Abstract
Existing work has documented certain important effects involving the way people think about statistical regularities in human behavior. When people observe a highly consistent behavioral regularity, they tend to treat this regularity as normative (e.g., thinking that there is something wrong with behavior that violates the regularity). As Heyes emphasizes in her fascinating target article, one way to explain this phenomenon is by developing a theory that focuses on people’s way of understanding social norms. However, people appear to show many of the same effects when they observe regularities that do not involve human behavior (e.g., when they observe regularities involving biological processes in other animals and in plants). Thus, another possible approach would be to explain this phenomenon in terms of a much more general psychological process that is not specific to social norms and simply involves people’s ordinary way of understanding statistical irregularities.
Imagine growing up in a culture in which everyone refrains from eating meat on Wednesdays. Now, suppose that one day you are talking with a coworker on a Wednesday night, and you see him eating a steak. You might immediately feel that he is doing something wrong.
What exactly leads us to make these sorts of judgments? In her fascinating and thought-provoking target article, Heyes introduces one way of conceptualizing this question. In this conceptualization, people’s way of making these judgments reflects a capacity they have for thinking about social norms. The key question then is whether this capacity is innate or whether it is acquired through learning.
Although Heyes’s way of conceptualizing the question is a promising one, I want to push back a bit and consider the prospects of a very different approach. I do not mean to suggest that Heyes’s conceptualization will necessarily turn out to be incorrect; the point is just that this other approach also seems promising and worth considering. Thus, before we make a choice between the two, it might be helpful to contrast them and see what evidence there might be for or against each.
At the core of this alternative approach is the observation that we seem to observe a similar phenomenon even if we look outside the domain of social norms. To see the issue here, consider a case that is almost exactly like the example with which we began, except that it does not involve social norms.
Imagine growing up in a culture in which every single time you see a tiger, it has black and yellow stripes. Now suppose that one day you encounter a tiger that is entirely black. You might immediately feel that there is something wrong with this tiger.
Notice that this example is not plausibly understood as a case of an agent violating a social norm. Indeed, there is a clear sense in which the two examples involve different types of wrongness: The first involves the kind of wrongness one finds for agents violating social norms, while the second involves the kind of wrongness one finds for biological systems violating norms of proper functioning. In short, if we have a theory that is only about people’s capacity for thinking about social norms, it would presumably not be able to explain people’s judgments in this latter example.
One possible view, then, would be that judgments like the one in the former example should be explained in terms of capacity for thinking about social norms, whereas judgments like the one in the latter example should be explained in some completely different way. This is certainly a reasonable suggestion. But of course, another possible approach would be to start by developing a theory that also provides an explanation for the latter example and then to try to explain the former as a special case of this much broader phenomenon.
In recent years, there has been a surge of research on how to understand judgments like the one we find in the second example. This research suggests that, quite generally, there is a tight connection between statistical representations and prescriptive representations. Thus, in the case of the tiger, you could imagine people having a representation of what tigers usually look like (a statistical representation) and then, separately, a representation of what it is good for tigers to look like (a prescriptive representation). But the mind does not seem to work like that. The mind does not seem to involve completely separate statistical representations and prescriptive representations. Instead, the way the mind seems to work is that people have hybrid statistical and prescriptive representations that represent both how things usually are and how they ought to be.
The evidence for this claim comes from a convergence across numerous different lines of research. We are getting evidence for this claim from careful research on the role of statistical and prescriptive considerations in intuitions about prototypicality (Bear & Knobe, 2017; Foster-Hanson & Rhodes, 2019), causation (e.g., Halpern & Hitchcock, 2015; Icard et al., 2017), natural-language modals (e.g., Phillips & Cushman, 2017), what first comes to mind (Bear et al., 2020), and in numerous other streams of research.
With all this in mind, we can come back to our original question. What is happening in people’s minds when they observe an agent violating a social norm and conclude that this agent has done something wrong? One possible strategy would be to explain this phenomenon by ignoring the broader connection between statistical and prescriptive representations and instead developing a theory that applies only to cases of social norms. Importantly, however, another possible strategy would be to explain this phenomenon in terms of a much more general connection within human cognition between statistical and prescriptive representations.
If we adopt that second explanatory strategy, our question becomes a very different one. We do not need a special explanation of the fact that when people see that an agent has violated a social norm, they think that the agent has done something wrong. That phenomenon just follows from something much more general about the relationship between representations of statistical norms and representations of prescriptive norms.
Instead, if we want to understand what is special about social norms, what we need is an explanation of the additional complexities that are unique to the case of social norms—that is, the features we observe for social norms that we do not also observe for other kinds of norms. This, too, would be a difficult and important task, but it would be a very different task from the one we would have faced if we had adopted the other strategy.
