一阶逻辑学习疑问：为何假设不能是开语句？

Great question—this is a super common point of confusion when first diving into formal systems for first-order logic, so let’s break it down in plain terms.

First, let’s align on key definitions to make sure we’re talking the same language:

• An open sentence (or open formula) is a logical statement with at least one free variable, like P(x) or x + y = 10. Its truth value isn’t fixed—it depends entirely on how we assign values to those free variables (e.g., x > 7 is true if x=9, false if x=4).
• A closed sentence (or proposition) has no free variables—think ∀x P(x) (all x satisfy P) or P(a) where a is a specific constant. These have a definite truth value in any given model.

Now, here’s why open sentences don’t work as hypotheses:

1. Open sentences lack fixed truth values #

Formal systems for first-order logic are built to preserve truth: if your hypotheses are true in a model, your derived conclusions should also be true in that model (this is called soundness). But open sentences aren’t definite claims—they’re more like "templates" that only become true or false when you plug in specific values for their free variables. If you start with a hypothesis that’s sometimes true and sometimes false, you can’t guarantee that any conclusions you draw will hold consistently across all interpretations.

2. They break core inference rules #

Take the Universal Generalization (UG) rule, which lets you derive ∀x P(x) if you’ve proven P(x) where x is a "general" free variable (not tied to any assumptions). If you allowed an open sentence like P(x) as a hypothesis, you could immediately apply UG to get ∀x P(x)—which is obviously absurd! For example:

Hypothesis: x is a prime number (open sentence)
Apply UG: All x are prime numbers (closed sentence)

This is a totally invalid inference, and allowing open hypotheses would make the formal system unsound (it would let you derive false conclusions from vague, non-definite premises). To prevent this, rules like UG explicitly require that the free variable in your derived formula doesn’t appear freely in any hypothesis.

3. Formal systems are designed for definite propositions #

First-order logic’s formal systems exist to study logical relationships between concrete, truth-apt statements (closed sentences). Open sentences are useful building blocks for those closed sentences (via quantifiers), but they aren’t standalone claims that can serve as stable premises for reasoning. Allowing open hypotheses would complicate the system’s semantics way beyond its intended purpose—we’d have to track variable assignments through every step of the proof, which defeats the point of having a clean, generalizable formal system.

In short, hypotheses need to be statements with fixed truth values to ensure the reliability and consistency of the formal system. Open sentences just don’t fit that bill.

内容的提问来源于stack exchange，提问作者Tantaros