Swinburne's The Existence of God: Bayesian Theism

Suppose several observations each seem to favor an explanation. One concerns why anything exists, another why nature follows intelligible laws, and another why conscious beings exist. Does putting them together strengthen a case, or merely repeat an assumption in different forms? Richard Swinburne's The Existence of God asks how considerations of this kind might combine in an argument for God.

His project is a cumulative argument. He examines whether theism provides a probable explanation of broad features of experience. The conclusion is intended to remain open to uncertainty. Following the argument requires separating a consideration that supports a hypothesis from evidence strong enough to make that hypothesis more probable than not.

Swinburne treats the existence of God as a claim about reality that can be assessed through explanatory reasoning. The proposed God is a personal agent with unlimited power and knowledge, subject to logical constraints, who creates and sustains the universe. Divine goodness is part of the religious conception he develops. The proposal is more specific than saying that the universe has an unidentified cause.

The personal element matters. An explanation in terms of a person appeals to capacities, beliefs, intentions, and reasons for acting. It differs from explaining an event through preceding physical conditions and laws. Swinburne asks whether a personal explanation can account for features that a scientific explanation presupposes.

Consider a hypothetical garden. One explanation of its arrangement concerns soil conditions and the growth of plants. Another concerns a gardener's intention to arrange paths and flowers. The explanations answer different questions and need not compete. This example illustrates personal explanation; it does not establish that the universe resembles a garden in the ways required to support a divine gardener.

The argument must do that additional work. We cannot infer a divine person merely from noticing that some ordinary events have personal explanations. Swinburne proposes criteria for assessing explanatory hypotheses and argues that theism performs well under them. The credibility of those criteria and their application form part of the case.

A valid deductive argument makes its conclusion follow necessarily from its premises. If the premises are true, its conclusion cannot be false. Swinburne's cumulative case instead concerns induction: premises can support a conclusion without guaranteeing it. The evidence can be genuine evidence even if the conclusion ultimately turns out to be mistaken.

He distinguishes two kinds of successful inductive argument. A C-inductive argument increases the probability of its conclusion. A P-inductive argument makes its conclusion more probable than not. The first expresses confirmation; the second expresses a stronger overall judgment. These definitions appear explicitly in his own discussion of arguments for God's existence.

A consideration can meet the first standard while falling short of the second. An explanation that begins with a very low probability might receive support without becoming likely overall. Thus saying that an observation favors theism does not yet establish that theism is probable. The cumulative project needs to address the starting assessment and the combined force of the evidence.

The distinction works in the other direction as well. A piece of counterevidence can lower the probability of a hypothesis without making it improbable overall. Reasoning about a cumulative case therefore requires tracking both the direction and the strength of an update. Counting favorable and unfavorable considerations is insufficient.

Swinburne's proposal is that several arguments can contribute to an overall probable conclusion. This does not mean that each traditional argument independently proves God or that every item discussed in the book supports theism. Questions about suffering, rival explanations, and the force of reported experiences belong inside the evaluation.

Bayesian reasoning distinguishes a hypothesis's probability before the evidence under consideration from its probability after that evidence. The first is a prior probability; the second is a posterior probability. A prior is relative to an evidential starting point. It need not represent what someone believed before encountering any experience whatsoever.

The likelihood of evidence given a hypothesis concerns how probable that evidence would be if the hypothesis were true. It differs from the probability that the hypothesis is true given the evidence. An explanation's ability to accommodate an observation does not by itself confirm it; confirmation requires comparing how probable the observation is under the hypothesis and under its negation.

For evidence to favor a hypothesis against its negation, it must be more probable given the hypothesis than given its negation, provided the relevant probabilities are defined and nondegenerate. A proposed explanation gains no advantage merely because the observation is compatible with it. The comparison asks whether the observation is more to be expected on that explanation.

For theism, this raises questions about divine intentions. An all-powerful being could produce many different worlds. To claim that a world like ours is probable given God, one needs reasons for thinking God would choose something relevantly like it. Power establishes ability; it does not by itself establish a likelihood of acting.

Swinburne appeals to goods God would have reason to realize, including conscious life and responsible agency. This supplies a proposed connection between the hypothesis and the evidence. A critic can question either the proposed goods or the inference from their value to what God would probably create. The probability framework makes that disagreement visible without deciding it.

Consider a numerical illustration of the method, rather than Swinburne's estimate of God's probability. Suppose a hypothesis starts with probability one fifth. Its negation then has probability four fifths, so the odds favoring the hypothesis are one to four. Odds compare the probability of the hypothesis with the probability of its negation.

Suppose an observation is three times as likely if the hypothesis is true as if it is false. Multiplying the initial odds by that likelihood ratio gives odds of three to four. Those odds correspond to a probability of three sevenths, about 0.429. The evidence supports the hypothesis, but it has not yet made it more probable than not.

Now suppose a second observation is twice as likely given the hypothesis, after taking the first observation into account, as given its negation with that same information. The updated odds become six to four, giving probability six tenths. Under these invented assumptions, the combined evidence makes the hypothesis more probable than not.

The numbers show the difference between confirmation and an overall probable conclusion. They do not measure theism. The likelihood ratios and starting probability have been stipulated to illustrate the arithmetic. Disagreeing with their use for a religious hypothesis would require evaluating the proposed probabilities and their grounds.

The conditioning in the second step matters. If both observations merely report the same underlying fact, treating them as independent can count the same support twice. A cumulative argument can handle dependent evidence, but it must assess the additional evidence in light of what is already known. Dependence is a reason for care, not a reason to abandon cumulative reasoning.

Evidence needs to be assessed alongside a hypothesis's prior probability. Swinburne gives simplicity a substantial role in that assessment. A hypothesis that postulates fewer entities, fewer kinds of entities, and fewer arbitrary limits can, on his account, deserve a higher prior than a more complicated rival, other things equal.

He argues that one divine person with unlimited power and knowledge is a simple explanatory proposal. A specified finite limit would invite a question about why the limit occurs at that particular point. His argument treats unlimited attributes as avoiding such arbitrary specification. This is a claim about explanatory description, not about physical size.

The proposal is contestable. A short phrase need not describe a metaphysically simple entity, and positing an agent with knowledge and intentions can seem to introduce substantial explanatory commitments. A critic can ask whether Swinburne's measure favors theism because it already accepts a controversial account of persons and properties.

There is a further question about why simplicity should track probability in an ultimate explanation. In familiar cases, we may have experience with successful and unsuccessful explanations. The existence of the universe provides a different setting. Swinburne argues for applying general explanatory criteria there; a reader must assess the extension, rather than assume that a familiar preference automatically settles it.

This is why the argument's dispute cannot be reduced to arithmetic. People could accept the same probability rules while disagreeing about the simplicity of theism, the available rivals, or the range of possible divine choices. A successful mathematical update would not independently vindicate the philosophical inputs.

Swinburne first draws attention to the existence of a physical universe. His question does not depend simply on whether the universe had a beginning. An everlasting universe would still, on his account, invite a question about why it exists. The proposed explanation concerns the existence of physical reality, rather than an event before its first moment.

Scientific explanations can explain one physical condition through others and through laws. Swinburne argues that this form of explanation leaves the existence of the ultimate physical conditions and laws unexplained. He then asks whether a divine person's sustaining activity supplies a better stopping point than accepting physical reality as a brute fact.

This is Swinburne's philosophical claim about explanation. It should not be presented as a scientific discovery that naturalism is impossible. A critic could question whether ultimate reality requires explanation, whether a personal explanation helps, or whether other forms of explanation remain available. Those are competing assessments of the proposed explanatory framework.

The orderliness of nature supplies a related consideration. Swinburne emphasizes general regularities that allow embodied agents to understand their environment and act within it. Predictable consequences make meaningful responsibility possible on his account. People can learn how to feed, shelter, help, or harm one another because events follow patterns.

This claim differs from treating every temporarily unexplained event as evidence for God. Swinburne's focus is the general existence and character of the regularities. A better scientific theory might explain a local pattern while leaving his larger question about why the most general laws hold. Whether that larger question supports theism remains a philosophical dispute.

Swinburne also asks whether a universe capable of supporting conscious, responsible beings is more to be expected given God. The explanatory link again concerns value: a good God would have reason to create beings capable of experience, action, and relationships. Those reasons are intended to make the hypothesis predictive enough to receive evidential support.

Fine-tuning arguments examine whether life depends on particular laws, constants, or initial conditions. Stating such an argument responsibly requires more than saying that any tiny change would eliminate all life. Claims about the relevant alternatives and their probability need specification. A possibility space and a way of comparing its members cannot simply be assumed from the fact that life exists here.

Swinburne's author-hosted lecture discusses evolution as a scientific explanation while asking why laws and initial conditions permit such a process. The theological explanation is intended to concern that wider setting. It does not require denying that evolutionary mechanisms operate. A critic can accept the mechanisms and dispute that the wider question favors a divine explanation.

Consciousness introduces a related but distinct question. Explaining behavior and explaining subjective experience may raise different issues. Swinburne treats conscious beings as relevant to the case for God. That attribution does not establish that neuroscience can never explain consciousness, or that current explanatory difficulties entail theism.

Alternative proposals also need fair treatment. A multiverse, a necessary physical order, or an unexplained fundamental reality would carry different commitments. Their merits cannot be settled merely by calling them speculative. A comparison must ask what each explains, what it assumes, and whether the probability judgments used against it are defensible.

Swinburne's case includes religious experience. He defends a principle of credulity: in the absence of counterevidence, things probably are as they seem to a subject in experience. He also defends a principle of testimony, giving reports of others a presumption of credibility in the absence of reasons for doubt.

These principles do not say that every apparent encounter is veridical or that every report is trustworthy. Their qualifications matter. A reason to suspect deception, impairment, or a misleading setting can change an assessment. The proposed application to religious experience must therefore consider defeaters rather than count reports as unqualified confirmations.

There is also a difference between an experience's force for the person having it and its evidential force for someone hearing about it. The second person receives testimony, with questions about what was experienced and how accurately it is reported. Identifying those stages helps avoid treating personal assurance as automatically transferable public evidence.

Religious diversity creates another question for the cumulative case. If reports support incompatible interpretations, an assessment must consider what they have in common and where their apparent content conflicts. The mere existence of many reports would not by itself determine which interpretation is correct. Swinburne's principles establish a proposed starting policy for assessment, not an exemption from comparison.

A probable explanation must be assessed against unfavorable as well as favorable considerations. Suffering raises questions about what a perfectly good, powerful God would permit. The argument cannot handle this by counting positive considerations and setting the rest aside. Its ambition is an assessment of the overall case.

The Existence of God itself discusses the problem of evil. It would be misleading to direct the reader only to a separate companion book, as though the question were absent here. Swinburne argues that goods such as responsibility and significant choices bear on what God could permit. Whether those considerations adequately address suffering remains part of the dispute.

His author-hosted lecture acknowledges pain and suffering as possible counterevidence for a person evaluating religious belief. The publisher's abstract of the earlier book edition reports his conclusion that evil does not diminish the probability on his assessment. These formulations should not be compressed into the claim that he simply concedes a specific negative numerical update. The adequacy of his treatment requires examining the particular argument.

An account of why some danger could accompany responsibility would not, without further premises, explain every actual harm. A critic can ask about apparently excessive suffering, unequal burdens, and alternatives that preserve valuable agency. Such questions challenge the proposed likelihood of the world given a good God. They engage the cumulative method directly.

The book asks a reader to evaluate a connected set of judgments. What exactly is the hypothesis? What does it lead us to expect? How probable are the observations without it? How should simplicity affect the prior? Have dependencies and counterevidence been accounted for? Each question supplies a place where agreement about one part can coexist with disagreement about another.

Swinburne's conclusion is that the combined case makes God's existence probable. A reader need not accept that conclusion to understand how the case is meant to work. The important distinction is between establishing a valid pattern of probabilistic reasoning and establishing credible values or comparisons within that pattern.

The opening question about several observations therefore has a qualified answer. They can support one another when they provide additional evidence under a defensible explanatory account. They can also repeat support, conceal disputed assumptions, or fail to outweigh contrary considerations. Swinburne's project earns assessment at the level of those connections, rather than through the number of arguments assembled.

• Richard Swinburne, The Existence of God, second edition (2004), especially its discussions of inductive arguments, explanation, evil, religious experience, and the balance of probability.
• Richard Swinburne, "Faith and the Existence of God: Arguments for the Existence of God" (1988), for his distinction between C-inductive and P-inductive arguments.
• Richard Swinburne, "The Existence of God," the author-hosted Oxford lecture, for a shorter presentation of explanation, simplicity, and the arguments from the universe and its order. This lecture supplements the book rather than replacing its full argument.