Normative Behaviourism: A Reply

Abstract

Normative behaviourism says that the measure of political principles is how we respond to them in practice, not how they appear to us in theory, but is that a sustainable distinction? Does normative behaviourism end up relying on mentalism, or even utilitarianism? Does it assume too much of the data we either have now or could ever have? Does it bind us to the status quo or presume the end of history? All these are plausible worries, though perhaps not fatal ones, provided one remembers at least two things: first, that we judge this approach by comparing it to the alternatives; second, that we keep on experimenting, both in politics and philosophy alike, including with normative behaviourism itself.

Introduction

Normative behaviourism is an odd way of doing political philosophy with a lot to say for itself, including the claims that some actions speak for themselves, if we would just observe them; that they speak more clearly than dishonest words or unreliable thoughts; and that they speak more loudly than any decision we imagine we would make from the comfort of our armchairs. So, instead of hypothetical thought experiments, try real decisions made by real people in real circumstances. Instead of clamshell auctions, children drowning in ponds, or runaway trolleys, try revolutions and crime waves. Instead of reflecting upon and rationalising whatever it is you find attractive in theory, look at what billions of people find awful in practice. Think here of those West European intellectuals who thought life in Communist Eastern Europe heaven, while those who lived there found it hell. This is a theory that puts their experience, their black market economy, and their uprisings, front and centre. It means less contemplation, more causation; less deduction, more induction; less introspection, more observation. It means less ‘mentalism’. ¹ Here, we judge political principles by how people react to them once they have had the time to live with them, just as we judge meals, not at first taste, but only after full digestion, and regardless of what the clever chef thought of them in advance. Here, if you want philosophical proof, check the political pudding.

All of this, and more, has been well understood by each author in this collection, and for that I’m truly grateful. ² Our work knows no compliment beyond critique, and no attention beyond argument, so all one ever asks of such things is that they be carefully done, which is certainly the case all round here. Much of it, if I’m honest, feeds doubts I have long harboured myself, and with which I continue to wrestle. Some of it, rather usefully, raises new worries I had not considered. All of it, without fail, helps move these ideas forwards, pushing me to define and defend, but also refine and amend, as required. Am I, for example, an unwitting mentalist, or even utilitarian in denial? Are the data-sets I use insufficiently strong or detailed? Does the way I use them make me unduly conservative? I hope not in each case, though that is hardly the point, as it would be too ironic if my ‘normative thoughts’ ³ were somehow the crux of the matter. Instead, I leave it to others to make of these problems what they will, once they have seen the reasoning below, or perhaps returned to the original text ⁴ - unless of course they make something better. By seeing where normative behaviourism has been so far, perhaps soon they see where it should go next. As someone who has always tried to go from ‘was’ to ‘ought’, ⁵ this would be more than welcome, for although it might push me into the past, it would pull my theory into the future.

Mentalism in Disguise and Utilitarianism Under Cover

Can we say that I think therefore I should? Or that I act therefore I ought? We can certainly say that some of my critics seem to think that I think this (Dowding, 2023; Erman and Möller, 2023; Miller, 2023). Erman and Möller, in particular, pin it down when they say that, for me, ‘a person’s justification status always depends on facts about that individual’, whether they be facts about the things they think or the things they do (Erman and Möller, 2023: 4). And that is certainly true, just as it is true that I think thinking has been given more importance by most contemporary political philosophers than they realise, or indeed than it ought to have. Nonetheless, Dowding speaks for most scholars when he says that political philosophy proper is ‘composed of arguments justifying conclusions, even if those arguments defend or modify moral intuitions’ (Dowding, 2023). My view, by contrast, is that although this fits their self-understanding, it misunderstands what they actually do, by underestimating the real weight given to those intuitions, or indeed other kinds of ‘normative thought’, such as Rawls’ ‘considered judgements’ (Floyd, 2017b).

Even here then, it seems, we have a gap between what people think they are doing and what they are actually doing, because if I am right, the real use made of these thoughts is foundational, not fleeting. Although we like to think that we are only interested in what we ought to think, and not really what we currently think, this is misleading, as we only ever reason to the former on the basis of the latter. It is because you think you would save an imaginary child drowning in a pond that you ought to accept a particular principle of justice. It is because you think you would turn a runaway trolley this way or that way that you should accept a particular principle of warfare. Of course, such claims get stacked up and refined via ideas such as ‘reflective equilibrium’, ⁶ but even so, the working material ‘mentalist’ political philosophers rely upon remains the same – they are trying to turn thoughts into oughts.

Even if that is true though, how can I possibly say that certain kinds of action count as a basis for what we ought to think about politics without in turn pointing to some principle – and in turn some pre-existing thoughts – that explain just why that is? In other words, even if I were right that most political philosophers are more ‘mentalist’ than they think they are, might I not be just as much of one myself, except now with a whole new level of self-misunderstanding? After all, how could I possibly think, as I do, that insurrection and crime levels are the true measure of a political system, and in turn the principles those systems express, without some explanation, beyond such behaviour, of why that is? (Floyd, 2017a: 185–197). I might, for example, have within me some unexamined notion that expressed dissatisfaction matters because preference satisfaction matters? Or the feeling that it matters as a measure of our well-being? Or the conviction that societies with less of these things are somehow happier ones? If so, I’d have a double-problem on my hands, because I wouldn’t just be an unwitting mentalist, but also a utilitarian in denial (Erman and Möller, 2023: 6–7; Miller, 2023: 4). ⁷

Again though, my claim is that real actions really do ground political principles, just as hypothetical ones ground them in familiar ‘thought experiments’, were they to be as reliable as we would need them to be – which they are not – within and across individuals (Floyd, 2017a: 99–165). Think here of the logic of such experiments, according to which we say that if we would do x in circumstances y, and if x expresses principle z, then we are duly bound to z whenever y pertains, unless there is some other and more appealing such experiment, or some other set of thoughts more dear to us, that point elsewhere. Here, in similar fashion, we say that because people really do perform A in response to B, they really should think C.

How though is one to sustain that idea? Consider, for example, Dowding’s claim that in charactering normative reasoning this way, I essentially conflate fleeting norms with fundamental morality, implying once more a confusion between what we currently think (or do) with what we really ought to think (or do) upon reflection (Dowding, 2023: 3ff). Or consider Erman and Möller’s use of an example that cuts right to the core of the problem, involving the principle ‘we ought not to torture babies for fun’, and in turn a person who somehow does not hold it. Do we really want to say to such a person that they are free of that principle just because they don’t believe in it or act in accordance with it? Or, as they helpfully put it, even if the principle somehow isn’t ‘justified’ for them, would they not still be ‘bound’ by it? (Erman and Möller, 2023: 4).

This example is especially helpful, however painful, because it allows me to double-down on a point that needs emphasising as much as possible: For me, as for Goethe, the beginning really is the deed. Things aren’t normative, they are normative for. Things aren’t justified, they are justified for. I don’t judge the stars for not caring about my feelings. I don’t ask sheep to sign social contracts. I don’t expect rocks to treat other rocks as they would themselves like to be treated, any more than I expect my cats to stop torturing babies. Every week they bring them in and every week they play them to death, without a hint of guilt or remorse, though I suppose it is possible, if they could take part in our debates, that they might say 1-year-old mice are somehow different, morally squeaking, from 1-year-old humans?

Either way, the comparison matters. We cannot expect people to follow principles for which they lack the foundations. My feeling unhappy, for example, is only a reason for you to help me if you care at some level about my happiness. Sure, I could say you should care, but that would only bite if I could find some other commitment underpinning it you do already hold – perhaps, for instance, some notion of being a good person, of showing humanity, of having a noble reputation, or of being a kind and generous soul. If, however, you just don’t care, then you really don’t care, just as someone who cares nothing of pain, or babies, or social stigma, cares nothing of our outrage at their torturous habits. Erman and Möller’s principle would be neither justified nor binding for them, with no gap at all between those concepts, even if we justify to ourselves our binding them, locking them up, and throwing away the key. Principles are justified for people only in virtue of facts about them. These facts bind, base, ground, root or found – call it what you will. More than a feasibility check or filter, they are fundamental.

There might though be one last way of putting this point, in order to, if not convince these critics, then at least clarify our differences. For normative behaviourism, political philosophy operates independently of some ‘external’ moral theory in much the same way that Rawls once claimed moral theory was independent of meta-ethical theories. ⁸ In his view, we start with our considered judgements and see where they lead, without worrying about their truth from the point of view of God or the universe, let alone the epistemologists. Here, in turn, we start with the preferences expressed by our actions and see where they take us. As for the goose, so for the gander-or even another animal altogether?

Imagine, for example, that you observe grizzly bears trying to catch salmon, or squirrels trying to find nuts, but then notice a few bears looking for fish up trees, or a few squirrels staring longingly into the water. Presumably, you could advise them on what to do next based on what generally works better for their kind, or even more generally on the starvation that both species can be observed as trying to avoid? This is an analogy I have used before (Floyd, 2017a), and have tried to refine since (Floyd, 2020), but it helps to rework it here, and not just in the face of that telling counter-example from Erman and Möller, noted above, but also because it brings with it further problems addressed below. What happens, for example, if not all squirrels want nuts? What happens if some bears, at least sometimes, find things they care about more than their own survival? What happens if we are simply unable, for whatever reason, to observe and learn everything we need in order to really know their preferences and where they lead? It is to these and other problems that I turn next.

Bad Data In, Bad Principles Out

Consider the following question from Miller: ‘Suppose I see someone driving a car off the end of a pier [. . .] what have I witnessed: was it a tragic accident, a suicide attempt, or a stunt for the next Bond film?’ (Miller, 2023: 2). This neatly captures what seems to be a rather big interpretive hole at the heart of normative behaviourism. How, after all, are we to understand human actions without asking what they mean to the people performing them? Beyond observing them, shouldn’t we find out what they have said or written about whatever it is they take themselves to be doing?

There is, of course, an important truth in this objection, yet it misunderstands our particular interest in the actions described. Think, for example, of the thousands of protestors in Tahir Square in 2011. Each of them was angry and all wanted change, despite both vagueness and disagreement on what should replace the status quo. That though is not a problem for normative behaviourism. All it needs to know is that this crowd was reacting against the system as it stood; that they hated it; that they wanted it gone. Ultimate intentions, their variations, and their varied interpretations, are immaterial here, because we do not need the pro, only the anti. What counts is the strength and spread of anger, as captured by the sheer volume of those involved. What counts is not what they said or would say about the government, but the fact they risked their lives to say it. What counts, put differently, is the cumulative demonstration of opposition to the system, despite variation in how they experienced it, and despite some being angrier than others (Baderin, 2023: 5). Anyone, after all, can say they hate something, or write about what they would fight for if given the chance, just as everyone, according to Mike Tyson, has a plan until they’re punched in the face. Talk, too often, is cheap, which is why the verdicts that matter to normative behaviourism are the ones we give with our feet, not our mouths.

Do we though really see these verdicts with anything like the clarity I claim? Consider, for example, Perez’s argument that normative behaviourism requires behavioural patterns of a certain quality, which for him means sufficient certainty, clear thresholds, a precise ranking among rival preferences, and even a solid theory of their background, in the sense of the conditions needed for us to know when our actions really are an expression of the relevant preferences (Perez, 2023: 3–5). When we look at insurrection, for instance, how much is enough to be problematic, and under what circumstances? Or, when we look at criminology data-sets, how should we compare insider trading with violent crime, let alone crime levels across two different countries with completely different criminal codes? Baderin, helpfully, provides a range of telling examples on this front, including the problem of a crime involving 10 people counting as one crime in some countries and 10 crimes in another (Baderin, 2023: 3). How then are we to observe and measure such things in the way I describe, when the way they are categorised, and the patterns they supposedly express, vary so widely?

These are tricky and insightful points, as well as a reminder of what genuinely skilled interdisciplinary work looks like. Comparative data-sets involving political systems, on the one hand, and crime and insurrection levels, on the other, seem of little use if we throw them all together without registering that both the things they compare and the conclusions they draw vary widely (Baderin, 2023: 2-3). Nonetheless, it is that very idea of comparison that helps us here, because the key question at this point is not how good the described evidence is, when compared to some perfect standard, but rather how good it is when compared to the alternative, which in this case means mentalism. In other words, if there are clear tendencies in the relevant literature, as based on the studies used so far, and if they really do show such behaviour rising or falling according to changes in the political principles in operation, then we have what we need, even if, naturally, we would like to have more, just so long as it is also true that our thinking doesn’t somehow offer a better picture. That then is roughly my position, together with the dependence it inevitably implies on the work of other authors, because although both thoughts and actions are messy, we do know enough, I take it, to think the former hopeless, in the sense of insufficient consistency across and within individuals, and the latter just about good enough, given that it does seem to tell us the only two things we really need to know here: (1) increase liberal-democracy and you are likely to reduce insurrection; (2) reduce inequality and you are likely to reduce crime. ⁹

We can then, I hope, agree that there is much indeterminacy in the data without saying there are no tendencies at all. Humans are not random; politics is not chaos; and although you might fear, as Modood does, that a ‘probability’ could never be enough for the kind of ‘universals’ I’m offering (Modood, 2023: 5), it is when you realise just how low the bar is set. Think here, for instance, of the probabilistic data we have on smoking and seatbelts, as used to restrict the former and enforce the latter. Naturally, we do not know exactly who would get cancer from smoking, any more than we know exactly who would die without a seatbelt, but that is fine, because we know that, for every million people involved, thousands of lives hang in the balance. Think of the wood, not the trees. Or, put differently, forget the individual driving off a cliff, or carrying a placard, or even lighting up a cigarette, and focus instead on the large aggregate patterns those actions produce. Which systems produce more insurrection and which less? Which systems produce more crimes and which fewer? And if, and only if, there are real tendencies here, of a kind that is missing in our thoughts, then use them as best you can, while always looking to run new experiments to improve those data – a point to which I return in a moment.

I wonder though, if you are still not convinced, whether it might help to imagine not using behaviour at all? This means thinking actions say nothing at all, or at least nothing of politics; that there is no cause-and-effect relationship between systems and behaviour, despite what the studies suggest; that human nature is just too malleable or empty to show any regularity with political inputs and behavioural outputs, in which case, surely, why worry about those inputs at all? Or, if nobody wants to go that far, perhaps instead they would say simply that regimes do tend to produce different consequences, and that this does matter somehow, but not quite as I describe? Consider, on this note, Dowding’s claim that it would be no great surprise if the poor wanted more equality (Dowding, 2023: 2), and in turn the notion that normative behaviourism might be mistakenly tracking poverty rather than principles and economics rather than politics. If so, this would be normatively interesting, but not normative behaviourism, as the route from behavioural facts to political principles seems closed. Yet is that really true? Presumably, if there are enough angry poor people to cause a revolution, or a group of people so poor that they cause a crime wave, then that must tell us something of the system responsible? Of course, you might say it only ‘tells’ us anything in virtue of some other standard that has so far gone unnoticed, whether mentalist, utilitarian or otherwise, but that was the objection of the previous section, to which we need not return again here.

What matters here is that certain data-sets exist and persist. Ignoring them because we want them to be either more certain or more comprehensive would be a mistake, just as it would be a mistake to ignore them because we do not like their conclusions, though that in turn would be a familiar kind of behaviour, just as it is a familiar behaviour for all of us, regardless of theoretical persuasion, to read and judge people based on their actions in our day-to-day lives. Family members, friends, even colleagues. Their decisions on what to write, when to attend meetings, whether to review articles or how often to attend open days. All of this we manage without an interpretive encyclopaedia, knowing full well that such actions tell us something of their character, and in turn what they really value, just as the actions of millions of people say something of the character of their institutions, and in turn the principles those institutions represent.

Think here, for example, of that egalitarian colleague of yours with a ‘special’ reason for sending their child to private school; of Prime Ministers who says they are not really at a ‘party’, and therefore not really breaking their own Covid-19 rules; of dictators who call their invasions ‘special operations’ rather than wars; or of any contemporary regime trying to describe incarceration as ‘education’. The story told by interpretivists, don’t forget, is that you need to accept the stories people tell of themselves, but why on earth, especially in politics, should we stick to that? ¹⁰ Hobbes was at least partly right when he said that republicans say more of human nature by locking their doors than they do in their speeches; just as the department at which I started my PhD said more of politics when they abandoned ‘deliberative democracy’, in favour of a return to ‘dictatorship’, than they did when demanding it in the first place. The experience of chaotic and nasty behaviour across a range of reasonable colleagues is what counted, not how appealing it all looked in advance. Civil war is evil, but also educational.

Saying, however, that we should focus on our reactions to principles in practice, including in particular those reactions expressed by the kick-backs of crime and insurrection, is still not remotely the same thing as saying that the theory is finished or perfect. It might transpire, for example, that contemporary China confuses the picture, especially when compared to what is still a fairly young form of democracy in the West (Miller, 2023: 5). It might turn out that normative behaviourism helps us in political philosophy but not in moral philosophy, in which case we would have to look very carefully indeed at both the mental and behavioural resources available in that domain (Baderin, 2023: 3–4). Either way, we chase as much data as we can, including from new ‘experiments’, as explained in what follows.

Normative Behaviourism as Neo-Conservatism?

According to Modood, normative behaviourism offers too minimal a view of the ‘good polity’ (Modood, 2023: 2). According to Baderin, it is ‘too demanding and too restrictive’ (Baderin, 2023: 2). According to Rossi, it presumes we’re at the ‘end of history’ ¹¹ (Rossi, 2023: 4). These are both perceptive and mutually reinforcing claims, because you might well worry here that, if we did stick to deeds over words, let alone thoughts, we would never get beyond the world as we have known it so far. No one, after all, has had a chance to react to the as-yet-unseen, let alone the as-yet-unimagined, so ‘demanding’ behavioural evidence as a standard of justification really does seem unduly ‘restrictive’. It might mean, for example, that we could never criticise Denmark, if progress beyond it is impossible. ¹² Or it might mean that the only way to achieve such progress would be to temporarily ignore normative behaviourism by ignoring the idea that principles can only be justified by past practice. That though would only raise a further worry: How on earth could we have come as far as we have done, including the wonders of modern Denmark, if we had ever followed in the past such a backwards-facing theory? Maybe we needed traditional normative argument? Maybe we needed mentalism? Maybe we even needed violence?

This is a serious set of problems, though not quite on the scale suggested, if only because mentalism too has a problem with how it would have fared in the past, given how intuitions and judgements could have differed and thus ‘restricted’ us in earlier eras. ¹³ Saying, in other words, that it couldn’t have saved us in the Middle Ages, is not the same thing as saying it can’t help us now. That qualification, however, matters less than a second point, which is that normative behaviourism, by treating the past as a series of ‘experiments’ on principles ¹⁴ (Miller, 2023: 4), is perfectly open to further experimentation in the future, and especially so if such experimentation has itself proven useful (Floyd, 2017a, 2020, 2022).

The key here is simply to start seeing successful political innovations as things that were put to the test in the past, and in turn how it would be wise to continue such testing in the future, all without turning to mentalism. That is, rather than trying to prove the truth or justice of some new principle right now, from the safety of our armchairs, we say only that new ideas, from the European Union (EU) to Universal Basic Income (UBI), should themselves be put on trial, as indeed they have been, with their results then carefully monitored in terms of the long-term behaviour they produce. Should those results be promising, in terms of how people react to them in practice, then carry on tweaking, testing, and adopting, but if not then stop.

There is though still more that needs to be said here, beyond these two points, in order to properly resolve the problems described. Building on the three critical claims noted earlier, consider further Rossi’s worry that our actions, including our reactions, could somehow be biased towards the status quo via ideological forces, or even the political system itself (Rossi, 2023: 3–4). If so, we might feel or seem ‘content’ when we should not be, either because we think we are in the best of all possible worlds or because we think no other one is possible. In this case, both ambition and imagination alike, for whatever reason, would be stunted by forces we do not appreciate, and we would never take to the streets, not because we are satisfied, but because we are resigned.

Now, again this too might be a problem for mentalism as well as normative behaviourism, given our thoughts could be as corrupted as our behaviour, but it seems a bigger one for the latter, because it seems impossible to argue our way out of the ideological hole described without some kind of mentalist project. Can we then find a more empirical answer? Once more, the idea of experimentation is crucial, and in particular the finding in several parts of the best literature on social-liberal-democracies (SLDs) that a key feature of their success is their relative ability to experimentally evolve (Floyd, 2017a: 177–245). They can, for example, move from smaller to larger franchises without collapsing. They can expand welfare without revolution. They can adopt liberal norms on art, religion and sexuality without war. They can even join the EU (or leave it) while testing out UBI in all its variations. Of course, climate change stalks them while pandemics rattle them, but even so, shaking isn’t breaking. Notice how they once fixed the ozone layer. Notice how they still develop and deploy vaccines better than their rivals. As with skyscrapers on fault-lines, there is a lot to be said for states, in a faulty world, that sway rather than shatter. We should then have hope without arrogance. Liberalism has changed and will continue to change. Its freedom might be boring, but it is no iron cage.

I do though, despite all this, accept a part of Rossi’s worry here, which is that, even if SLDs don’t put us in boxes, they might still limit our thinking outside of them. We must then push experimentation as far as we can, especially once freed of that truly ‘restrictive’ idea that principles have first to be justified from the armchair, and including experiments way beyond those described so far, given that normative behaviourism itself might have to evolve. Maybe, for example, we should study online behaviour in ever richer virtual worlds, or even build cyber polities ourselves, before sitting back and watching the results? Maybe we should look at migration patterns, noting carefully which political factors push and pull people around the world? Maybe we should even revisit Durkheim and track suicide rates, including in response to – if indeed they do respond to – particular systems of digital and migratory regulation, as well as our wider political and economic system? Each of these ideas and more, I would suggest, is worth exploring, both in terms of the quality of the causal patterns available and in terms of the significance of the actions involved, as compared to the ‘special’ cases, as I have described them so far, of insurrection and crime.

We might even need to change, at least sometimes, the way that we talk, by no longer ‘restricting’ ourselves purely to complicated arguments in books and seminars, and instead trying a more ‘public’ form of political philosophy in which we experiment with, for example, fiction and humour, much as we once tried dialogues and meditations, seeking ‘engagement’ and even ‘impact’ in turn (Floyd, 2019, 2022). This would mean spreading ideas and opening minds, pushing imagination and curiosity, but also feeding a broader experimental culture, including as regards the core variables of normative behaviourism, just so long as we remember that the acid test of it all remains how people find such things once they have had to live with them. Or, put differently, ‘yes please’ to a thousand hypotheses, but still ‘no!’ to confusing them with the results.

All of which then takes us right back to Modood and Baderin’s distinctive worries about the ‘minimalism’ of the theory so again the key points needs emphasising. Where you are vague because the data are vague, experiment. Where you are stuck without ideas, reach out and broaden your audience and collaborators. Where you think ideology might have a hold of you, look comparatively across space and time, and see what bold new things either have been done or could be done, either physically or virtually. Think fast but test slow. Start small but build up. Admit that progress can be painfully incremental, but take solace in avoiding the greater pain of war and those collapses that flow easily from over-confidence, whether from mentalists on the one hand, or Crusaders, Jacobins and Soviets on the other.

Conclusion

Although it is easy to think like a mentalist, and say that normative behaviourism is unwittingly utilitarian, dependent on bad data, and excessively conservative, it is also easy to think like a normative behaviourist, and say that institutions express ideas, that their measure is how we react to them in practice, and that we should keep trying new ones in order to get the best reactions possible. Deeds not words, said the suffragettes, knowing that talk and thought are cheap and chaotic. Don’t explain your philosophy, embody it, said the stoics, knowing that actions and institutions are the only manifestations of our commitments that really count. Radicals, sure, but also realists (Floyd, 2022), and thus bearers of a spirit I have tried to ‘embody’ myself in recent years when connecting normative behaviourism to international politics (Floyd, 2016b), continental philosophy (Floyd, 2016b) and even ideal theory (Floyd, 2020), knowing there are always new combinations and conundrums to explore. What if, for example, our wonderful natural preferences ruin our wonderful natural environment? What if artificial intelligence leads to artificial behaviour? What if civic education could alter either challenge? (Floyd, 2022, 2023b) All these notions and more need careful testing. Hunting new virtues without old vices. Playing with principles without trying to prove them in advance. Considering new values without ever thinking consideration enough.