Behaviour and Thoughts: For a Pluralistic Model of Empirically Informed Political Philosophy

Abstract

Is Political Philosophy Impossible? develops a distinctive and powerful vision of empirically informed political philosophy: one that gives a central role to data about what people do, rather than what they think or say. Here, I offer some critical reflections on this ‘normative behaviourist’ account of how, and why, we should integrate normative theorizing with empirical research. I suggest that normative behaviourism is at once too ambitious and too restrictive concerning the role of social scientific data in political philosophy. On the one hand, it implicates philosophy in complex and contested issues in criminology, and developing the approach to address more fine-grained normative problems would place unrealistic demands on the empirical data. On the other hand, the emphasis on crime and insurrection excludes alternative valuable forms of empirical evidence from normative theorizing. I conclude by defending a more modest and pluralistic picture of data-sensitive political philosophy.

Introduction

Political philosophers ask how we ought to organize our political lives. In Floyd’s (2017: 6) terms, the discipline’s ‘organizing question’ is ‘how should we live?’. ¹ To generate meaningful and convincing answers to this question, he contends, we must go beyond standard methods of armchair philosophy and investigate patterns in citizens’ real-world behaviour. The first move in this argument involves a critique of the dominant approach to political philosophy that Floyd terms ‘mentalism’. Political philosophers, he maintains, have tried and failed to justify answers to the organizing question by looking for patterns in our thoughts. We disagree too much, within and between ourselves, to reach convincing and reasonably determinate conclusions by systematizing our normative judgements. Second, Floyd suggests that a behaviourally focussed approach offers a viable way forward in the face of the problems of mentalism: we can ‘“read off” normativity from behaviour by reading off the political preferences expressed by the behaviour of large groups of people over long periods of time’ (242). The twin building blocks of this ‘normative behaviourist’ political philosophy are political insurrection and crime. These behaviours, Floyd argues, are forceful negative verdicts on the systems that produce them. Thus, the best political regime is the one that tends to minimize their occurrence in the long run.

There are two powerful ideas at the centre of normative behaviourism. First is the notion that actions speak louder than thoughts or words (1). Second is the idea that some actions speak particularly loudly, namely, those that show us what people reject or find intolerable, rather than what they merely prefer. However, as a model of empirically informed political philosophy, I think that normative behaviourism is ultimately too demanding, and too restrictive: it sets unrealistic expectations of the relevant social science, and it rules out alternative productive ways of integrating empirical evidence into normative theorizing. I begin these brief comments by considering Floyd’s treatment of crime and crime data, suggesting that normative behaviourism is underpinned by unwarranted assumptions in criminology. I then raise some concerns about what we might lose by confining normative theorizing within the parameters of the normative behaviourist model.

Crime and Crime Data: Contested Empirical Foundations

Normative behaviourism treats acts of insurrection and criminality as signals that people reject the political system under which they live. ² These behavioural messages are compelling, Floyd suggests, because crime and insurrection involve significant personal risk and people tend only to engage in them when they find their current lives ‘unbearable’ (169). Floyd goes on to highlight the positive association between inequality and crime, a pattern that is taken to reveal ‘expressed political preferences’ (168) against inequality. Thus, we can justify a broadly egalitarian political system, while side-stepping interminable philosophical debates about the value of equality.

This normative behaviourist analysis of crime rests, unacknowledged, on controversial assumptions about why individuals engage in criminal acts. First, it is grounded in the claim that crime is a response to intolerable conditions, a claim that implicates political philosophy in deep and wide-ranging debates within criminology. ³ It is also unclear how the normative behaviourist approach can account for white-collar crime, often committed by individuals for whom life seems far from ‘unpalatable’ (22). Second, it sometimes appears that a rational choice theory, according to which individuals commit crime when the expected benefits outweigh the expected costs, is working in the background of normative behaviourism. ⁴ For example, Floyd positions his approach in relation to the prisoner’s dilemma and employs the language of incentives and payoffs for engaging in crime and insurrection (206–208). But again, the controversial nature of this framework, particularly the tension with emotional or sociological perspectives on criminal behaviour (e.g. Katz, 1988), is unexamined. Third, normative behaviourism groups different types of crime together. Crime in general is ‘normatively relevant behaviour’ (181) that grounds political principles by showing us what those who live under particular regimes are (un)willing to tolerate (178). Criminologists are again divided as to whether there is any plausible general theory that can make sense of diverse criminal acts.

Thus, normative behaviourism ties the foundations of political philosophy to controversial positions on large and difficult problems in criminology. The normative behaviourist might respond here that it does not matter why people commit crime, just the rates at which they do so across different political systems. We need then only observe the macro-level associations between crime rates and features of the political context; we do not have to commit to any explanation of these relationships at the micro-level. However, this reply would require the normative behaviourist to provide an alternative explanation of the normative significance of crime, which is independent of any account of why people engage in criminal behaviour. If crime (contra the picture in Is Political Philosophy Impossible?) is not to be taken as a signal that individuals find their lives intolerable, then why should we prefer systems that produce less crime? ⁵ The obvious answer is that, all else equal, there are good moral reasons to want to minimize crime. For example, where there is less crime, individuals will be better able to make plans free from threats of violence or intrusion. But then normative behaviourism would no longer be concerned with criminal behaviour as a signal of citizens’ political preferences, but rather with what crime means for our ability to lead decent lives. Crime would play a very different justificatory role, one that appears indistinct from a thin form of ‘mentalism’.

There is a second layer to the empirical complexity of normative behaviourism, because it involves making comparisons of crime data to assess which political regimes people find intolerable. Each of the major data sources – law enforcement agency records and victimization surveys – has its own limitations. For example, police-recorded crime is influenced by reporting and detection rates as well as varying recording practices: if 10 people together rob a bank, in some countries we have one crime; elsewhere, we have 10. This creates significant challenges for cross-national comparison of crime rates. For example, much of the variation between European countries in levels of recorded crime may be explained by different counting rules (Aebi, 2008).

The broader issue here concerns the robustness of any apparent relationships between crime rates and features of the political regime. Are these associations likely to be affected by cross-national differences in recording practices, or other factors that cannot be adequately controlled for? When it comes to the inequality–crime relationship, normative behaviourism can rely on a significant body of supportive evidence, at least in relation to violent crime (e.g. Choe, 2008; Enamorado et al., 2016; although cf. Neumayer, 2005). But here the causal story has been bolstered significantly by the availability of time series and regional data, which is unlikely to be available in the same quantity or quality for other aspects of political systems. Thus, the prospects for using evidence on levels of criminal behaviour to adjudicate further debates in political philosophy are uncertain.

The challenge here is less fundamental than my prior objection to normative behaviourism’s controversial criminology, since it concerns the prospects for operationalizing the approach rather than its basic underlying assumptions. Thus, it remains open to the normative behaviourist to identify or to generate further empirical evidence linking levels of crime to political regime characteristics. Alternatively, the normative behaviourist method might be extended to consider other behavioural signals beyond crime and insurrection. However, the closer we look, the more empirically contentious and demanding the normative behaviourist method appears.

What Normative Behaviourism Cannot Do

Normative Behaviourist Moral Philosophy?

I have briefly raised some questions about the internal workings of the normative behaviourist method, focusing on the treatment of crime. This second part of my response points to some valuable forms of philosophical inquiry that normative behaviourism seems to rule out. First, there is an important question about the implications for non-political ethics. While his stated focus is political philosophy, Floyd’s concerns about mentalism in philosophy appear quite general. More than this, much of his evidence about the instability of our normative thoughts comes from work in experimental moral philosophy (142–150). It is unclear why these data would represent a fundamental challenge to mentalism only in political philosophy. Thus, if there is a deep problem with mentalism, it is at least as evident in non-political ethics.

Crucially, however, it is not clear that the same normative behaviourist response is available outside of the political domain. Crime and insurrection data are said to reveal the tolerability of the overarching political system, but they do not deliver verdicts on individual conduct within that broad structure. Thus, to do normative behaviourist moral philosophy, we would need to identify parallel behaviours in the spheres of non-political ethics: high-risk actions that reveal what kinds of norms or systems people find unbearable in practice. Consider, for example, some central questions in medical ethics: When should doctors resuscitate extremely premature infants? Should clinicians take age into account when allocating scarce healthcare resources? How would we go about tackling these ethical questions in line with the commitments of normative behaviourism? Perhaps there are behavioural signals from doctors or patients we could use, such as levels of compliance with alternative regulatory systems. But it is not clear that such an approach would be workable given data limitations, or even equivalent in principle to the crime and insurrection-based model. More generally, the burden of proof rests with the normative behaviourist: to explain why mentalism remains a viable approach to moral philosophy, despite its failings as method for political philosophy, or to show how to do moral philosophy in behaviourist mode.

Fine-Grained Normative Problems

We should also consider what normative behaviourism might preclude within political philosophy. Floyd argues that we have sufficient behavioural data to justify a particular kind of political system. Specifically, the regime that tends to minimize crime and insurrection is democratic, sets limits on inequality and gives significant space to key individual freedoms. However, much of the work of political philosophy goes on at a level significantly below these three broad features. We consider the appropriate form and limits of self-government; we try to work out how egalitarian our political community should be, and in what senses; and we think about what shape liberal freedoms should take. Thus, we are typically not trying to answer political philosophy’s ‘organizing question’ all at once. Instead, we seek to make progress on small pieces of the puzzle. For example, does conditionality in welfare serve equality? If we care about individual liberty, should we oppose policies that ‘nudge’ individuals towards better choices? What role should referenda play in democratic systems? It is at this more fine-grained level that much of the valuable work of political philosophy lies. This is not to deny the value of normative projects that seek an overarching vision of politics. It is simply to highlight how much of political philosophy would be precluded if we were to confine our work to ‘grand theory’ mode.

The question then is whether we can tackle more fine-grained problems within the parameters of the normative behaviourist method. One potential response here is that, while this may not yet be possible, we could do so in future if we generate more evidence of the right kind (181–182). However, the data insufficiency problem appears deep, rather than temporary or contingent. In attempting to apply the normative behaviourist method to more detailed normative questions, I worry that we would be asking of social science things it is not able to do. Consider, for example, testing for a potential causal relationship between nudging policies and levels of crime or insurrection. We lack any systematic data on the prevalence of nudges in different contexts, or indeed any measurement method. More fundamentally, even if we could generate data that allowed us to extend the normative behaviourist method, as the issues become more fine-grained there is an increasing likelihood that we will simply fail to uncover any clear behavioural patterns. This suggests that the normative behaviourist will ultimately have to withhold judgement on many of the specific normative problems that are currently the primary focus of political philosophy.

These reflections on the prospects for taking normative behaviourism beyond the level of broad regime characteristics bring me back to my earlier observation about the empirical demandingness of the method. Normative behaviourism presents itself as a more grounded and realistic way of doing political philosophy: an approach that demands we turn away from abstract thought experiments and get to grips with data about how citizens behave under real-world conditions. But we also need to be wary of an alternative kind of lack of realism here, in what, as political philosophers, we expect empirical social science to be able to deliver.

Diverse Models of Empirically Informed Political Philosophy

In committing to the normative behaviourist approach, we also risk excluding other valuable models of empirically informed political philosophy. I agree strongly with Floyd that empirical evidence can make important contributions to normative inquiry, and this includes a significant role for data about how people behave under different types of political systems. However, there are multiple productive ways of integrating empirical evidence into normative theorizing, and diverse kinds of data that political philosophers can usefully draw on. For example, we might turn to evidence about public opinion to assess the feasibility of our normative ideals. We might draw on ethnographic research to reframe normative problems or to uncover fruitful new areas of normative inquiry (Longo and Zacka, 2019). Perhaps some of these alternative approaches can be integrated with normative behaviourism. However, Floyd’s scepticism about mentalism appears in deep tension with any method that gives public opinion data an important place in political philosophy.

By focusing our attention exclusively on large-scale behavioural data, and according it one kind of justificatory role, normative behaviourism narrows the parameters of empirically informed political philosophy. One area where this narrowing seems particularly problematic concerns a lack of attention to the diversity of lived experience. While Floyd emphasizes the need to ground normative theory in the actions of those who have lived under a political regime, he does not consider how individuals acquire normative insights through their particular experiences and social position. Political philosophers should sometimes turn to empirical evidence in order to understand, and give due weight to, the views of those occupying epistemically privileged standpoints. For example, addressing normative questions about disability requires taking seriously the experiences and perspectives of disabled individuals. To fail to do so is to violate important principles of epistemic responsibility and humility: ‘know the subject that you are using to make a philosophical point and . . . know what you don’t know’ (Kittay, 2009: 614). In contrast, in normative behaviourism it does not seem to matter who the behavioural signals come from; we all count for the same as potential law-breakers, or insurrectionists.

Conclusion

I have offered two sets of critical reflections on normative behaviourism, viewed as a distinctive model of empirically informed political philosophy. First, I highlighted some controversial empirical assumptions bound up with treatment of crime. Second, I suggested that there are more things we can usefully do in political philosophy, including more valuable ways to engage with empirical evidence, that go beyond the model that Floyd defends. Political philosophers should, as Floyd highlights, make use of behavioural data to advance our normative work. But we should not become normative behaviourists.