Everyone who makes a choice has to avoid unwanted arbitrariness as much as possible. This is a fundamental principle in ethics, in the sense that it applies to all moral choices. This anti-arbitrariness principle is the reason why dictatorship and discrimination are morally wrong.
Definition of unwanted arbitrariness
Unwantedness means being incompatible with someone’s largest consistent set of that person’s strongest subjective preferences. A subjective preference is a conscious value judgments or evaluation that has a subjective strength (to be distinguished from e.g. a mere unconscious behavioral disposition). For example, being told a lie is incompatible with a preference for knowing the truth. If something is not logically compatible with the largest consistent set of your strongest preferences, it cannot be consistently wanted by you. Everything that is compatible, can be consistently wanted. Saying that you cannot consistently want something is the same as saying that you can reasonably object it.
Arbitrariness means selecting an element (or subset) of a set without using a selection rule. A selection rule is a rule that logically determines the selection. It is an if-then statement that consists of a set of conditions. For example: “If element X has conditions A and B or not C, then select X.” If the question “Why selecting element X instead of element Y?” has no answer that refers to a selection rule (for example if the only answer is “Therefore!”), then selecting X is arbitrary.
Combining the above definitions of unwantedness and arbitrariness, we can define unwanted arbitrariness as making a choice without following a selection rule, whereby the consequences of that choice cannot be consistently wanted by at least one person. Here, a choice can be defined as a conscious decision. Making a choice means consciously selecting an element from a choice set, the set of eligible options.
Formulations of the anti-arbitrariness principle
The anti-arbitrariness principle states that:
When making a choice, we have to avoid unwanted arbitrariness as much as possible.
To avoid arbitrary exclusion of choices, this principle applies to all possible choices, including very specific actions (“Sit at seat 5 on bus 42 at 1 pm Friday”), to more general choices (“Use public transport”), to justifications (“Take a seat when the seat is empty and you paid for a ticket”), to higher level moral choices (“Choose the action allowed by a contractualist ethic”), to even very basic choices of premises and logical deduction rules used in justifications (“Use deontic logic to determine the validity of an argument”). For practical reasons, we do not have to consider impossible choices (“Avoid unavoidable unwanted arbitrariness”).
To avoid arbitrary exclusion of people who have to respect this principle, the principle applies to everyone who is able to make choices based on selection rules. When someone cannot make a choice, that is an exception, but not an arbitrary exception because it is justified using an ‘ought implies can’ rule: “If you cannot do something, you have no obligation to do it.”
But this anti-arbitrariness principle does not yet say what happens if we don’t avoid unwanted arbitrariness. Also, the ‘as much as possible’ hints at the possibility that sometimes unwanted arbitrariness may not be avoidable. Therefore, we can give a more exact formulation of the anti-arbitrariness principle, in a strong and a weak version.
Anti-arbitrariness principle, first formulation, strong version. If you do not avoid avoidable unwanted arbitrariness when making a choice, you are not allowed to make that choice.
Anti-arbitrariness principle, first formulation, weak version. If you cannot avoid unwanted arbitrariness when making a choice, you are allowed to make that choice but other people may make other choices from the same choice set (i.e. you have to tolerate that other people make other choices).
We can give another formulation of the anti-arbitrariness principle:
For every choice you make, you have to be able to give a justification rule such that you and everyone can consistently want that everyone follows that rule in all possible (including hypothetical) situations (i.e. you and everyone can accept the consequences of a universal compliance by everyone of the justification rule).
A justification rule for (im)permissibility of a choice should be used in a logical deduction. Therefore, it is basically an if-then statement that consists of a set of conditions: “If conditions C apply, then it is permissible to choose X.”
As with the first formulation, this second formulation of the anti-arbitrariness principle also comes in two versions.
Anti-arbitrariness principle, second formulation, strong version. If, when making a choice, you cannot give a justification rule of which you would accept universal compliance, then you are not allowed to make that choice nor follow that rule.
Anti-arbitrariness principle, second formulation, weak version. If, when making a choice, you cannot give a justification rule of which everyone would accept universal compliance, then you must accept or tolerate that other people make other choices from the same choice set and follow other justification rules for making those choices.
There are many similarities between the two formulations of the anti-arbitrariness principle, such that they can be said to be roughly equivalent.
First, there is a correspondence between the selection rule and the justification rule. The first formulation works with a selection rule to avoid arbitrariness. In the second formulation, arbitrariness is avoided by the justification rule and by the idea that if you may follow that rule in a specific situation, then everyone may follow that rule in all possible situations. Suppose that the “everyone” and “all possible situations” were no requirements. Replacing them by “some people” and “some situations” would introduce arbitrariness, because arbitrary subsets of the sets of all people and all situations can be chosen.
Second, both formulations look for what can be consistently wanted. The condition “everyone can consistently want that everyone follows that rule in all possible situations” is the opposite of unwanted arbitrariness. Suppose you choose option A arbitrarily and person Y is in a position P in which s/he cannot consistently want that arbitrary choice. If we consider everyone and all possible situations, this includes the situation where person Y chooses A and you are in the same position P that Y had, in which case you cannot consistently want A.
A third similarity between the two formulations, is that they both come in a weak and a strong version. Unwanted arbitrariness may not always be avoidable, because there may always be someone who cannot consistently want a choice that cannot be based on a selection rule. Similarly, it may not be possible to find a justification rule of which everyone can accept universal compliance. In these cases, people must tolerate that other people make other choices.
A final similarity between the two formulations of the anti-arbitrariness principle, is that they both apply to all possible choices, including the choice of selection and justification rules (in particular the choice of conditions in those rules). That means a selection meta-rule should be given to select the selection rule from the set of all selection rules. Similarly, a justification meta-rule should be given to that justifies the chosen conditions in a justification rule. With the application to all possible choices and the resulting necessary inclusion of such meta-rules (and higher order meta-meta-rules), the anti-arbitrariness principle becomes perhaps the most fundamental principle in ethics.
An example might give some clarification. Consider the situation of taking a seat on the bus. If you choose to take a seat, the rule could be: “If you are white, you may take the seat,” or “If you have permission by person X, you may take the seat.” But the choice of these conditions is arbitrary (they refer to an arbitrary skin color or person). A better rule would be: “If the seat is empty and you have permission by the people who have a special relationship with the seat, you may take the seat.” We have to specify what counts as a special relationship. This can again be done by considering relationships of which everyone can consistently want that they are part of the conditions in the justification rule. Examples of such a special relationship could be ‘being the owner of the bus’ or ‘having reserved the seat’. Having permission could mean ‘having paid for a ticket’.
As the anti-arbitrariness principle deals with choices and rules, we are confronted with two important questions. Who decides or chooses the choices and rules? And who is affected by the choices and rules? These two questions relate to the dual problems of dictatorship and discrimination. The next two sections discuss how the anti-arbitrariness principle implies the non-dictatorship and non-discrimination principles.
Implication 1: non-dictatorship
The non-dictatorship principle says that no-one should have the unconditional power to always unilaterally make decisions that affect other people. A dictatorship clearly violates the anti-arbitrariness principle, because the choice for the dictator is arbitrary (as the dictator’s power is unconditional, no rule was followed to grant that power), and unwanted (when there are affected people who do not want the decisions made by the dictator).
The non-dictatorship principle can also be applied to moral theories. These theories are logical systems of ethical principles that represent moral intuitions or values. There are different moral theories, such as a deontological rights ethic, a consequentialist utilitarian welfare ethic, a libertarian ethic or pluralist ethics that combine several ethical principles. But which theory should we choose? The anti-arbitrariness principle sets strong constraints on a moral theory. The theory should be coherent in the sense that it should be constructed following some rules, such as:
- One should not arbitrarily limit the ethical principles to an arbitrary group of objects, beings or individuals.
- One should not arbitrarily give weaker moral intuitions stronger priority. One should not arbitrarily change or exclude basic moral judgments.
- One should not arbitrarily allow inconsistencies and gaps in the ethical system.
- One should not arbitrarily introduce ambiguous or vague principles that one can interpret and apply arbitrarily in concrete situations.
- One should not arbitrarily add artificial, complex, ad hoc constructions and exceptions to save the moral theory from counterintuitive implications.
These construction rules for a coherent theory can be consistently wanted. If for example I allow inconsistencies, gaps, ambiguities or arbitrary exceptions in my theory, then I have to accept that your moral theory also contains such things. With such an incoherent theory, you can easily justify choices that I cannot consistently want.
To avoid dictatorship, everyone is allowed to construct a coherent moral theory that best fits one’s moral intuitions and values. Incoherent theories are impermissible. But there are many possible coherent moral theories. We do not have a rule that determines which of those coherent theories is the best. If we are against unwanted arbitrariness, we have to recognize that every equally coherent moral theory is equally valid. I cannot say that my coherent theory, based on my moral intuitions, is better than yours if both our theories are equally coherent. I prefer my theory, but I cannot impose my theory upon you, because what would make me so special that I would be allowed to do that? And the same goes for you and everyone else. It would be an unwanted kind of arbitrariness if I claim that my moral theory is special without good reason.
So picking one of the coherent moral theories always involves unavoidable arbitrariness. The non-dictatorship principle says that we should democratically choose which moral theory to apply. And if you follow a coherent moral theory without being able to give a justification rule that selects that theory, you should tolerate that other people follow other coherent moral theories. We should be tolerant towards all other coherent ethical systems, no matter how much they go against our own moral intuitions.
A choice for an incoherent system, on the other hand, does not have to be condoned, because you can give a justification rule “If the theory is incoherent, then it is impermissible to choose it,” and everyone can consistently want that everyone follows this rule. If you choose to follow an incoherent theory, I am allowed to reject that theory and impose my theory on you, and you are not able to complain. You are not able to give reasonable or justified counterarguments against the imposition of my ethical principles, because by following your incoherent theory, you are acknowledging that unwanted arbitrariness and hence arbitrary exclusion are allowed. That means it is also permissible to arbitrarily exclude your moral theory and ignore your moral views and ethical principles. You can only give a valid complaint or argument if you accept the anti-arbitrariness principle. Without that principle, any critique becomes invalid and complaints become impossible.
As the ethical systems of e.g. racists, rapists or religious fundamentalists contain inconsistencies, avoidable arbitrariness, unscientific beliefs and vague principle, they can easily be rejected. If your ethical system is more coherent than theirs, then you can rightfully say that your ethical system is better than theirs and then you may oppose their incoherent systems.
The prohibition of incoherent theories allows us to avoid an extreme form of moral relativism that says that all moral theories, including incoherent ones, are equally valid. This extreme relativism implies that everything would be permissible, and we cannot consistently want that. The non-dictatorial claim that coherent moral theories are equally valid is a kind of weak moral relativism, which is a consequence of the anti-arbitrariness principle.
How do we deal with that plentitude of coherent ethical systems that are equally valid? Everyone (who is able to do so) constructs their own coherent ethical systems, and we can aim for a consensus or democratic compromise between everyone’s system by using a democratic procedure. In a democracy, everyone has one vote, or everyone’s vote is equally important, because we cannot say that one vote (one coherent theory) is better than someone else’s. But those who cannot provide a coherent moral theory that does not contain unwanted arbitrariness, lose their vote. Or in other words: in this moral democracy it is not allowed to vote for parties who have incoherent moral theories, such as racist parties. Those parties cannot participate in elections.
Implication 2: non-discrimination
Arbitrary discrimination of individual (or group) A relative to B is a systematically different treatment of A and B, whereby
- B gets more advantages than A,
- A has a lower moral status than B (e.g. A has less intrinsic value or weaker rights than B) in the sense that one would not tolerate swapping positions (treating A as B and B as A), and
- there is no justification or the justification of the difference in treatment refers to morally irrelevant criteria (properties that are not acceptable motives to treat A and B differently in the concerned situation), whereas A and B both meet the same morally relevant criteria to treat and value them more equally.
The first two conditions reflect unwantedness. The discriminated person A does not want the disadvantage, but also the person who discriminates does not want swapping positions of A and B. The third condition reflects arbitrariness, i.e. the lack of a justifying rule. Discrimination is based on arbitrariness, and this arbitrariness is avoidable and unwanted, because the discriminated people do not want their negative treatment, their arbitrary exclusion from the moral community.
The anti-arbitrariness principle specifies what counts as morally irrelevant criteria. A criterion or property is morally irrelevant when it is
- arbitrary (there is no non-circular rule that selects the property out of a multitude of similar kinds of properties),
- contingent (not present in all possible worlds) or not intrinsic (it does not refer solely to the individual possessing the property), or
- ambiguous (the property is inherently difficult to detect, define or delimit or it is non-empirical, which means there are no scientific criteria and methods – not even in principle – to clearly see whether the property is present).
Of course if the property is arbitrary, it violates the anti-arbitrariness principle. If the property is contingent or not intrinsic, there is also an arbitrariness in the sense that the property depends on contingent (accidental) circumstances that could have been different in another possible world. Someone’s rights should not depend on contingent properties, because we can ask the question: why should that individual get rights in this possible world but not in that? And if the property is difficult to detect or delimit, there is the risk that one arbitrarily assigns the property as one pleases.
Examples of criteria that are morally irrelevant because of the above reasons, are: physical characteristics and appearances (e.g. skin color, behavior, gender), genetic properties (e.g. race, ethnicity, genetic kinship), preferences (e.g. sexual, political), supernatural properties (e.g. having a soul) or belonging to an arbitrary group.
As a concrete example of the non-discrimination principle, consider the choice of moral community: the subset of all entities in the universe that have moral status (in the sense of e.g. having moral rights). Consider only living beings. According to the biological classification, we can classify living beings in a vertical taxonomic hierarchy, with the taxonomic rank ‘life’ at the top, followed by ranks such as ‘domains’ (e.g. eukaryotes), ‘classes’ (e.g. mammals), ‘orders’ (e.g. primates), and finally the taxonomic rank ‘populations’ (races, subspecies) at the bottom. A white supremacist first chooses the lowest level in this hierarchy (the populations or ethnic groups), and then picks a subset at this level (the ethnic group of whites). Similarly, a speciesist first selects the level of the species, and then selects a specific species (e.g. Homo sapiens) as the moral community. If no selection rules were followed, these two choices involve respectively vertical and horizontal arbitrariness. We can first ask the non-trivial question: “Why choosing a species and not e.g. a biological order or a phylum?” And at the level of the species, we can ask: “Why choosing Homo sapiens (humans) and not e.g. Sus scrofa (pigs)?” One could answer: “Because most humans have the capacity for moral thought”, but it is possible that this answer also applies to some levels up or down in the hierarchy. If for example there are less than 14 billion primates alive, containing more than 7 billion humans with the capacity for moral thought, then the majority of primates have this capacity. Hence, one could equally well first select the level of orders and then the order of primates. By selecting a biological group as a moral community, it is not easy to avoid arbitrariness.
The definition of discrimination means you can avoid discrimination in three ways: either treating A and B equally, tolerating swapping their positions or justifying the preferential treatment using non-arbitrary criteria.
If you tolerate swapping the positions of A and B, you give them equal moral value. This implies that some kinds of partiality are not (yet) discriminatory. Consider a burning house dilemma where you can either save Alice or Bob from the flames. Suppose you want to save Bob first because he is your child, whereas Alice is a child from another country, with another skin color. Non-discrimination does not imply that you should flip a coin and give each child an equal 50% survival probability. You are not a racist or sexist (at least not necessarily) if you want to save Bob, as long as you do not condemn someone else who wants to save Alice. If you criticize someone who saved Alice using arbitrary criteria such as skin color or gender, then you discriminate and then it becomes racism or sexism.
Considering the above, we can formulate the following ethical principle of tolerated partiality: when helping others, you are allowed to be partial in favor of one individual or group (e.g. your own child), as long as you tolerate someone else’s choice to help the other party (e.g. another child). In this sense, saving your child is not inconsistent with the claim that all children have an equal moral value. Two children can have a different personal values for you, but they inherit an equal moral value when a tolerated symmetry (swapping their positions) is satisfied. Having a stronger empathic connection for one individual or having a stronger inclination to save one individual instead of the other, and acting on those feelings, is not necessarily discrimination.
This principle of tolerated partiality can be derived from the unwanted arbitrariness principle: everyone should tolerate your preference for saving the people you hold dear, even if your selection of those people is arbitrary (e.g. from my perspective), because everyone can consistently want to be able to save the people they hold dear.
What if you do not tolerate swapping the positions of Alice and Bob? Suppose Bob is your child and Alice is the name of my car. You would not tolerate me saving the car. The definition of arbitrary discrimination implies that to avoid discrimination, there must be a valid reason or justification, based on non-arbitrary criteria, why one entity (the child) is more important or valuable than the other (the car). In this example you can easily give a valid reason: the child has preferences to be rescued, to keep on living and to avoid the pain from the flames, whereas the car does not care at all about being burned or rescued.
Similarly, suppose you give a piece of chocolate to Bob, a child, instead of Alice, a dog. You have a non-arbitrary justification: chocolate is unhealthy for dogs. Being able to safely eat chocolate is a non-arbitrary criterion, because both the dog and the child prefer safe food. Non-discrimination does not say that we must treat everyone the same and give everyone the same food.
However, some reasons are invalid. For example, the reason to save Bob instead of Alice because Bob belongs to a certain social group or believes in a certain God. Those invalid reasons refer to arbitrary criteria, such as skin color, religious beliefs or group membership. A white supremacist might help Bob instead of Alice based on their skin colors, but what does skin color have to do with a preference to help or being helped? Skin color is but one bodily characteristic, and it is arbitrary to claim that this particular characteristic relates to subjective preferences.