Arguments from Observation

The First Way begins from motus, change understood as the reduction of potentiality to actuality. A thing cannot actualise the same potential in itself in the same respect: what is merely potentially hot does not, just by being potential, make itself actually hot. The Second Way begins from an order of efficient causes and denies that a thing can be the efficient cause of itself, since it would have to be prior to itself. Neither Way uses the premise “everything has a cause”. The proposed terminus is precisely not a changing or derivatively acting item.

Aquinas need not be tracing a temporal sequence back to the first event. Medieval interpreters distinguish accidentally ordered series, where earlier causes can cease while later agents continue, from essentially ordered series, where an instrument acts only through a concurrent source. A parent can die while a child later becomes a parent; but a stick moves a stone only while the hand moves the stick. Aquinas's point is that a hierarchy made entirely of borrowed causal power never supplies a non-borrowed source, however many members it contains.

This reading answers a superficial “what caused God?” response: God is not introduced as another changing member but as non-derivative actuality. It does not prove the argument sound. Critics can reject the act–potency analysis, deny that every derivative series needs a first member, or propose a fundamental physical reality rather than a personal God. Nor does “and this everyone understands to be God” establish omniscience, goodness or worship-worthiness. Aquinas develops divine attributes later; a fair evaluation judges the limited conclusion the Ways claim before assessing the additional bridge to classical theism.

A contingent being can fail to exist and depends on conditions beyond itself; a necessary being cannot fail to exist, though philosophers disagree about whether necessity belongs to a being or only to propositions. Aquinas observes generation and corruption and argues that if everything were merely possible not to be, then at some time nothing would have existed; nothing could then begin, so the existence of anything now requires something necessary. He distinguishes necessary beings whose necessity is caused from a being necessary through itself.

J. L. Mackie's scope objection targets the move from “each thing at some time does not exist” to “there is a time at which nothing exists”. It resembles inferring that because every runner eventually leaves the track there must be a moment when no runner is on it; overlapping lives block the conclusion. An infinite succession of contingent things could ensure that something always exists. This criticism defeats a temporal formulation, not every contingency argument. A Leibnizian version asks why the entire contingent order exists at all, even if it has no temporal beginning.

The deeper issue is the Principle of Sufficient Reason (PSR): every fact has an adequate explanation of why it is so rather than otherwise. A strong PSR drives the argument beyond contingent explanations to a necessary ground. A critic may accept local scientific explanation while rejecting a demand for an explanation of the totality. The theist then cannot merely repeat “everything needs an explanation”, because that invites the same demand for God; the claim must be that contingent reality needs an external explanation whereas necessary reality explains itself or has its sufficient reason in its nature. Whether that distinction is coherent is the real point of dispute.

In the 1948 BBC debate, F. C. Copleston treats the world as the aggregate of objects none of which contains the reason for its existence. Even an infinite series of such beings would remain unexplained, because adding dependent members does not produce independence. He argues to a necessary being whose reason lies within itself. Bertrand Russell resists both the application of “necessary” to an existing subject and the demand for an explanation of the universe as a whole. His position that the universe is “just there” treats the totality as a brute fact.

The fallacy of composition occurs when a property of every member is invalidly transferred to the whole: every human has a mother, but humanity does not have one mother. If every object has a cause, it does not follow without argument that the universe has a cause. Copleston can reply that his argument is not a simple transfer. He is asking what explains the existence of all the contingent members; the collection does not acquire a sufficient reason merely by being named a universe. Russell can answer that explaining each event in relation to others may be all coherent explanation requires.

Neither side wins by definition. “Brute fact” is parsimonious because it adds no transcendent entity, but it abandons a question which the PSR makes rationally pressing. “Necessary being” preserves explanatory ambition, but only if necessary existence is intelligible and explanatory rather than a label for where inquiry stops. Compare where each view places ultimacy: an unexplained cosmos or an allegedly self-explanatory divine reality. The dispute is about explanatory standards before it is about any particular scientific discovery.

Aquinas observes non-rational natural bodies acting regularly towards ends: processes tend to characteristic outcomes rather than every possible result. Since what lacks awareness cannot direct itself knowingly to an end, he concludes that an intelligent being directs nature, as an archer directs an arrow. This is grounded in Aristotelian final causation. It does not depend on organisms looking like human machines or on discovering a rare arrangement among otherwise random objects.

Paley's argument is analogical and abductive. A watch's ordered parts are arranged to produce a function; their coordination is best explained by intelligence even if the watch is broken, unfamiliar or able to reproduce. Organisms display an even richer coordination of parts for functions, so Paley infers a designer. The argument is often represented as “complex things have designers”, but complexity alone is insufficient—a pile of stones can be complex. Specified complexity, purposive arrangement or functional adaptation supplies the intended evidential feature.

The distinction matters under criticism. Darwinian natural selection directly competes with Paley as an explanation of biological adaptation: heritable variation filtered by differential reproduction can accumulate design-like fit without foresight. It does not directly refute Aquinas's metaphysical claim that law-governed tendencies themselves require intelligence, though it may encourage a non-teleological interpretation. Conversely, accepting evolution does not prevent a theist from treating the whole lawful process as created; that is a theological compatibility claim, not a scientific alternative to selection.

Calling an argument a posteriori identifies its starting point, not the strength of its conclusion. Aquinas begins from change, causation and contingency; Paley begins from organised complexity; Tennant and Swinburne focus on intelligible order, beauty or life-permitting conditions. In every case, observed features are combined with a metaphysical principle about what counts as an adequate explanation. No one sees a first cause or designer in the same way that one sees a moving object. The conclusion is an inference beyond what is immediately observed.

This matters for the common contrast between inductive and deductive reasoning. Aquinas presents the Ways in compressed demonstrations: if the premises about derivative actuality and ordered causes are necessary, a first source follows. The premises are nevertheless known through experience and include contestable metaphysical judgements. Paley's watch reasoning is more clearly analogical and probabilistic. Later fine-tuning arguments are often abductive: theism is proposed as the best explanation among chance, necessity and selection effects. A student should classify each argument rather than assigning one label to the whole topic.

Explanatory scope and explanatory power must also be separated. A divine cause has enormous scope because it could account for the existence and order of everything. Whether it has power depends on what predictions or expectations theism supplies. If any possible universe is compatible with an unlimited creator, invoking God may not make this universe more expected. Conversely, physical explanations can be precise about processes while leaving the existence of laws or initial conditions unexplained. The arguments operate at different explanatory levels.

The best assessment therefore asks comparative questions. What exactly is the evidence? Which rival hypotheses are live? Would the evidence be less surprising if theism were true? Does the proposed explanation introduce equal or greater complexity? A conclusion that an argument is "not proof" says little, because most reasoning about unobserved causes is non-demonstrative. The serious issue is whether the observation raises the probability of a divine explanation and, if so, how far it identifies the God of classical theism.

Aquinas' First Way uses Aristotle's distinction between potentiality and actuality. Change occurs when a real potential is actualised: wood that is potentially hot becomes actually hot through something already hot or able to produce heat. Nothing can be potential and actual in the same respect at the same time, so what is changing cannot be the complete actualiser of that very change. Aquinas is not denying self-movement in ordinary speech; an animal moves itself because one actual part or power initiates change in another.

The relevant regress is often misunderstood as a line of earlier events. Aquinas can allow that a father is generated by a father through an indefinitely extended accidental series: an earlier member need not remain active for a later one to exercise its own power. An essentially ordered series is different. A hand moves a stick which moves a stone; the stick has instrumental power only while the hand acts. If every member possessed only borrowed actualising power, adding further instruments would never supply a source. The argument seeks a non-derivative actualiser sustaining change now, not the first event at a date in the past.

This distinction answers the reply that an infinite universe has no first moment, but it does not make the conclusion automatic. A critic can deny that all causal power is derivative in the required hierarchical sense, treat fields or fundamental laws as non-instrumental, or accept an infinite explanatory structure. Aquinas must show that finite changing beings depend here and now upon a source whose actuality is not received, not merely assert that an infinite series sounds impossible.

Even if a first actualiser is granted, further work is needed to reach God. "Unmoved" initially means that the source does not acquire the actuality by which it actualises this series. Classical theism develops this into pure actuality, immateriality, necessity and unity. Those deductions depend on rejecting composition and unrealised potential in the first cause. A precise essay can credit the First Way with a sophisticated present-dependence argument while asking whether its act-potency metaphysics is the only adequate account of change.

The Second Way begins from efficient causes. Nothing can be the efficient cause of itself because it would have to be prior to itself in the relevant explanatory order. Remove a cause and its dependent effect is not produced; therefore a series made entirely of caused causes needs a first member. As in the First Way, "first" is principally first in dependence rather than simply earliest in clock time. A cause can be temporally simultaneous with its effect, and an existing world could still depend continuously on a source.

The premise that nothing causes itself is more limited than the slogan "everything has a cause". Aquinas never applies a universal causal rule and then exempts God. His claim is that what receives or derives existence and causal power cannot be its own complete source. The proposed first cause is not an arbitrary exception but a being whose actuality is non-derived. A critic can still deny that such a contrast is coherent, but the school-level objection "Who caused God?" misses the form of the argument unless it shows that the proposed God has the dependent features which trigger the principle.

A deeper question concerns what efficient causation explains. Physical causes ordinarily transform an existing state under laws. Aquinas ultimately uses causal language to explain the existence of finite substances, not merely changes within the universe. A physical account of earlier conditions can be complete at its own level while the metaphysical question "why does this dependent order exist at all?" remains. Russell's reply is that demanding a further cause for the total order may be illegitimate; explanation properly terminates in the universe or its fundamental facts.

The dispute is thus about stopping rules. The theist stops at a necessary source which explains without requiring the same kind of explanation. The naturalist stops at a fundamental physical reality or brute fact. Neither can win by accusing the other merely of stopping. One must ask whether divine simplicity and necessity are intelligible and independently motivated, and whether the physical terminus still exhibits contingency. The more the first cause is enriched into a personal creator, the more attributes require arguments beyond the bare causal structure.

Aquinas observes that ordinary things are generated and corrupted; they are able not to exist. His compressed text then appears to argue that if everything were contingent, there would have been a time when nothing existed, after which nothing could begin without a cause. Therefore at least one necessary being exists. Critics such as J. L. Mackie point out that even if each thing fails to exist at some time, it does not follow that there is one time at which all fail to exist. The expiry dates of contingent beings need not coincide. This is a quantifier shift, not merely an empirical doubt.

Defenders can read the argument more charitably. Medieval "possible not to be" may indicate an intrinsic tendency to corruption, and an infinite past of universally perishable beings was thought to entail eventual non-being. That premise is questionable under modern probability and cosmology. More promisingly, the Way later distinguishes necessary beings whose necessity is caused by another from a being necessary through itself. The deepest target may be an order whose members do not contain the sufficient ground of their own existence, whether or not they overlap eternally.

On that modal reading, the argument resembles Leibniz. A totality of contingent facts remains contingent even if it has no temporal beginning; describing each internal dependence does not yet explain why the entire network exists. A necessary ground would explain the actuality of one complete contingent order rather than another or none. The critic can reject the Principle of Sufficient Reason, deny that the totality forms one further fact, or propose that the laws and basic state are metaphysically necessary.

Necessity must also be defined. Logical necessity belongs to propositions whose denial is contradictory; metaphysical necessity concerns what could not have failed to exist given its nature. Aquinas is not claiming that "God does not exist" is an obvious contradiction available from the concept alone. His a posteriori route aims to discover a source which lacks received existence. Evaluation should not mix this conclusion with Anselm's ontological claim. The temporal formulation is vulnerable, while the dependence formulation survives as a serious dispute about whether contingent reality can be ultimate.

In their 1948 radio debate, Frederick Copleston defines a contingent being as one which does not contain within itself the reason for its existence. He argues that the collection of such beings also lacks an internal sufficient reason and must be explained by a being which has the reason for existence in itself. His use of Leibniz's Principle of Sufficient Reason is not the claim that every event has an earlier event-cause; it is a demand that there be an adequate explanation for positive reality.

Russell resists both the move to a totality and the unrestricted principle. A class of human beings has a mother for each member without the class having one mother. Likewise, explaining individual events need not require another entity which causes the universe as a whole. His famous brute-fact response denies that every meaningful "why" question has an answer. The universe is not obviously the sort of item for which an external causal explanation applies, especially if causal concepts are learned within it.

Copleston replies that this is not a straightforward fallacy of composition. If every book in a collection is borrowed, the collection contains no unborrowed source of ownership; adding borrowed items does not create self-sufficiency. The property at issue is explanatory dependence, which may transfer from every member to the whole more plausibly than properties such as having a mother or weighing little. Russell can answer that the universe is not merely a set of objects: laws and relational structure may explain members mutually without forming one additional dependent object.

The debate exposes a methodological divide. Copleston sees intelligibility as requiring a terminus whose existence is self-explanatory. Russell regards the demand for total explanation as an optional metaphysical preference, not a rule established by experience. Theism wins only if the Principle of Sufficient Reason is defensible and "necessary being" explains more than the phrase "brute universe". Naturalism wins only if stopping at fundamental reality is not an evasion of a legitimate dependence question. Neither side can settle this by repeating its preferred stopping point.

Hume's account of causation undermines easy movement from observed events to a necessary first cause. Experience presents constant conjunction: one type of event regularly follows another. It does not present a visible power or necessary connection binding them. The expectation that similar causes will produce similar effects arises from habit. Extending causal reasoning to the origin of the universe is therefore precarious, because there is no repeated experience of universes beginning against which the inference can be calibrated.

This does not show that causal reasoning is useless. Hume relies on it in ordinary life and inquiry. The point is that causal principles are not demonstrable a priori and should remain proportioned to observed regularity. "Whatever begins to exist must have a cause" feels compelling, but denying it does not generate a formal contradiction like denying that a triangle has three sides. A defender may treat the principle as a necessary metaphysical truth grounded in the intelligibility of change rather than an empirical generalisation. The burden is then to justify that stronger status.

Hume also resists transferring rules from parts to wholes. Every event within the universe may have a cause while the universe does not have a further cause; explaining the twenty particles may explain the group without a twenty-first explanation. Yet the defender can distinguish aggregative facts from the obtaining of the complete contingent order. Internal causes explain why one state follows another, but not why there is a causally structured reality. The argument survives only by formulating that second question without treating the universe as an ordinary event within a wider timeline.

Kant later adds that causality organises possible experience and cannot simply be projected beyond all experience to a transcendent being. Aquinas' metaphysical causation is not identical to Kant's category of temporal event-causation, so the objection requires careful targeting. A strong judgement asks whether causal language can be analogically extended to existential dependence. If it cannot, the cosmological conclusion exceeds its evidence; if it can, a first cause need not be an earlier item inside the cosmos.

In Dialogues Concerning Natural Religion, Hume's Philo asks whether the world is sufficiently like a human artefact for reasoning by analogy. We have repeated experience of houses and machines being made, but no experience of worlds being produced. The universe may resemble an organism or self-ordering system more than a machine, and selecting the machine analogy can build the desired conclusion into the description. The more unlike the cases, the weaker the analogical inference.

Even if order supports intelligence, the effect limits the cause. A finite, mixed and imperfect world does not license an infinite, perfect designer. The design could be an early attempt, the work of several cooperating deities, or the product of generation rather than manufacture. Poor design and suffering make perfect goodness especially difficult to infer. The Epicurean hypothesis adds a selection effect: given enough rearrangements of matter, stable orders may eventually arise and persist, so observers unsurprisingly find themselves within one.

An abductive defender can reformulate the argument: theism need not resemble a watchmaker case exactly if it makes order more probable than rival hypotheses. But probability must be comparative. The defender must specify evidence, background knowledge and alternatives; saying simply that chance is impossible creates a false dilemma between one-step randomness and design. Natural laws, evolutionary selection and cosmological models are structured alternatives. Hume's lasting contribution is to police the strength of the conclusion: no cause should be assigned more attributes than the observed effect warrants.

F. R. Tennant develops an anthropic argument from the universe's intelligibility and suitability for rational moral agents, and an aesthetic argument from humans' capacity to appreciate beauty beyond immediate survival. These are cumulative rather than demonstrative: several features are said to fit more naturally within theism. Richard Swinburne similarly treats temporal order—the conformity of events to simple laws—as evidence which a personal explanation might unify. The question becomes which worldview makes a law-governed, consciousness-permitting universe less surprising.

Contemporary fine-tuning arguments focus on the narrow ranges of constants or initial conditions compatible with complex embodied life. The evidence is not that the universe was designed “for us” in every detail. It is conditional: if values had differed, life of our general kind would not arise. Responses include physical necessity, a multiverse with observer selection, unknown deeper laws and design. The anthropic principle reminds us that observers can observe only observer-compatible conditions; it explains selection bias but, by itself, not why there is a life-permitting domain to select.

Probability claims face a measure problem because we do not know a distribution over possible constants or divine intentions. A multiverse may increase the chance that some region permits life, yet requires a mechanism and may generate its own fine-tuning questions. Design can explain intentional selection, but predicts life only if there is reason to expect a designer to value embodied agents. The strongest conclusion is modest: fine-tuning may contribute to a cumulative case, while underdetermining both the identity of a designer and whether personal explanation is better than future physical explanation.

Paley asks a reader to imagine finding a watch on a heath. Its parts are arranged and adjusted to produce motion and indicate time; even if the mechanism is imperfect or not fully understood, purposive coordination warrants an inference to a maker. Organisms exhibit a much more elaborate fit of parts to functions, so Paley infers an intelligent designer. The argument does not depend only on visual resemblance between a watch and an eye. Its deeper principle is that specified functional organisation is characteristically produced by intelligence.

Hume's Dialogues, published before Paley, questions the comparison. We have extensive experience of machines being made but no experience of worlds being made. The universe may resemble an organism, vegetable or self-ordering system as much as an artefact. Like effects justify only like causes: a finite, mixed and changing world would at most support a finite, multiple or inexperienced designer. Cooperation among many human makers makes polytheism no less analogically available than one perfect God.

A modern defender can recast the reasoning as inference to the best explanation. We routinely infer unseen agency from organised traces without observing that particular production event. The issue becomes whether natural processes can generate the relevant organisation. This avoids requiring the universe to be literally a machine, but it also opens direct competition with evolutionary and self-organising explanations. Design can no longer be inferred from complexity before alternative mechanisms are considered.

Paley anticipated limited objections: ignorance of every component or evidence of imperfection does not erase obvious function. A broken watch can still have a maker. However, poor and harmful design changes the attributes supported. If natural structures contain waste, predation and avoidable suffering, the evidence does not lead straightforwardly to an omnipotent and perfectly good designer. The argument's strongest conclusion may be modest intelligence behind some order; classical theism requires independent reasoning about unity, power and goodness.

Natural selection explains how populations acquire the appearance of purposive fit without foresight. Heritable variation affects reproductive success; cumulative selection retains variants suited to an environment. Complex organs need not appear in one chance leap because intermediate functions can be advantageous. This directly challenges Paley's use of biological adaptation as evidence that an intellect arranged each part for its role.

Calling evolution "chance" is misleading. Mutation includes stochastic elements, but selection is a non-random differential process. The Epicurean image of particles accidentally assembling a finished eye understates cumulative retention. Conversely, evolution does not imply that every trait is optimal. Historical constraints, trade-offs, drift and changing environments predict improvised structures. Such imperfections support Darwinian history and weaken an argument from perfect engineering.

A theist can accept evolution as the means by which divine purposes are realised. That preserves compatibility but changes evidential force. If the same biological data are expected under unguided evolution, they do not independently favour special design merely because a believer can place the process within providence. Theistic evolution becomes a wider metaphysical interpretation of laws, existence and intelligibility rather than an alternative biological mechanism. It should not compete with natural selection in filling the same explanatory gap.

Evolution has less direct force against cosmological arguments and cosmic fine-tuning. It presupposes replicators, stable regularities and life-permitting chemistry; it does not explain why there is a contingent law-governed reality. Yet invoking God to explain those conditions must avoid a retreat into whatever science has not yet explained. A durable design argument identifies a principled feature—perhaps the existence of an intelligible order—rather than moving its target whenever a natural mechanism is discovered. Darwin defeats Paley's inference from adaptation more decisively than it defeats every argument from observation.

Hume's Epicurean hypothesis imagines finite particles rearranging through unlimited time until a stable order happens to arise. The hypothesis anticipates a selection effect: observers necessarily find themselves in a region able to sustain observers, so their observation of order is not an unbiased sample. Contemporary multiverse proposals give this reasoning a new form. If physical reality contains many domains with varied constants, at least some life-permitting domains may be unsurprising even when any one is rare.

The anthropic principle by itself is not a productive cause. It explains why we cannot observe an observer-prohibiting universe, but not why a life-permitting domain exists. Its force depends on a mechanism or theory which generates sufficient variation. Likewise, a multiverse is not a cost-free answer invented solely to avoid God; some inflationary or cosmological models may motivate it. But its measure, testability and explanatory scope remain disputed. Comparing design with "pure chance" while ignoring selection is incomplete, and invoking infinitely many universes without evidence can also weaken parsimony.

Fine-tuning arguments ask how likely a life-permitting range would be under different hypotheses. Theism may predict ordered embodied knowers, making the conditions less surprising. Yet this depends on prior assumptions about what a perfect God would choose. A creator could make disembodied minds, different laws or no creation; the probability assignment is not straightforward. Naturalism may regard constants as brute, necessary under a deeper theory, or environmentally selected. None of these possibilities is established merely by naming it.

Swinburne's cumulative approach treats simple theism as a broad personal explanation of laws and conscious life. Critics challenge whether an infinite mind is simpler than a physical theory. The dispute is not settled by counting entities: theoretical simplicity concerns kinds, adjustable assumptions and explanatory unification. Fine-tuning can raise a meaningful comparative question, but claims of precise probability exceed what is known about the space of possible laws. Its conclusion should remain probabilistic and conditional.

F. R. Tennant develops an accumulative case from the world's intelligibility, its capacity to produce moral agents and its beauty. These are not identical to Paley's biological mechanisms. Mathematical order makes rational inquiry possible; consciousness introduces subjects able to know and value; beauty appears to exceed what survival alone requires. Taken together, Tennant argues, they fit a purposive creator better than an indifferent accident.

Each strand is contestable. Minds adapted within the world would be expected to grasp some regularities on naturalism, while aesthetic preferences may develop from pattern recognition, social signalling and environmental association. Calling a landscape beautiful describes a relation between world and valuing subject, not necessarily an intention placed in the object. Nevertheless, the existence of consciousness and successful abstract mathematics remains philosophically striking even if evolutionary accounts explain particular capacities.

John Stuart Mill reverses the evidential direction. Nature includes predation, disease, disaster and extravagant suffering that a morally decent human designer would try to prevent. If natural order reveals character, it seems at least as compatible with limited power, indifference or competing designers as with perfect goodness. Hume's character Philo similarly restricts the conclusion to what the mixed effect warrants. One cannot infer infinite perfection from a finite and morally ambiguous world.

A theist may incorporate free will, soul-making, stable laws and eschatological fulfilment, but these are additional explanations, not data supplied by design itself. A cumulative case must cumulate adverse as well as favourable evidence. The strongest assessment asks whether beauty and intelligibility remain more expected under theism after apparently gratuitous suffering is included. Design arguments can contribute to a wider case, but observation alone underdetermines whether the source is one, unlimited, personal and good.