How to Win Arguments

When someone makes a claim, I tend to start working backwards through what would have to be true for it to follow. This happens while they are still talking, often before I have decided whether I agree. My attention goes to an assumption, a change in the meaning of a word, or a step that seems to have been passed over. I do not consciously switch into this mode, and I suspect it explains some of the difficulty I have with conversations that other people seem to find quite straightforward.

If a conclusion depends on something I think is wrong, continuing as though I accept the reasoning feels dishonest. Pointing it out can seem to me like helping someone make their thought more precise. People often seem to experience the interruption as pedantry, coldness, or an attempt to take apart something they were trying to communicate. My intention gives me an account of why I intervened; I still need to consider what the intervention did to the conversation.

A statement can be doing several things at once. Someone saying, 'People like me always get taken advantage of,' may be describing repeated experiences, expressing exhaustion, or offering an explanation of why those experiences keep happening. The word 'always' makes a claim that a counterexample could defeat, but producing one would still leave the experiences behind that claim to be understood. If I treat the whole sentence as a universal proposition, I may correct its literal scope while failing to address the experience that prompted it.

When the explanation is offered as a reason that others should accept a conclusion, its assumptions become available for examination. A belief can help someone organise their experience and still require better evidence. Equally, recognising that a belief serves a psychological purpose does not establish that it is false. I would have to examine the claim itself, and remain open to the possibility that the person has seen a pattern I have missed.

Winning an argument can mean securing agreement through confidence or persistence, leaving an opponent unable to reply, or giving an audience a satisfying conclusion. Whether the reasons justify the conclusion requires an assessment that can come out differently from the interaction. I am interested in whether the reasons survive examination by someone who has no wish to agree.

An argument offers reasons for accepting a conclusion. Two questions follow: how well supported are those reasons, and what do they allow us to infer? In a valid deductive argument, the premises cannot all be true while the conclusion is false. Establishing that the premises are true is a further task. A valid argument with true premises is sound; valid form alone leaves the truth of the premises open. A false premise removes that argument's claim to soundness, although the conclusion might still be true for another reason. If someone says a particular vehicle must have an engine because all vehicles have engines, a bicycle is enough to defeat the general premise. It tells us nothing decisive about the particular vehicle. I cannot move from finding a fault in their reason to assuming I have established the opposite conclusion.

Much of what we argue about also depends on evidence that supports a conclusion without making it necessary. Repeated observations may give us a good reason to expect a pattern to continue, while leaving room for a case we have not encountered. Here the questions concern the quality and range of the observations, plausible alternatives, and how much confidence they justify. Demanding deductive certainty from such an argument would apply a standard it was never equipped to meet.

The claim that people should trust their intuition makes these differences easier to see. One possible reason is that an intuition feels compelling, and that compelling feelings are meaningful. Even if we grant that a feeling is meaningful, its meaning might concern what someone fears or expects. A further reason is needed to show that it accurately identifies what is happening. The word 'meaningful' cannot carry us all the way from having a strong response to having a reliable judgement.

A more careful account would ask how the judgement was acquired. Kahneman and Klein identify a sufficiently predictable environment and adequate opportunity to learn its regularities as conditions for skilled intuition. Experience needs to give someone a chance to recognise relevant cues and learn what follows them. Feedback matters because it can expose mistaken expectations, although repeated feedback cannot create a dependable pattern in an environment that lacks one.

Consider a hypothetical mechanic who has repeatedly heard a particular engine fault, investigated it and checked whether the repair worked. There is a reason to take that person's immediate recognition seriously: it has developed through encounters in which an expectation could be checked. The same confidence in judging a stranger's intentions would require a separate account of what was learned and how its accuracy was established. Expertise in one task gives us no automatic measure of reliability in another.

When we are trying to judge accurately, evidence that an intuition reliably tracks the relevant pattern gives us a reason to give it weight. That requires attention to the task, the opportunities for learning and the evidence of accuracy. Even in a familiar domain, an immediate judgement can be checked when the consequences warrant it. A compelling feeling by itself gives us too little information about how it was formed or when to reconsider it.

Arguments that seem to repeat can become more intelligible once those conditions are made explicit. One person may use 'trust' to mean acting without further investigation; another may mean treating the impression as a reason to look more closely. They could agree about the value of experience while disagreeing about what action it permits. Until the difference is stated, each reply can appear to miss something the other person considers obvious.

Philosophy spends considerable time on this kind of clarification. A question about what can be known may already assume a standard of evidence, and a claim about what exists may depend on which forms of observation count. Examining those commitments helps identify what the disagreement actually concerns. Definitions organise the inquiry; claims about the world still need evidence capable of supporting them.

Counterexamples need similar care. One contrary case can refute an unrestricted claim about every member of a category. A claim about what usually happens requires attention to frequency and conditions, so a single exception may already fit it. If I want to challenge an argument fairly, I need to identify the strength of the claim being made before deciding what would count against it. Otherwise I can create an easier position to defeat than the one the person holds.

There is work for both people here. The speaker can make their assumptions explicit, and the listener can check that the reconstruction is one the speaker recognises. A disputed premise may then lead to evidence, a narrower formulation, or an admission that the available information does not settle it. Sometimes disagreement concerns values as well as facts: people can agree about likely consequences and still differ over which losses they are prepared to accept.

My own habit of auditing arguments needs to remain answerable to that process. Noticing a weakness quickly can make it tempting to stop listening before I have understood the position. Any claim I make about someone else's assumptions is itself an interpretation that they should have an opportunity to correct. Applying the method consistently also means examining the reasons for conclusions I find congenial, including the belief that my intervention is helping.

A disagreement has progressed when we can identify what supports a conclusion, what would weaken it and what remains uncertain. Agreement may follow, or we may reach a more precise account of why we disagree. I still find it difficult to leave an unsupported step alone. But if I want the conversation to improve anyone's understanding, I have to give the other person enough room to explain it, and leave my own position open to the same examination.

Further reading: Kahneman and Klein (2009), Conditions for Intuitive Expertise: A Failure to Disagree; Stanford philosophy notes: Validity and Invalidity, Soundness and Unsoundness.