Your complimentary articles

You’ve read all of your two complimentary articles for this month. To have complete access to the thousands of philosophy articles on this site, please

If you are a subscriber please sign in to your account.

To buy or renew a subscription please visit Subscriptions.

If you are a print subscriber you can contact us to create an online account.

You can register for a free account to have four complimentary articles per month. We will occasionally email you a newsletter, from which you can unsubscribe at any time. We do not sell personal data or otherwise disclose personal information to other organisations.

Metaphysics

What is Metaphysics?

Sajid Bijapure asks what it does, and what it might do in the future.

The word ‘metaphysics’ did not begin as a name for something mysterious. The word originally referred, quite literally, to the book by Aristotle that came after (meta in Greek) his writings on nature (the Physics). Yet over time metaphysics has come to denote a class of questions that reappear no matter how far scientific explanation advances. For instance, what must the world be like for anything to exist at all? What does it mean to cause, to persist, or to be possible?

What does doing metaphysics involve? I want to argue that it is best understood not as a special seperate subject, but as a recursive procedure built into human understanding. Metaphysical questions arise when the very concepts used to explain the world require explanation in their own right. Once seen this way, metaphysics appears both inescapable, and, surprisingly perhaps, partially automatable.

So let’s push metaphysics up a level. We can call this enquiry ‘metametaphysics’, or if you prefer, ‘meta²physics’.

Science, or Metaphysics?

Metaphysics asks the most fundamental questions about reality – questions that lie beneath, behind, or prior to scientific explanation. So it does not ask, ‘How fast does light travel?’ or ‘What causes cancer?’, it asks, ‘What is time, space, causation, identity, or possibility?’, ‘What is existence?’, or ‘What does it mean to be a thing?’. It asks what it means to exist at all, what makes something the same over time, whether time flows or is fixed, and what it means for one event to bring about another. Such questions also appear in physics, psychology, ethics, and everyday reasoning. What distinguishes them as metaphysical is not their subject matter, then, but the level at which they’re asked. They target the background assumptions that make explanation, description, and inquiry possible in the first place.

Metaphysics is different from science. Science assumes certain metaphysical ideas without proving them, such as the existence of an external world, the uniformity of natural laws, or causation itself. Metaphysics asks, why these assumptions are justified – or if they are. Metaphysics ranges over questions about existence, identity, time, causation, and possibility, but it is not defined by any one of these topics. In general, metaphysics asks, ‘What must the world be like for anything to exist, change, or be known at all?’ We can see then that metaphysics is not anti-science. Rather, it is pre-scientific and meta-scientific.

Many times it’s hard to determine if a question is scientific or metaphysical. For instance, in physics one may ask metaphysical-looking questions such as, ‘Is spacetime fundamental or emergent?’ or ‘Is the universe deterministic?’. Or about consciousness neuroscientists may ask, ‘Is consciousness physical, emergent, or a distinct type of thing?’, ‘Can machines be conscious?’. About identity or ethics a sociologist or anthropologist may ask, ‘What counts as a person?’ and ‘When does a life begin or end?’ These are all metaphysical questions wearing modern scientific clothes.

Asking Metaphysical Questions

We don’t begin with a discipline called metaphysics then generate questions for it. We begin with curiosity-driven questions, and only later notice some family resemblances among them and give those clusters of questions the label of ‘metaphysics’. So, the real issue for us is not ‘What is metaphysics?’, but rather. ‘What kind of curiosity keeps generating questions that later get grouped together as ‘metaphysical’?’

First, what do questions we’ve labelled as ‘metaphysical’ all have in common? Is it their subject matter? (Mathematical questions have their subject matter in common, as they deal with numbers and calculations and their logical implications). Or do we call a question ‘metaphysical’ because we have arrived at it by using certain procedures? Or are metaphysical questions a combination of these? Or something else?

As I say, in mathematics, the subject matter is clear. Numbers, operations, functions, there is a recognizable domain. But metaphysics lacks a clean domain. Metaphysical questions range over time, existence, causation, possibility, identity, mind, laws of nature, and further. These topics also appear in physics, psychology, linguistics, ethics. So, metaphysics cannot be defined by the topics alone. Is it a special procedure, then? Partly, but not merely. For instance, is it ‘focusing on one concept and defining it’? This might be one aspect, but definition-seeking happens everywhere, in philosophy and outside it: physics defines ‘mass’; biology defines ‘species’; law defines ‘person’… So, definition-hunting alone does not make a question metaphysical.

So what is distinctive about it? One thing is that metaphysical questioning refuses to accept any definition as final if it relies on empirical measurement, institutional convention, linguistic habit, or theoretical usefulness. It asks instead, ‘What must be the case for this concept to apply at all?’ So, the real commonality is at the level of question, not with the content or method. The shared feature of metaphysical questions is that they target the conditions of the possibility of concepts, practices, and explanations. (This insight is already visible in Immanuel Kant’s work.) Metaphysics does not depend on pure subject matter, or pure procedure; the metaphysical move occurs when the scientific or empirical explanation itself becomes the object of inquiry. Metaphysical questions are not defined by what they are about, then, but by what they are beneath. They arise when we turn our attention from empirical answers to the frameworks that make answers possible.

Let’s see a contrast that illuminates this point:

Physical question: What causes lightning?

Scientific-metaphysical boundary: What is causation in physical theories?

Metaphysical question: What does it mean for one event to cause another? Is causation a feature of reality – or of our description?

We do not begin with concepts like ‘cause’. We begin with curiosity. One could ask ‘Why does lightning occur?’ without ever using the word ‘cause’. The concept ‘cause’ appears later as a linguistic and conceptual compression of patterns we notice. Metaphysics begins when that compression itself becomes questionable – when we start asking: What exactly are we doing when we say ‘A causes B’? But this reflective turn, not the concept itself, is the metaphysical move.

Image © Athamos Stradis 2026 for more: www.instagram.com/athamos.stradis

Why Does Metaphysics Feel Inescapable?

The concept of causation is not forced by reality, but introduced by us as a tool. However, once a conceptual tool exists, two things happen: it stabilizes inquiry – it lets us generalize, predict, and explain; and it invites reflection. This gives rise to metaphysics.

Even when we try to avoid metaphysics in our systematic enquiry into reality, we smuggle it in. For instance, the realist assumes a mind-independent world, the stable identity of objects, and law-like regularity. Or the skeptic assumes distinctions between appearance and reality. Even anti-metaphysicians such as Rudolf Carnap still assume criteria of meaning and rules for legitimate questioning. But metaphysics resurfaces because assumptions cannot justify themselves. The motivation is not idle abstraction, but comes from conceptual breakdowns (‘this explanation no longer works’), boundary friction between disciplines, contradictions between equally successful theories, or reflective unease about hidden intellectual commitments. Metaphysics is what happens when curiosity turns inward and asks, ‘What had to be presupposed for my position to make sense at all?’

Normally, to understand something is to express it in terms of things already known and agreed upon. This implies that explanation borrows from prior concepts. Therefore, explanation is structurally regressive. Metaphysics arises not because philosophers are obsessive, but because understanding itself is recursive: it’s analogous to a feedback loop in which we repeatedly use the output of a function as its next input. When this recursive procedure is applied to concepts, initial curiosity gives rise to questions, conceptual articulation gives rise to new concepts, reflection on the concepts gives rise to metaquestions, and new conceptual frameworks lead to further reflection… Sometimes the loop stabilizes (for example, with working definitions in science), and sometimes the loop diverges (into metaphysics, ontology, epistemology). But this happens in other fields of enquiry too, like mathematics. So, why does this recursive loop not plague mathematics or engineering (usually)?

The recursive process appears in mathematics when foundational concepts such as infinity or sets are questioned, and it disappears again once new primitives are stipulated. Thus mathematics stabilizes inquiry by giving fundamental logical definitions. Metaphysics lacks such external stopping rules. Instead, when explanation turns back upon its own concepts, and no practical criterion halts the process, metaphysical inquiry begins. So we see that metaphysical questions arise because understanding is itself a recursive process. Concepts introduced to explain phenomena become, in turn, objects requiring explanation. Metaphysics is the name we give to inquiry when conceptual feedback fails to converge on a definitive understanding.

Let’s not view this as a defect, but as a feature. This tendency is how science advances (until it stabilizes) and how philosophy deepens (when it doesn’t stabilize). Even language evolves in this way. So metaphysics is not confusion. It’s reason encountering its own architecture.

The Positivist Turn, & Turn Again

Early twentieth-century logical positivism, associated with Rudolf Carnap and A. J. Ayer, among others, was motivated by a deep irritation with exactly the kind of endless conceptual looping I’ve hinted at here. Their core diagnosis implied that metaphysical questions persist because philosophers misuse language. So, the proposed cure was not better metaphysics, but better language hygiene. For this, logical positivists introduced a criterion intended to halt the metaphysician’s recursive process, and called it the verification principle. It says that a proposition (or phrase) is meaningful if and only if it is empirically verifiable or is analytically true (true by definition, as we might say). If the proposition is neither, it’s not false, it’s meaningless. On this view: ‘What causes lightning?’ is meaningful (because empirical), whereas ‘What is causation?’ is a pseudo-question, and ‘Why is there something rather than nothing?’ is a linguistic illusion. Metaphysics, the logical positivists concluded, is not deep, it’s confused.

Yet the verification principle itself uses various concepts, such as ‘verification’, ‘empirical’, ‘meaning’, and ‘analytic’. These concepts are neither empirically nor logically verifiable in the positivist sense. So the ‘meaning criterion’ cannot justify itself. The verification principle either violates its own rule, or it becomes just another output fed back as input for further questioning – exactly our previous loop. It turns out to be just another gear in the metaphysics machine.

Some logical positivists locate the trouble with understanding in words like ‘cause’, ‘being’, ‘substance’. But we can locate it earlier. We might say that curiosity precedes language; concepts arise to stabilize inquiry somewhat; and reflection questions those stabilizers. Language merely records the recursion. So, eliminating certain words does nothing. The process merely reoccurs with new vocabulary. So, while logical positivism says that metaphysics is a misuse of language, and its cure is the elimination of the abuse, our recursive model says that metaphysics is a feedback loop of understanding, and has no cure, only management. The later Ludwig Wittgenstein wrote, “Philosophical problems arise when language goes on holiday” (Philosophical Investigations, 1953). Is this a complete description of philosophy, though? Even if we fix language so that the meaning of all words becomes clear, not confusing, the question ‘What makes this explanation explanatory?’ remains. The urge to re-input outputs into inputs persists. Wittgenstein diagnoses symptoms. We search for the underlying mechanisms.

It seems that logical positivism misunderstands the motivation behind metaphysics. It assumes that metaphysics is an optional indulgence; but we can see that it’s a structural consequence of seeking understanding. We do not ask metaphysical questions because we like abstraction, or because language seduces us. We ask them because concepts cannot ground themselves.

Arguably, logical positivism collapsed not because it was found to be self-refuting, not because metaphysicians out-argued it, but because science itself kept generating foundational (that is, metaphysical) questions, on the topics of the nature of spacetime, laws, probability, identity etc. Suppressing metaphysics crippled reflection rather than clarified it, so the metaphysics question loop inevitably reasserted itself.

Turning the Metaphysical Crank Automatically

Simply put, expressing one concept in terms of other concepts is a procedure, and metaphysics consists in repeatedly applying this procedure to concepts. Therefore, metaphysics is a well-defined cognitive procedure. This means it should be automatable – and that metaphysicians are now dispensable. This is a position that metaphysics itself forced into existence!

But is it true? Metaphysical inquiry clearly does exhibit rule-following behavior. Abstraction, redescription, recursion, consistency checking, conceptual recombination, are clearly applied procedures. They’re also exactly the kinds of repetitive operations at which computers excel. This is why automated theorem-proving works, and why large language models can already do nontrivial philosophical work.

The assumption is, ‘If an activity can be described as a procedure, then it can be fully automated.’ This sounds right – but it quietly itself smuggles in metaphysical assumptions. Computers can execute procedures, iterate (repeat) them flawlessly, and thus explore huge conceptual spaces. But metaphysics is not just ‘Apply procedure P repeatedly’: it also involves deciding which procedure is worth applying, where, and when to stop applying it. That decision is not given by the procedure itself. If metaphysics were only recursion, then endless question-looping would be metaphysical success. But actual metaphysical practice involves pruning, abandonment, reframing, declaring questions ‘ill-posed’, and tolerating ambiguity selectively. These are value-based judgements, not mechanical calculations. For instance, if we ask ourself, ‘When should the recursive process stop?’ then we have several possible answers. When it converges?When it becomes useless? When it conflicts with experience? When it violates intuition? When it becomes boring? When it becomes dangerous? No answer is supplied by the procedure itself. So automation doesn’t eliminate metaphysics – it relocates it.

Suppose now we build a ‘Metaphysics Engine’: it recursively analyzes concepts, flags circularity, generates alternative frameworks, and ranks conclusions by coherence. Now we ask, ‘Why this engine? Why these primitives? Why this ranking metric?’ Those questions are metaphysical questions about the machine, which the machine cannot answer because it has to presuppose certain answers to these questions in order to work at all. So automation cannot end metaphysics; it just pushes it up a level.

This mirrors the earlier attempts to end metaphysics. Rudolf Carnap tried to replace metaphysics with formal frameworks; but Alan Turing showed that rule-following does not exhaust meaning, and Ludwig Wittgenstein showed that rule application presupposes rule-applying practices. So metaphysics didn’t die – it reappeared at the meta-level.

The deepest problem here is that procedures don’t generate new values. Metaphysics is not only about conceptual consistency: it’s also about what we care to make sense of, what seems intelligible, and what demands clarification. A metaphysician chooses where confusion or clarity occurs. That choice is not computational, it is existential, cultural, historical. It is a human choice. Caring is contingent, intelligibility is person-relative, and agreement is not guaranteed, even among experts. Therefore, these values cannot serve as foundations. And to be computable they must themselves be formalized.

Metaphysics is procedural enough to be partly automatable, sure. But it’s wrong to say that AI makes metaphysicians unnecessary. A better formulation is, ‘Computers can perform metaphysical labor, but humans must perform metaphysical judgment.’ In other words, if metaphysics is understood as a recursive procedure – as ‘using concepts to explain concepts’ – then much of it can indeed be automated; but what cannot be automated is the decision of which conceptual recursions are worth pursuing, which should be abandoned, which matter to human understanding, and when we have understanding.

A Future For Metaphysics

This is, however, refusing the traditional philosophical hope of a final foundation for metaphysics. I’m saying that there’s no privileged stopping point, there is only a hierarchy of meta-levels. Machines can handle any finite level, but not consider all the levels together, as a human might.

Moreover, since different criteria for choice generate different metaphysics, the real task for metaphysics is not simply choosing answers, but studying the space of criteria. Thus we have a reframing from metaphysics to metacriteria, from questions to selection rules, and from truth to ‘stability under criteria’. Here we stop arguing which answer is right, and start asking, ‘What kinds of criteria generate which kinds of worldviews?’ I also said that humans must perform metaphysical judgment. No two humans agree on all judgements. But under our framework, disagreement becomes data, or evidence of multiple criteria sets, or a mapping problem, not a failure. This makes metaphysics formally analyzable: once framed this way, we can ask precise questions concerning:

Consistency: Does a criteria-set generate contradictions?

Convergence: Does recursion stabilize or diverge? (That is, do we end up with a final answer, or not?)

Expressive power: How many conceptual frameworks does a set of criteria generate?

Computational cost: Is exploration tractable or explosive? (In other words, will the answer be computationally affordable?)

Equivalence: Do different criteria sets lead to equivalent results?

Machines aren’t a threat here; they’re the point. Under this view, humans do not adjudicate metaphysical truth, they propose (or inherit) criteria for investigation, and machines explore the consequences exhaustively. Philosophy becomes the study of conceptual generators rather than conceptual verdicts. This is not anti-human, but it is post-anthropocentric, in that it shrinks the human role drastically, but not to zero. What remains for metaphysicians is:

a. Deciding which criteria are worth modelling (pragmatically, politically, ethically)

b. Proposing initial criteria for metaphysical computations (often culturally inherited)

c. Spotting equivalence classes of results

None of these activities claim authority. They are inputs, not conclusions.

This proposal dissolves old metaphysical anxieties: ‘Is this true?’ is ill-posed under this model, and ‘Is this meaningful?’ is criterion-relative. The criteria for metaphysical computations themselves become testable via generation behavior. Here, metaphysics stops pretending to deliver the world, and becomes a theory concerning possible worlds under logical constraints.

If metaphysics is a recursive procedure applied to concepts, then pushing the procedure to higher levels is an inevitability. But rather than seeking final metaphysical truth, we should study the criteria by which metaphysical ideas are generated. Different criteria yield different conceptual worlds; but here, as I say, disagreement becomes data rather than failure.

On this view, philosophy does not compete with science, nor does it deliver reality ‘as it really is’. Its task is simultaneously more modest and more demanding: to map the space of conceptual generators, or in other words, to define the rules by which intelligibility is produced, and reproduced. Here metaphysics survives not as a doctrine, but as the architecture of understanding.

Sajid Bijapure is a mathematics graduate and a retired bank clerk.