Moral Empiricism

Is morality prewired into our minds? The idea that morality is built into us is an old one in philosophy, and it has seen a resurgence of late. Indeed, the prevailing systematic account of how we acquire complex moral representations is a nativist view inspired by arguments in Chomskyan linguistics. If the statistical learning accounts I’ve defended in Part II of this book are right, we have the beginnings of an empiricist account of important aspects of human morality. This chapter offers a sustained defense of a moral empiricist view in the face of the Chomskyan challenge.

Rational Rules and Normative Propriety

To what extent are our moral beliefs rational? This chapter argues that the statistical learning accounts offered in Part II provide reason to think that many central beliefs about socio-moral rules are acquired through rational processes and that this confers on the beliefs an essential kind of rational credential. The beliefs are doxastically rational. However, even if the beliefs about the rules are doxastically rational, the rules themselves might be counterproductive for achieving our ends. Another way to evaluate the propriety of a rule is in terms of ecological rationality, that is, how well the rule works given our minds and environments. This chapter articulates a number of factors that contribute to the ecological rationality of a rule, and argues that act-based rules are plausibly are more ecologically rational than consequence-based rules.

Priors

People seem to expect that rules will be act-based rather than consequence-based. For instance, when people learn a new rule, given minimal evidence, they tend to think that the rule prohibits people from producing a certain consequence, rather than that the rule dictates that such a consequence should be minimized. One explanation is that people have an innate bias to think rules are act-based. This chapter suggests an alternative empiricist proposal. The expectation that rules will be act-based might be explained as a prior that is itself the result of earlier learning. In particular, given that most rules that people have learned are act-based rules, the prior for act-based rules might be acquired through a process of overhypothesis construction.

Status

Moral judgments are often regarded as universally true, whereas judgments of taste are taken to be only true relative to some group or individual. How could such meta-evaluative assessments be acquired? This chapter argues that people use consensus information to arrive at such assessments, and that it is rational to do so. Statistical inference mandates a trade-off between the extent to which a hypothesis fits the data, and the extent to which the hypothesis is flexible in its ability to fit a wider range of data. If almost everyone agrees in their judgments, this provides some reason to endorse a universalist hypothesis, according to which there is a single fact that the majority is tracking. So if almost everyone thinks that a certain action is wrong, the high consensus provides some evidence that it’s a universal truth that this action is wrong. The inference that it’s a universal truth that an action is wrong can also ground the judgment that the action is wrong in a way that is independent of authority. Thus, this might also provide an explanation for the acquisition of the moral/conventional distinction.

Rationality and Morality

To what extent is morality based on reason? To answer this question, we need to clarify which aspect of morality is under investigation, and which notion of reason is in play. Recent work in moral psychology has attempted to debunk central aspects of moral judgment and metaethical judgment. However, rational processes might play a vital role in the acquisition of moral systems. This chapter sets out the basic idea of processes and suggests that one kind of process—statistical learning—is especially significant for moral learning. Statistical learning processes are both empiricist, in that they are domain-general processes, and they are rational, in that they conform to the rules of probability theory. Thus they are poised to provide alternatives to both nativist and sentimentalist accounts of our moral psychology.

Is It Rational to Be Moral?

Why should all rational agents be moral? This is one ancient and challenging question about moral motivation. But there is another perhaps more tractable question about moral motivation. Why as a matter of fact are most of us motivated by moral considerations? What is it about the kind of creature I am that inclines me to be moral? Moral judgments (e.g. that it’s right to give to a certain charity) seem to be directly motivating. This chapter argues that even non-moral normative judgments often are directly motivating. A primary form of rule representation automatically carries with it motivation force.

Rationalism, Universalism, and Relativism

People seem to regard some norms (e.g., about the wrongness of armed robbery) as universally true, and other norms (e.g., about which side of the road to drive on) as true relativized to some context or group. This chapter considers whether such meta-evaluative beliefs are rational. There are reasons to doubt that the belief in universalism about norms is evidentially rational. Nonetheless, universalism seems to be a default setting in normative cognition. That is, when we acquire norms, we tend to presuppose that they hold universally. This chapter argues that even though the default assumption of universalism might not be evidentially rational, it is ecologically rational to have a bias in favor of universalism. However, people are not completely locked into universalism. Under conditions of low consensus regarding some norm, people do move away from universalism and adopt some form of relativism, at least at a reflective level. And this is plausibly an evidentially rational response to low consensus.

Scope

Commonsense, as well as experimental psychology, indicates that there are subtle distinctions in the normative domain. Many people, both adults and children, think that it’s worse to produce a bad consequence than to allow it and that it’s worse to produce a bad consequence with intent than to produce it with mere foreknowledge. People also often think that it’s forbidden to treat people of their own community in a certain way, but not that it’s forbidden to treat people in other communities in that way. It has been unclear exactly how these distinctions arise in ordinary moral thought. This chapter draws on the “size principle,” which is implicated in word learning, to explain how children would use scant and equivocal evidence to acquire these aspects of moral systems.

Closure

When people learn normative systems, they do so based on limited evidence. Many of the possible actions that are available to an agent have never been explicitly permitted or prohibited. But people will often need to figure out whether those unspecified actions are permitted or prohibited. How does a learner resolve this incompleteness? It seems that at least for many people in many contexts, there is an assumption that if an action-type is not expressly forbidden, then acts of that type are permitted. This “closure principle” is one of Liberty. But how might such a principle be acquired? This chapter argues that the statistical technique of pedagogical sampling provides an answer. If one is taught a rule system via a set of prohibitions, this provides reason to think that the set of actions in the domain that are not mentioned in the prohibitions are permitted.

The Wrong and the Bad

The fundamental element of a cognitive account of moral judgment will be some form of representation. Two kinds of value representations need to be distinguished: value representations and rule representations. Value representations register valence for particular courses of action. For instance an organism might represent touching a certain wire as a bad thing. Such representations feed directly into behavior. Rule representations involve more than registering a bad value. They can involve complex representations composed of abstract concepts like impermissible, harm, and knowledge. This chapter argues that value representations cannot provide an adequate account of moral judgment, and that we must invoke rule representations to explain key features of moral judgment. In particular, it is difficult to capture the distinctive nature and specificity of wrongness judgments without adverting to structured rules.