Argument From Bayesian Probability
👍
P1: If a hypothesis explains the data better than alternatives (i.e., yields a higher posterior probability), it is rational to accept that hypothesis.
P2: Theism (God exists) is a hypothesis with a non-zero, reasonable prior probability P(G) > 0 based on metaphysical, historical, and experiential grounds.
P3: On the hypothesis of theism (G), the following are highly probable:
a) A life-permitting universe (fine-tuning)
b) Conscious, rational beings
c) Objective moral knowledge
d) Religious experiences
e) Mathematical discoverability and intelligibility of nature
i.e., P(D | G) is very high
P4: On the hypothesis of atheism/naturalism (¬G), these features are extremely improbable or unexplained:
a) Fine-tuning must be brute, uncaused, or one lucky outcome
b) Consciousness and normativity lack a grounding
c) Moral realism becomes inexplicable
d) Religious experience is just cognitive malfunction
So, P(D | ~G) is very low
P5: Therefore, Bayes’ Theorem implies that P(G | D) the posterior probability that God exists given the data is significantly higher than P(G) alone.
P6: Rational belief should proportion to posterior probability, not just prior assumptions.
C: Therefore, belief in God is epistemically justified and rational, since the data we observe is far more expected under theism than under atheism.
Objections:
- “But priors are subjective, what if someone gives P(G) = 0?”
A prior of 0 is irrational unless one has logical certainty of God’s nonexistence. Since theism is logically possible and supported by multiple philosophical arguments (e.g., cosmological, moral, modal), assigning a zero prior is unwarranted and dogmatic.
- “This argument doesn’t prove God just makes belief more reasonable.”
Correct, it is a cumulative probabilistic case, not a deductive proof. However, when combined with other arguments (e.g., contingency, morality, consciousness), Bayesian reasoning helps formulate this data into a coherent worldview.
- “Multiverse explains fine tuning!”
Even if a multiverse exists, it must be finely tuned to produce universes. Also, multiverse models lack empirical confirmation and shift the problem of ordered structure back a level they don’t eliminate the need for an explanation.
Note: Richard Swinburne’s method extracted an almost 90% chance of God’s existence. But that’s a discussion for another day