Cognitive or Non-cognitive? The Religious Language Debate Weighed: the Cognitivist Account, ‘God Exists’ Read Both Ways, Cognitive Content, Contradictions, Which View Is Most Satisfactory, and an Indian–Western Synthesis

Nature of Religious Language: Analogical and Symbolic; Cognitivist and Non-cognitive.

Official Syllabus: Paper II, Section B, Unit 10

After analogy and symbol, verification and falsification, bliks, policies and language-games, one question remains: is religious language cognitive or non-cognitive? Does “God exists” state a fact that is true or false, or does it do some other job? This final article sets out the cognitivist account in full — realism, Hick, Mitchell, Crombie, Swinburne and William Alston’s case for literal talk about God — reads “God exists” both ways, asks whether religious language carries cognitive content, whether either account falls into contradiction, and which view is most satisfactory. It closes with a graded synthesis drawing on both Western and Indian thought: religious language is cognitive at its core, analogical and symbolic in its reach, expressive and committing in its use, and silent at its limit.

Context & Problem

The Two Families — a Recap

The Quasi-cognitive Middle

• Analogical (Aquinas, Ramsey) and symbolic (Tillich) theories claim that religious language gives some sort of vague or imperfect knowledge of God — more than feeling, less than literal description. They are best counted as cognitive in intention, non-literal in form.

Core Doctrine

The Cognitivist Account

1. Realism:

• A statement is true if it corresponds to the state of affairs it attempts to describe. Most theists are realists about God: they may not know whether a given statement about God is true, but they hold that there is a truth to be known.
• Religious language, on this view, provides special knowledge about transcendental realities; verification and falsification are possible at least in theory.

2. John Hick — faith as cognitive:

• Religious statements like “God exists” and “God is omnipresent” are not expressions of feeling but descriptions of God.
• Faith is cognitive: the believer’s awareness of God rests on direct experience — he apprehends God in and through his material and social environment. Moments of ordinary life have religious significance; religious experience is “the whole experience of a religious person” — he feels the pressure of divine command in conscience and sees God’s grace in the beauty of nature.
• Hick concedes that this experience cannot demonstrate God’s existence — but neither can we demonstrate the existence of the real world; both are matters of interpretative trust (“experiencing-as”).
• Religious statements are verifiable after death — eschatological verification (Article 4).

3. Basil Mitchell and Ian Crombie:

• Religious statements are significant articles of faith: real assertions, against which evil counts, though not decisively (Article 5).
• “God is love” would be falsified by utterly pointless suffering — so it is a real claim (Crombie).

4. Richard Swinburne:

• Meaning requires understanding, not verifiability or falsifiability — the toys in the cupboard (Articles 4–5).
• Belief in God can be supported by a cumulative case of arguments and experience that make it more probable than not.

5. William Alston — literal talk about God:

• There are statements about God with truth values, and they are foundational to religious life: they make explicit the rationale for petitionary prayer, confession, thanksgiving and worship. Without “God hears prayer” being true, petition loses its point.
• Much God-talk is literal, not metaphorical:
  1. Negative statements — “God is immaterial, not restricted to one place, not dependent on anything” — are literal.
  2. Statements about God’s effects — “God comforted me in my distress”, “God forgives the truly repentant” — report effects literally, without detailing the divine action itself.
  3. Very abstract terms — “exists”, “powerful”, “not dependent on anything” — apply to God and creatures in the same sense.
• Partial univocity: more concrete terms (action, intention) can be analysed into an abstract core that applies univocally and a concrete part that does not (Article 2).
• Beyond literal speech: what partial univocity cannot say is left to metaphor, symbol, model and “pointing in a certain direction”. Jesus, asked how to pray, did not give a philosophical definition; he said “Our Father” — a metaphor.

6. How “God” refers:

• Descriptive reference: we pick out God by descriptions that fit God alone — “the omniscient knower”, “the omnipotent agent”, “the source of being for everything else”.
• Direct reference: we also refer to God as the one encountered in prayer, worship and sacrament, standing at the end of a chain of transmission going back to the experiences of the founders of a tradition — as people refer to “Aristotle” without being able to describe him uniquely.
• Why it matters:
  1. If reference is fixed mainly by descriptions, the attributes “call the shots” — what lacks them is not God.
  2. If reference is fixed mainly by experience, people with very different descriptions could still be referring to the same Reality — which opens a door to religious pluralism (Unit 09).

7. “Exists”, “fact” and “real”:

• Does “exists” have one meaning? “Do flying fish exist?” asks about organisms in the sea; “Does the square root of minus one exist?” asks about mathematical conventions; “Does the Freudian superego exist?” asks whether to accept a picture of the psyche. What does “Does God exist?” ask?
• Hick’s suggestion: the common core of “existence”, “fact” and “reality” is “making a difference“. To say God exists is to claim that the universe differs in some specific way from how a God-less universe would be — in its past, present and future (the eschatological difference).

“God Exists” — the Cognitivist and Non-cognitivist Readings

• The key difference: for the cognitivist, “God exists” can be false — and that possibility is what makes it a claim worth staking a life on. For the non-cognitivist, it cannot be false in that way — it is a way of living or speaking.

Does Religious Language Carry Cognitive Content?

Arguments that it does:

1. Believers intend it: ordinary believers are realists who mean “God loves us” as true (Hick). A theory that denies this is a reform proposal, not a description.
2. Worship presupposes it: praise, thanksgiving and petition make sense only if someone is there to hear (Alston).
3. Ethics depends on it: the loving life is rational only if reality is as faith says — the house on rock (Hick against Braithwaite).
4. History matters: claims about the Buddha’s enlightenment, Jesus’ resurrection, the Prophet’s revelation are historical in form; believers care whether they happened.
5. Some God-talk is plainly literal: negations, statements about effects, abstract predicates (Alston).
6. Reasons are given: religious beliefs are argued for and against — proofs, problem of evil, religious experience — which only makes sense if they are cognitive.
7. The verification and falsification criteria have collapsed, so they cannot rule out cognitive content (Articles 4–5).
8. Indian thought agrees: scripture is a pramāṇa (means of knowledge) in Nyāya and Vedānta; Brahman-knowledge is knowledge (jñāna), not a feeling.

Arguments that it does not (or not fully):

1. No empirical test in this life, and believers often allow nothing to count against their claims (Flew).
2. Transcendence: if God is beyond all categories, literal content is impossible (via negativa, neti neti, Tillich).
3. Functions dominate: much religious language is praise, vow, story, ritual (Braithwaite, Wittgenstein).
4. Diversity: conflicting religious claims suggest cultural expression rather than knowledge (non-realists).

Conclusion: religious language does carry cognitive content, but that content is partial, analogical and often symbolic, and it is always embedded in expressive and practical uses. “Cognitive” does not mean “scientific“.

Does Either Account Fall into Contradiction?

• The charge against cognitivism (Braithwaite, Flew): it claims factual status for statements that nothing could test — a “fact” that makes no difference (Article 6).
  • Reply: some facts are verifiable only in the end (Hick) or in principle (Crombie); evil does count (Mitchell); and strict empiricism is itself disputable.
• The charges against non-cognitivism:
  1. Braithwaite’s ethics makes “Lying is wrong, but I intend to lie” a contradiction (Hick).
  2. Tillich needed one literal statement (“God is Being-itself”) to deny that any are literal.
  3. Phillips must say that “true” and “exists” mean something different in religion — yet “Jesus rose” is true if and only if Jesus rose.
  4. The embedding problem: we reason with religious sentences — “If God loves us, we should love one another; God loves us; so we should love one another.” This is a valid inference only if “God loves us” means the same thing in the “if” clause (where nobody is expressing a feeling or a vow) and in the plain statement. A purely expressive reading cannot explain why the reasoning is valid. (Philosophers call this the Frege–Geach problem; it troubles all non-cognitive theories.)
  5. All claims become one: if every Christian doctrine declares the same agapeistic policy, “God created the world” and “Christ rose” would mean the same — absurd.
• Verdict: cognitivism faces an epistemological difficulty (how do we know?); non-cognitivism faces logical and descriptive difficulties (it cannot explain reasoning with religious sentences or what believers mean). The first is the lighter burden.

Which View of Religious Language Is Most Satisfactory?

Criteria for judging a theory:

1. Fidelity — does it describe how believers actually use religious language?
2. Coherence — is it free of self-refutation and contradiction?
3. Transcendence — does it respect the otherness of the divine (avoid anthropomorphism)?
4. Content — does it allow religious statements to say something (avoid emptiness)?
5. Range — does it explain the many uses — assertion, praise, prayer, story, ritual?
6. Resilience — can it meet the verification and falsification challenges?
7. Cross-cultural reach — does it illuminate other traditions?

The answer — no single theory, but a layered view:

• No single theory is fully satisfactory, because religious language does several jobs at once. Each theory captures one layer and fails when it claims to be the whole.
• The most defensible combination:
  1. A cognitive core: some literal claims — negations, effects, abstract predicates (“exists”, “powerful”) — and real assertions held as articles of faith (Alston, Mitchell).
  2. Analogy for the divine perfections — good, wise, loving — in a related, higher sense (Aquinas, Ramsey).
  3. Symbol, metaphor and myth for depth, devotion and the transformation of the whole person (Tillich).
  4. Expressive and committing uses — praise, vow, policy, outlook — recognised as additional, not substitute, meanings (Braithwaite, Hare).
  5. The grammar insight of the language-game view — God is not an object among objects; religious words must be understood in their use — without its fideist isolation (Wittgenstein).
  6. A frank recognition of limit — negation and silence at the edge of speech.
• In one sentence: the positivist criterion was too narrow and has been abandoned even by its authors; the non-cognitivist rescues concede too much, since believers do not take themselves to be merely expressing attitudes; the best account combines analogy for content, the language-game point about grammar for use, and Mitchell’s insistence that a faith which nothing could count against has ceased to be faith and become obstinacy.

Where the Debate Stands Today

• Face value: the default view in philosophy of religion today is that religious statements are what they appear to be — truth-apt assertions — unless strong reasons show otherwise.
• Verificationism is regarded as refuted; the objections to it are widely judged conclusive.
• Non-cognitivism survives mainly in revised forms — such as religious fictionalism, which treats religious talk as a valuable fiction that people may engage in without believing literally (a modern relative of Braithwaite’s stories).
• Apophatic (negative) theology has enjoyed a revival, as has interest in metaphor as a genuine way of knowing — “speaking about one thing in terms which are seen to be suggestive of another“.
• Comparative philosophy has brought Indian, Buddhist and Islamic theories of language (lakṣaṇā, apoha, two truths) into the discussion.

The Other Side — What Non-cognitivists Rightly Insist On

• Religious language is never merely informative: “God is love” is a claim, but it is also a call, a comfort and a commitment.
• The meaning of religious words is learnt partly by living them — grace, agape, bhakti, śaraṇāgati are fully understood only within a form of life.
• Treating religion as bad science misreads it (Wittgenstein against Frazer).
• Critics too must face the tests: Freud’s claims are no more verifiable than the believer’s; the critic plays a language-game of his own.
• When we speak of “loving God“, the word “love” is stretched — “father”, “shepherd”, “love” are shipped into a new context and slightly adjusted; the believer must not suppose that her private intention fixes their meaning. “It is not that there is nothing to be sensibly communicated; the fault lies not in us, but in our language.“

Critical Evaluation

Critics and Defenders

Cross-Tradition Comparison

The Synoptic Table

Indian and Western — One Graded Scheme

Synthesis (Siddhānta)

• The cognitivist account — realism, Hick’s cognitive faith and eschatological verification, Mitchell’s articles of faith, Crombie, Swinburne, and Alston’s literal and partially univocal talk about God, with a theory of reference by description and by experience — holds that religious statements are truth-apt claims about a reality independent of us.
• “God exists” read cognitively states a fact that could be false; read non-cognitively it expresses feeling, intention, outlook, a move in worship or a pointer to Being-itself.
• Religious language does carry cognitive content — partial, analogical and symbolic — embedded in expressive and practical uses. “Cognitive” does not mean “scientific”.
• Cognitivism’s difficulty is epistemological (how do we know?); non-cognitivism’s are logical and descriptive (contradictions, the embedding problem, misdescription of believers). The first burden is lighter.
• No single theory is fully satisfactory. The best account is layered: a cognitive core, analogy for perfections, symbol for depth, expressive and committing uses, the grammar insight of language-games, and negation and silence at the limit.
• India’s abhidhā, lakṣaṇā, vyañjanā and mauna, with verification in anubhava, offer a strikingly complete version of this synthesis — showing that the problem of religious language, and its best solution, are universal.

“The language of faith is the language of symbols.” — Paul Tillich