Many believe that moral education is a good candidate for a mechanism that causes moral progress. For example, Morrow (2020, p. 503) argues that “plausible theories of moral progress must provide an account of the pathways and processes by which progress occurs” and then cites “moral education, achieved through exposure to compelling texts, images, or oral narratives” as an example of such a process. Hermann (2017, p. 46) agrees: “the education of capacities and virtues […] can contribute to the achievement of moral progress.” So do Schinkel and de Ruyter (2017, p. 121): “that moral progress is possible is a foundational assumption of moral education.” Indeed, Gilead (2009) shows the idea that “education should facilitate constant moral progress” (p. 94) already enjoyed a high degree of prominence in eighteenth-century educational thought, and argues, after tracing the idea’s influence to the current day, that “generating moral progress [f]or many thinkers concerned with moral education […] is still a viable aim” (p. 105).

At first glance, this makes sense. A major goal of moral education is the improvement of the moral beliefs, values and behaviors of young people (e.g., Halstead, 2010; White, 2016). All of these improvements also constitute important types of moral progress (e.g., Buchanan and Powell, 2018; Sauer et al., 2021). Moreover, moral education is a form of social learning, and there are good reasons to think that social learning processes shape episodes of progressive moral change (see below).

Despite this, we will argue that it is doubtful whether moral education often is a cause of moral progress. Instead, we will argue that in many circumstances, the main effect of moral education is not to bring about moral change (progressive, or otherwise), but to provide stability or continuity. In addition, we will argue that when the conditions are right for moral education to cause large-scale moral change, it is far from clear that this change will always, or even most of the time be progressive moral change.

This is not to say that moral education is irrelevant for the achievement of moral progress. In particular, if we are right, then moral education may often help stabilize and sustain past moral gains, and so enable future generations to build upon them further. We are not doubting that moral education can be an important enabling condition of moral progress. Instead, moral education does not often cause moral progress—or so we will argue.

Our paper has five sections. We start with conceptual clarifications (Sect. 1). Section 2 makes the case that social learning is an important mechanism for moral change. Section 3 argues that despite this, moral education is often (but not always) more of a force for moral stability than for moral change. Section 4 then sketches several scenarios in which moral education could contribute to moral change, and asks whether we should expect these changes to be progressive. We finish with a few concluding thoughts about philosophical implications (Sect. 5).

1 Definitions

What is moral education? Here, we define moral education as any activity explicitly aimed at or geared towards improving the moral beliefs, values or behaviors of children and adolescents (e.g., Halstead, 2010; Nucci et al., 2014; White, 2016). All sorts of activities can lead to moral improvement, but in order for one to count as moral education, moral improvement has to be the point. The two groups that are most involved in moral education so understood are parents and teachers.

In this paper, we are interested in moral education as a form of social moral learning. For our purposes, social learning is learning that results in the social transmission of information, which occurs when one individual who has acquired a piece of information (or a behavior), expresses that information (for example, says the information out loud, or behaves in a way that depends on the information), and this exerts a lasting causal influence on the rate at which another individual acquires the same information (see, Hoppitt & Laland, 2013, p. 4).Footnote 1Social learning contrasts with asocial learning, which does not involve social transmission (for example, learning a piece of information by trial and error). Social learning is social moral learning when the information that is transmitted is moral information (for example, information about moral rules, values or expected behaviors; cf., Cushman et al., 2017, p. 2; Railton, 2017, p. 172).

Since the aim of moral education is to improve moral beliefs, behaviors and so on, at first glance, it may seem obvious that all moral education is social moral learning. However, while almost all moral education likely involves the transmission of some moral information, this aspect is more central to some accounts of moral education than others. For example, skills-based moral education often aims to bring about moral improvement not via direct transmission of moral information, but instead through the development of cognitive abilities to approach, think and reason about moral questions and problems (Watson, 2016, pp. 153–154). Contrast this with goods-based education, where the transmission of moral information in the form of moral beliefs, understanding or knowledge is a lot more central (Watson, 2016, pp. 152–153).

In this paper, we deal with moral education to the extent that it involves the social transmission of moral information. We hope that our discussion will be of interest to many; however, depending on your favorite approach to moral education, your mileage may vary.

2 Social Learning is Important For Moral Change

Moral progress is moral change for the better (e.g., Jamieson, 2002, p. 318; Sauer et al., 2021). This means that in order to understand the causal mechanisms of moral progress, we should try to understand the mechanisms of moral change. Social moral learning is one such mechanism.

To start, we will borrow from Schinkel and de Ruyter (2017) and introduce a distinction between individual-level and group-level moral change (progressive, or otherwise). Individual-level moral change refers to “the moral-psychological development of an individual” (p. 124) and is perhaps better known as moral development, in particular in the moral psychology literature (see, Schinkel & de Ruyter, 2017, p. 133). In contrast, group-level moral change occurs in larger group of individuals, for example, all of humanity or a particular cultural group, state, society, or organization. Recent discussions of moral progress have tended to focus on this notion; we will do so, as well.

Our definition of moral progress does not say anything about what it is that undergoes moral change. Existing literature provides many options. Some writers focus on individuals (e.g., Buchanan & Powell, 2018, pp. 54–58; Kumar & Campbell, 2022; Singer 2011). To them, moral progress happens when the moral beliefs, values, attitudes, behaviors, feelings or motivations of (enough) individuals improve. Others focus on features of large groups (societies, cultural groups, all of humanity), like political institutions or social norms (e.g., Anomaly, 2017; Macklin, 1977; Sunstein, 1996) or on non-moral properties like health, wealth or safety (e.g., Evans, 2017).

In this paper, we are concerned with individual-centric moral change (progressive, or otherwise), for two reasons. First, moral education targets individuals, and so it makes sense to think that if moral education causes any (group-level) moral change at all, it will do so by changing the moral beliefs, values, behaviors, etc. of (enough) individuals. Second, individual-centric moral change likely facilitates or enables many other types of moral progress, and so even authors who doubt that individuals are the ultimate locus of moral change should still consider changes in their moral beliefs, values, behaviors, etc. (see, Hermann, 2019; Klenk & Sauer, 2021).

Understood in this way, for social learning (or any other process) to be a cause of (group-level) moral change, it has to be an effective cause of individual-level moral change. Indeed, this is true almost by definition. But that is not enough. Instead, these individual-level changes then needs to scale up to the level of the group. When only a single individual changes their mind about one of their moral beliefs, for example, few would speak of moral progress or regress. Instead, these labels apply only when the moral beliefs of a large enough proportion of people in a group all change in a similar way (see, Buchanan & Powell, 2018, p. 60). Therefore, just because a process causes individual people to change their moral beliefs, values, behaviors, etc., does not mean that the group these people are members of also undergoes moral change. For example, a process may lead lots of people to change their mind about a moral issue, but still fail to affect the proportions of people in the population who belief this or that about the issue if the distribution of these changes is symmetrical (that is, the number of people who change their mind from A to B is the same as the number of people who change their mind from B to A). In such a case, individual moral development does not add up to group-level moral change.

Social moral learning meets both conditions. One powerful way to investigate if social learning leads to individual-level change is to study how people acquire new information in the laboratory. Much of this research focuses on young children. In line with the idea that social learning plays an important role in their acquisition of various beliefs and behaviors, research suggests that children are “predisposed to rapidly and automatically acquire huge amounts of information from other people” (Mesoudi, 2011, p. 15; for reviews, see, Csibra & Gergely, 2009; Tomasello, 2016b).

Many researchers suggest that children are sponges for moral or at least proto-moral information in particular. At 14 months of age, they already have the necessary cognitive apparatus in place to be able to start learning social rules and conventions (Tomasello et al., 2005), and make use of this apparatus in various ways (Prinz, 2005). For example, children between the ages of three and eight have been found to imitate the costly punishment decisions of a model in a third party punishment game (Salali et al., 2015) and to imitate altruistic behaviors, like charitable donating of an adult model (Midlarsky & Bryan, 1972; for a review, see, Bryan & London, 1970). Moreover, preschool children also already take part in social learning from the teacher’s side, enforcing some norms, for example, through protest (for a review, see, Schmidt & Tomasello, 2012) and punishment (McAuliffe et al., 2015; Riedl et al., 2015).

This leaves the question whether individual-level changes acquired through social learning ever scale up to the level of the group. Two lines of argument say “Yes.” The first line of argument starts by noting that there are many substantial differences between different cultural groups. Many of these differences cannot be explained in terms of non-cultural factors like individual learning or genetic differences alone. Instead, so goes the argument, in order to explain them, we need to appeal to cultural evolution and the transmission of information through social learning (e.g., Boyd et al., 2011; Mesoudi, 2011, chap. 1; Richerson & Boyd, 2005, chap. 2).

Here is one illustration of this type of argument. Boyd et al. (2011, pp. 10919–20; see, also, Henrich, 2016, chap. 3) describe a number of technologies, skills and social arrangements that allowed the Central Inuit to survive and thrive in very harsh environments above the Arctic circle. These examples include the know-how to make special clothes from caribou skins and to build snow houses from snow blocks in order to stay warm in sub-zero temperatures, the skills and weapons required to successfully hunt for seal and caribou, and extensive knowledge about the local environment. Boyd et al. then argue that this wealth of adaptive information is far too complex for it to be plausible that the Central Inuit developed and acquired it because of their special genes, or through individual learning alone. Instead, Boyd et al. think that it is much more plausible to think that this body of information developed over many generations and was transmitted and refined via cumulative social learning.

Can social moral learning provide a plausible explanation for moral differences between cultures (for a review, see, Graham et al., 2016)? One famous example (for other examples, see, Henrich et al., 2005, 2006; Rice & Feldman, 1997) concerns different levels of violent behavior, and tolerance of violent behavior, found between the Northern and Southern US (Nisbett & Cohen, 1996). Evidence suggests that in the South, individuals are more likely to be disposed to respond violently to threats and insults, and they are more likely to be prepared to excuse such behavior and to find it permissible. Nisbett and Cohen argue that these differences are best explained culturally: Southern ‘cultures of honor’ emerged from herding economies, where it was more important to signal one’s willingness to react aggressively to external threats to one’s entire livelihood being stolen.

A second line of argument makes use of formal models (see, Mesoudi, 2011, pp. 16–17). These models typically simulate a population of individuals who need to survive in certain environmental conditions. Individuals are simulated to use one of three different strategies to achieve this. For some individuals, their behavior is determined by their ‘genes’, meaning that these individuals get assigned a behavior at birth that does not change throughout the simulation. Other individuals are individual learners. Individual learners try out different behaviors during their lifetime and stick with the one that results in the highest payoff. Finally, social learners copy the behavior of other individuals in the population in one way or another. Formal models then set two or more of these different types of individuals to compete against one another, often for many generations, and observe which strategy (genes, individual learning, social learning) does best. Different authors have argued that in many situations, social learning turns out to be most adaptive (e.g., Aoki et al., 2005; Boyd & Richerson, 1995).

Again, there are examples of formal models which suggest that acquiring moral information in particular via social learning would have been adaptive. One important example is the evolution of human cooperation. It has been argued that it is an evolutionary puzzle how cooperative behavior, in particular in one-shot interactions with strangers (which is what happens a lot in larger cultural groups), could have evolved because it seems that such behavior would be evolutionarily unstable and subject to free-rider problems. Some authors have argued that this puzzle can be solved if we add cultural evolution and social learning into the mix (e.g., Boyd & Richerson, 2009; Tomasello, 2016a, chap. 4): The social norms and corresponding norm psychology that we inherited via cultural evolution (and gene-culture coevolution) help stabilize cooperative arrangements through institutions of punishment and other adaptive normative behaviors (see, Henrich & Muthukrishna, 2021).

3 Moral Education And Moral Change

We have suggested that social learning plays an important role in group-level moral change. However, moral education (to the extent that it involves the transmission of moral information) is just one type of social learning. Is moral education in particular an important cause of moral change?

If so, then the two conditions we looked at for social learning in general should again hold. First, moral education has to be effective in bringing about individual-level moral change, that is, at changing the moral beliefs, values, dispositions or behaviors of individual learners. While we think that there is a case to be made that the available empirical evidence on this question is mixed (e.g., Meindl et al., 2018; see also, Harris, 2009; Caplan, 2018), we will not pursue this point here. Instead, we will grant that moral education is often an effective tool to transmit new moral information at the individual level.

Second, we need to consider if those individual-level changes ever scale up to group-level moral change. In this section, we will argue that this is not the case for moral education in many circumstances. Instead, moral education first and foremost leads to moral stability and conservation.

Let us explain.

We make three assumptions about moral education. First, the moral information that educators attempt to transmit mirrors the moral beliefs, rules, values, etc. of their group. What we mean here is that the distributions of the moral beliefs, rules, values, etc. that educators (as a group) attempt to transmit are (about) the same as the distributions of these same moral beliefs, rules and values in the population at large. So, for example, if a proportion of the general population has certain moral beliefs (e.g., that slavery is wrong; that women should have the same rights as men; that animals deserve better treatment), (about) the same proportion of educators will attempt to transmit that moral belief. Second, no one educator (or small group of educators) has an out-sized influence. That is, all educators reach similar proportions of the pool of potential learners; no one educator reaches, for example, ten times more learners than every other moral educator. Third, old and new moral information have the same likelihood of being transmitted. In particular, this means that educators who aim to transmit moral information that deviates from the status quo are not systematically more (or less) successful.

In situations where this is a fair characterization of moral education, it is hard to see how moral education would contribute to moral progress. To illustrate this, we can use a simulation (code available at: https://osf.io/qpsug). For example, we can look at a group of 1000 educators, each of whom teaches 10 prospective learners. Let us call the proportion of people in the overall population who hold some new moral belief Ppop. Because of our first assumption, the proportion Pedu of educators in who will attempt to transmit that belief then is Ppop= Pedu. To make things easier, we model prospective learners as blank slates; that is, learners have no moral beliefs prior to moral education.

We then simulate moral education. Each educator attempts to transmit either the new or the status quo moral belief to their group of prospective learners. We then set a probability padopt that moral education will be effective for a given learner. To make things easier, we will choose this value uniformly for all learners and educators (say, padopt = 0.5). Thus, after moral education, some learners will have adopted the old moral belief, while other learners will have the new belief (or no belief at all). If the proportion of learners who end up with the new belief is larger (smaller) than Ppop, then our simulation will have resulted in group-level moral change—the overall proportion of people in the population with the new moral belief will have increased (decreased). We can repeat this process for many generations of educators and learners. Figure 1A shows the results for 15 generations and a few different initial values of Ppop.

Fig. 1
figure 1

Simulation results (details in the main text). Sub-figure A shows the situation where our three assumptions all hold (the moral information that educators attempt to transmit reflects the distribution of moral information in their group; no one educator or small group of educators has an out-sized influence; old and new moral information have the same transmission probability); B relaxes the first assumption, C relaxes the second assumption, and D relaxes the third assumption (again, see the main text for details).

From Fig. 1A, we see that our simulated moral education indeed did not result in group-level moral change. If so, then moral education indeed does not have a major role to play in bringing about any moral change (progressive, or otherwise), in situations where our three assumptions are true. Instead, in these situations, moral education is better understood as a force for moral stability or continuity.

Bear in mind that we are not trying to suggest that moral progress does not occur, only that moral education, if it works the way we describe it, does not play a major role in bringing it about. The main reason, we suspect, why many people tend to think otherwise (see above) is that the point of moral education is to improve the moral beliefs and values of the nascent generation. But this process merely brings said generation up to speed: it does not push the frontier of moral development in an ameliorative direction.

4 Moral Education And Moral Progress

In situations where our three assumptions are plausible, it is hard to see how moral education can contribute to group-level moral change (progressive, or otherwise). While we think that these assumptions are often true, it is clear that there are circumstances in which they are not. In such circumstances, moral education can contribute to group-level moral change. Figures 1B-D illustrate this. Each shows the results of the same simulation as Fig. 1A, except that each relaxes one of our three assumptions. Figure 1B shows what happens when some educators are twice as likely to transmit new moral information compared to what would be expected if moral education reflected the distribution of moral information in the general population. In Fig. 1C, one educator transmitting new moral beliefs ends up with 50 times the number of potential learners compared to all other educators. And in Fig. 1D, educators who attempt to transmit new moral information are have a higher rate of success (padopt = 0.75) than educators who transmit status quo moral information (padopt = 0.50).

This leaves us with two questions. One, when do our assumptions about moral education fail to hold? Two, in those circumstances, would the group-level change that moral education brings about be progressive in these circumstances, rather than regressive? It is one thing to find situations in which moral education could contribute to group-level moral change; whether this moral change will also be progressive is another matter.

4.1 First Assumption

Recall that according to our first assumption, the moral information that educators attempt to transmit mirrors the moral beliefs, values, etc. of the larger group that they are members of.

When it comes to moral education in the family, we think that most of the time, this assumption is true. It stands to reason that in terms of their moral views, parents and other members of the family of a potential learner are a representative subset of the overall population—at least, we know of no empirical data that would suggest otherwise. If so, then to the extent that parents teach what they know, family-based moral education should more or less mirror the moral beliefs, values, etc. of the wider population.

The case for school-based moral education is different. Just like math teacher’s beliefs about geometry or calculus are unlikely to be representative of the general population, ethics teachers may have systematically different moral beliefs, values, etc. from other people. We see two main ways for the first assumption to be violated here.

First, most schools are run by large institutions, like the church or the state. These institutions can (and do) influence or outright dictate the moral education curriculum. Because teachers are typically required to adhere to the contents of the school curriculum at least to some degree (e.g., Adolph, 2015, pp. 18–24; Kauffman, 2005, pp. 2–3; Priestley et al., 2015, pp. 134–135), large institutions often have a measure of influence (sometimes a lot of influence) over what moral educators teach. When they do, education may lead to group-level moral change if this curriculum differs from the distribution of moral information in the wider population.

To give a well-known historical example, “schools played an important role in spreading Nazi ideas to German youth” (United States Holocaust Memorial Museum, 2022) during the reign of the German National Socialist Party. The state exerted massive influence on the content of all education, including through textbooks, the indoctrination of teachers, and rewriting the school curriculum (Flessau, 2018, pp. 19–30; United States Holocaust Memorial Museum, 2022). Moral education was a central part of these efforts. Students were taught “love for Hitler, obedience to state authority, militarism, racism, and antisemitism” (United States Holocaust Memorial Museum, 2022), likely leading many more young people to adopt these beliefs and ideas than otherwise would have, and so contributing to large-scale moral change (regress, in this case).

When large institutions influence moral education, will this often result in moral progress? This is very hard to say in general. However, we think that there is some reason to doubt it. Large political and religious institutions on the whole tend to be quite conservative, and are invested in maintaining the status quo, that is, in keeping the distribution of moral beliefs, values and so on the way they are (e.g., Acemoglu & Robinson, 2012; Elert & Henrekson, 2016).

A second possibility is that moral educators differ systematically and in morally relevant ways from the wider population. This may lead to group-level moral change via school-based moral education when teachers have a significant say in what moral information they transmit. Such differences could arise for a few reasons. For example, in most places, school teachers have to go through a significant amount of training before they are ever allowed to enter the classroom. This training may shape aspiring teachers in ways that are reflected in the content of the moral education they choose to provide later on. Instead, the people who end up becoming moral educators may differ in relevant ways from the average person due to a selection effect: those who want to become teachers tend to be a certain type of person, or it may take a certain type of person to complete teacher training.

Are these the kind of differences that would turn moral education into a force for progress? Again, this is impossible to say in general; however, a few hopeful signs point in this direction. Like moral education itself, teacher training tends to be run by large institutions. If we are right to suggest that such institutions are often biased towards the status quo, then this may also apply to teacher training. At the same time, research suggests that in many Western countries, universities, which are in large part in charge of training new teachers, are a lot more politically liberal than the wider population (e.g., Abrams & Khalid, 2020; Broćić & Miles, 2021; Werfhorst, 2020). Moreover, there is evidence that many of the general qualities that help people succeed in higher education (whether this is teacher training or something else) like intelligence, openness to experience, conscientiousness, and critical thinking ability correlate with liberal political views (Jost et al., 2003; Sibley et al., 2012; however, see, Malka et al., 2017). If so, then this would provide some reason to suspect that the people who make it through teacher training and become teachers will tend to have more liberal values than the average person, which may in turn skew the moral education curriculum in that direction. Many would consider the moral change resulting from such an education to be moral progress (e.g., Buchanan & Powell, 2018; Sauer et al., 2021).

4.2 Second Assumption

According to our second assumption, no individual educator or small group of educators has an out-sized influence on the pool of potential learners. Unlike for the first assumption, it is not so easy to come up with scenarios where this assumption fails. Time and biology both constrain the number of children that any individual parent can attempt to educate within their lifetime, so it will be rare for any one parent to have an out-sized influence on the population of prospective moral learners. The same strikes us as reasonable for school-based moral education, on similar grounds.

Still, one intriguing avenue for moral education to violate our second assumption is when it functions as a form of status-biased learning. Status bias is an example of a model bias. Social learners do not just adopt every new cultural variant they encounter, but instead prefer to learn from certain individuals, or to learn certain variants (for a review, see, Kendal et al., 2018). Model bias favors the acquisition of one cultural trait instead of another due to which model the trait is learned from. Status bias is one important example of model bias, which is where learners prefer to learn and adopt cultural traits from individuals who have high social status (Chance, 1967; Henrich & Gil-White, 2001). Other examples of model bias include a preference to learn from individuals based on their success, the learner’s familiarity with them, their knowledge, their sex or their age (Kendal et al., 2018, Fig. 1A).

Henrich and Gil-White (2001, pp. 166–172) distinguish two kinds of status bias: prestige and dominance. Both can underwrite high social status. However, where prestige rests on merit in the eyes of other people, and so engenders admiration for and a desire to be close to prestigious individuals, dominance rests on the threat or use of force, and so is instead associated with fear and a desire for distance. A paradigmatic example of individuals who enjoy high social status mostly due to their prestige are celebrities (e.g., actors, musicians, royalty); instead, politicians, military figures or religious leaders often derive their high social status in most part from their position of dominance in society.

If moral education were status-biased, it is not hard to see how it could make a contribution to the trajectory of moral change (cf., Ellickson, 2001; Sunstein, 1996). High status moral educators are visible to a large group of potential learners, and so have the potential to influence many more potential learners than all other educators. The moral information that is transmitted by these educators may thus shape the trajectory of moral change (see, Fig. 1C).

One question is whether moral education is often status-biased. While some point out that high status individuals may be an important piece of the moral education puzzle (e.g., Hammond et al., 2022), this is not because these individuals are themselves thought to be active participants in moral education, but instead because young people often look up to them and use them as role models or moral exemplars. Nevertheless, there are likely exceptions. For example, Kohlberg describes Martin Luther King Jr. as an example of a public moral educator committed to improving people’s ideas about justice (see, Power 2015, pp. 191–192).

Does status-biased moral education lead to progress? Since this depends in large part on the specifics of what is taught, this is impossible to say in general. Still, there may be ways to get at least a sense of the answer. Huemer (2016, pp. 2000–2007) suggests that high status individuals often got to where they are because they are cognitively more sophisticated than most people, in particular in terms of skills like critical reasoning and overcoming biases; this, Huemer thinks, can explain many recent episodes of moral progress such as declines in violence and developments towards emancipation and inclusion. If Huemer is right that high status individuals often differ in terms of their cognitive abilities from the general population, then depending on what these differences are, it may open up the possibility to make some predictions about the course that future moral change will take. For example, we have already seen that compared to conservatives, political liberals do better on measure of things like critical reflection and intelligence, and score lower on things like cognitive rigidity and dogmatism (Jost et al., 2003). If this is right, then then status-biased moral education may, more often than not, swing in the direction of progress.

On the other hand, it may be that who ends up with social status has nothing to do with cognitive ability and the like, or that such traits are just morally neutral. In that case, this would justify doubts and anxiety about our ability to predict the direction of status-biased moral learning. Also, while Huemer may have had a point about certain past episodes of moral change (which is what he is talking about), this could be changing fast. One argument for this is that the modern media environment has opened up many new avenues to achieve high social status, many of which do not seem to place a premium on cognitive ability and the like (see, Barkow et al., 2012).

4.3 Third Assumption

Our third assumption about moral education is that both new and status-quo moral information have about the same chance of being transmitted. One way for this assumption to be violated is if educators who attempt to transmit a certain piece of moral information are more successful at this than educators who attempt to transmit a different piece of information, for example, because they are more capable reasoners or more agreeable, empathetic or charismatic. We already mentioned that empirical research suggests a correlation between liberal views and cognitive ability. A growing body of studies also suggests that liberal political views may be positively correlated with empathy (for a review, see, Morris, 2020). In contrast, research does not find a correlation between political views and traits like agreeableness or extroversion (see, Sibley et al., 2012). Still, if people with higher cognitive ability and more empathy are on the whole more persuasive, then liberal educators may tend to be more successful compared to their more conservative colleagues. All this is assuming that the moral education someone provides tends to reflect their own moral views.

Instead of having to do with the educators, another route is that there is something about certain pieces of moral information themselves that make them spread more effectively. In this case, moral education would be content-biased. Social learning is content-biased when it favors the acquisition of one cultural trait instead of another due to the content of the trait itself (e.g., Claidière & Sperber, 2007).

Researchers have proposed a number of possible content biases. One that most deserves a closer look has to do with our emotions.Footnote 2 The idea is that information that elicits an emotional reaction from the learner is more likely to be transmitted compared to information that does not (e.g., Eriksson & Coultas, 2014; Nichols, 2002). Emotion-biased transmission could be particularly relevant to the transmission of moral information, because emotions likely have an important part to play in human moral psychology (for a review, see, Avramova & Inbar, 2013). We know of one study that lends credence to this idea: Nichols (2002) investigated the role of emotion bias in moral change in particular focused on disgust, and found that “sixteenth-century etiquette norms prohibiting disgusting actions were much more likely to survive than other sixteenth-century etiquette norms” (p. 234).

Does emotion-biased transmission lead to moral progress? Various authors have argued for the importance of emotions in pushing moral change towards progress (e.g., Appiah, 2010; Hu, 2021; Sauer, 2019). At the same time, it is far from clear that emotions-biased transmission would often, or even most of the time, result in moral progress. No one emotion will consistently bias transmission in the direction of or away from progressive information. For example, at first glance, empathy may look like a safe bet for progress (Masto, 2015; Sorrell, 2014). After all, empathy makes people feel the emotions that they think someone else experiences (see, Cuff et al., 2016, p. 150), and so it makes sense to think that empathy will often lead us in the direction of compassion, care, lending a helping hand and of a greater circle of moral concern. However, the reality is likely much more complicated: instead of a force for good, empathy in many circumstances may actually be unhelpful, ineffective, even destructive (Bloom, 2016). On the flip-side, while the intuition that negative emotions like disgust, outrage or shame will always bias the trajectory of moral change in all sorts of morally questionable directions appears sensible, many disagree and think that these emotions can often shape change towards progressive ends (e.g., Hu, 2021; Sauer, 2019; Spring et al., 2018).

Another content bias that may push the transmission of moral information towards progress is a preference for information that coheres better with or is more easily supported by the learner’s (moral) reasoning and reflection. A number of authors have proposed a role for this type of bias in causing moral progress. For example, Kumar and Campbell (2022; also, see, Singer 2011, pp. 118–119) argue that “reality has an inherent progressive bias,” in the sense that “when people form accurate beliefs about the world around them and those who inhabit it, they tend to re-evaluate their moral feelings and norms in ways that lead them rationally toward greater inclusivity and equality” (both, p. 195). Pinker (2011, pp. 671–696) puts forward a similar view, arguing that when the socio-economic conditions are right and levels of literacy, cosmopolitanism and education in a group are high, reasoning often leads people to commit fewer acts of violence.

While there are many open empirical questions about these proposals, Pinker’s argument is consistent with recent work in sociology, showing that across a wide range of countries, higher levels of economic and physical security predict shifts towards emancipative and self-expression values (Inglehart, 2018; Welzel, 2013). Emancipative values lead people to support for freedom of choice and equality of opportunity (Welzel, 2013, pp. 57–104); self-expression values favor gender equality, tolerance of outgroups, freedom of expression and democratic participation (Inglehart, 2018, pp. 1–37). This work outright supports Pinker’s view if (part of) the explanation for these findings is that when people are more secure and better off, they engage in more reasoning, or their reasoning becomes more effective.

5 Conclusion

We have argued that moral education seldom is a cause of group-level moral change (progressive or otherwise), but instead a force for moral stability or continuity. Does this make moral education pointless or bad? No. Perhaps it sounds paradoxical, but stability can be a force for progress, too. Of course, stability can be a vice. When the status quo is untenable, change needs to happen. Still, moral progress requires not only that progressive gains be made, but also that these gains are sustained, so that upcoming generations can build upon them. Moral education may be able to help with this, which would likely make it an important enabling condition for moral progress.

Of course, this idea is not new. Gilead (2009) argues that historically, moral education used to be seen mainly as a tool for stability and continuity. It was only in the 18th century, Gilead claims, when many thinkers began to see the point of moral education to be a cause of progress. We think that this was a mistake, but not because we think that progress is an unworthy or unattainable goal, or because we reject the notion of moral progress altogether (see, 2009, pp. 102–105). Instead, our point is just that moral education is often not cut out for the job of being an agent for change.

While we think that moral education is often not a cause of large-scale moral change, there are exceptions. Moral education can be a force for change when moral educators transmit information that differs from the moral beliefs and values of their society, when a small group of like-minded educators are disproportionately influential, or when different pieces of moral information transmit at different rates. It is one thing, however, for moral education to contribute to moral change; it is another for that change to be progressive. While we have pointed out scenarios where moral education may be a force for progress, we think that it is very unlikely that moral education will always or even most of the time, contribute to moral progress. Instead, whether it does will depend on the details of each episode of moral change.

We have pointed to a number of these details, some of which may be quite informative. For example, we have argued that moral education can play a role in progress if the institutions that run it skew the curriculum away from the status quo and in the direction of improvement. This underlines the important role that institutions have to play in moral progress (e.g., Anomaly, 2017; Sauer, 2019).

Our investigation also reveals the potential relevance of social learning biases. Both content bias and prestige bias may play a role in turning the fruits of moral education progressive. In the scholarship on moral progress, few have focused on these or other social learning biases, or indeed, on the role of social learning more generally. We think that this should change because there are signs that (biased) social learning may often play a powerful role in shaping moral change. For example, a number of researchers argue that status bias can cause large-scale cultural shifts (e.g., Chance, 1967; Henrich & Gil-White, 2001), and some have developed detailed models of how status-biased social transmission can play an important causal role in moral and social change in particular (Ellickson, 2001; Sunstein, 1996). In light of this, future research should attempt to better understand these biases, their potential roles in moral progress, and how they may be influenced or harnessed to make and sustain progressive gains.