Skip to content

Maybe two parts of reality instead of three?

By Dave Elder-Vass

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: Elder-Vass, Dave (2022) Maybe two parts of reality instead of three? Critical Realism Network,


Roy Bhaskar’s three domains of reality – the empirical, the actual and the real – are staples of many expositions of critical realism. In their provocative blog post, Tom Fryer and Cristián Navarrete (2022) argue that we should stop using them because they are redundant and confusing. This post engages constructively with their argument. It positions the three domains as a rhetorical device rather than a fundamental element of Bhaskar’s ontology. Then it asks what we might want to retain from the idea and how we might improve on it as a way of representing critical realism’s ontology.

It argues that the three domains embody two significant distinctions. The first, between the empirical and what we might call the actual-but-not-empirical, is highly problematic. The distinction is contingent and arbitrary and serves no significant ontological role. Even as a rhetorical device, it is directed against a form of empiricism that hardly anyone endorses and thus serves no significant discursive role. By contrast, the second distinction, between the actual and the real-but-not-actual, is essential to critical realist ontology. Causal mechanisms can be real without being actual in two different senses, both of which are important to our understanding of causality. Still, the overlapping structure of Bhaskar’s actual and real domains is confusing to explain, and the paper suggests an alternative: would it be more useful to distinguish between the actual and the potential as two ontological subsets, or aspects, or perhaps domains if the term is useful, of reality?

The three domains

As Tom and Cristián point out, Bhaskar summarises the three domains in a table in A Realist Theory of Science, reproduced below, which appears twice, once on page 56 in a section entitled “A Sketch of a Critique of Empirical Realism”, but also on page 13 – the second page of the Introduction. This positioning already tells us something. First, that it was formulated as part of the critique of empirical realism, rather than as independent thesis about ontology. Second, that he found it important enough, and presumably clear and striking enough, to feature it very prominently at the beginning of the book, which no doubt explains why so many critical realists have been influenced by it and see it as significant for explaining Bhaskar’s work.

Domain of RealDomain of ActualDomain of Empirical
Figure 1 – Bhaskar’s three domains: populating entities (Bhaskar, 1978, p. 13)

The three domains perform at least two functions for him in the book. First, they combine and systematise two key insights into causality. One, that causal mechanisms may operate unobserved (the empirical/actual distinction). And two, that they may operate without having their characteristic effect, due to counteracting mechanisms, or even lie dormant and not operate at all when their triggering conditions are absent (the actual/real distinction).

Second, the significance of these two insights for him at the point where the diagram appears is that they are used to criticise competing understandings of science. The three domains operate as a rhetorical device that neatly and symmetrically expresses his critique of alternative perspectives on the operation of mechanisms. Those who ignore what happens unobserved can be accused of empiricism, which is a failure to recognise the domain of the actual (or strictly speaking, the actual-but-not-empirical, since the empirical is a subset of the actual). Those who ignore the unactivated mechanisms in the world can be accused of actualism, which is a failure to recognise the domain of the real (or, strictly, the real-but-not-actual).

Incidentally, apart from Bishop George Berkeley in the eighteenth century, most empiricists are quite happy in practice to accept that things can happen unobserved and their focus on experience is epistemological rather than ontological. The first critical argument, then, is a neat rhetorical flourish, but it is not particularly significant for the critique of contemporary empiricism. The second is much more interesting.

My sense, then, is that the three domains were originally more of an expository device for clarifying his critique of the empiricist understanding of science rather than being intended to express his own ontological argument from first principles. But the trouble is that they have been used to do both of those jobs. It is therefore worth asking, as Tom and Cristián have, whether the diagram and the concept of the three domains really work as a way of explaining critical realism’s ontological position.

The empirical and the actual

Read as a summary of an ontological argument, rather than as a critique of competing accounts of science, the three domains represent two distinctions. This section discusses the first one, the distinction between Bhaskar’s empirical domain, composed of experiences, and everything that is actual-but-not-empirical, which I shall refer to as ABNE. Those other elements might include events that no-one experiences, but also those events that someone does experience, understood as events as they are in themselves, as distinct from the way we experience them. It also includes things or objects (I could argue the case for that from first principles, but Bhaskar also says so himself at the bottom of page 32).

We can read the argument in one of two ways, depending on whether we think of the empirical as a domain in Bhaskar’s ontology or an object of critique defined in terms that would be acceptable to the empiricists of Bhaskar’s story.

In Bhaskar’s ontology the distinction between the empirical and the actual seems to rest on the claim that my experience of an event may be distinct from the event itself – the event as it occurred independently of my observation of it. This is problematic as a distinction between ontological domains, because my experience is itself also an event. Granted, it is an event that happens in my head and so it is a different event than the one outside me that I am observing, but it is nevertheless an event. It is also a different kind of event, but then there are many different kinds of event, so why would we mark this kind out as ontologically different? From the point of view of an analysis of science, the reason for distinguishing our experience from that which is observed is clear: observation is a crucial element of the process of science. Once we move from the critique of empiricism to the business of establishing a realist ontology, however, science is just one small corner of the universe and it is by no means clear why we need to see one aspect of it as a separate ontological domain.

For the putative empiricists that Bhaskar is criticising, on the other hand, the distinction between an event and our experience of it collapses: “whatever men currently experience is unquestionably the world” (I confidently hope that he really meant people, regardless of gender, rather than men) (p. 58). There is no event beyond our experience of it for the thoroughgoing empiricist, since our experience is the event. This could be critiqued by asserting the view of experiences described in the previous paragraph, but Bhaskar also develops his transcendental argument: for experimental science to make sense, it must tell us something about events in the world beyond the experiences of scientists, or what would be the point of doing it? But if there are events other than those the scientists experience, then this variety of empiricist philosophy of science is incoherent.

But this version leaves us with an arbitrarily anthropocentric set of domains. Bhaskar says something similar himself on pages 16 and 58. Let’s see first why it is anthropocentric, and then wonder why he would admit to it later. It is anthropocentric because it assumes (and Bhaskar is quite explicit about this – see pages 32 and 34) that it is the domain of specifically human experience. But why should the experiences of human beings be marked out as sufficiently ontologically special as to deserve a whole domain of reality? What about, for example, the experiences of dogs or earthworms or intelligent aliens if there are any? Does this mean we will also need a domain of the Martian empirical, a domain of the canine empirical, and so on, in a completed ontology?

I don’t think we will, because I think this distinction should remain firmly in the space of his critique of empiricist philosophy of science, and not at all in his substantive ontology. In the critique it makes perfect sense: the issue at stake is what is involved in humans studying the universe, and so human experience has a special place in the argument. And it makes sense for a second reason, too. His critique is framed as an immanent critique of empiricism, which is formed by taking the empiricist argument and showing how it is incoherent. He therefore takes the special place of human experience as a starting point because it is part of the argument he is seeking to expose. And that, of course, is why he is prepared to criticise the privileging of experience as anthropocentric: it is an element of the empiricist object of his critique.

The net result, though, is that there doesn’t seem to be a good reason for distinguishing a separate domain of the empirical to contain experiences in a critical realist ontology.

The actual and the real

The second distinction, between the actual and non-actual elements of the real, is much more important. The actual includes all of the things we have just been discussing – experiences, events, and things, as a minimum. Personally I think this means that it includes everything that exists in a material form and everything that happens to anything that exists in a material form. According to Bhaskar’s table, the non-actual part of the real, the real-but-not-actual or RBNA, includes mechanisms. As he often identifies mechanisms with powers it may also include those. 

This distinction, by contrast with the first one, is both profound and fundamental to the critical realist account of causality, although there is again some ambiguity (of a different kind) about exactly what we should include in it. Let me focus on the key question, which I’ve illustrated sometimes with the example of a laser pointer. A laser pointer is a small handheld device which projects a dot of light onto a surface when a button on the device is pressed. There is a mechanism, a process of interaction involving electrical wires, switches, light emitting diodes, and other things, which operates when the button is pressed (subject to certain other conditions being satisfied, such as the device having a charged battery in place). As a result of the mechanism, the device has the causal power to project a dot of light onto a surface. When such a device exists and the button is not being pressed, there is a sense in which the mechanism is a real potential of the device, but the events that constitute the mechanism are not occurring and so the mechanism is not actual. The mechanism is real but not actual. When the button is pressed, the process starts to run, the mechanism is actualised, and the power is exercised.

This suggests a challenge to Bhaskar’s division between the domains: if a mechanism is constituted by interactions between the parts of a material thing, then those interactions would seem to be events and thus to belong in his domain of the actual. This is an argument I worried about in one of my early papers (see the references at the end), and I’ve heard several other people worry about it too. But I don’t think it undermines Bhaskar’s point. Yes, at times mechanisms become actualised as events, but it is fundamental to explaining causality to recognise that things have mechanisms (and the powers that they produce) as potentials even when they are not operating. And this, it seems to me, is why we need some space in our ontology that accommodates non-actual mechanisms and powers.

If you’re not yet convinced, consider the ambiguity I promised you earlier, which can be illustrated using a slightly different case: Was the potential of laser pointers to project a dot of light onto a surface real before the first laser pointer had been created? I think it is obvious that it was, and thus we have what might seem to be a second kind of potential: not a potential that is implicit in actually existing things, but a potential that could be realised if and when we created such things. That’s something real that seems very clearly not actual. (What’s less clear to me is whether Bhaskar intended to include this second understanding of mechanisms in the real but not actual when he formulated his three domains.)

An ontology that only has room for the actual – things that exist and the events that occur to them – will be unable to explain causality because it will have no space for the unrealised potentials of the world that only become activated in certain circumstances.

The actual and the potential

The consequence is that I’d be happy with Tom and Cristian’s proposal to stop talking about the three domains, but I do think it’s essential to continue to talk about two of them: the actual and the non-actual real. Still, “the non-actual real” or the “real-but-not-actual” are hardly terms that slip off the tongue. Part of the problem is that they label one of the most fundamental and distinctive elements of critical realist ontology, not in terms of what it is but in terms of what it is not. But Bhaskar doesn’t even give the real-but-not-actual a name in RTS. It’s hardly surprising that people struggle to make sense of it.

That would be much simpler and clearer if we could have a positive name for the RBNA (or at least the element of it that I have focused on here) that encapsulates the significant contribution it makes. I can’t claim to have thought through the alternatives in any detail, but how about ‘the potential’? Then we could just say that ‘the real includes both the actual and the potential’, using ‘the actual’ as Bhaskar did and ‘potential’ as I have above. I don’t mind whether we call these domains or something else, but it is worth stressing that, unlike Bhaskar’s domains, they do not  overlap with each other. This does leave open the question of whether there is anything else that is real but not actual, but that’s a topic for another day (and some of my old blog posts).

So, then, let’s return the three domains to what was probably their original purpose: as a rhetorical device for expounding Bhaskar’s critique of empiricist philosophies of science, rather than a fundamental ontological structure. But let’s rescue the distinction between the actual and the potential, since it is crucial to the critical realist enterprise.


My thanks to Tom and Cristián for commenting on my first draft and to the participants at the online CR discussion group in June 2022 for their questions, challenges, and ideas on both Tom and Cristián’s paper and my response to it.


Bhaskar, R. (1975). A realist theory of science. Leeds Books.

Elder-Vass, D. (2007). Re-examining Bhaskar’s three ontological domains: the lessons from emergence. In C. Lawson, J. Latsis, & N. Martins (Eds.), Contributions to Social Ontology (pp. 160–76). Routledge. Open access version available here.

Fryer, Tom & Navarrete, Cristián (2022) Let’s stop talking about the three domains of reality, Critical Realist Network,

Leave a Reply

Your email address will not be published.

%d bloggers like this: