer’s ‘synthetic philosophy’ (Turner, 1980: 381).
Differentiation is a process of decomposition
Because Parsons, like Spencer, adopts a fixed catalogue of functional exigencies, he continues to treat differentiation as a process of decomposition, arguing that differentiation occurs when a single, multifunctional unit divides into units that are each primarily oriented to a different AGIL exigency (Parsons, 1966: 22; Parsons and Smelser, 1956: 66, 61). The differentiation process does not, then, introduce any kind of functional novelty; rather differentiated entities are, taken together, ‘functionally equivalent to the original unit’ from which they disaggregate (Smelser, 1959: 2). This is evident in Parsons’ most common example of the differentiation process: the differentiation of households from productive firms during the Industrial Revolution. In Parsons’ account, these two units, which were, under ‘classical peasant agriculture’, a ‘single undifferentiated collectivity’, performing adaptation and pattern-maintenance functions in equal measure, gradually disaggregate into two functionally distinct entities: economic firms, oriented to the ‘A’ function, and a more differentiated family unit, oriented to the ‘L’ (Parsons and Smelser, 1956: 16). At no point in this process do new functions emerge. Instead, units merely become more firmly placed in a single AGIL subsystem.
Here, again, we see a repetition of Spencer’s ideas. In accordance with the AGIL paradigm, Parsons holds that every system contains four – and only ever four – functional subsystems, each made up of those units (or the parts thereof) that work to address a particular AGIL exigency (Johnson, 1973: 208). As in Spencer, differentiation does not, then, create these subsystems, but merely makes the analytic distinctions between them more evident on the empirical level. As Parsons and Smelser (1956: 83) explain, ‘Functional subsystem categories correspond more closely to organizationally differentiated sectors of the social structure as the society approaches greater structural differentiation.’ Put differently, the idea is that as units specialize in particular functional dimensions, or as societies develop specialized apparatuses to deal with particular problems. their ‘concrete structure of collectivities and roles’ will increasingly reflect the distinctions posited by the AGIL model. (ibid.: 83; Mayhew, 1982: 30).
Parsons is thus clear that differentiation cannot produce a ‘fifth’ functional subsystem, but can only progress further at the sub-functional level – that is, by sharpening distinctions between AGIL dimensions within an existing subsystem. For example, units focused on the ‘A’ function, participating in the ‘adaptation’ subsystem, might come to focus on a single AGIL exigency of that subsystem, as when a unit comes to cater primarily to the ‘capitalization and investment’ system (AA) of the economy (A), rather than making a functional contribution to all four of the economy’s AGIL needs (Parsons and Smelser, 1956: 44). In this regard, Parsons in fact presents a more rigid (and more classical) vision of decomposition than Spencer. While the latter figure sometimes suggests that the distinction between sustaining, regulative, and distributive structures can repeat itself within these three subsystems (claiming, for example, that the distributive system ‘eventually gets a regulating apparatus peculiar to itself’ [1898: Vol. 1, 548]), he is not bound to this conception, and speaks of a diverse variety of distinctions at the sub-functional level. In Parsons, no such moderation is present: all cases of differentiation must, at any system level, occur along AGIL lines.
Differentiation is a criterion of evolutionary progress
What motivates this view? Like Spencer, Parsons is ultimately driven by a desire to develop nomothetic (or what he terms ‘dynamic’) knowledge: generalizations that ‘permit deductive transitions from one aspect or state of a system to another’ (Parsons, 1951: 20). While his own work generally remains at the level of conceptualization, and repeatedly defers the task of formulating a system of analytic laws (see Parsons, 1949: 751), it nonetheless operates with this aim in view, and is therefore oriented to the needs of comparative inquiry. In fact, Parsons plainly states that theoretical concepts should create the potential for empirical comparisons (1951: 20), contrasting his own approach to a more ‘historistic’ position in which the ‘meaningfulness … of a structural unit or process could be defined only within one very specific system’ (Parsons, 1977: 283).
Parsons thus adopts the decomposition model for the same reasons as Spencer. Because he wishes to treat differentiation as a comparative variable, his description of this process is designed, as Smelser (1967: 35) puts it, to be ‘general enough to encompass phenomena in highly diverse social settings’. The AGIL paradigm is consciously oriented to this particular need, in that it is ‘not bound to any particular structural type of society’; speaks of analytic elements that are ‘distinguishable … in any society [or] social system’; and offers – by allowing one to assess the extent to which a particular structural pattern corresponds with a ‘generalized theoretical scheme’ – a useful framework for the ‘comparative treatment of social structures’ (Parsons and Smelser, 1956: 83–4).
While contemporary sociologists continue to develop such schemes, there is at least one aspect in which Parsons’ approach to comparative inquiry is resolutely classical (and more deeply evocative of Spencer), namely, his idea that social comparisons should not only develop predictive generalizations, but must also make judgements about which societies are more ‘advanced’ than others (Parsons, 1966: 110). This is clearly stated in the 1966 volume Societies: Evolutionary and Comparative Perspectives, a work that, though claiming to have ‘reconsidered the idea of social evolution in the context of the major theoretical and empirical advances that have accumulated since the earlier evolutionists [including Spencer] wrote’ (ibid.: 109), in fact argues for a surprisingly outdated concept of evolutionary change. For example, Parsons (ibid.: 22) claims that evolution is a fundamentally directional process, proceeding from ‘simple to progressively more complex forms’; that an ‘evolutionary perspective’ requires both a ‘criterion of … direction’ and a ‘scheme of stages’; and that his aim is to ‘distinguish between broad levels of advancement’. Such statements show little trace of modern evolutionary thought, but in fact repeat one of Spencer’s most significant mistakes: the conflation of organic evolution (i.e. phylogenesis) with the process of organismic development (i.e. ontogenesis; Wortmann, 2012: 377).
For this reason, Parsons, like Spencer, needs to establish a criterion of evolutionary progress. His choice, said to be ‘congruent with that used in biological theory’, is what he calls the ‘enhancement’ or ‘generalization’ of ‘adaptive capacity’ (Parsons, 1966: 110, 22): the ability of a system to develop ‘modes of adaptation’ that allow it to cope with ‘increasingly wide ranges of exigencies’ (Parsons and Platt, 1973: 13–14). Differentiation is, however, still central to this conception; indeed it is, Parsons believes (following the AGIL paradigm), one of the four specific processes that collectively lead to an evolutionary advance (these being adaptive upgrading [A], structural differentiation [G], inclusion [I], and value generalization [L]).
Though its categories are more consciously constructed, this model of evolution remains quite close to Spencer’s. Differentiation is still seen as a crucial feature of the ‘evolutionary change cycle’ (Parsons, 1966: 22), and still recognized as a key element in ‘“growth”, “advancement”, and “civilization”’ (Smelser, 1959: 1). There are, however, further similarities. For example, Parsons (1966: 22) explicitly describes two of the four evolutionary processes – inclusion and value generalization – as processes that integrate differentiated units, repeating Spencer’s claim that integration is an essential feature of evolutionary change. Likewise, the process of ‘adaptive upgrading’, by which roles or collectivities come to perform their functions in a more efficient or capable manner, is, in Parsons’ (ibid.: 22) view, intrinsically linked to structural differentiation, it being differentiation that ultimately facilitates productive improvements. This argument is already present in Spencer; indeed it simply restates common claims about the benefits of the division of labour.
Because of this, Parsons also follows Spencer in his thoughts about why differentiation may be favoured as a path of development. As we have seen, Spencer describes differentiation as a direct, teleological response to a loss of equilibrium: typically brought on by changes in a system’s environment. In his view, a more differentiated system has a greater variety of tools by which to address or neutralize these changes; being more internally heterogeneous, it can better respond to a greater variety of external conditions (Spencer, 1855: 376–9). In Parsons’ work, the very same thesis applies. Change, he argues, is a fundamentally adaptive process, which presupposes a ‘need or demand factor’ (Parsons, 1973: 83), and arises in response to ‘disturbances of sufficient magnitude to overcome … stabilizing or equilibrating forces or mechanisms’ (Parsons, 1977: 37; see also Smelser, 1959: 2). While differentiation is only one possible response to such disturbances (which can, Parsons [1961b: 75] believes, also be addressed by ‘mechanisms of control’, the ‘isolation of disturbing forces’, or the ‘radical dissolution of the system’), Parsons shares Spencer’s view of its general advantages. Systems with greater internal variety can, he believes, cope with higher levels of external variety, and can thus, in comparison to systems with a relatively simpler structure, deal with a more diverse array of environmentally imposed challenges (Parsons, 1977: 231; Parsons and Platt, 1973: 13–14).
Differentiation is accompanied by integration
As the previous section indicates, Parsons believes that a differentiated social structure must contain ‘integrative mechanisms which interrelate the functionally differentiated subsystems’ (Parsons, 1966: 24), or which ensure that they ‘[avoid] disruptive interference with each other’ and ‘[contribute] to the realization of certain shared collective goals’ (Parsons and Shils, 1951: 197). Parsons, then, considers the same basic issues as Spencer: first, the need for mechanisms that facilitate cooperation (allowing systems to make use of each other’s outputs); second, the need for mechanisms that motivate cooperation (countering the ‘centrifugal forces’ that ‘divide and fragment’ societies [Parsons, 1968: 140]).
In Parsons’ framework, the former end is met by the so-called generalized symbolic media of interchange: action-based equivalents to the organic mechanisms that coordinate physiological components (such as ‘enzymes, hormones, and neural processes’ [Parsons and Platt, 1973: 23]). Making a rather Spencerian distinction between ‘information-processing’, ‘alimentary’, and ‘energy-utilization’ systems in organisms, Parsons argues that systems contain elements that do not themselves make a ‘substantive contribution’ to an AGIL function, but instead facilitate the activity of functional components: ensuring that units ‘primarily centred in different functional subsystems’ can garner the necessary resources to carry out their roles (ibid.: 23). This is said to occur through exchange: units in one subsystem receive ‘inputs’ from others and deliver ‘outputs’ to them in turn, a process that is mediated by the development of system-specific media (these being, in social systems, money [A], political power [G], influence [I], and value commitments [L]; Parsons, 1968: 149).
The theory of symbolic media is one of the more unique and ambitious aspects of Parsonian theory, and is difficult to summarize in a short space. At the same time, it is possible, even in this brief summary, to see traces of Spencerian thinking. As we have seen, Spencer’s theory of integration also relies on analogies to physiological processes, and on a generic distinction between those mechanisms that perform functions and those mechanisms that merely facilitate these performances. While Parsons’ theory of symbolic media contains some interesting points of distinction (such as the idea that media may be transferred between units, or that there must be – in accordance with the AGIL paradigm – a 1:1 ratio between subsystems and media), his general viewpoint is clearly evocative of Spencer’s perspective.
The same may be said for Parsons’ thoughts on the mechanisms that motivate cooperation. For Parsons, integration of this kind occurs via the two evolutionary processes outlined above: inclusion and value generalization. Inclusion is handled by the integration subsystem (whose function is to maintain the ‘complementarity’ of differentiated units [Parsons, 1968: 140]), and involves two distinct processes: those by which structural components outside of a system ‘are brought into a status of fuller integration in the system and hence with its other components’; and those by which ‘elements which have become separate from others by segmentation or differentiation’ are retained within the system rather than being ‘extruded’ (Parsons, 1977: 293). Value generalization, by contrast, issues from the pattern-maintenance subsystem (whose function is to maintain the stability of value patterns [Parsons, 1961b: 38]) and specifically targets the normative integration of differentiated units. The special feature of this process is the development of values that, while providing units with a common point of orientation and identity, are broad enough to allow a wide range of activities to be understood as legitimate implementations of a society’s dominant value pattern (Parsons, 1968: 140; Parsons, 1977: 307).
Once again, we see some conceptual innovations, but also echoes of Spencer, particularly when examples are given. For example, just as Spencer argues that a sense of national identity, and a corresponding love of country, is an important source of ideational cohesion in industrial societies (Spencer, 1898: Vol. 2), so too does Parsons (1971: 92) argue that the notion of citizenship – in dissociating national belonging from ethnic attributes – facilitates the development of more inclusive societal communities. Likewise, Parsons’ idea that the pluralistic value patterns of liberal democracy are a major evolutionary advance, allowing for common identification with ‘fundamental societal value-commitments’ to exist alongside tolerance for a broad range of functionally distinct activities (Parsons, 1968: 159; 1977: 309), is clearly anticipated by Spencer, whose defence of the liberal community rests on the thought that liberal values, by combining the ‘highest individuation’ with the ‘greatest mutual dependence’, characterize societies in their most advanced or perfected form (Spencer, 1851: 441; see also Hiskes, 1983).
Parallels and problems
The above discussion presents several substantive parallels between the work of Spencer and Parsons, and shows that Spencer, far from being a primitive forerunner for subsequent theorists, articulated much of what Parsons would later present as novel theory. Indeed, I would go so far as to say that on the topic of differentiation there are many aspects in which the work of the former has clear advantages of that of the latter. For example, Turner (1985: 73–9) has argued that Spencer’s account of the causes of differentiation is in fact more developed that the one offered by Parsons, in that it not only identifies some empirical factors that may encourage differentiation (such as population growth, resource scarcity or ecological concentration), but shows, in comparison to Parsons’ rather ‘primitive and poorly thought out evolutionism’ (Abrutyn, 2014: 2), much more interest in the mechanisms by which such change occurs. Parsons does, of course, offer conceptual innovations in other areas, but even then, it is questionable whether he really advances beyond Spencer’s position. The theory of symbolic media, for example, is generally recognized to have ‘suggested as many new problems as it resolved’ (Fox, Lidz, and Bershady, 2005: 13), while the AGIL paradigm, in forcing one to conceptualize differentiation via the exact same set of theoretical distinctions, at every possible level, creates an approach whose rigid formalism seems far more classical than anything offered by Spencer.
These differences should not, however, obscure our central point, which is that Parsons’ theory of differentiation is highly similar to the one offered by Spencer. Though this fact is little recognized today, the parallels between these two accounts are, for those who know Spencer, difficult to ignore; in fact they eventually became so obvious that Parsons himself was, in defence of his own framework, ultimately forced to revise his claim that Spencer’s thought was dead (Wearne, 2006). Indeed, his little-known introduction to a 1961 reprint of Spencer’s The Study of Sociology, presents a complete about-face, with Parsons now arguing that Spencerian theory, far from lacking significance in the contemporary discipline, contains ‘much that is surprisingly modern and relevant to our own time’, is ‘very close to the position of modern “functional” theory’, and provides ‘very much of the framework of a satisfactory sociological scheme’ (Parsons, 1961a: v, vii, x).
Though Parsons’ admissions have, like the correction of an erroneous headline, done little to offset his earlier pronouncement, they not only offer strong support to my overall argument about the relationship between these two figures (to which we can add similar judgements by Sorokin [1961], Andreski [1969], and Turner [1985]), but also suggest some interesting follow-up questions. For example, one prominent feature of the 1961 introduction is that Parsons, while admitting that his theoretical framework shares some striking similarities with Spencer’s, consistently stops short of treating Spencer as an influence on his own work. Instead, the notion of convergence – so important to The Structure of Social Action – seems more apt, with Parsons giving the distinct impression that he independently landed on a number of ideas which just so happened to be central features in Spencer’s system of thought. 4
If we trust Parsons on this point, and are not satisfied that such significant parallels could arise from sheer coincidence, it is worth asking why Parsons’ account of differentiation repeats so many of Spencer’s claims. One can of course point to the impact of intermediary figures (most notably Durkheim, whose ideas were more Spencerian than he cared to admit [Perrin, 1995]), as well as to a shared background in biology, but I ultimately believe that the similarities outlined above are tied to the fact that both figures begin from a similar set of metatheoretical assumptions, which guide their theories of the differentiation process in specific and limited directions (Taylor, 2024). Put differently, it is my belief that Spencer and Parsons are joint participants in what Larry Laudan (1977) refers to as a research tradition: a theoretical enterprise marked by a common set of metaphysical and methodological commitments.
Similarities of this kind are easily seen in our above comparison. On the metaphysical level, both Spencer and Parsons believe that societies should be seen as self-regulating systems. On the methodological level, both hold that sociology should uncover generalizable knowledge about how these systems function. Taken together, these positions form the basis of a distinctly functionalist research tradition (Taylor, 2024), and it is, I believe, by virtue of their shared participation in this tradition that Spencer and Parsons have produced remarkably similar accounts of functional differentiation. For example, the notion that societies are self-regulating systems clearly influences the way that both theorists conceptualize the differentiation process, in that it underlies the thought that differentiated components work to the benefit of society as a whole, and makes questions of integration – or of the ‘mutual dependence’ of differentiated elements and the ‘advantages that accrue from this’ (Parsons, 1961a: vii) – a significant point of thematic concern. Likewise, the idea that sociology should look to develop nomothetic knowledge encourages both theorists to adopt the decomposition paradigm, or to orient their thinking on the differentiation process around the needs of comparative inquiry.
While the notion of a shared research tradition cannot explain every parallel between the two theorists, the concept can help us explain some important points of convergence, as well as some problems that are common to both figures. As Laudan argues, traditions, by establishing a ‘general ontology and methodology for tackling … the problems of a given domain’ (1977: 89), and by providing ‘assumptions about basic causal processes and entities, whose existence and operation the specific theories take “as given”’ (ibid.: 92), will inevitably constrain the formation of specific empirical theories, excluding certain possibilities while encouraging others. This is evident in our above comparison. Although Parsons obviously looks to advance beyond Spencer, he does not make a sufficient break with the metatheoretical assumptions of Spencerian theory. As a result, he is pushed towards an account of the differentiation process that not only repeats Spencer’s claims, but struggles with the same set of conceptual and logical problems.
Three such problems seem especially prominent. First, the mereological idea that differentiated units address the needs of a societal whole has been successfully problematized by decades of trenchant critique. Merton (1968: 80–2), for example, famously argued that this position – the so-called ‘postulate of functional unity’ – runs ‘repeatedly contrary to fact’, and diverts attention from the ‘disparate consequences’ of supposedly ‘functional’ phenomena. 5 Opposing physiological analogies, critics have noted that societies are markedly less concrete than organisms, such that determining what it means for a society to maintain equilibrium, or even to preserve its existence, is not only a difficult task, but perhaps a fundamentally misguided one. As Smelser (1985: 119) would later note, one cannot assume that societies have a single dominant value pattern, such that all members agree on ‘boundaries’, ‘goals’, or ‘needs’. Statements invoking these terms must therefore be recognized as expressing the subjective preferences of particular actors, rather than speaking to an objective set of system requirements (Rueschemeyer, 1977: 5).
Second, the decomposition paradigm has faced some prominent challenges; in fact it was – in the form of criticisms against Milne-Edwards’ doctrine of the physiological division of labour – already under sustained attack at the time of Spencer’s writing (D’Hombres, 2012). For opponents of this idea, it was plain that functions could vary in accordance with the complexity of organisms: complex organisms did not, they argued, merely have more sophisticated tools with which to perform their work, but had different kinds of work to perform (ibid.: 13). Bracketing the question of whether social entities can in fact ‘function’, it seems clear that sociological applications of the decomposition paradigm face the same critique. While this paradigm provides a useful foundation for comparative analysis, the idea that all societies contain identical kinds of activity, differing only in how they distribute this activity, simply lacks verisimilitude, being too dependent on the notion of functional unity, too proscriptive in its claims about how change will unfold, and too reliant on arbitrary decisions about what exactly constitutes a ‘functional pre-requisite’.
Finally, we must consider some basic problems with the proposed relationship between differentiation and evolution. With the ascendance of the Darwinian paradigm came a fundamental shift in the latter term, which no longer implies that certain organisms, species, or forms of organization are intrinsically superior to others. The link between differentiation and evolution thus seems misguided from the outset. As Darwin himself recognized, the adaptive value of a structural trait must be established in relation to a specific environment or ecological parameter, rather than in some absolute sense (Peel, 1969: 175). While differentiation is often favourable, a comparative lack of differentiation can, in certain environments, be an evolutionary advantage. Even if we merely wish to argue, with Spencer and Parsons, for a ‘tendency law’ – that more differentiated entities will, on average, survive in a greater variety of environmental conditions – we immediately face a range of conceptual difficulties. To rank entities in terms of their ‘adaptive capacity’, one would have to assess all possible environments in which these entities could exist, and to ‘know with some confidence what [their] likely future environmental exigencies are’ (Granovetter, 1979: 501, 505). Judgements of this kind are thus highly impracticable, and simply too speculative for us to be confident about conclusions formed on their basis.
Conclusion
That these criticisms are applicable to both Spencer and Parsons is, in my view, a consequence of the fact that they do not stem from incidental decisions, or from missteps in their implementation of what are essentially sound principles, but are deeply entwined with the metatheoretical positions on which the two theories are founded. For example, the postulate of functional unity flows directly from the idea that societies are self-regulating systems, and will be an issue for any theory that begins from this ontological position. Likewise, the decomposition paradigm is clearly informed by the needs of comparative inquiry, and will be difficult to avoid if one wishes to treat differentiation as a generic comparative variable (i.e. to rank all societies on a single scale of ‘functional specificity vs. functional diffuseness’ [Parsons and Smelser, 1956: 34]).
It is in light of this point that the present piece can demonstrate its ultimate merit. While I have certainly been motivated by a desire to give Spencer his due – or to develop Andreski’s (1969: xiii) claim that Spencer ‘said more clearly, as well as somewhat earlier’, what many mid-century theorists ‘claim as their discoveries’ – the purpose of this piece is not merely to correct the historical record. Instead, by arguing that Spencer and Parsons are joint participants in a shared research tradition, and for this reason arrive at many of the same ideas, arguments, and obstacles, I have attempted to provide an illustrative example of the way that theories are constrained by their metatheoretical commitments, and to show how certain difficulties should be expected to emerge if a theory of differentiation is founded on the distinctive commitments of the functionalist research tradition.
From this comes an important lesson for current work. If future theories of differentiation are to avoid the problems of earlier accounts, they must start from a different set of metatheoretical assumptions. It is not enough to make minor adjustments to existing approaches, or to attempt a conceptual ‘shoring up’; one must instead rebuild from a new foundation. As such, we must not only move beyond the functionalist tradition, but beyond those forms of more contemporary thought that have sought to revive or reform it. 6 While difficult, this process is clearly underway, and one can be inspired by a range of new approaches that attempt to begin with alternative assumptions. In particular, one should note the influence of Wortmann (2012), whose evolutionary model of differentiation speaks of the diversification of ‘social forms’ rather than the disaggregation of functional subsystems; Schwinn (2001), whose notion of ‘differentiation without society’ rejects both systems theory and functional analysis; and Schimank (2005), whose ‘action-centred’ theory of the differentiation process breaks with the notion of societal-level needs.
Though these promising theories will, in repudiating aspects of functionalist thinking, inevitably pull us away from those of Spencer or Parsons, this does not mean that we should forget these figures entirely. Indeed, one sees many contemporary accounts in which the insights of these theorists are – appropriately detached from the metatheoretical assumptions of the functionalist tradition – given fresh life. For example, both Turner (1985) and Abrutyn (2016; 2021) have been able to reformulate Spencer’s ideas on the causes of differentiation into a more systematic and formal set of empirical propositions, with the latter arguing that Spencer offers a set of explanatory principles (about, for example, the relationship between population size and structural complexity) that ‘remain valid and worth holding on to’ (ibid.: 152). Moreover, these theorists have jointly shown that Spencer’s thoughts on the mechanisms of change – in particular his view that agents may consciously work to find new structural variants that resolve real or imagined ‘selection pressures’ impinging on social groups – provide an important supplement to neo-Darwinian models of sociocultural evolution (cf. Offer, 2010; 2015: 353), and can be successfully separated from his orthogenetic views on the directionality of change (Abrutyn, 2014; Turner and Abrutyn, 2017).
In a similar vein, Offer (2015: 354) has recently argued that Spencer’s reflections on the ‘social organism’ remain relevant for contemporary thinking, and can be rediscovered as a worthwhile contribution to debates about the ontology of the ‘social’. Offer (2019) is, however, keen to show that the benefits of Spencer’s ideas lie less in their relationship to a systems-oriented functionalism, à la Parsons, and more in their place within a tradition of thinking on the notion of ‘spontaneous order’. By making such links, Offer (2019) demonstrates that Spencer can still offer ‘original and stimulating’ ideas on such topics as the interrelation of holism and individualism (cf. Zafirovski, 2000), as well as the relationship between sociology and biology. Considering that differentiation theorists continue to debate the relative merits of ‘systems-theoretical’ vs ‘action-theoretical’ approaches (Schimank, 2005; Taylor, 2024), and continue to draw on biological thinking (e.g. Luhmann, 2013; Wortmann, 2012) it seems that Spencer still has much to offer, and that elements of his thought could, separated from the functionalist tradition, and harmonized with the demands of an alternative metatheoretical approach (see Taylor, 2024), prove instructive for the future development of differentiation theory.
Efforts of this kind show the merits of engaging with earlier theorists, even as we seek to move beyond them. Without a clear map of the intellectual paths that others have already taken, it is all too easy to retread the same ground, to discover the same insights, and to find ourselves arriving at the same dead ends (Turner, 1985: 153). As the above analysis has shown, such tendencies are all too evident in Parsons’ account of functional differentiation, and future work should take care to avoid the same mistakes. Let us then treat this as a cautionary tale, and recognize that differentiation theory can continue to benefit from an engagement with Spencer and Parsons. The present piece is offered with this thought in mind, and with the firm belief that we can, by understanding their respective contributions to differentiation thinking, not only learn from past mistakes, and gain confidence in future directions, but recover key insights that are worthy of preserving in alternative traditions of thought.
