Philosophy & Critical Thinking: Argument reconstruction

What you'll be able to do: Turn a philosopher's prose, or your own, into numbered premises and a labelled conclusion, and tell the difference between an argument that's invalid and one that's merely unpersuasive.

Why prose hides the argument

Philosophical prose is written to persuade, not to expose its own skeleton. A paragraph can smuggle in an unstated premise, beg the question, or shift between two senses of a word, and none of it shows on the page as an error, prose doesn't have to declare its logical form. Reconstruction is the X-ray: restate the passage as premises, numbered P1 to Pn, each a standalone claim, followed by a labelled conclusion.

Once it's numbered you can ask two questions a paragraph blurs together: is the form valid, does the conclusion follow if the premises are true, and are the premises actually true, which is soundness. Most first-term confusion is answering one question while believing you've answered the other.

Worked example: the ontological argument

Anselm's argument in the Proslogion (ch. 2), reconstructed:

P1. God is defined as "that than which nothing greater can be conceived."
P2. It is greater to exist in reality than to exist in the understanding alone.
P3. Suppose God exists only in the understanding, not in reality.
P4. Then a greater being is conceivable, one existing in reality too, which would itself be "that than which nothing greater can be conceived."
P5. But P4 contradicts P1: nothing can be greater than the thing defined as that than which nothing greater can be conceived.
C. Therefore God exists in reality, not only in the understanding.

The argument is valid in form. Gaunilo and Kant didn't attack the logic, they attacked a premise. Kant's move, that "existence is not a predicate," denies P2: adding "exists in reality" to a concept doesn't make the concept greater the way adding a property does, because existence isn't a property that stacks onto a description at all. That's the whole discipline of the distinction in one example, you can grant the form is airtight and still reject the conclusion, because a valid argument with a false premise proves nothing.

Worked example: the problem of evil

The logical problem of evil, in the form associated with Epicurus and used by Hume in the Dialogues Concerning Natural Religion:

P1. If God is omnipotent, God could prevent all evil.
P2. If God is omniscient, God would know about all evil.
P3. If God is omnibenevolent, God would want to prevent all evil.
P4. Evil exists.
C. Therefore no being that is omnipotent, omniscient and omnibenevolent exists.

Valid in form. Theodicies, the free-will defence, soul-making theodicy. Don't attack the logic either. They add a qualifying premise to P1 or P3 (God permits some evil as the necessary cost of free will, or of soul-making), and the entire debate since Plantinga's Free Will Defense has been about whether that added premise survives scrutiny. Reconstruction shows you exactly which premise is under dispute, instead of leaving you with a vague sense that "religion has an answer to this."

How to run it

1. Take the passage, source text or your own draft, and paste it whole.
2. Ask for reconstruction as numbered premises and a labelled conclusion, not a summary.
3. Check each premise is one standalone claim, not two joined by "and" (a hidden extra premise) or a rhetorical question.
4. Ask separately: is this valid as reconstructed, and which single premise is doing the most disputable work?
5. Attack that premise specifically. Not the argument in general.

Reconstruct this argument as numbered premises (P1, P2..) followed by a labelled conclusion (C). Don't summarise, don't evaluate yet, and don't add a premise that isn't in the passage. If a step needs a missing premise to be valid, say so separately and mark it as supplied by you, not present in the text. Then tell me: is the argument valid as reconstructed, and which single premise is doing the most disputable work?

An argument doesn't only have this shape

Classical Indian logic (the Nyaya school) reconstructs an inference in five members, not two premises and a conclusion: pratijna (thesis) , "there is fire on the hill"; hetu (reason) , "because there is smoke"; udaharana (example) , "wherever there is smoke there is fire, as in a kitchen"; upanaya (application) , "the hill has smoke of that kind"; nigamana (conclusion) , "therefore there is fire on the hill." The example clause does work a bare P1.Pn form leaves implicit: it's where the inference rule itself gets licensed, not merely asserted. Reconstructing an Nyaya inference into flat Western premises and reconstructing it in its own five-part form are not the same exercise, the collapse loses something. "Argument" is a culturally specific container, not a neutral one, and that's worth noticing once, not forgetting.

Pitfalls

1. Summary dressed as reconstruction. If you can't point to which sentence became which premise, you paraphrased, you didn't reconstruct.
2. Adding premises silently. Every added premise needs to be marked as added; otherwise you're defending or attacking an argument nobody made.
3. Stopping at validity. Valid tells you the form works. It says nothing about whether you should believe the conclusion.
4. The tell: you can produce the numbered list but can't say, without checking again, which premise is false if you reject the conclusion. If reconstruction only relocates a verdict you already held , "I reject the conclusion, so something must be wrong somewhere" , you've decorated the exercise, not done it.

Try this today

Take the last philosophical claim you half-believed without being able to defend it. Reconstruct it yourself, in your own words, as numbered premises, before you read anyone else's take on where it goes wrong.