Skip to content

There Aren’t Really Three Domains: or, Metaphor Is Great, Except When It’s Not

By Ruth Groff

A PDF of this piece is available to download here. If you’d like to join the debate by contributing a piece of your own, email us.

Suggested citation: Groff, Ruth (2022) There Aren’t Really Three Domains: or, Metaphor Is Great, Except When It’s Not, Critical Realism Network,

Thank you to Tom Fryer and Cristián Navarette for their post on Roy’s metaphor of ontological stratification – i.e., the so-called ‘three domains’ of reality – and for Dave Elder-Vass’ and Tobin Nelhous’ subsequent responses.  Thanks too to Dave, for the invitation to join the conversation.

One big problem with metaphors is that a certain type of reader – of which there seems to be no shortage – is tempted to reify them.  A particularly egregious example of this is Marx’s metaphor of base, structure and superstructure.  160 years on, it’s still necessary to point out that Marx was talking about society, not about buildings.  While he used an architectural metaphor to say something important about society, he was not saying that societies are buildings.  If nothing else, societies – unlike buildings – are composed of social relations (themselves composed of, but not reducible to, individuals).  For this reason alone, the metaphor only goes so far: social relations do not actually sit atop other social relations as the 8rd floor sits atop the floors below it.  Marx’s point was not that society is a three-story building.  Or even a pyramid of acrobats or cheerleaders.  Rather, the claim (correct or not) was that it’s possible to distinguish between the social relations that constitute different institutions and practices, and that some social relations have a greater effect upon the nature and operation of the whole than do others.

Tom and Cristián object to Roy’s metaphor partly on the grounds that it’s redundant.  We have perfectly adequate, non-metaphorical ways to say the same thing, they observe, and to do so in an arguably more accessible way.  I don’t disagree with them, but I think that the objection is not pointed enough.  There’s always a non-metaphorical way to say the metaphorical thing.  Nor is it simply that the metaphor – with its accompanying table of categories – is confusing, as they also contend.  The precise problem (in my view) is that, as with Marx’s metaphor, readers tend to interpret Roy’s metaphor of ontological stratification literally, or at least close to literally, and in so doing they misconstrue, or even miss altogether, the ideas that the metaphor is meant to capture.  It’s the same, as it happens, with Roy’s talk of transitive and intransitive ‘dimensions,’ which grew out of a distinction made early in RTS between the transitive and intransitive objects of scientific inquiry.

If Marx’s base-structure-superstructure metaphor has led generations of readers to imagine that society is a parking garage, Roy’s ontological stratification metaphor leads readers to conceive of society as a kind of curiously disaggregated rock face, where each of the levels of sedimentation is somehow spatially discrete, and the deepest level is not visible.  Indeed, with the help of the table that accompanies the metaphor, readers go on to assign phenomena to levels that have now become boxes – in a manner akin to sorting clean laundry.  All of this is unfortunate – and, as Tom and Cristián suggest, leads to convoluted thinking.  I would add that in addition to being convoluted, the thinking is often rote, technically dogmatic. 

I agree with Tom and Cristián that an obvious remedy is to simply state the ideas in a non-metaphorical way – at least by default (rather than the other way around, with the non-metaphorical expression coming as an afterthought, as is so widely done at present). 

But what are the non-metaphorical ideas?  On this point, I probably come closer to Dave than to either Tom and Cristián or to Tobin.  The driving question of A Realist Theory of Science is “How can it be that so-called ‘laws of nature’ hold even though the regularity in terms of which they are defined by Humeans and Kantians alike does not obtain?”  How is it that laws hold ‘transfactually,’ to use Roy’s term?  Roy’s answer, of course, was that regularity, such as it is, is epiphenomenal.  It is the powers of things, as he had it (others might say that it is the powerful things themselves; I say that, for instance[1]), that, when expressed or exercised, sometimes (depending upon the nature of thing and the kind of power) yield outcomes that occur predictably, or regularly – which outcomes, when they do occur, are sometimes observed, though not always. 

This core claim sustained a reconceptualization of ‘laws of nature’, such that the answer to the question of how they hold transfactually is that they are best conceived as being a function of the powers (of things), and powers do not go away when, for one reason or another, they have not been expressed, or actualized – let alone if they have merely not been observed.  The metaphor of ontological stratification is supposed to capture this bit of neo-Aristotelian ontology.  ‘The Real’ represents the idea that (as per Aristotle) powers exist even when they are not being exercised.  ‘The Actual’ represents the idea that any given power may be exercised, even if being so is not a condition of its existence.  ‘The Empirical’ represents the idea that being observed by a subject is also not a general existence condition for (the) powers (of things).  Finally, the formula ‘R>A>E’ tells us that at any given moment not all powers are actualized, and not all actualized powers are observed.

Having transposed the metaphor back into non-metaphorical language, we can see that while the relevant content comes as an answer to a precisely formulated question about laws, it contains quite a bit of metaphysical content.  It is at once an endorsement of Aristotle and a rejection of both Hume and Kant.  It connects, via realism about powers, to a theory of causation.  It has modal implications (in this regard Dave is absolutely right, in my view, to emphasize the significance of the divide between R and A, implying as it does a commitment to the phenomenon of real potentiality).  But the metaphor, as I suggested at the outset, is not a reliable short-cut for any of this.  It really only does the work of imparting said content if one is already cognizant of the content.  If you want to understand Bhaskar on causation, you can’t just read it off of the concept of ‘ontological stratification.’  You have to deal with the whole account.  Admittedly this sucks a little, since RTS isn’t all that well-written, and working through it calls for more interest in philosophy than most readers who aren’t philosophers can be reasonably expected to muster.

Let me say a quick word in response to Tobin’s piece, before I cast my vote on the metaphor.  Tobin argues that the ‘domain’ of the Empirical should be replaced by the domain of the Semiotic.  As I see it, critical realists whose concern is to explain social phenomena can’t avoid the phenomenon of human subjectivity, even if their focus is on the efficacy of high speed rail rather than on pop culture (or religion, or poetry, or school systems), since critical realists think of social phenomena as being, in part – though only in part – constituted by meanings.  Attention to semiotics is therefore required.  But I don’t think that there is a non-metaphorical intransitive ‘domain’ or ‘dimension’ any more than there are any other non-metaphorical ‘domains,’ and while Tobin would probably agree with this, it seems to me, further, that the ‘domain’ of the Empirical really does need to be defined in terms of observation, given what Roy was doing with the metaphor.  The claim in RTS is not that subjectivity is exhausted by perception.  It’s that laws of nature – reconceptualized in terms of causal powers – are not existentially contingent upon the outcomes produced by the exercise of powers being observed.  To my mind, then, Tobin is right to direct us to the semiotic features of social phenomena, but wrong to try to incorporate that analysis into Roy’s talk of the Real, the Actual and the Empirical.  The metaphor presents us with enough difficulties as it is, without making it be a mixed metaphor.

So, should we retain Roy’s metaphorical three-domains-talk, or not?  I suppose that my answer is that we don’t need the metaphor, but we do need the metaphysics that it’s meant to encapsulate.  Perhaps this is to say that what we should do is de-emphasize the metaphor.  Stop treating it as catechism.  The substantive analysis of RTS does not reduce to “Bhaskar believes that there are three domains of reality.”  The sentence is superficially true, but the metaphor is just that: a metaphor.  It’s not literally true.  Whether or not you find it to be a useful heuristic for communicating key aspects of a particular neo-Aristotelian ontology, the important thing is the metaphysics, not the device.

[1] Ruth Groff, “What Does the Doing?  Powers, Things and Powerful Things,” in (ed. Anne Sophie Meincke) Dispositionalism: Perspectives from Metaphysics and Philosophy of Science, Synthese Library, 2020.

2 thoughts on “There Aren’t Really Three Domains: or, Metaphor Is Great, Except When It’s Not”

  1. Hi Ruth,

    Sorry for a long reply, but I think I need to clear up what I was and wasn’t saying.

    First, I agree that metaphors can create problems, though I’m skeptical that metaphors can be =completely= avoided and that =totally= non-metaphoric language is feasible: there are all sorts of arguments against that, ranging from some of Bhaskar’s points to cognitive science’s. I also agree with you and Dave that Bhaskar’s concept of the empirical “domain” concerns the perceptual and subjective aspects of social activity, including science, a point that needs to be preserved.

    However, I wasn’t at all emphasizing the semiotic features of social phenomena. I’m arguing that the limits of Bhaskar’s thinking shouldn’t be the limits of CR. For the sake of relative brevity, I’ll focus on one aspect. Animals other than humans have experiences or perceptions of the world around them, and they respond to those perceptions; if they didn’t, they wouldn’t survive even the first round of Darwinian selection! CR needs a way to theorize the reality of non-human perception.

    My proposal to reconceptualize the “domain” of the empirical (qua domain) to one of the semiosic aims at that. The empirical/subjective doesn’t disappear as a category, feature or realms of social practices: it’s merely the tiny area where humans are (just as humans occupy a minuscule part of the biological kingdom Animalia) — a vital area to =us=, obviously, but the universe doesn’t actually revolve around us, and neither should our metaphysics. (Semiosic events are of course a minuscule portion of all the events that occur, in keeping with Bhaskar’s argument.) So I’m not mixing metaphors, I’m seeking a wholesale, anti-anthropocentric reframing of the entire domain. Note that if one thinks Saussure had the right theory of semiosis, then the universe indeed revolves around humans, which is one reason (of several) I think we should chuck it.

    Perhaps one might simply extend the phrase “empirical domain” to encompass the experiences of all living beings, but this still requires a departure from Bhaskar. In addition, I suspect that the term retains anthropomorphism, whereas “semiosis” emphasizes that there is a real mechanism or power involved. In humans it’s become an incredibly sophisticated mechanism, but it’s still semiosis.

    Re the “domains” metaphor itself, I might mention that Peirce had metaphysical concepts strikingly similar to CR’s domains, but he merely called them “Categories,” which is about as non-metaphorical as you can get. But without a reasonably close reading, they could become just as siloed as “domains” can be. Let’s face it: language poses risks!

  2. D’oh! I did of course bury the lede. If you’re assuming Saussurean semiology, in which all signs are arbitrary and social, you’ll end up thinking I’m just talking about the importance of signs in social activity. I’m working from a Peircean view, in which all sorts of things are signs: say, a rustling sound that warns a deer of a nearby predator, pheromones, the color of leaves as an indicator of a plant’s health, chemical markers telling the COVID virus to attach to a host, etc. A more structured concept than information, and a much broader concept than language.

Leave a Reply

Your email address will not be published.

%d bloggers like this: