Trusting in reason. Martin Hollis and the roots of plural rationality

There are purely predictive expectations and normative expectations. I expect the sun to rise tomorrow, my old alarm clock to go off, and a person of habit to repeat a certain behaviour. In these cases, trust is akin to a prediction. But when I lend a book, entrust a child to a teacher, sign a contract, enter into a care relationship, or take part in a democratic process, I do not merely predict that the other person will do something. I expect the other person to recognise the moral obligation to do it. Interpersonal trust always contains a normative element. It is not just about what the other person is likely to do, but what the other person knows they are obliged to do, because I expect it of them.

Rational choice theory struggles to account for this factor. It can explain why it pays to appear trustworthy, why it is useful to build a good reputation, why sanctions encourage compliance with the rules, and why, in repeated interactions, it is prudent not to betray others straight away. All this is true and important. But it remains insufficient. Because in all these cases, the constraint continues to depend on a calculation. I behave well as long as it suits me, as long as I fear punishment, as long as I need a reputation, as long as the future weighs heavily enough on the present. But as soon as conditions change, so does the strength of the constraint. The promise becomes a contract with an invisible clause: subject to better opportunities. Martin Hollis calls this logic ‘prudence’. And he does not dismiss it at all. Prudence is a virtue of practical intelligence, one of the virtues celebrated by the Enlightenment and commercial society. It looks ahead, takes consequences into account, avoids blind immediacy, and calculates the reactions of others. Without prudence, social life would be chaotic. Yet prudence becomes dangerous when it claims to be the whole of reason. When every constraint is reduced to convenience, every promise to strategy and every relationship to an exchange, then relationships cease to be rational. They simply become more suspicious. Prudence can manage trust, but it cannot fully generate it. It can exploit it, discipline it, protect it with incentives and sanctions. But if left to its own devices, it ends up consuming it.

The problem, then, is not that human beings are too rational. It is that they can be rational in a rather limited way. A rationality reduced to individual calculation is not false because individuals never calculate. It is false because it claims to explain everything that matters about social action. But many things that matter do not arise from isolated calculation. They arise from mutual recognition, from belonging to shared practices, from the ability to uphold a commitment even when the opportunity to break it arises. They arise, to put it in Nietzsche’s terms, from the ability to be ‘animals capable of making promises’. It is here that Hollis’s work paves the way for a line of research that would find a particularly original formulation in John Roemer. In *How We Cooperate* (Yale University Press, 2019), Roemer distinguishes between a John Nash-style rationality and a Kantian rationality. In the first case, each person chooses the best response given the choices of others. Others are parts of the strategic environment. They represent constraints, threats, opportunities and information. In the second case, however, each person asks themselves what behaviour would make sense if everyone involved in the same situation were to adopt it. The categorical imperative and the principle of universalizability. What should we do together, and what part falls to me? Roemer formalises what Hollis had laid out philosophically.

To understand cooperation, it is not enough simply to attribute more generous preferences to individuals. It is not enough to say that, alongside selfishness, there exist sympathy, altruism and benevolence. Of course they exist. But cooperation is not merely a matter of preferences. It is a matter of the form of reasoning. In certain situations, an individual does not make the right choice if they think of themselves as an isolated individual reacting to others. They make the right choice only if they see themselves as part of a joint action. This is a crucial difference. Two people carrying a piece of furniture together do not each choose the best movement based on the other’s anticipated movement. They must understand what they are doing together. What the shared aim is. Two musicians playing in a quartet do not separately maximise their own sound. They participate in a joint performance. A medical team dealing with an emergency is not a collection of individuals devising independent strategies. It is a collective entity organised into different roles. In all these cases, the ‘I’ does not disappear. But it changes position. It is no longer the absolute starting point of the action. It is part of an operational ‘we’.

The perspective outlined by Martin Hollis helps us to see why this transition is so difficult for the modern economy. Methodological individualism has played an extremely powerful role. It has freed social analysis from vague collective entities, from dangerous forms of organicism, and from communities imagined as entities superior to individuals. But, in doing so, it has often impoverished the description of collective action. It has conceived of the ‘we’ as a mere sum of ‘I’s’. It has treated the collective result as the outcome of separate choices. It has instilled the suspicion that any reference to shared goals conceals confusion, rhetoric or authoritarianism. Of course, the ‘we’ can become oppressive; it can erase differences, impose identities and even stifle dissent. But a society that can no longer conceive of any ‘we’ ends up being unable to conceive even of the goods that keep it alive.

There are, in fact, certain goods that cannot be produced by an individual alone. Trust is one of them. Knowledge is another. Public health, security, democratic quality, environmental sustainability, social peace and the credibility of institutions are too. These are goods we can enjoy individually, but which must be produced collectively. A just society must therefore do more than simply distribute opportunities. It must also safeguard the conditions for cooperation. It must make it reasonable to trust. It must prevent those who follow the rules from appearing systematically naive, those who contribute from feeling exploited, and those who keep a promise from being considered less intelligent than those who find a way to evade them. No institution can survive on private virtue alone. But no institution can survive on incentives and sanctions alone either. If everything must be monitored, if every cooperative behaviour must be bought or threatened, then the moral and administrative cost of living together becomes extremely high.

This point is particularly relevant to contemporary democracies. Many of them suffer not only from inequality or a scarcity of resources. They suffer from the erosion of the cooperative foundations of public life. Citizens continue to demand services, rights, protections and opportunities. But they often doubt that others are doing their bit. They doubt that institutions are impartial. They doubt that the sacrifices required are shared fairly. They doubt that loyalty is recognised. When this happens, trust does not disappear because of some psychological flaw in citizens. It disappears because it becomes less reasonable.

The crisis of trust, then, cannot be resolved with moralistic appeals. It is not enough to ask people to be more responsible, more generous, more civic-minded. We must build institutions that make the link between individual contribution and the common good visible. We must reduce the inequalities that turn the cooperation of the most vulnerable into an opportunity for the strongest to gain an advantage. We must demonstrate that the rules apply even to those who have the means to circumvent them. We must ensure that public promises are not perceived as empty rhetoric. Trust does not arise from the rhetoric of trust. It arises from repeated experiences of reliability. This is why Martin Hollis’s contribution is so important for a theory of justice as well, because justice is not merely about the distribution of goods, income, opportunities and positions. It is also about the quality of the reasons a society makes available to its members. A society can be formally free and yet continually foster mistrust. It can proclaim equality and yet make opportunism the norm. It can celebrate responsibility and yet reward exploitation. It can invoke merit and yet create environments in which individuals learn to view others primarily as obstacles, competitors or threats.

That would be a symmetrical and equally serious mistake. His ambition is more demanding. It is to show that reason can only live up to that trust if it ceases to identify itself with individual prudence. Reason must be able to recognise obligations, promises, reciprocity and shared goals. It must be able to say not only ‘if you want this, do that’, but also ‘this is a rule we can recognise together’. It must be able to speak not only in the first person singular, but also in the first person plural.

John Roemer takes up this insight and applies it to the economics of cooperation. His idea of ‘Kantian optimisation’ is not so much an uplifting call to be better people. It is a proposal that explains why, in certain contexts, the rationality of action depends on the possibility of universalising one’s strategy within a group of similar individuals. It is not a matter of sacrificing the self to the collective. It is about understanding that, for certain goods, the ‘I’ can truly pursue its own good only within a rule that others can also adopt. Cooperation is not the opposite of self-interest. It is a more demanding form of understanding self-interest, when self-interest depends on goods that no one produces alone. In the end, the question we started with returns in a more precise form. How can human beings trust one another? Not because they are always good. Not because they are always in control. Not because calculation always drives them to cooperate. They can trust one another when they live within institutions and practices that make it reasonable to recognise common constraints. They can make promises when a given word is not treated as a sentimental remnant, but as a structure of freedom. They can cooperate when reason is not severed from reciprocity.

Perhaps this is the most relevant lesson that leads us from Martin Hollis to John Roemer. It is not a question of choosing between reason and trust. It is a question of saving reason from its most impoverished form. For a society made up solely of prudent individuals may survive for a time, and perhaps even appear to function. But sooner or later it will discover that it has consumed the very asset that made it possible. It will discover that no rule, no contract, no institution can hold up for long if everyone learns to ask themselves only when it is convenient to respect it. And then it will realise, perhaps too late, that some of the most rational things we can do cannot be expressed by beginning with ‘I’.