What is Systemic Racism?

Neil Shenvi is an evangelical Christian chemist and blogger who’s notable online for arguing that critical theory is incompatible with Christianity. He’s recently published a post criticizing the concept of systemic racism. While Shenvi agrees that racial injustice exists, he thinks the concept of systemic racism is “fatally flawed,” which I take it roughly means he thinks it’s not a useful analytical concept. I don’t really think of myself as a “critical theorist,” but I think he’d regard me as sort of critical-theory-adjacent, and anyway, I think it provides an opportunity for me to sort out my own thoughts on this.

After surveying some glosses on systemic racism, Shenvi concludes that the term “systemic racism” is supposed to refer primarily to “ideologies and policies that produce racial disparities.” I want to discuss two of his criticisms of the concept. One, given quickly near the end, is that “racial disparities do not entail injustice.” (He cites the racial disparity among NBA players.) The other, larger one is essentially that systemic racism is too blunt of a concept: the term carries a kind of normative weight that doesn’t apply to everything the term itself applies to. (Perhaps there are some other criticisms, but these are the ones I’ll focus on.)

Rather than trying to speak for the people Shenvi cites, I’ll explain my own view, in light of some of the scholarly work I’ve done on egalitarianism more broadly. (Of course, I don’t think my view is idiosyncratic; I think it represents a way of articulating what people are typically getting at.) This will allow me to say something about the first point. I’ll then say something about the second point.

Which disparities do egalitarians care about?

The first thing to do in response to Shenvi’s proposed definition (“ideologies and policies that produce racial disparities”) is to get clear on what we mean by “disparity.” There are an infinite number of ways in which there could be “disparities” between people, and obviously egalitarians should care about some and not others. An example from Marilyn Frye: a billionaire playboy breaks his leg while skiing. I’m not a billionaire, but (let’s suppose) I’m sort of naturally content, so I was already as happy as he was. Plus, I didn’t break my leg. So now there’s a disparity in happiness between us. But while the suffering of the billionaire is too bad in itself, presumably no one thinks the disparity between us–the fact that I’m happier than him–is problematic in the way that certain other disparities are.

So egalitarians owe an account of exactly which disparities are objectionable, and why. Inspired by people like Elizabeth Anderson and Iris Marion Young, I’ve defended a “social egalitarian” answer to this. (It may be helpful to read the hyperlinked piece.) What egalitarians should primarily care about is that people relate to one another as equals. The disparities they fundamentally oppose are social disparities–the oppressive hierarchies in which some people are dominated, disregarded, disrespected, and exploited. Part of the aim is to provide a sort of unifying theory of what’s wrong with other, more concrete disparities, such as inequalities in wealth, access to healthcare, treatment by the criminal justice system, etc. These often result in things which are bad in absolute terms: lacking healthcare is bad, even if other people also lack healthcare. But they’re objectionable as disparities insofar as, and because, they cause, or constitute, or are expressions of, oppressive hierarchies. So, for instance, many unjust elements of the criminal justice system (police brutality, disproportionate punishments, lack of adequate legal defense, etc.) would be bad even if they fell upon people equally. But the racial disparities are objectionable primarily insofar as they demonstrate lack of equal concern for the interests of Black people and serve to maintain (in Michelle Alexander’s words) a “racial caste system.”

Core-dependent homonymy

One other thing: I’m going to suggest that the term “systemic racism” exhibits core-dependent homonymy, so I’d better explain that. Let’s use a classic example from Aristotle. To say that an organism is “healthy” is to say something like that it’s functioning well biologically and psychologically, that it’s free from disease and injury, or whatever. We can also say that walking is healthy, that kale smoothies are healthy, that a complexion is healthy, etc. But “healthy” applied to these other things can’t mean the same thing as “healthy” applied to an organism: kale smoothies don’t function biologically or psychologically at all, and they’re “free from injury or disease” only in the sense that milkshakes or rocks or cigarettes are.

At the same time, it’s not like “healthy,” applied to an organism and “healthy” applied to smoothies are unrelated in the way that, say, “bank” applied to a river bank and “bank” applied to a financial institution are. Instead, they’re unified in this way: the “core” notion comes from healthiness as applied to an organism, and then various secondary notions are about how things relate to healthiness in this core sense. Foods, activities, environments, mental states, etc. are “healthy” in the sense that they are conducive to healthiness in the core sense–they promote or maintain the healthiness of the organism. Meanwhile, a complexion, say, or a value that falls within a certain range on a blood test, could be “healthy” if it indicates that you are healthy in the core sense. That’s core-dependent homonymy.

What is systemic racism?

With all that in place, I’d say something like this: systemic racism, in its core sense, refers to unjust racial hierarchies qua unjust racial hierarchies. This is an important concept to have–and the strong term “racism” is merited–precisely because these are the main thing that egalitarian racial justice opposes. There is then (at least one) secondary sense which is about how other things relate to this core notion: institutions, policies, resource distributions, attitudes, patterns of behavior, etc. are all systemically racist insofar as they cause, constitute, sustain, are expressions of, etc., unjust racial hierarchies. This inherits the term “racism” via core-dependent homonymy, and is an important concept to have precisely because it picks out the social factors that, in one way or another, have to be addressed in order to achieve equitable social relations.

Of course, there are other ways to use the term “racism,” and they may well be useful in other contexts (for instance, if we’re trying to evaluate an individual’s moral character). But insofar as we’re concerned in our socio-political theorizing with egalitarian social justice, this will be the sense of fundamental interest to us in doing that kind of theorizing.

This definition actually accords with Shenvi’s thought that not everything which causes racial disparities should count as systemically racist. Technically, we can even imagine cases where it counts as “not systemically racist” things which exacerbate present disparities that do result from systemic racism. The life expectancy gap between whites and Blacks in the US results from systemic racism. Suppose some lethal childhood disease disproportionately affects children of European descent. We cure it. The life expectancy gap increases. Curing the children might somehow be implicated in systemic racism. (If this sounds weird to say, hold that thought until the next section.) Perhaps that disease, rather than some other, worse one, was researched because it disproportionately affected white children; perhaps better health among white kids exacerbates the racial wealth gap, increasing disparities in political power and social status between racial groups; perhaps, etc. In fact, in a society shot through with structural oppression, we should sort of expect that all kinds of things are going to be implicated in systemic oppression in one way or another. But the point is that it’s not like keeping these kids from dying is itself any kind of problem for racial equity: going that route renders us vulnerable to the “leveling down” objection (think “Harrison Bergeron”).

But at the same time, I think this does explain why the sorts of disparities which people actually object to are objectionable. And more broadly (as Elizabeth Anderson argues) I think the social egalitarian picture captures the heart of what participants in egalitarian social movements are after. What they are demanding is, first and foremost, the ability to live as full and equal members of society.

What is the upshot of calling something “systemically racist?”

But this brings us around to Shenvi’s major point. If all kinds of stuff are (at least in the secondary sense) systemically racist–including things which are very good in other ways, and which are only very indirectly, unintentionally, and slightly connected to unjust racial hierarchies–isn’t “systemically racist” sort of a misleading term? Specifically, Shenvi’s thought seems to be something like: look, if something is racist, we ought to destroy it. But we shouldn’t destroy everything that’s systemically racist, in the sense at issue here. Shenvi says:

When we designate some phenomenon as “systemic racism” we are implicitly passing a moral judgment on it and are indicating that such systems need to be dismantled. When people say that racial disparities in wealth, in prison sentencing, or in police shootings are manifestations of “systemic racism,” they are not making purely descriptive observations. They are assuming that there is something deeply wrong with these systems.

Unfortunately, many other systems meet this same definition for systemic racism.

Take marriage. People tend to marry intra-racially, not inter-racially. Because whites have higher incomes and greater wealth than blacks, the combined income and wealth of a white married couple will exceed the income and wealth of a black married couple, allowing the white couple to invest, buy a home, save, and build additional wealth relative to the black couple. Thus, according to this definition, marriage is a manifestation of “systemic racism.”

Private property is another example. There is a tremendous wealth gap between whites and blacks, a large part of which comes from home ownership. If private property were abolished, a large portion of this disparity would vanish. Conversely, to fail to abolish private property is to perpetuate this racial wealth disparity. Private property therefore meets our definition of “systemic racism.”

Don’t worry about the specific examples here. The question to ask is: in calling something systemically racist, are we implicitly indicating that it needs to be dismantled?

In the core sense, the answer is yes. In a just society, racial hierarchies will no longer exist. But Shenvi’s examples are, if anything, supposed to be examples of systemic racism in the secondary sense. This complicates the question.

Something could be systemically racist, in the second sense, solely in virtue of its effects. And, in general, if something has a bad effect, there are three things you could do to stop that: you might destroy it, you might change it so that it no longer has the bad effect, or you might change the things around it so that it no longer has the bad effect (or some combination of these things).

Consider health. Doing cocaine is unhealthy (in the secondary sense); the solution is to stop. Doing certain exercises incorrectly is unhealthy (in the secondary sense), because it risks injuring you. The solution there might be, not to stop doing the exercises, but to do them correctly. Taking a medication that interacts with alcohol might be unhealthy (in the secondary sense) given that you are a heavy drinker. But the solution there might not be to change anything about the medication at all, but instead to cut back on drinking.

We can say the same thing about things which are systemically racist in the secondary sense. Legal segregation served no purpose except as a constituent and sustainer of racial hierarchy, and had to be destroyed. Electoral systems have been and remain systemically racist, whether openly or in more subtle ways (witness the many ways Republicans attempt to disenfranchise minorities). But the solution there is not to abolish elections, but to fix the system. Finally, suppose the cure I discussed earlier did exacerbate the racial wealth gap. The cure is extremely good in itself, but has a problematic consequence because of external economic inequalities; obviously the solution would be to address these external factors in some other way, rather than to let the kids die.

Is the term misleading or confusing?

Of course, Shenvi could say that, while all this follows from the definition of “systemic racism” given above, it’s nonetheless somehow misleading or confusing. An obvious initial worry might be that the fact that there are different senses of “systemic racism” risks introducing ambiguity: people may not realize there are different senses, it may not be clear which sense is being employed, etc.

Of course, as an analytic philosopher, I’m always in favor of drawing fine distinctions, using words in laboriously exact ways, etc. But I don’t see why the problem here is any worse than with other terms, and other terms generally seem to serve their function well enough. Many people wouldn’t be able to explicitly articulate the different senses of “health” and how they are related via core-dependent homonymy, but they can nonetheless understand well-enough what people mean when they use “health.” I don’t see why the same shouldn’t be true here.

A second claim–closer to Shenvi’s actual objection–would be that the secondary sense of “systemic racism” is misleading because in ordinary language, saying that something is racist entails that it should be dismantled, whereas this isn’t true of saying something is systemically racist in the secondary sense. But I just don’t think the claim about ordinary language is right. The electoral system in the South when Black people were not allowed to vote was uncontroversially racist. I don’t mean this in any rarefied academic sense: it’s totally correct to say, in ordinary English, “that system was racist.” Even conservatives agree that sentence is true. But no one thinks that, in saying this, you’re saying that the electoral system should have been abolished, rather than fixed.

What’s true is this: there are contexts where saying that something is systemically racist, even in the secondary sense, carries the implicature that you want it to be abolished (or at least seriously altered), so that, if that isn’t what you mean, it makes sense to cancel the implicature. Suppose you say “It’s great that they cured that disease,” and I respond with “It’s systemically racist, because it exacerbates the racial wealth gap.” There, I didn’t literally say that I thought the cure wasn’t great and shouldn’t have been distributed. What I literally said may well be literally true. But you would naturally wonder why, out of all the literally true things I could have said, I said that one, and would probably assume I was opposed to the cure. So if I did have some reason to mention the systemic racism angle at all, it would make sense to cancel the implicature by following it up with something like “But of course I’m glad that the cure exists and it should be distributed; my point is just that…” Similarly, suppose a movement makes “X is systemically racist” its slogan. Obviously, we’d suppose they were saying something was seriously wrong with X and were advocating for either its abolition or thorough reform. If they weren’t, it might make sense to choose a different slogan.

This isn’t a problem with the concept of systemic racism, though; this is just a general feature of human language. If you ask me whether I would like a salad and I respond with “Eating salad poses a threat to your health,” what I have said is literally true (I could choke on the salad). But you would obviously think I was refusing the salad, and more generally that I was communicating that people shouldn’t eat salads because it was too dangerous. If this isn’t what I meant, it would make sense to clarify. But there’s no problem with the term “threat to your health;” it’s just generally a fact about human communication that literally true statements can carry false implicatures, and that it’s important to keep this in mind and respond accordingly. Or suppose a movement makes “X is killing people” its slogan. Naturally, we’d assume they were advocating for the abolition or reform of X. But, of course, the mere fact that something is killing people doesn’t entail that it should be abolished, or even changed: I guess salads occasionally kill people, when they choke on them. This isn’t a problem with the term “killing people.” More broadly, any term which suggests that something is in some way seriously bad or having a bad effect will carry this kind of implicature in many possible cases; if it’s a problem for “systemic racism,” it’s a problem for all such terms. But what would a language that eliminated this possibility even look like?

So, to sum up: on my view, the concept of systemic racism is an important one to have, in light of its usefulness for an egalitarian understanding of what we’re up against. Like any term once it enters common usage, there is a potential for ambiguity, misunderstanding, etc. It’s nice to clarify what our terms mean, and what we think they’re referring to, as much as possible (and I think this is one useful thing philosophers can do). But at the same time, it really only looks like there’s some special problem for “systemic racism” here if we hold it to a different standard than other terms in common usage.

Reflections on My Debate with Ben Watkins

I recently had a fun YouTube debate on philosophy of mind with Ben Watkins. I’m thankful to Ben for debating, and to Cameron Bertuzzi for moderating. The debate is here, and the slides I used are here. These things always feel like they need to take about twice as long as they actually do, so I have a few further thoughts. I don’t have the time just now to write something explaining the background concepts, so parts of this may not make much sense if you don’t have some experience with philosophy of mind. But I’ll start with an overview of what I take the major points of contention in the debate to have been. I think that might help people who had some trouble following the more technical aspects.

Since this is my first blog post, I’ll add that you can subscribe for email updates using the little widget at the side of the page.

I basically pursued two main lines of argument. The first, the transplant argument, is contained within the slides above, so I won’t rehash it here. The ultimate takeaway of that argument is supposed to be that reflecting on my nature as an enduring subject of experience shows that I have to be simple and indivisible, and this (pretty much) rules out my being material. Ben’s response to this was to deny that I am an enduring subject of experience at all. He instead endorses a “no self” view, saying that all there is to being me is psychological continuity over time (though, as came out in the debate, I had difficulty seeing exactly what his concept of “psychological continuity” is–contrary to what he said, it didn’t seem to me to be compatible with Parfit’s understanding).

A second argument, which I didn’t frame as an explicitly different argument but which sort of came out over the course of the debate as I learned about Ben’s view, was this: both Ben and I agree that the mental and the physical are distinct. And both Ben and I agree that there are (complicated, and in some ways very arbitrary-seeming) law-like correlations between mental and physical states. However, Ben thinks these are, essentially, brute necessities: they are not self-evident, and (as far as I could tell) they don’t have any deeper explanation. On the other hand, I think these correlations represent contingent psycho-physical laws which are explained by divine action. Therefore they have a deeper explanation. So I think my view has an advantage here by having fewer complicated, arbitrary, unexplained posits (at least, it has an advantage if theism is not otherwise too implausible–admittedly, this implicates many other issues). On my view, theism and dualism are sort of mutually supporting here: together they can provide the best view about about consciousness and its relation to the material world, so this gives us reason to accept both. I think at least part of Ben’s response to this was that, even if this was some advantage, dualism still has problems with mental causation.

This brings us to Ben’s arguments. He cites (I would say) three main kinds of consideration which are supposed to favor the view that the mental, while not reducible to the physical, necessarily supervenes upon it (i.e., there can’t be any mental differences without physical differences). The first kind is an alleged in principle problem with dualistic causation. We can divide this into two parts. One, the interaction problem, asks how it’s possible for an immaterial soul to interact with a material body. My answer to this was that there is no “how” involved: it’s a basic form of causation, which everyone has to believe in for some cases. The other, the pairing problem, asks why souls are paired with the bodies they’re paired with, rather than other bodies. I actually didn’t address this, and say a bit about it below.

A second class of considerations are supposed to pose an empirical problem for dualistic mental causation: essentially, we haven’t observed such causation (though we’ve observed plenty of purely physical causation), and such causation would violate the laws of nature. I made two points in response. One was that, as far as I could tell, this was just as much a problem for Ben’s view as for mine, given that he agrees that the mental is distinct from the physical: irreducibly mental properties or events causing things in the physical world violates causal closure just as much as immaterial substances doing so. The other was just that I wasn’t sure we’d expect to have observed immaterial causation even if it was going on in our brains, so that I doubt this provides much evidence against its occurring.

A third was that the close, law-like correlations between mental and physical states are best explained by a necessary supervenience relation. Here, I made three points. The first was that observing law-like correlations generally doesn’t make us conclude that the correlations are necessary (e.g., the significant majority of philosophers agree that the laws of nature are contingent). The second was really a form of what I called my second strand of argument above: if we say these are contingent correlation explained by divine action, we have less in the way of complicated, arbitrary, unexplained posits. A third was that standard conceivability arguments (involving philosophical zombies, etc.) count against the necessary supervenience thesis.

All that being said, here are a few reflections:

A) I think a pivotal moment in the debate takes place around 56:58-57:45, when Ben agrees that my argument for dualism works if we reject the “no self” view in favor of the view that I am an enduring subject of experience. My argument in the slides was focused on trying to establish that conditional (if we are enduring subjects of experience, then…), since I assumed most people watching would accept the existence of a self (thought of as a subject of experience). But since Ben ultimately granted the conditional, I guess the real action was in the antecedent.

Relatedly: the way I set up the personal identity question in presenting my argument during the debate presupposed that, if I’m a material object, then my persistence conditions are those of whatever material object I am. While I think this is what a materialist should say, this isn’t uncontroversial, and in fact it’s not true on (ordinary versions of) the psychological theory of identity: there, it’s more like being me is an office which could be held by different objects at different times. For the sake of time, accessibility, etc., you always have to sort of smooth over certain things in a context like this. But now I regret having smoothed that over, partly just out of generally not liking to say things that are technically false and partly because the psychological criterion wound up factoring into the debate later.

B) Notice that my “second strand of argument” above counts against Ben’s position (which involves necessary supervenience), but is actually neutral between substance dualism and ordinary forms of property dualism (according to which the connection between physical and mental properties is contingent). Since the debate was billed as being about whether the soul existed, I focused on that. If the debate had instead been primarily about the argument from consciousness for theism, though, I could have merely defended the ordinary, contingent form of property dualism, arguing that, even if mental and physical properties inhere in the same substance, theism is still the best explanation of the correlations between them (including the fact that mental properties are instantiated at all).

This would have sidestepped the self/no-self question, and also would have avoided at least some of Ben’s arguments (e.g., the pairing problem). So, oddly, it may be in some ways easier to get from consciousness to theism than from consciousness to the soul.

C) Later on in the debate, Ben expressed the view that it was implausible to say (as I had) that animals have souls, but we didn’t really have time to get into why he thought that. I wish we had. My argument for substance dualism implies that subjects of experience in general are immaterial souls, and so implies that any animals who have conscious experiences are immaterial souls. (In fact, it also applies to any artificial intelligences that might someday have conscious experiences, etc.) But why should that be a problem? I think the most you can say is that it might be historically unusual in certain ways–but I think this historical tendency was for (epistemically, and often morally) bad reasons.

D) I realized after the debate that, although Ben mentioned the pairing problem a few times, I never directly addressed it. If I had, basically I’d have made the points Andrew Bailey, Josh Rasmussen, and Luke Van Horn make in this paper.

E) I didn’t pursue because I was worried the audience would get lost, but I wish we’d been able to get into whether the necessary a posteriori relations between the mental and the physical which Ben believes in are supposed to obtain between mental properties and functional properties, or between mental properties and specific neurological properties, or what. Basically, my thought is that answering “functional properties” commits you to weird judgments about what sorts of things are necessarily conscious. On the other hand, saying “specific neurological properties” makes these relations much more complicated and arbitrary, so that it’s less plausible to posit them without offering some deeper explanation for them (as I try to do). This seems particularly true when we consider that, to account for multiple realizability, the mental properties will probably have to be supervening of some complicated disjunction of different neurological properties.