The Individual and the Aggregate
Science fluently speaks in aggregates. It has a rich vocabulary for averages and distributions, fields and networks, populations and societies, genres and historical periods. Yet this vocabulary can become brittle when directed back towards the particular. Stable patterns at the aggregate level may coexist with considerable variation among individual cases; what holds for a population need not hold for any one individual within it. In thermodynamics, for example, the complex movements of vast numbers of particles appear random. Yet, at the macroscopic level they give rise to stable regularities in temperature and pressure. Such regularities hold out a promise of prediction and control – a promise that becomes politically charged when the elements being aggregated are people rather than particles. The Studienkolleg 2026/27 explores the movement and the tensions between these levels in both directions:
How do we pass from individuals to aggregates – and what allows us to return from aggregate knowledge to an individual case?
From Individuals to AggregatesThe first movement leads from individuals to aggregates. What counts as an individual – and consequently as an aggregate – varies across disciplines. By any means, an individual may be understood as an entity that can be singled out, distinguished from others, and tracked as it persists through time. Aggregates, however, are not simply given. Before individual cases can be counted, compared, or combined, decisions must be made about the categories under which they are treated as equivalent or commensurable. The same individuals can therefore be grouped into different aggregates, each making certain features visible while setting others aside. Yet aggregation does more than summarise: interactions among individual elements can give rise to collective patterns and properties that no element possesses on its own. | From Aggregates Back to IndividualsThe reverse movement – from aggregates back to individuals – is not its mirror image. We possess rigorous scientific languages for describing averages, distributions and population-level effects, but we lack an equally coherent language for what these quantities mean for an individual case. In forming an aggregate, individual differences become mingled together: some are averaged out, others are obscured by the categories through which cases are combined. They cannot simply be recovered by reversing the procedure. Statements about population-level patterns are therefore not automatically statements about individual members. Applying aggregate results to individuals requires substantive assumptions rather than a mere act of deduction. The elusive notion of “individual risk” makes this difficulty especially visible. What does a probability derived from a population mean for this particular person? Aggregate knowledge can nevertheless inform judgment in a singular case – but only through additional assumptions, contextual knowledge and interpretation.
|
Quantification, Legibility, and Power
These two movements extend beyond the production of knowledge. The Studienkolleg will also examine quantification historically and sociologically, asking how collecting numbers and classification became embedded in modern science, state administration, and public decision-making.
Collecting numbers never was innocent or a mere technical business. By classifying, counting, and tabulating individuals, modern states render the populations they govern legible in synoptic form. This “Politics of Large Numbers“ (Alain Desrosières) was reinforced in the 19th century by a fascination with “statistical laws“ whose apparent regularity seemed to reveal an underlying stability of society. Adolphe Quetelet gave this idea its most influential form in the figure of l’homme moyen, the „average man“. For Quetelet, this figure was not merely a statistical summary but a representative social type, while individual variation appeared as deviation or error. The average could thus become not only a description of a population but also a norm against which individuals were measured. The underlying ambition remains uncannily familiar today, as new computational tools extend quantification into ever more areas of life.
From Datasets to Decisions
Machine learning gives this movement from individuals to aggregates and back again a new form. Many systems are trained to minimize average prediction error on a dataset and are then used to classify and rank particular people, make decisions about them, and allocate resources. This industrialisation of decision-making through algorithms may make decisions about individuals appear to follow automatically from data, obscuring choices embedded in categories, objectives, and thresholds. This raises the question of whether the movement between aggregates and individuals is ever as frictionless as it appears.