Embracing the modern, only to find that it bites

After a conflict almost as old as sociology itself, structure and agency began the century at relative peace. Syntheses led by Anthony Giddens's structuration theory had even nudged coexistence towards symbiosis. Agents needed structures to enable and constrain them; structures needed agents to reproduce and change them. While economics urged rejection of any "macro" models lacking "micro" foundations in individual behaviour, social theory found an historic compromise in which people were still products of their context as well as its progenitors.

With neoliberalism flourishing all around it, a reawakening of agency was perhaps inevitable. But, as three of Europe's leading social theorists quickly show, there is little consensus on what sort of individuals are back in control.

Patrick Baert, after an extensive critique of top-down theories of knowledge, calls for a closer assessment of what people want from what they (and social scientists) know. Knowledge is sought as a way of comprehending their world and getting on with it. Sociology's task is to help them articulate the assumptions and understandings that work, and to identify and modify those that do not.

Philosophy of the Social Sciences uses chapter-length assessments of earlier approaches to social explanation - Durkheim's naturalism, Weber's interpretivism, Popper's falsificationism, various strands of critical realism and the Frankfurt School's critical theory - to argue that they share a common fault. All assume a stock of social knowledge awaiting excavation and explanation by whichever methodology does this best. But knowledge can also be created in the process of acquiring and using it, approachable by various methodologies, and defining of the observer as much as explaining what they observe.

Baert's subtitle, "Towards Pragmatism", flags up a boldly unfashionable solution to the paradox of sociologists crediting people with knowledge and reflexivity, while generating theory they find ever harder to use or understand. An action-based framework is derived from John Dewey and George Mead via Richard Rorty, and found at work in such recent developments as Ian Hodder's post-processual archaeology and James Clifford's cultural anthropology. Baert calls for a recognition that individuals seek to understand the world, and themselves, using observations and experience to articulate and challenge their previous assumptions. He urges social scientists to do the same.

Harvie Ferguson perceives the same need to restore sociology's focus on the practical perceptions of ordinary people. But the search for philosophical guidance takes him more across the Channel than the Atlantic, to the phenomenology of Husserl and Heidegger and its postwar existential refinements.

Phenomenological Sociology 's starting observation is that unease at the loss of certainty and unity in social vision was there at sociological inquiry's beginning, not dropped on it at the end. An earlier phase of global transport and communication growth prompted richer Europeans to assemble collections of natural and human-made artefacts. These "monuments to the contingent and arbitrary" challenged Enlightenment belief in a single underlying logic for society, and one best way of arranging its economy and politics.

But multiple logics, while perhaps satisfactory to intellectual theorists trying to differentiate their product, are anathema to people in the process of everyday life, who can handle its complexities only with a very clear picture of what they and their situation are about. Ferguson views sociology's task as understanding the combination of inner traits and outer influences that give people such "totality" and the way their totalisations drive society. He charts the human-level ideas that accompanied the move to larger scale: from industrialisation's bombardment of new commercial, scientific and artistic forms, through the privations of bureaucratically enforced conceptual order, to the traumas of war and of peacetime return to juxtaposition of underprivilege and overchoice. The aim is less to argue the relative merits of Husserl, Heidegger and their interpreters than to clarify their overlaps and divergences and establish their relevance to a world whose rising flow of information seems only to erode its knowledge stock.

Peter Hedstrom also sees social phenomena rooted in individual thought and action, but of a very different kind, for which new investigative methods are now available. By assigning primacy to desires, beliefs and opportunities ("DBO theory") in guiding human action, he roots his analysis in the rational choice made familiar by mainstream economics and game theory.

Dissecting the Social makes the case for modelling individual agents as a sequence of actions, induced by regular confluence of different preferences and choice availabilities; tracking their interactions and feedbacks over time through computer simulation; and "calibrating" this model against real-life instances of the situation until it can reliably predict a subsequent instance. More precisely, the social consequences of DBO-driven action can be traced out using empirically calibrated agent-based (ECA) models, which work by distilling each individual in the model into a computational "triplet" of desires, beliefs and actions (DBA).

Hedstrom seeks early on to distance himself from the "unfortunate instrumentalist tendency" of many rational-choice applications. Even in economics, the approach has come under fire for expediently exaggerating individuals' knowledge and calculating power, and ignoring the lapse from harmonious "equilibrium" if they act on mistaken or inconsistent expectations. He is happier to claim allegiance to Raymond Boudon's generative modelling and James Coleman's mathematical sociology. ECA modelling aims to distil complex social reality into a more tractable on-screen virtuality, while allowing simulated souls a diversity and adaptability denied to economists' "representative agents". It uses data to reality-check a model grounded in individual choice, rather than to find a statistical link among aggregates that subsumes choices under heroic information and computation assumptions, or buries them altogether.

Simulations thus trace out the consequences of different starting points and policy interventions, drawing attention to the sensitivity of results to procedural assumptions, in contrast to traditional regressions, which draw attention to central predictions and often overlook the sensitivities that make for very wide margins of error.

Instead of provoking a straight fight, asking these protagonists why their prescriptions differ so glaringly would probably invite a diamond analogy, taking different lenses to different facets. Hedstrom would accuse the others of pursuing description in place of analysis, or of producing inapplicable analysis by depicting either anecdotal individuals who are at the mercy of idiosyncrasy, or generalised individuals who dissolve into abstraction. Ferguson, replying for his phenomenological tradition, would contend that binary choices of stylised agents in a simulated, computer-generated world cannot approach real people's totalising moments, or give useful clues as to how these combine.

Baert, arguing for an inherent and desirable plurality of approaches, might concede that both the analytical and phenomenological have their place. But his book accuses both of failing to address what concerns most individuals in most practical situations. Sociologists should probe living agents'

actual thoughts and beliefs before turning to simulations and statistical representations and engage with the practical theories in use around them before overriding these with often less practical theories of their own.

Clinging to "unscientific" beliefs and rejection of bold expert counter-intuitions become more explicable once social knowledge comes to be seen as an instrument, not something inherent.

All three books approach their main messages with contextualisation and appraisal of sources that give them value for teaching purposes. Baert's scene-setting is the most extensive, and the opening chapters of his new book complement the earlier Social Theory in the 20th Century. While he again gives balanced accounts of Durkheim, Habermas, Popper and Weber, they serve this time as groundwork for two final chapters recommending, for the 21st century, a narrowing of the gap between scholars and constituents of society. Hedstrom's retrospection focuses more narrowly on other analytical approaches, contrasting "covering law" and statistical models unfavourably with his more flexible mechanism-based explanation.

Ferguson discusses Husserl and Heidegger extensively, but too technically to serve as an introduction; his briefer treatments of Derrida, Dilthey, Durkheim and Weber are more accessible. The biggest complaints over treatment are likely to come from critical realists, whose quest for deeply structured acts loosely linked to surface events is dismissed explicitly by Baert as too foundational and metaphysical, and implicitly by Hedstrom as too vague about the mechanisms and resistant to quantification.

Should we be heartened or disappointed at the disparity of conclusions from authors ostensibly tackling the same questions on how societies work and how sociologists should approach them? There are plenty of points of near-convergence. Phenomenology led, via Merleau-Ponty and Sartre, to an existentialism whose uncertain, contingent world might have lent itself to pragmatic approaches; and Ferguson's "Dialogue" chapter, tackling Hans-Georg Gadamer's hermeneutics and Alfred Schutz's common-sense intersubjectivity, steps close to the ground where Baert urges his discipline to relocate.

But Sartre reacted against the implied egocentrism by rediscovering external structure. And Husserl, Ferguson's starting point, opposes the view of knowledge as instrumental creation by maintaining that social reality is still "out there", rewarding objective probing. Baert's treatment of knowledge as action comes conceptually close to Hedstrom's view that desires, beliefs and opportunities must come together for individual action to result. But the analytic approach, with its assumption that theory must lead to explanation and prediction, requires that individual actions be electronically propagated to see what patterns form on the screen, and into what behavioural proportions the virtual population divides after sufficient iterations. The pragmatic approach asks how such reasoning might help in negotiating the world outside the computer.

Sociology is in shock after embracing the "modern" and discovering that it bites. The world its pioneers analysed as ordered, structured and underpinned by economic logic disintegrated even as elaborate general theories were being built on top of it. The less constrained, more informed individual action it aimed to promote has removed the regularities that had promised to turn analysis into prediction and control.

The remarkable difficulty of making a richer world into a freer one, and the ambiguity of attitudes towards new-found freedom, is eerily evoked by Jose Maurício Domingues's recent Modernity Reconstructed : social sciences' destruction of the certainties of religion, and failure to replace these certainties with the equivalent certainties of Newtonian natural science, "left history without a direction, unable to fill in the gap opened by the collapse of sacred universal standards". Sociology has not abandoned its struggle to fill that gap. These contrasting contributions, each meriting attention, ensure there is no danger of the struggle forging unity.

Alan Shipman is a freelance academic and author, most recently of Knowledge Monopolies , and an occasional participant-observer in social science teaching at Cambridge University.