For this post, I set out to describe and analyze what arguably constitutes the most fundamental disagreement in philosophy: realism versus anti-realism, not only in metaethics but in general. In one form or another, this disagreement also shows up in the philosophy of mind, personal identity, aesthetics, epistemology, the philosophy of physics, and so on. Belief in different types of realism is correlated (see the appendix), so I see significant benefits to also discussing other types of realism. For instance, moral philosophers on both sides of the realism/anti-realism divide have drawn comparisons from different domains in support of their positions.[1] To address those comparisons, prior discussion of those other domains is helpful. However, I tried to structure this post to be more than a (second) introduction. Even though my aim is to neutrally describe the core differences between realism and anti-realism, in doing so I will already present some of my main arguments for anti-realism. My most persuasive “argument” against (moral) realism isn’t any single knockdown objection, but rather my overall impression that when we go from realism to anti-realism, we don’t have to give up anything worth wanting. I expect (moral) realists to disagree with that sentiment, in part because I could imagine that many may not have been motivated to explore the option space for moral reasoning under anti-realism (“to make anti-realism work”). I wrote this post to make what I perceive to be underappreciated points about how realism comes with some surprisingly non-trivial claims, and how anti-realism doesn’t have to mean throwing up one’s hands saying “anything goes.”
Summary
We tend to have the feeling that disagreements on what’s “moral,” “conscious,” “(epistemically) right,” and so on, go beyond semantics, or are unambiguous. The discussion between realists and anti-realists is about the degree to which this feeling is accurate.
Anti-realists don’t deny that domains in question (morality, philosophy of mind, epistemology, etc.) have some “structure.” Instead, they disagree with the realists on how to interpret and relate to that structure.
While the typical realist believes in the notion of an unambiguous speaker-independent (“objective”) reality, the typical anti-realist would consider this a confusing concept because we have to interpret reality, and interpretations about a domain can only be formed on top of subjective evaluation criteria.
There’s a category of realisms built around what David Chalmers has called “bedrock concepts:” concepts like “moral” or “conscious” or “(epistemically) right.” Bedrock concepts are assumed to be meaningful, but also inexplicable—in the sense that we cannot explain them in neutral terminology. Anti-realists are generally skeptical of bedrock concepts, whereas many realists cannot imagine doing philosophy without them.
For many types of realism, the only arguments in favor are appeals to widely distrusted essentialist intuitions. (There are also so-called naturalist versions of realism that don’t rely on bedrock concepts; I won’t focus on them in this post.)
Not all types of realism rest on thin assumptions. Arguably, bedrock concepts related to (various kinds of) normativity, as well as bedrock concepts related to consciousness (“qualia”), are backed by stronger arguments. Realists about normativity argue that normative anti-realism is self-defeating, and realists about qualia argue that qualia anti-realism is impossible.
Points of agreement
Discussions on the following domains all have something in common:
The presence or absence of consciousness
Morally good or bad actions
Beautiful or ugly art
The preservation (or lack thereof) of personal identity (for instance after going through a teletransporter)
Epistemically right or wrong reasoning
The truth or falsity of scientific theories
Discussing the topics above tends to be accompanied by the impression that we are accomplishing something substantive, i.e., that we can make unidirectional intellectual progress and that corresponding disagreements can go beyond personal opinion. (The strength of this intuition may vary depending on the person or the field in question.) Very crudely, “realism versus anti-realism” is the philosophical discussion about the degree to which that impression is appropriate.[2]
Going by connotations alone, we might at first think that realism means that a domain in question is real, whereas anti-realism implies that it’s something other than real (e.g., that it’s merely imagined). Although accurate in a very loose sense, this interpretation is misleading.
It’s misleading in that it suggests the discussion is analogous to conversations about the existence of God. But philosophers call this “theism versus atheism,” instead of “realism versus anti-realism about God.” There’s a cliché about atheists continuing to talk about God even after they quit believing. Still, this cliché only refers to “talking about God” to convince others (or to show off one’s reasoning)—which is different from continuing to “talk about God” in the interest of figuring out God’s true nature. Typically, when someone stops believing in God, they also stop talking as though God exists. As far as private purposes are concerned, atheists don’t generally refine their concept of God; they abandon it.[3]
Going from realism to anti-realism works differently. (Of course, not all anti-realists think alike; and many people may change their opinions in the other direction, too.) Rejecting realism for a domain neither entails erasing the substance of that domain, nor (necessarily) its relevance. Anti-realists will generally agree that the domain has some relevance, some “structure.” For instance, anti-realists about morality may still be interested in (and moved by)[4] moral arguments, anti-realists about epistemic facts may still care about how to reason, and anti-realists about consciousness may still be interested in the latest expert survey on invertebrate sentience.
Points of disagreement
There’s agreement on the domain in question having some structure. The disagreement comes down to the correct way to relate to that structure. While the realists believe that there are speaker-independent facts directly about the domain in question, the anti-realists deny this.
The existence of “speaker-independent facts” would mean that the domain in question lends itself to only one interpretation: the interpretation that holds true because it’s true.[5] Another way to express this sentiment is that there’s an interpretation that exists “prior to thought” (Boyd, 1998).
Not all anti-realists think alike, but those who endorse anti-realism about most (or even all) the contested domains are especially likely to reject the notion of speaker-independent standards altogether. This rejection results in a philosophical framework grounded in merely speaker-dependent standards.[6] In such a framework, we need to specify our criteria of evaluation before we can meaningfully discuss what’s true or false about a domain. Once those evaluation criteria (“standards”) are stipulated, then we can evaluate claims. In this sense, the claims can be true or false, but only in relation to subjectively (“speaker-dependently”) chosen standards.
Figure 1. Duck-rabbit illusion (public domain)
From an anti-realist perspective, there’s a degree of ambiguity to everything. Consider the duck-rabbit illusion (above) for illustration. Although there are two different ways to interpret the figure, we can still distinguish between true and false things to say individually about the duck or the rabbit. For instance, choosing to focus on the “duck interpretation,” it would be wrong to claim that the feathery animal has its beak wholly closed. Anti-realism about the duck-rabbit illusion doesn’t mean that we can’t make valid claims about the different animal interpretations; it only means that there’s no point in asking which of the two animal interpretations is the right one. Neither the duck interpretation nor the rabbit interpretation adds anything further to the underlying patterns of black and white.
Some people equate anti-realism with the nihilistic sentiment “all interpretations are wrong.” However, the way I think about it, anti-realism is best summarized as follows:
We cannot avoid the use of interpretations. Insofar as interpretations can be right or wrong, there can be more than one right interpretation.
We can apply this perspective to domains like morality, consciousness, or epistemology. To give an example, consider (again) anti-realism about morality. Looking at some of the longstanding disagreements in normative ethics, it can appear as though different sides are talking past one another, answering different questions. For instance, when I was involved in animal advocacy, I noticed that some moral arguments worked well for some people, while other people seemed to consider them beside the point. Having always looked at morality from a consequentialist lens, it felt only natural to me to frame my inquiries into normative ethics as inquiries about what it means to make the world a better place (for all morally relevant beings). However, I discovered that this framing was far from self-evident to everyone. In particular, many people think about morality more in terms of norms or social contracts that people would agree on in order to co-exist peacefully and productively in society. This perspective certainly captures some aspects of the folk concept “morality,” but it leaves out other aspects. (Of course, the same can be said for consequentialism.) For a long time, I thought that at least one side had to be wrong, that at least one party doesn’t get what morality is about. I had a viscerally intense sentiment that “either morality includes caring intrinsically about others, or it doesn’t.”[7]
By thinking of it as a dichotomy, I had committed myself to a metaethical assumption: moral realism. Moral anti-realists can point out that I should have at least considered the possibility that some discussion participants may be talking past one another, “playing different games.” To make the analogy more explicit:
Duck-rabbit illusion: There’s a sense in which the duck is “real.” But we could also take a different perspective, in which case we’d see a rabbit. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d see in the figure is shapes of black and white.
Normative ethics: There’s a sense in which consequentialist obligations to avoid purchasing meat from factory-farmed animals are “real.” But we could also take a different perspective (according to which morality is about hypothetical contracts between people), in which case we’d see no obligations toward animals. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we see in morality is people making up various justifications for things that (in one way or another) appeal to them.
The analogy becomes increasingly more strained for some of the other domains that are commonly discussed, but I think the key aspects still apply:
Consciousness: There’s a sense in which bees are “conscious.” But we could also take a different perspective (for instance, one that highlights higher-order cognition, drawing a conceptual boundary to exclude hidden qualia),[8] in which case we’d (probably?) see bees as not conscious. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d ever see is various cognition-like processes of different complexity, with differing degrees of intentionality, self-reference, and so on.
Personal identity: There’s a sense in which I’m still “the same person” after I exit a teletransporter. But we could also take a different perspective (for instance, one that focuses primarily on physical connectedness), in which case we’d view the person exiting as a different person from the one who entered. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d ever see in personal identity thought experiments are varying degrees of physical or psychological similarities and temporal differences.
Aesthetics: There’s a sense in which all exhibits in modern art galleries are “art.” But we could also take a different perspective (for instance, one that focuses on whether something is pleasant to look at or whether it took special skills to produce), in which case we’d only view a subset of things in modern art galleries as “art.” And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d ever see is patterns, shapes, and forms that leave various impressions on different onlookers.
These are strained analogies in many ways, but the key point is that there seems to be an ever-present element of arbitrariness when it comes to the aspects of reality to which we want to draw attention. According to my anti-realist perspective, reality simply is, but interpretations always add something. Deep down, all interpretations are arbitrary and we can always take on the “stubborn” perspective to say that there’s not even a question that needs to be answered. Still, we’ll find that we often care about how to interpret a domain and where to draw categories, and that we bring certain expectations already to the table. It is that second sense in which anti-realism is not arbitrary.
(Re)interpreting realist discourse
Realists about some of the above domains might object that the proposed analysis misses the point! (At least when it comes to the domain that they are realists about.) Unlike in the duck-rabbit illusion, the way we talk about normative ethics, consciousness, and epistemology seemingly postulates the existence of some speaker-independent reality. At the very least, anti-realists should be able to explain what (if anything) they think might be going wrong with that.
In the context of moral anti-realism, there are distinctions between three different ways of rejecting the idea that our discourse describes actual features of a speaker-independent reality. This terminology can be confusing because it packs together linguistic claims about the nature of moral discourse and substantive claims about whether or not there is a speaker-independent reality. Very roughly, these three ways are:
Non-cognitivism:
Linguistic claim: Moral claims, even though they may appear as though they are about a speaker-independent (moral) reality, are best interpreted as expressions of attitudes, not statements of fact. (See, for instance, emotivism.)
Substantive claim: People aren’t talking about a speaker-independent reality, therefore(?) it doesn’t exist.
Error theory:
Linguistic claim: Moral claims purport to describe a speaker-independent (moral) reality and can have truth-values (though always false according to error theory’s substantive claim).
Substantive claim: The concept of a speaker-independent (moral) reality doesn’t make sense. Depending on how one looks at it,[9] all (first order) moral claims are either false because they are about something meaningless, or they are meaningful but doomed to be false. Either way, people who make first-order moral claims are committing an error.
Non-objectivism:
Linguistic claim: Moral claims (at least some of them)[10] are best interpreted according to fixed evaluation criteria. For instance, moral claims could be interpreted as claims about what a speaker would come to value if they were aware of all relevant information. But the whole point behind non-objectivism is that there’s not just one possible interpretation. Alternatively, another non-objectivist evaluation criterion might be to interpret moral claims as claims about what conforms with a specific moral theory, such as consequentialism or contractualism.
Substantive claim: Even anti-realists can adopt the notion of “moral facts,” provided that we think of them as facts about a non-objective (speaker-dependent) reality, instead of facts about a speaker-independent (objective) one.
Though these terms are primarily used in the moral context, they can be applied analogously to other domains. For instance, a non-cognitivist about aesthetics might say that when someone says, “This painting is beautiful,” they are not making a proposition that can be true or false, but merely communicating a positive attitude toward the painting.
Non-cognitivism, error theory, and non-objectivism are largely compatible with one another— even though they aren’t always presented this way. Their respective linguistic claims are conceptually distinct (and therefore separable) from the substantive claims (e.g., Kahane, 2013). Moreover, we can’t assume that folk discourse always fulfills the same function (so non-cognitivism can be an accurate interpretation for some moral claims, but false for others).
The version of anti-realism I’m arguing for in this sequence is a blend of error theory and non-objectivism. It seems to me that any anti-realist has to endorse error theory (in some sense at least) because realists exist, and it would be uncharitable not to interpret their claims in the realist fashion. However, the non-objectivist perspective seems importantly correct as well:
When someone makes claims like “double-dipping is wrong” or “sharks are most likely conscious,” it would be pedantic to point out that because realism about morality (or consciousness) may be wrong, the person is making an error. Even if those realisms are wrong, the statements still convey information. Charitably interpreted, “double-dipping is wrong” conveys the belief that the activity violates pro-social norms that Decent People would upon reflection adopt. Similarly, “sharks are most likely conscious” conveys the empirical belief that shark cognition is relevantly similar to the cognition of other animals commonly thought of as conscious.
(That said, in contexts where precision matters a lot, anti-realists may want to choose their words carefully and make explicit—to both themselves and others—the evaluation criteria they have in mind.)
Loosely applied non-objectivism works passably as a (re)interpretation of many instances of folk discourse precisely because anti-realism doesn’t mean “anything goes.” The domains under scrutiny do have some structure! And by specifying evaluation criteria that interest us, we can highlight and investigate various aspects of that structure.
Yes, there really is morality, and we can locate it in reality—either as a set of facts about the well-being of conscious creatures, or as a set of facts about what an ideally rational and perfectly informed agent would prefer, or as some other set of natural facts
But in another sense, pluralistic moral reductionism is ‘anti-realist’. It suggests that there is no One True Theory of Morality.
The different “sets of facts” describe various ways in which morality has structure; each set of facts is highlighted by different evaluation criteria.
We can apply the perspective of pluralistic moral reductionism analogously to other domains. Pluralistic reductionism is an accurate description for the brand of anti-realism I’m arguing for in this sequence.[11]
Why realists and anti-realists disagree
After explaining how I think realists and anti-realists disagree, I also want to address why they disagree.
Verbal disputes and the method of elimination
In the paper Verbal Disputes, David Chalmers (2011) introduced what he called the method of elimination (Yudkowsky’s “rationalist taboo” is a version of the same procedure) to help determine whether a disagreement is substantive or merely verbal. Unlike substantive disputes, merely verbal disagreements only arise because the discussion participants fail to notice that they have a different vocabulary. A notorious example is two people discussing whether a tree falling in the forest makes a sound if no one is there to hear it.
The method of elimination involves restating a given claim in neutral terminology. For instance, in the discussion about the falling tree, the word ‘sound’ can be “eliminated” (or “tabooed”) to determine whether or not the disagreement is substantive. (The apparent dispute disappears once it is discovered that one party would replace “sound” with “air vibrations,” whereas the other person would have replaced it with “auditory processing.”)
Applied to notorious concepts such as “moral,” “right” or “conscious,” the method of elimination forces us to make apparent our standards of evaluation. While anti-realists would typically consider the provision of such standards a requirement for using language properly, this is where the realists take a different route. Because realists are interested in only the (speaker-independently) right standards, rephrasing the claim in question risks getting it wrong.
Using the example of the moral term “ought,” Chalmers describes this situation as follows:
It is plausible that once all moral terms are gone, no disagreement can be stated. We might agree on all the nonmoral properties of the relevant actions, but still disagree on whether it is right. In the case of [previously analyzed disputes about the concepts] ‘semantics’, ‘physicalism’, and so on, this situation suggested a verbal dispute. Should we likewise diagnose a verbal dispute here? Intuitively, the answer is no. For all we have said, moral disputes are substantive disputes. Instead, we have simply exhausted the relevant vocabulary. It appears that at a certain point (perhaps once we have fixed on the appropriate moral “ought”), we have reached bedrock: a substantive dispute involving a concept so basic that there is no hope of clarifying the dispute in more basic terms.
The judgment about when and where “bedrock has been hit” is left to the discretion of the speaker:
At a certain point, when the method of elimination is applied, a proponent may say “That’s bedrock”. Here, the thought is that we have reached a point where only cognate expressions can be used to state the issue, and where there is no hope of finding a relevant disagreement at an “underlying” level. Further applications of the method will just lead to wheel-spinning, and ultimately to vocabulary exhaustion.
Bedrock concepts
Based on the notion of “hitting bedrock,” Chalmers coined the term bedrock concepts. Bedrock concepts are thought to sit at the hearts of domains like consciousness, morality, and epistemology. They have in common that any attempts to (re)define them with terminology from outside their particular domain will trigger an intuition that this rephrasing endeavor misses the point.[12]
At most, we can “define” bedrock concepts in a circular way (e.g., “Doing good is about performing the morally right actions”), or by ostension. For example, Derek Parfit defined the concept of irreducibly normative reasons by pointing at examples (Parfit, 2011a):
Like some other fundamental concepts, such as those involved in our thoughts about time, consciousness, and possibility, the concept of a reason is indefinable in the sense that it cannot be helpfully explained merely by using words. We must explain such concepts in a different way, by getting people to think thoughts that use these concepts. One example is the thought that we always have a reason to want to avoid being in agony.
The need for bedrock concepts
The disagreement between realists and anti-realists is primarily a disagreement about bedrock concepts. Different realisms[13] are built around bedrock concepts. While anti-realists are generally skeptical of bedrock concepts, realists cannot imagine doing (good) philosophy without them.
That said, even the realists would typically agree that bedrock concepts are weird. Without them, philosophy would be simpler—we could follow Wittgenstein’s (1999(1921)) maxim, “What can be said at all can be said clearly.” By incorporating bedrock concepts into our philosophical repertoire, many things change. Laying out what constitutes philosophical progress then becomes a bedrock concept as well (“yes, doing philosophy is related to figuring out what’s right—but how do we do that?”). Arguably, that’s a steep price to pay.
For some bedrock concepts such as “art” or “identity” in “the ship of Theseus,” the case for incorporating them into our philosophical repertoire is weak. Even though we may have the essentialist intuition that concept like “art” or “identity” have a single correct meaning, it seems (to me at least) that if we were to drop this intuition, nothing of value would be lost. It seems to me that we don’t hold those intuitions for good reasons, but rather because they seem to derive from a lack of introspective discernment about how our minds form categories.
However, not all bedrock concepts can be rejected this easily. The strongest argument for realism is that when it comes to adopting some bedrock concepts, we may not have a choice. Below, I discuss this in more detail.
Candidate bedrock concepts ordered by their perceived degree of philosophical indispensability
I’ve come up with a somewhat subjective, very crude ranking of bedrock concepts ordered by how easy it seems to me to do philosophy without them. The lower the tier, the closer it is to the proverbial bedrock (if there is such a thing). So, while I think that the candidate bedrock concepts in Tier 4 and Tier 3 can be rejected safely without much discussion, I find the debate between realists and anti-realists more interesting for Tier 2, Tier 1 and Tier 0 bedrock concepts.
Plato’s idealism about forms for human-made objects (e.g., this page lists “table” and “house” as examples of things about which Plato was a realist)
I think the case for bedrock concepts is the weakest when it comes to human-made objects. For any human-made object, we can ask humans why they made it—the answer will describe the speaker-dependent purpose of the object. There’s not much more to say. We may still have the intuition that “Is it really a table?” is a well-defined question, but absent further evaluation criteria about what we mean by this, the question is clearly underdetermined. (We can also think of language itself as a human-made object: words get their meaning from how we use them. If context is lacking, so is meaning.)
Before Darwin, it would have seemed more intuitive to me to be an essentialist about animal species.[14] Some philosophers have also invoked realism about the properties of the source objects for our sensory perception (for instance, realism about colors, sounds, beauty, etc.). While I understand the intuition behind (e.g.) colors existing independently of color perception, this view seems untenable to me because of the Darwinian arguments for anti-realism about sense appearances. With the advance of our scientific understanding, the reductionist worldview became intellectually satisfying despite our anti-reductionist intuitions.
Metaphilosophical realism (realism about the proper philosophical method)
On the one hand, it seems that our intuitions about normativity as a bedrock concept are not too different from the intuitions in favor of Tier 3 bedrock concepts. On the other hand, normative realists have one additional consideration to go by. Namely, they can argue that in addition to normative anti-realism being (perhaps) counterintuitive, it is also self-defeating. To give an analogy, consider imagining myself as a Boltzmann brain. Though logically consistent, the prospect that I’d cease to exist in the next flicker of a moment would render everything I’m about to do or think about pointless—my actions wouldn’t carry any relevance if the scenario were true. Arguably, there’s a sense in which imagining normative anti-realism is a bit like that: Though logically consistent, it’s pointless according to existentialist sense of things mattering to us. Because the alternative appears to be self-defeating,[17] it has been argued that we should adopt normative realism.
Tier 1: Qualia (“irreducible sensations”)
Examples:
Qualia realism
I’m categorizing qualia as Tier 1 candidate bedrock concepts because the case for incorporating them into our philosophical repertoire is particularly strong (at least as far as different types of realism are concerned). Tier 3 and Tier 1 differ as follows: If a person looks at tree leaves, the Tier 3 candidate bedrock concept is the “greenness” of the leaves—a claimed, objective property that some objects (e.g., leaves) either have or lack. By contrast, the Tier 1 candidate bedrock concept in that scene would be the quale “green”— the “what-it’s-likeness” of seeing green as an onlooker.
Being mistaken about an object’s true nature seems conceivable, but it seems impossible to make the same sort of mistake about the content of one’s immediate moment of consciousness. If I believe that, right this moment, something in my visual field looks green, I can conclude that I’m seeing green with certainty. Because we experience consciousness directly and immediately, the usual anti-realist arguments don’t seem to apply. Arguably, consciousness anti-realism is inconceivable because it contradicts our sensory experience.[18]
While I listed qualia realism as the sole example of Tier 1 realism, it’s worth pointing out that other types of realism could use qualia realism as their backbone. For instance, philosophers who endorse a theory of well-being according to which conscious experience is the only thing of (dis)value might argue that moral realism can be derived from realism about intrinsically morally valuable or disvaluable qualia (e.g., Hewitt, 2008). While I don’t intend to take a stance on qualia realism per se in this sequence, I’m planning to write a post (or post section) on arguments that go from qualia realism to realism about morality.
Tier 0: Something irreducible
Examples:
Existence realism
At the lowest tier and closest to ultimate philosophical bedrock (if there is such a thing), we find reality itself. Even if qualia aren’t quite what we think they are, and even if there’s no speaker-independently correct way to reason, we might want to assume—if only for the sake of preserving our sanity—that there’s at least something that goes beyond subjective standards. (Existence anti-realists may point out that things are not as they appear, but whether or not they have lost their minds is for readers to determine.)
Summary and conclusion
Some types of realism seem to be backed by nothing more than widely distrusted essentialist intuitions. However, normative realism, qualia realism, and existence realism have further considerations for their support. For instance, normative anti-realism is something we arguably wouldn’t care about if it was true, and qualia anti-realism is something we arguably can’t imagine to be true.
Realists and anti-realists may disagree about the validity of those considerations, and the degree to which they may force us to adopt the costs of incorporating bedrock concepts (e.g., irreducible normativity or the notion of qualia) into our philosophical repertoire. In addition, realists and anti-realists may disagree about the severity of those costs, perhaps because they think differently about the workability and palatability of an anti-realist philosophical framework.
Outlook on future posts
In the next post in this sequence (“3. Against Irreducible Normativity”), I will argue that normative bedrock concepts (especially moral ones, but the arguments apply generally) are best to be rejected. I will argue that (1) there’s a strong sense in which we can already tell that they make clear-minded philosophical progress impossible, and (2) that the alternative—normative anti-realism—can be richer and more palatable than one might at first think.
In posts 4 and 5, I will then address the possibility of a wager for irreducible normativity, as well as its counterintuitive consequences. In post 6, I will discuss arguments that also apply to naturalist moral realism. In particular, the post discusses a challenge for the concept of “moral uncertainty.” In post 7, I plan to address versions of moral realism built on top of qualia realism. Finally, in posts 8 and 9, I sketch I reason about morality given that I suspect moral realism to be false.
Appendix: Realisms are correlated
According to the 2010 PhilPapers survey (Bourget & Chalmers, 2014), belief in different realist positions strongly correlates (as does belief in various types of anti-realism).
“Moral realism” (note that this also includes naturalist versions of moral realism) correlates with the following views:
Aesthetic value: objective (0.411)
Abstract objects: Platonism (0.335)
Laws of nature: non-Humean (0.329)
Science: scientific realism (0.32)
Libertarianism about free will (0.24)
Different types of realism also correlate with one another:
“Mind: non-physicalism” correlates with “Free will: libertarianism” (0.386)
“Aesthetic value: objective” correlates with “Laws of nature: non-Humean” (0.228)
“Time: A-theory” correlates with “Mind: non-physicalism” (0.23) and Aesthetic value: objective (0.164)
To get a sense of the strength of these correlations relative to other comparisons, I’m listing the five most strongly correlated answers in the entire study:
“Moral judgment: cognitivism” correlates with “Meta-ethics: moral realism” (0.562)
“Metaphilosophy: non-naturalism” correlates with “Mind: non-physicalism” (0.497)
“Analytic-synthetic distinction: yes” correlates with “A priori knowledge: yes” (0.467)
“Meta-ethics: moral realism” correlates with “Aesthetic value: objective” (0.411)
“Mind: physicalism” correlates with “God: atheism” (0.393)
Though far from perfect, these correlations indicate that different realist (or anti-realist) positions have things in common. Correspondingly, we should expect that there are underlying factors that make some philosophers more predisposed toward realist (or anti-realist) positions.
Acknowledgements
I’m thankful for comments by Jesse Clifton and Sofia Davis-Fogel. Many of the ideas discussed in this article were things I first encountered when I read Brian Tomasik’s essay The many fallacies of dualism (Tomasik, 2013).
Bourget, D. and D. Chalmers. (2014). What do philosophers believe? Philosophical Studies, 170(3):465–500.
Boyd, R. (1988). How to be a moral realist. In G. Sayre-McCord (ed.), Essays on Moral Realism. Ithaca: Cornell University Press. 181–228.
Brogaard, B. (2010). Color. In: Oxford Annotated Bibliographies Online. <www(.)oxfordbibliographies(.)com/view/document/obo-9780195396577/obo-9780195396577-0021.xml>.
Case, S. (2016). Normative Pluralism Worthy of the Name Is False. Journal of Ethics and Social Philosophy, 11(1):1–20.
Chalmers, D. (2011). Verbal Disputes. Philosophical Review, 120 (4):515–566.
Cuneo, T. (2007). The Normative Web: An Argument for Moral Realism. Oxford: Oxford University Press.
Hewitt, S. (2008). Normative Qualia and Robust Moral Realism. PhD thesis. New York University.
Hume, D. ((1990)1757). Dialogues Concerning Natural Religion, Bell, M. (ed.). London: Penguin.
Kahane, G. (2013). Must Metaethical Realism Make a Semantic Claim? Journal of Moral Philosophy, 10(2):148-178.
Parfit, D. (1984). Reasons and Persons. Oxford: Oxford University Press.
Parfit, D. (2011a). On What Matters, Volume I. Oxford: Oxford University Press.
Ruse, M. (2010). The biological sciences can act as a ground for ethics. In Ayala, F. and R. Harp (eds.), Contemporary Debates in Philosophy of Biology. New York: Wiley-Blackwell. 297–315.
Singer, P. (1973). The Triviality of the Debate Over ‘Is-Ought’ and the Definition of ‘Moral.’ American Philosophical Quarterly, 10(1):51–56.
Tomasik, B. (2013). The many fallacies of Dualism. reducing-suffering(.)org. <reducing-suffering(.)org/the-many-fallacies-of-dualism/>.
Williams, B. (1979). Internal and External Reasons. In R. Harrison (ed.), Rational Action. Cambridge University Press. 101–113.
Wittgenstein, L. (1999(1921)). Tractatus logico-philosophicus. Frankfurt am Main: Suhrkamp.
For instance, the moral realist Terence Cuneo (2007) has argued that anti-realism about morality also entails anti-realism about all other types of irreducible normativity. Similarly, the anti-realist Michael Ruse (2010) has argued that because of what we know about the evolutionary origins of our moral intuitions, the case for moral realism is no stronger than the case for aesthetic realism.
Note that I’m grouping the above domains only because realism about them is an appealing sentiment at least superficially. As I will argue in later sections, I think that there are vast differences in plausibility between various types of realism.
There are interesting exceptions to this, though usually they get categorized as (lukewarm) theism: Instead of abandoning the concept ‘God,’ certain theists seem to water it down until it makes almost no predictions about reality.
I’m probably oversimplifying this, but according to the Stanford Encyclopedia of Philosophy article on moral anti-realism, the difference between speaker-independent facts and speaker-dependent ones is tricky to explain. They write:
So many debates in philosophy revolve around the issue of objectivity versus non-objectivity that one may be forgiven for assuming that someone somewhere understands this distinction. There certainly exists a widespread intuitive imagery associated with the duality that is sufficiently vivid to motivate heartfelt philosophical commitments, but, once approached directly, the distinction nevertheless proves extremely difficult to nail down. It is likely that part of what is causing confusion is that there are a number of non-equivalent ways of drawing the distinction, some of which are better suited to certain subject areas than others.
For normative contexts, the distinction between speaker-independent and speaker-dependent facts roughly corresponds to Immanuel Kant’s distinction between categorical and hypothetical imperatives. The same difference is also described with what we can call “reasons terminology.” Realists believe there are irreducibly normative reasons (or so-called “reasons proper”), whereas anti-realists only think in terms of instrumental reasons. Williams (1979) provides a discussion of (and arguments for) “reasons anti-realism.”
This doesn’t mean that morality can’t be multifaceted. Derek Parfit, who was a moral realist, argued that the different normative-ethical theories of consequentialism, contractualism, and deontology are ways of “climbing the same mountain” from three different sides (Parfit, 2011a, p. 419).
Consciousness anti-realists are sometimes associated with statements like “qualia don’t exist.” I want to make clear that I interpret the term “hidden qualia” in a neutral sense, compatible with both realism and anti-realism about consciousness. Consciousness anti-realists think that whether or not hidden qualia “count” depends on where/how we want to draw the boundaries.
This footnote is only about semantics, but I don’t understand the notion of “meaningful, but necessarily false.” The way I think about it, claims about a speaker-independent moral reality are always false because the concept of such a reality makes no sense in the first place.
I’m not entirely sure what the non-objectivists’ linguistic claim is. Some non-objectivists may indeed want to commit themselves to the view that all moral discourse is non-objectivist. However, others may think about it in terms of offering a more constructive alternative to error theory. It is this second notion of non-objectivism that I’m interested in.
Muehlhauser left open whether we should think of pluralistic moral reductionism as realism or anti-realism. By contrast, according to definitional choices I made and explained in my previous post, I’d call it “moral anti-realism” myself.
The full picture is a bit more complicated because not all types of realism are built around bedrock concepts. Generally, realism about a domain means that there’s room for only one interpretation. Bedrock concepts at the heart of the domain would make sure of that—which is why most realisms (so-called “non-naturalist realisms”) are built around bedrock concepts. However, philosophers have also proposed the idea of naturalist realism. For instance, I might identify as a naturalist realist about electrons. If I understand what a single electron is, I’ll also understand what other electrons are like. There’s no open question about where to draw the boundary because as far as we know, something either is an electron, or it isn’t. Facts about electrons are reducible to physical facts. As I’ve mentioned in my previous post, for domains such as morality, naturalist versions of realism are sometimes criticized for being “too watered down” compared to non-naturalist realism. I described the criteria that could convince me of a naturalist version of moral realism in my previous post. In short, I consider naturalist realism about a domain true if the domain’s structure, its “features in the option space,” are sufficiently salient and unambiguous that there exists little to no disagreement about evaluation criteria. That is, no matter how they look at it, philosophically sophisticated reasoners will come to interpret the domain the same ways, provided that they can reason about it under idealized conditions. I devote later posts this sequence to discussing why I think those criteria for naturalist realism can’t be met for the domain of morality.
There were philosophers who questioned this view even before Darwin. For instance, Hume’s Dialogues Concerning Natural Religion (Hume, 1757), especially chapters 7 and 8, contain arguments against the need of a creative intelligence to explain nature’s design and orderliness. Hume even anticipated some of the key features of natural selection.
These examples make up different flavors of normative realism. Some philosophers argue that there are distinct kinds of normativity, while others maintain that there’s only one kind (Case, 2016).
A distinction that’s often made is between self-oriented (“prudential”) reasons for action and other-regarding (“moral”) ones. For reasons that will become apparent later on in this sequence, I consider this distinction to be overrated and potentially misleading. Therefore, I will be using “moral realism” in an unusually broad sense, to refer to realism about at least some irreducibly normative reasons for action (whether self-oriented reasons, other-regarding reasons, or both).
In the post Realism and Rationality, Ben Garfinkel expressed a related sentiment with respect to the reasons we may have for believing something:
If (normative) anti-realism is true, then it can’t also be true that we should believe that anti-realism is true. Belief in anti-realism seems to undermine itself.
Anti-realists about consciousness would argue that appearances are deceptive, and that the realists are smuggling in too many connotations. Consider the following dialogue (adapted from Brian Tomasik’s writings on consciousness, where he has a similar dialogue):
Alice: “What do you mean by ‘consciousness’?”
Bob: “What I’m experiencing this very moment: the way it feels like something to be me.”
Alice: “Okay, let’s note this down: Bob’s brain state at time 14:04:28 EST. Is that all you mean by consciousness?”
Bob: “No! I didn’t mean to point to just this one instance. You got right that this particular moment was an instance of consciousness. But I also meant this moment right now, and all moments like it, present and future!”
Alice: “Okay, I’m adding that down as well: Bob’s brain state at time 14:04:38 EST, and all configurations of matter exactly like it.”
Bob: “You’re still not getting it! I’m not just talking about the specific instances, I’m talking about what those instances have in common: the extremely salient feature that it feels like something to be me. I’m puzzled why you pretend that you don’t understand it.”
Alice: “I think that’s an illusion.”
Bob: “What do you mean? Of course that’s no illusion! The fact that when I’m awake, it feels like something to be me is the most certain fact there is.”
Alice: “Sure, but the only thing you can be certain of is that the state your brain had just that second ago was what you refer to as ‘feeling like something.’ I’m happy to grant you that—it’s what I already wrote down. What I’m denying is that there is a clear way to extract from single examples a universal concept of ‘consciousness’ that determines, once and for all, which aspects of reality make up consciousness, and which ones don’t.”
Bob: “Okay, I think I see what you’re doing. You’re saying I’m not allowed to be completely certain that ‘it feels like something’ picks out a property that brain states of all kinds either have or lack? That I can only ever point to the brain state I’m currently in and say this or that about it, but that if we were to try and extrapolate what I would say about an entirely novel brain state, we can conceivably doubt whether my initial examples pick out a large and well-specified cluster in the space of all possible brain configurations in the same way as ‘This substance is what I mean by water’ manages to successfully pick out a vast space of configurations of H2O molecules in all kinds of shapes and forms?”
Alice: “Exactly. Why did you assume that the feeling you call ‘being conscious’ picks out brain states in that way? I mean, I can see that it’s natural to also think of other people as conscious, but why did you think that once you go to examples where some features remain similar while other features start to vary a lot or become more distant, the examples would continue to be clear cut?”
Bob: “It seems like that’s part of my feeling of being conscious! I can’t imagine there being edge cases.”
Alice: “Okay, but that’s a different claim now. Initially, you said the fact that it feels like something to be you is the most certain fact there is. Now you moved from that statement to a statement about edge cases. I don’t see how the latter statement is any different from people claiming that it’s inconceivable to them that there may not be a clear fact of the matter whether a certain action is moral or immoral, or whether the expression ‘Theseus’s ship’ is to be interpreted this way or that way.”
Why Realists and Anti-Realists Disagree
Last updated: 20/1/2022
This is the second post in my sequence on moral anti-realism (see my previous post). This second post should work perfectly when read as a standalone piece.
My motivation to write this post
For this post, I set out to describe and analyze what arguably constitutes the most fundamental disagreement in philosophy: realism versus anti-realism, not only in metaethics but in general. In one form or another, this disagreement also shows up in the philosophy of mind, personal identity, aesthetics, epistemology, the philosophy of physics, and so on. Belief in different types of realism is correlated (see the appendix), so I see significant benefits to also discussing other types of realism. For instance, moral philosophers on both sides of the realism/anti-realism divide have drawn comparisons from different domains in support of their positions.[1] To address those comparisons, prior discussion of those other domains is helpful. However, I tried to structure this post to be more than a (second) introduction. Even though my aim is to neutrally describe the core differences between realism and anti-realism, in doing so I will already present some of my main arguments for anti-realism. My most persuasive “argument” against (moral) realism isn’t any single knockdown objection, but rather my overall impression that when we go from realism to anti-realism, we don’t have to give up anything worth wanting. I expect (moral) realists to disagree with that sentiment, in part because I could imagine that many may not have been motivated to explore the option space for moral reasoning under anti-realism (“to make anti-realism work”). I wrote this post to make what I perceive to be underappreciated points about how realism comes with some surprisingly non-trivial claims, and how anti-realism doesn’t have to mean throwing up one’s hands saying “anything goes.”
Summary
We tend to have the feeling that disagreements on what’s “moral,” “conscious,” “(epistemically) right,” and so on, go beyond semantics, or are unambiguous. The discussion between realists and anti-realists is about the degree to which this feeling is accurate.
Anti-realists don’t deny that domains in question (morality, philosophy of mind, epistemology, etc.) have some “structure.” Instead, they disagree with the realists on how to interpret and relate to that structure.
While the typical realist believes in the notion of an unambiguous speaker-independent (“objective”) reality, the typical anti-realist would consider this a confusing concept because we have to interpret reality, and interpretations about a domain can only be formed on top of subjective evaluation criteria.
There’s a category of realisms built around what David Chalmers has called “bedrock concepts:” concepts like “moral” or “conscious” or “(epistemically) right.” Bedrock concepts are assumed to be meaningful, but also inexplicable—in the sense that we cannot explain them in neutral terminology. Anti-realists are generally skeptical of bedrock concepts, whereas many realists cannot imagine doing philosophy without them.
For many types of realism, the only arguments in favor are appeals to widely distrusted essentialist intuitions. (There are also so-called naturalist versions of realism that don’t rely on bedrock concepts; I won’t focus on them in this post.)
Not all types of realism rest on thin assumptions. Arguably, bedrock concepts related to (various kinds of) normativity, as well as bedrock concepts related to consciousness (“qualia”), are backed by stronger arguments. Realists about normativity argue that normative anti-realism is self-defeating, and realists about qualia argue that qualia anti-realism is impossible.
Points of agreement
Discussions on the following domains all have something in common:
The presence or absence of consciousness
Morally good or bad actions
Beautiful or ugly art
The preservation (or lack thereof) of personal identity (for instance after going through a teletransporter)
Epistemically right or wrong reasoning
The truth or falsity of scientific theories
Discussing the topics above tends to be accompanied by the impression that we are accomplishing something substantive, i.e., that we can make unidirectional intellectual progress and that corresponding disagreements can go beyond personal opinion. (The strength of this intuition may vary depending on the person or the field in question.) Very crudely, “realism versus anti-realism” is the philosophical discussion about the degree to which that impression is appropriate.[2]
Going by connotations alone, we might at first think that realism means that a domain in question is real, whereas anti-realism implies that it’s something other than real (e.g., that it’s merely imagined). Although accurate in a very loose sense, this interpretation is misleading.
It’s misleading in that it suggests the discussion is analogous to conversations about the existence of God. But philosophers call this “theism versus atheism,” instead of “realism versus anti-realism about God.” There’s a cliché about atheists continuing to talk about God even after they quit believing. Still, this cliché only refers to “talking about God” to convince others (or to show off one’s reasoning)—which is different from continuing to “talk about God” in the interest of figuring out God’s true nature. Typically, when someone stops believing in God, they also stop talking as though God exists. As far as private purposes are concerned, atheists don’t generally refine their concept of God; they abandon it.[3]
Going from realism to anti-realism works differently. (Of course, not all anti-realists think alike; and many people may change their opinions in the other direction, too.) Rejecting realism for a domain neither entails erasing the substance of that domain, nor (necessarily) its relevance. Anti-realists will generally agree that the domain has some relevance, some “structure.” For instance, anti-realists about morality may still be interested in (and moved by)[4] moral arguments, anti-realists about epistemic facts may still care about how to reason, and anti-realists about consciousness may still be interested in the latest expert survey on invertebrate sentience.
Points of disagreement
There’s agreement on the domain in question having some structure. The disagreement comes down to the correct way to relate to that structure. While the realists believe that there are speaker-independent facts directly about the domain in question, the anti-realists deny this.
The existence of “speaker-independent facts” would mean that the domain in question lends itself to only one interpretation: the interpretation that holds true because it’s true.[5] Another way to express this sentiment is that there’s an interpretation that exists “prior to thought” (Boyd, 1998).
Not all anti-realists think alike, but those who endorse anti-realism about most (or even all) the contested domains are especially likely to reject the notion of speaker-independent standards altogether. This rejection results in a philosophical framework grounded in merely speaker-dependent standards.[6] In such a framework, we need to specify our criteria of evaluation before we can meaningfully discuss what’s true or false about a domain. Once those evaluation criteria (“standards”) are stipulated, then we can evaluate claims. In this sense, the claims can be true or false, but only in relation to subjectively (“speaker-dependently”) chosen standards.
Figure 1. Duck-rabbit illusion (public domain)
From an anti-realist perspective, there’s a degree of ambiguity to everything. Consider the duck-rabbit illusion (above) for illustration. Although there are two different ways to interpret the figure, we can still distinguish between true and false things to say individually about the duck or the rabbit. For instance, choosing to focus on the “duck interpretation,” it would be wrong to claim that the feathery animal has its beak wholly closed. Anti-realism about the duck-rabbit illusion doesn’t mean that we can’t make valid claims about the different animal interpretations; it only means that there’s no point in asking which of the two animal interpretations is the right one. Neither the duck interpretation nor the rabbit interpretation adds anything further to the underlying patterns of black and white.
Some people equate anti-realism with the nihilistic sentiment “all interpretations are wrong.” However, the way I think about it, anti-realism is best summarized as follows:
We can apply this perspective to domains like morality, consciousness, or epistemology. To give an example, consider (again) anti-realism about morality. Looking at some of the longstanding disagreements in normative ethics, it can appear as though different sides are talking past one another, answering different questions. For instance, when I was involved in animal advocacy, I noticed that some moral arguments worked well for some people, while other people seemed to consider them beside the point. Having always looked at morality from a consequentialist lens, it felt only natural to me to frame my inquiries into normative ethics as inquiries about what it means to make the world a better place (for all morally relevant beings). However, I discovered that this framing was far from self-evident to everyone. In particular, many people think about morality more in terms of norms or social contracts that people would agree on in order to co-exist peacefully and productively in society. This perspective certainly captures some aspects of the folk concept “morality,” but it leaves out other aspects. (Of course, the same can be said for consequentialism.) For a long time, I thought that at least one side had to be wrong, that at least one party doesn’t get what morality is about. I had a viscerally intense sentiment that “either morality includes caring intrinsically about others, or it doesn’t.”[7]
By thinking of it as a dichotomy, I had committed myself to a metaethical assumption: moral realism. Moral anti-realists can point out that I should have at least considered the possibility that some discussion participants may be talking past one another, “playing different games.” To make the analogy more explicit:
Duck-rabbit illusion: There’s a sense in which the duck is “real.” But we could also take a different perspective, in which case we’d see a rabbit. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d see in the figure is shapes of black and white.
Normative ethics: There’s a sense in which consequentialist obligations to avoid purchasing meat from factory-farmed animals are “real.” But we could also take a different perspective (according to which morality is about hypothetical contracts between people), in which case we’d see no obligations toward animals. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we see in morality is people making up various justifications for things that (in one way or another) appeal to them.
The analogy becomes increasingly more strained for some of the other domains that are commonly discussed, but I think the key aspects still apply:
Consciousness: There’s a sense in which bees are “conscious.” But we could also take a different perspective (for instance, one that highlights higher-order cognition, drawing a conceptual boundary to exclude hidden qualia),[8] in which case we’d (probably?) see bees as not conscious. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d ever see is various cognition-like processes of different complexity, with differing degrees of intentionality, self-reference, and so on.
Personal identity: There’s a sense in which I’m still “the same person” after I exit a teletransporter. But we could also take a different perspective (for instance, one that focuses primarily on physical connectedness), in which case we’d view the person exiting as a different person from the one who entered. And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d ever see in personal identity thought experiments are varying degrees of physical or psychological similarities and temporal differences.
Aesthetics: There’s a sense in which all exhibits in modern art galleries are “art.” But we could also take a different perspective (for instance, one that focuses on whether something is pleasant to look at or whether it took special skills to produce), in which case we’d only view a subset of things in modern art galleries as “art.” And if we wanted to play stubborn, we could adopt a third perspective yet and claim that all we’d ever see is patterns, shapes, and forms that leave various impressions on different onlookers.
These are strained analogies in many ways, but the key point is that there seems to be an ever-present element of arbitrariness when it comes to the aspects of reality to which we want to draw attention. According to my anti-realist perspective, reality simply is, but interpretations always add something. Deep down, all interpretations are arbitrary and we can always take on the “stubborn” perspective to say that there’s not even a question that needs to be answered. Still, we’ll find that we often care about how to interpret a domain and where to draw categories, and that we bring certain expectations already to the table. It is that second sense in which anti-realism is not arbitrary.
(Re)interpreting realist discourse
Realists about some of the above domains might object that the proposed analysis misses the point! (At least when it comes to the domain that they are realists about.) Unlike in the duck-rabbit illusion, the way we talk about normative ethics, consciousness, and epistemology seemingly postulates the existence of some speaker-independent reality. At the very least, anti-realists should be able to explain what (if anything) they think might be going wrong with that.
In the context of moral anti-realism, there are distinctions between three different ways of rejecting the idea that our discourse describes actual features of a speaker-independent reality. This terminology can be confusing because it packs together linguistic claims about the nature of moral discourse and substantive claims about whether or not there is a speaker-independent reality. Very roughly, these three ways are:
Non-cognitivism:
Linguistic claim: Moral claims, even though they may appear as though they are about a speaker-independent (moral) reality, are best interpreted as expressions of attitudes, not statements of fact. (See, for instance, emotivism.)
Substantive claim: People aren’t talking about a speaker-independent reality, therefore(?) it doesn’t exist.
Error theory:
Linguistic claim: Moral claims purport to describe a speaker-independent (moral) reality and can have truth-values (though always false according to error theory’s substantive claim).
Substantive claim: The concept of a speaker-independent (moral) reality doesn’t make sense. Depending on how one looks at it,[9] all (first order) moral claims are either false because they are about something meaningless, or they are meaningful but doomed to be false. Either way, people who make first-order moral claims are committing an error.
Non-objectivism:
Linguistic claim: Moral claims (at least some of them)[10] are best interpreted according to fixed evaluation criteria. For instance, moral claims could be interpreted as claims about what a speaker would come to value if they were aware of all relevant information. But the whole point behind non-objectivism is that there’s not just one possible interpretation. Alternatively, another non-objectivist evaluation criterion might be to interpret moral claims as claims about what conforms with a specific moral theory, such as consequentialism or contractualism.
Substantive claim: Even anti-realists can adopt the notion of “moral facts,” provided that we think of them as facts about a non-objective (speaker-dependent) reality, instead of facts about a speaker-independent (objective) one.
Though these terms are primarily used in the moral context, they can be applied analogously to other domains. For instance, a non-cognitivist about aesthetics might say that when someone says, “This painting is beautiful,” they are not making a proposition that can be true or false, but merely communicating a positive attitude toward the painting.
Non-cognitivism, error theory, and non-objectivism are largely compatible with one another— even though they aren’t always presented this way. Their respective linguistic claims are conceptually distinct (and therefore separable) from the substantive claims (e.g., Kahane, 2013). Moreover, we can’t assume that folk discourse always fulfills the same function (so non-cognitivism can be an accurate interpretation for some moral claims, but false for others).
The version of anti-realism I’m arguing for in this sequence is a blend of error theory and non-objectivism. It seems to me that any anti-realist has to endorse error theory (in some sense at least) because realists exist, and it would be uncharitable not to interpret their claims in the realist fashion. However, the non-objectivist perspective seems importantly correct as well:
When someone makes claims like “double-dipping is wrong” or “sharks are most likely conscious,” it would be pedantic to point out that because realism about morality (or consciousness) may be wrong, the person is making an error. Even if those realisms are wrong, the statements still convey information. Charitably interpreted, “double-dipping is wrong” conveys the belief that the activity violates pro-social norms that Decent People would upon reflection adopt. Similarly, “sharks are most likely conscious” conveys the empirical belief that shark cognition is relevantly similar to the cognition of other animals commonly thought of as conscious.
(That said, in contexts where precision matters a lot, anti-realists may want to choose their words carefully and make explicit—to both themselves and others—the evaluation criteria they have in mind.)
Loosely applied non-objectivism works passably as a (re)interpretation of many instances of folk discourse precisely because anti-realism doesn’t mean “anything goes.” The domains under scrutiny do have some structure! And by specifying evaluation criteria that interest us, we can highlight and investigate various aspects of that structure.
Luke Muehlhauser’s sequence on No-Nonsense Metaethics built toward this same point for the domain of morality, spelled out in the post Pluralistic Moral Reductionism:
The different “sets of facts” describe various ways in which morality has structure; each set of facts is highlighted by different evaluation criteria.
We can apply the perspective of pluralistic moral reductionism analogously to other domains. Pluralistic reductionism is an accurate description for the brand of anti-realism I’m arguing for in this sequence.[11]
Why realists and anti-realists disagree
After explaining how I think realists and anti-realists disagree, I also want to address why they disagree.
Verbal disputes and the method of elimination
In the paper Verbal Disputes, David Chalmers (2011) introduced what he called the method of elimination (Yudkowsky’s “rationalist taboo” is a version of the same procedure) to help determine whether a disagreement is substantive or merely verbal. Unlike substantive disputes, merely verbal disagreements only arise because the discussion participants fail to notice that they have a different vocabulary. A notorious example is two people discussing whether a tree falling in the forest makes a sound if no one is there to hear it.
The method of elimination involves restating a given claim in neutral terminology. For instance, in the discussion about the falling tree, the word ‘sound’ can be “eliminated” (or “tabooed”) to determine whether or not the disagreement is substantive. (The apparent dispute disappears once it is discovered that one party would replace “sound” with “air vibrations,” whereas the other person would have replaced it with “auditory processing.”)
Applied to notorious concepts such as “moral,” “right” or “conscious,” the method of elimination forces us to make apparent our standards of evaluation. While anti-realists would typically consider the provision of such standards a requirement for using language properly, this is where the realists take a different route. Because realists are interested in only the (speaker-independently) right standards, rephrasing the claim in question risks getting it wrong.
Using the example of the moral term “ought,” Chalmers describes this situation as follows:
The judgment about when and where “bedrock has been hit” is left to the discretion of the speaker:
Bedrock concepts
Based on the notion of “hitting bedrock,” Chalmers coined the term bedrock concepts. Bedrock concepts are thought to sit at the hearts of domains like consciousness, morality, and epistemology. They have in common that any attempts to (re)define them with terminology from outside their particular domain will trigger an intuition that this rephrasing endeavor misses the point.[12]
At most, we can “define” bedrock concepts in a circular way (e.g., “Doing good is about performing the morally right actions”), or by ostension. For example, Derek Parfit defined the concept of irreducibly normative reasons by pointing at examples (Parfit, 2011a):
The need for bedrock concepts
The disagreement between realists and anti-realists is primarily a disagreement about bedrock concepts. Different realisms[13] are built around bedrock concepts. While anti-realists are generally skeptical of bedrock concepts, realists cannot imagine doing (good) philosophy without them.
That said, even the realists would typically agree that bedrock concepts are weird. Without them, philosophy would be simpler—we could follow Wittgenstein’s (1999(1921)) maxim, “What can be said at all can be said clearly.” By incorporating bedrock concepts into our philosophical repertoire, many things change. Laying out what constitutes philosophical progress then becomes a bedrock concept as well (“yes, doing philosophy is related to figuring out what’s right—but how do we do that?”). Arguably, that’s a steep price to pay.
For some bedrock concepts such as “art” or “identity” in “the ship of Theseus,” the case for incorporating them into our philosophical repertoire is weak. Even though we may have the essentialist intuition that concept like “art” or “identity” have a single correct meaning, it seems (to me at least) that if we were to drop this intuition, nothing of value would be lost. It seems to me that we don’t hold those intuitions for good reasons, but rather because they seem to derive from a lack of introspective discernment about how our minds form categories.
However, not all bedrock concepts can be rejected this easily. The strongest argument for realism is that when it comes to adopting some bedrock concepts, we may not have a choice. Below, I discuss this in more detail.
Candidate bedrock concepts ordered by their perceived degree of philosophical indispensability
I’ve come up with a somewhat subjective, very crude ranking of bedrock concepts ordered by how easy it seems to me to do philosophy without them. The lower the tier, the closer it is to the proverbial bedrock (if there is such a thing). So, while I think that the candidate bedrock concepts in Tier 4 and Tier 3 can be rejected safely without much discussion, I find the debate between realists and anti-realists more interesting for Tier 2, Tier 1 and Tier 0 bedrock concepts.
Tier 4: Human-made objects
Examples:
“The ship of Theseus” and other expressions of object identity over time
Plato’s idealism about forms for human-made objects (e.g., this page lists “table” and “house” as examples of things about which Plato was a realist)
I think the case for bedrock concepts is the weakest when it comes to human-made objects. For any human-made object, we can ask humans why they made it—the answer will describe the speaker-dependent purpose of the object. There’s not much more to say. We may still have the intuition that “Is it really a table?” is a well-defined question, but absent further evaluation criteria about what we mean by this, the question is clearly underdetermined. (We can also think of language itself as a human-made object: words get their meaning from how we use them. If context is lacking, so is meaning.)
Tier 3: Sense appearances
Examples:
Plato’s idealism about forms for natural objects
Aesthetics realism
Realism about colors (Brogaard, 2010)
Before Darwin, it would have seemed more intuitive to me to be an essentialist about animal species.[14] Some philosophers have also invoked realism about the properties of the source objects for our sensory perception (for instance, realism about colors, sounds, beauty, etc.). While I understand the intuition behind (e.g.) colors existing independently of color perception, this view seems untenable to me because of the Darwinian arguments for anti-realism about sense appearances. With the advance of our scientific understanding, the reductionist worldview became intellectually satisfying despite our anti-reductionist intuitions.
Tier 2: Irreducible normativity
Examples:[15]
Moral realism[16]
Realism about self-oriented reasons for action
Realism about other-regarding reasons for action
Realism about epistemology
Realism about decision theory
Metaphilosophical realism (realism about the proper philosophical method)
On the one hand, it seems that our intuitions about normativity as a bedrock concept are not too different from the intuitions in favor of Tier 3 bedrock concepts. On the other hand, normative realists have one additional consideration to go by. Namely, they can argue that in addition to normative anti-realism being (perhaps) counterintuitive, it is also self-defeating. To give an analogy, consider imagining myself as a Boltzmann brain. Though logically consistent, the prospect that I’d cease to exist in the next flicker of a moment would render everything I’m about to do or think about pointless—my actions wouldn’t carry any relevance if the scenario were true. Arguably, there’s a sense in which imagining normative anti-realism is a bit like that: Though logically consistent, it’s pointless according to existentialist sense of things mattering to us. Because the alternative appears to be self-defeating,[17] it has been argued that we should adopt normative realism.
Tier 1: Qualia (“irreducible sensations”)
Examples:
Qualia realism
I’m categorizing qualia as Tier 1 candidate bedrock concepts because the case for incorporating them into our philosophical repertoire is particularly strong (at least as far as different types of realism are concerned). Tier 3 and Tier 1 differ as follows: If a person looks at tree leaves, the Tier 3 candidate bedrock concept is the “greenness” of the leaves—a claimed, objective property that some objects (e.g., leaves) either have or lack. By contrast, the Tier 1 candidate bedrock concept in that scene would be the quale “green”— the “what-it’s-likeness” of seeing green as an onlooker.
Being mistaken about an object’s true nature seems conceivable, but it seems impossible to make the same sort of mistake about the content of one’s immediate moment of consciousness. If I believe that, right this moment, something in my visual field looks green, I can conclude that I’m seeing green with certainty. Because we experience consciousness directly and immediately, the usual anti-realist arguments don’t seem to apply. Arguably, consciousness anti-realism is inconceivable because it contradicts our sensory experience.[18]
While I listed qualia realism as the sole example of Tier 1 realism, it’s worth pointing out that other types of realism could use qualia realism as their backbone. For instance, philosophers who endorse a theory of well-being according to which conscious experience is the only thing of (dis)value might argue that moral realism can be derived from realism about intrinsically morally valuable or disvaluable qualia (e.g., Hewitt, 2008). While I don’t intend to take a stance on qualia realism per se in this sequence, I’m planning to write a post (or post section) on arguments that go from qualia realism to realism about morality.
Tier 0: Something irreducible
Examples:
Existence realism
At the lowest tier and closest to ultimate philosophical bedrock (if there is such a thing), we find reality itself. Even if qualia aren’t quite what we think they are, and even if there’s no speaker-independently correct way to reason, we might want to assume—if only for the sake of preserving our sanity—that there’s at least something that goes beyond subjective standards. (Existence anti-realists may point out that things are not as they appear, but whether or not they have lost their minds is for readers to determine.)
Summary and conclusion
Some types of realism seem to be backed by nothing more than widely distrusted essentialist intuitions. However, normative realism, qualia realism, and existence realism have further considerations for their support. For instance, normative anti-realism is something we arguably wouldn’t care about if it was true, and qualia anti-realism is something we arguably can’t imagine to be true.
Realists and anti-realists may disagree about the validity of those considerations, and the degree to which they may force us to adopt the costs of incorporating bedrock concepts (e.g., irreducible normativity or the notion of qualia) into our philosophical repertoire. In addition, realists and anti-realists may disagree about the severity of those costs, perhaps because they think differently about the workability and palatability of an anti-realist philosophical framework.
Outlook on future posts
In the next post in this sequence (“3. Against Irreducible Normativity”), I will argue that normative bedrock concepts (especially moral ones, but the arguments apply generally) are best to be rejected. I will argue that (1) there’s a strong sense in which we can already tell that they make clear-minded philosophical progress impossible, and (2) that the alternative—normative anti-realism—can be richer and more palatable than one might at first think.
In posts 4 and 5, I will then address the possibility of a wager for irreducible normativity, as well as its counterintuitive consequences. In post 6, I will discuss arguments that also apply to naturalist moral realism. In particular, the post discusses a challenge for the concept of “moral uncertainty.” In post 7, I plan to address versions of moral realism built on top of qualia realism. Finally, in posts 8 and 9, I sketch I reason about morality given that I suspect moral realism to be false.
Appendix: Realisms are correlated
According to the 2010 PhilPapers survey (Bourget & Chalmers, 2014), belief in different realist positions strongly correlates (as does belief in various types of anti-realism).
“Moral realism” (note that this also includes naturalist versions of moral realism) correlates with the following views:
Aesthetic value: objective (0.411)
Abstract objects: Platonism (0.335)
Laws of nature: non-Humean (0.329)
Science: scientific realism (0.32)
Libertarianism about free will (0.24)
Different types of realism also correlate with one another:
“Mind: non-physicalism” correlates with “Free will: libertarianism” (0.386)
“Aesthetic value: objective” correlates with “Laws of nature: non-Humean” (0.228)
“Time: A-theory” correlates with “Mind: non-physicalism” (0.23) and Aesthetic value: objective (0.164)
To get a sense of the strength of these correlations relative to other comparisons, I’m listing the five most strongly correlated answers in the entire study:
“Moral judgment: cognitivism” correlates with “Meta-ethics: moral realism” (0.562)
“Metaphilosophy: non-naturalism” correlates with “Mind: non-physicalism” (0.497)
“Analytic-synthetic distinction: yes” correlates with “A priori knowledge: yes” (0.467)
“Meta-ethics: moral realism” correlates with “Aesthetic value: objective” (0.411)
“Mind: physicalism” correlates with “God: atheism” (0.393)
Though far from perfect, these correlations indicate that different realist (or anti-realist) positions have things in common. Correspondingly, we should expect that there are underlying factors that make some philosophers more predisposed toward realist (or anti-realist) positions.
Acknowledgements
I’m thankful for comments by Jesse Clifton and Sofia Davis-Fogel. Many of the ideas discussed in this article were things I first encountered when I read Brian Tomasik’s essay The many fallacies of dualism (Tomasik, 2013).
My work on this post was funded by the Center on Long-Term Risk.
References
Bourget, D. and D. Chalmers. (2014). What do philosophers believe? Philosophical Studies, 170(3):465–500.
Boyd, R. (1988). How to be a moral realist. In G. Sayre-McCord (ed.), Essays on Moral Realism. Ithaca: Cornell University Press. 181–228.
Brogaard, B. (2010). Color. In: Oxford Annotated Bibliographies Online. <www(.)oxfordbibliographies(.)com/view/document/obo-9780195396577/obo-9780195396577-0021.xml>.
Case, S. (2016). Normative Pluralism Worthy of the Name Is False. Journal of Ethics and Social Philosophy, 11(1):1–20.
Chalmers, D. (2011). Verbal Disputes. Philosophical Review, 120 (4):515–566.
Cuneo, T. (2007). The Normative Web: An Argument for Moral Realism. Oxford: Oxford University Press.
Hewitt, S. (2008). Normative Qualia and Robust Moral Realism. PhD thesis. New York University.
Hume, D. ((1990)1757). Dialogues Concerning Natural Religion, Bell, M. (ed.). London: Penguin.
Kahane, G. (2013). Must Metaethical Realism Make a Semantic Claim? Journal of Moral Philosophy, 10(2):148-178.
Parfit, D. (1984). Reasons and Persons. Oxford: Oxford University Press.
Parfit, D. (2011a). On What Matters, Volume I. Oxford: Oxford University Press.
Ruse, M. (2010). The biological sciences can act as a ground for ethics. In Ayala, F. and R. Harp (eds.), Contemporary Debates in Philosophy of Biology. New York: Wiley-Blackwell. 297–315.
Singer, P. (1973). The Triviality of the Debate Over ‘Is-Ought’ and the Definition of ‘Moral.’ American Philosophical Quarterly, 10(1):51–56.
Tomasik, B. (2013). The many fallacies of Dualism. reducing-suffering(.)org. <reducing-suffering(.)org/the-many-fallacies-of-dualism/>.
Williams, B. (1979). Internal and External Reasons. In R. Harrison (ed.), Rational Action. Cambridge University Press. 101–113.
Wittgenstein, L. (1999(1921)). Tractatus logico-philosophicus. Frankfurt am Main: Suhrkamp.
For instance, the moral realist Terence Cuneo (2007) has argued that anti-realism about morality also entails anti-realism about all other types of irreducible normativity. Similarly, the anti-realist Michael Ruse (2010) has argued that because of what we know about the evolutionary origins of our moral intuitions, the case for moral realism is no stronger than the case for aesthetic realism.
Note that I’m grouping the above domains only because realism about them is an appealing sentiment at least superficially. As I will argue in later sections, I think that there are vast differences in plausibility between various types of realism.
There are interesting exceptions to this, though usually they get categorized as (lukewarm) theism: Instead of abandoning the concept ‘God,’ certain theists seem to water it down until it makes almost no predictions about reality.
In the essay “The Triviality of the Debate over ‘Is-Ought’ and the Definition of ‘Moral,’” Peter Singer (1973) convincingly pointed out that whether people find moral arguments—such as, for instance, the drowning child argument—personally motivating or not does not seem to depend on metaethical assumptions.
I’m probably oversimplifying this, but according to the Stanford Encyclopedia of Philosophy article on moral anti-realism, the difference between speaker-independent facts and speaker-dependent ones is tricky to explain. They write: So many debates in philosophy revolve around the issue of objectivity versus non-objectivity that one may be forgiven for assuming that someone somewhere understands this distinction. There certainly exists a widespread intuitive imagery associated with the duality that is sufficiently vivid to motivate heartfelt philosophical commitments, but, once approached directly, the distinction nevertheless proves extremely difficult to nail down. It is likely that part of what is causing confusion is that there are a number of non-equivalent ways of drawing the distinction, some of which are better suited to certain subject areas than others.
For normative contexts, the distinction between speaker-independent and speaker-dependent facts roughly corresponds to Immanuel Kant’s distinction between categorical and hypothetical imperatives. The same difference is also described with what we can call “reasons terminology.” Realists believe there are irreducibly normative reasons (or so-called “reasons proper”), whereas anti-realists only think in terms of instrumental reasons. Williams (1979) provides a discussion of (and arguments for) “reasons anti-realism.”
This doesn’t mean that morality can’t be multifaceted. Derek Parfit, who was a moral realist, argued that the different normative-ethical theories of consequentialism, contractualism, and deontology are ways of “climbing the same mountain” from three different sides (Parfit, 2011a, p. 419).
Consciousness anti-realists are sometimes associated with statements like “qualia don’t exist.” I want to make clear that I interpret the term “hidden qualia” in a neutral sense, compatible with both realism and anti-realism about consciousness. Consciousness anti-realists think that whether or not hidden qualia “count” depends on where/how we want to draw the boundaries.
This footnote is only about semantics, but I don’t understand the notion of “meaningful, but necessarily false.” The way I think about it, claims about a speaker-independent moral reality are always false because the concept of such a reality makes no sense in the first place.
I’m not entirely sure what the non-objectivists’ linguistic claim is. Some non-objectivists may indeed want to commit themselves to the view that all moral discourse is non-objectivist. However, others may think about it in terms of offering a more constructive alternative to error theory. It is this second notion of non-objectivism that I’m interested in.
Muehlhauser left open whether we should think of pluralistic moral reductionism as realism or anti-realism. By contrast, according to definitional choices I made and explained in my previous post, I’d call it “moral anti-realism” myself.
See also G.E. Moore’s open-question argument.
The full picture is a bit more complicated because not all types of realism are built around bedrock concepts. Generally, realism about a domain means that there’s room for only one interpretation. Bedrock concepts at the heart of the domain would make sure of that—which is why most realisms (so-called “non-naturalist realisms”) are built around bedrock concepts. However, philosophers have also proposed the idea of naturalist realism. For instance, I might identify as a naturalist realist about electrons. If I understand what a single electron is, I’ll also understand what other electrons are like. There’s no open question about where to draw the boundary because as far as we know, something either is an electron, or it isn’t. Facts about electrons are reducible to physical facts. As I’ve mentioned in my previous post, for domains such as morality, naturalist versions of realism are sometimes criticized for being “too watered down” compared to non-naturalist realism. I described the criteria that could convince me of a naturalist version of moral realism in my previous post. In short, I consider naturalist realism about a domain true if the domain’s structure, its “features in the option space,” are sufficiently salient and unambiguous that there exists little to no disagreement about evaluation criteria. That is, no matter how they look at it, philosophically sophisticated reasoners will come to interpret the domain the same ways, provided that they can reason about it under idealized conditions. I devote later posts this sequence to discussing why I think those criteria for naturalist realism can’t be met for the domain of morality.
There were philosophers who questioned this view even before Darwin. For instance, Hume’s Dialogues Concerning Natural Religion (Hume, 1757), especially chapters 7 and 8, contain arguments against the need of a creative intelligence to explain nature’s design and orderliness. Hume even anticipated some of the key features of natural selection.
These examples make up different flavors of normative realism. Some philosophers argue that there are distinct kinds of normativity, while others maintain that there’s only one kind (Case, 2016).
A distinction that’s often made is between self-oriented (“prudential”) reasons for action and other-regarding (“moral”) ones. For reasons that will become apparent later on in this sequence, I consider this distinction to be overrated and potentially misleading. Therefore, I will be using “moral realism” in an unusually broad sense, to refer to realism about at least some irreducibly normative reasons for action (whether self-oriented reasons, other-regarding reasons, or both).
In the post Realism and Rationality, Ben Garfinkel expressed a related sentiment with respect to the reasons we may have for believing something: If (normative) anti-realism is true, then it can’t also be true that we should believe that anti-realism is true. Belief in anti-realism seems to undermine itself.
Anti-realists about consciousness would argue that appearances are deceptive, and that the realists are smuggling in too many connotations. Consider the following dialogue (adapted from Brian Tomasik’s writings on consciousness, where he has a similar dialogue): Alice: “What do you mean by ‘consciousness’?”
Bob: “What I’m experiencing this very moment: the way it feels like something to be me.”
Alice: “Okay, let’s note this down: Bob’s brain state at time 14:04:28 EST. Is that all you mean by consciousness?”
Bob: “No! I didn’t mean to point to just this one instance. You got right that this particular moment was an instance of consciousness. But I also meant this moment right now, and all moments like it, present and future!”
Alice: “Okay, I’m adding that down as well: Bob’s brain state at time 14:04:38 EST, and all configurations of matter exactly like it.”
Bob: “You’re still not getting it! I’m not just talking about the specific instances, I’m talking about what those instances have in common: the extremely salient feature that it feels like something to be me. I’m puzzled why you pretend that you don’t understand it.”
Alice: “I think that’s an illusion.”
Bob: “What do you mean? Of course that’s no illusion! The fact that when I’m awake, it feels like something to be me is the most certain fact there is.”
Alice: “Sure, but the only thing you can be certain of is that the state your brain had just that second ago was what you refer to as ‘feeling like something.’ I’m happy to grant you that—it’s what I already wrote down. What I’m denying is that there is a clear way to extract from single examples a universal concept of ‘consciousness’ that determines, once and for all, which aspects of reality make up consciousness, and which ones don’t.”
Bob: “Okay, I think I see what you’re doing. You’re saying I’m not allowed to be completely certain that ‘it feels like something’ picks out a property that brain states of all kinds either have or lack? That I can only ever point to the brain state I’m currently in and say this or that about it, but that if we were to try and extrapolate what I would say about an entirely novel brain state, we can conceivably doubt whether my initial examples pick out a large and well-specified cluster in the space of all possible brain configurations in the same way as ‘This substance is what I mean by water’ manages to successfully pick out a vast space of configurations of H2O molecules in all kinds of shapes and forms?”
Alice: “Exactly. Why did you assume that the feeling you call ‘being conscious’ picks out brain states in that way? I mean, I can see that it’s natural to also think of other people as conscious, but why did you think that once you go to examples where some features remain similar while other features start to vary a lot or become more distant, the examples would continue to be clear cut?”
Bob: “It seems like that’s part of my feeling of being conscious! I can’t imagine there being edge cases.”
Alice: “Okay, but that’s a different claim now. Initially, you said the fact that it feels like something to be you is the most certain fact there is. Now you moved from that statement to a statement about edge cases. I don’t see how the latter statement is any different from people claiming that it’s inconceivable to them that there may not be a clear fact of the matter whether a certain action is moral or immoral, or whether the expression ‘Theseus’s ship’ is to be interpreted this way or that way.”