Why Do Your Beliefs Sometimes Clash? The Hidden Rules of Thinking
Ali’s Flying Problem
Imagine Ali. He’s convinced he’s Superman and that Superman can fly. Yet he also insists he can’t fly. Something is off with his thinking. If each belief stood alone, we could simply ask: does he have good evidence? But Ali’s problem isn’t just about evidence. It’s about how his beliefs fit together. Believing both that he can fly (since Superman can) and that he can’t fly seems like a special kind of mess. This is the puzzle of structural rationality.
To see why it matters, consider Jo. Jo intends to kill Jay in the most violent way possible. She believes a cleaver is the most violent weapon. Yet she does not intend to use the cleaver. Something is wrong with Jo’s thinking. If we said she is irrational only because she lacks good reasons, we might conclude that her nasty goal gives her a strong reason to pick up the cleaver. That’s bizarre — a cruel goal shouldn’t create reasons to commit horrible acts. The philosopher Stephen Darwall (b. 1946) noticed this in the 1980s. He argued we need a kind of irrationality that is about how our mental states hang together, not about whether each one is supported by reasons. He called these “principles of relative rationality.” Today we call it structural rationality.
The idea was developed further by T. M. Scanlon (b. 1940) and John Broome (b. 1947). They showed that separating coherence from reasons lets us say someone is irrational for having contradictory beliefs or mismatched intentions, without having to say those messed-up attitudes give them good reasons to do or believe anything. This move protects us from bootstrapping — the strange idea that a crazy belief could force a reasonable intention.
So structural irrationality is not about getting the facts wrong. It’s about the shape of your mind.
Two Kinds of Rationality: When Good Reasons Aren’t Enough
Philosophers often draw a line between two kinds of rationality. Substantive rationality is about correctly responding to your reasons. If you have strong evidence it will rain but you don’t take an umbrella, you’re substantively irrational — your reasons point one way, and you go another. If you intend to go to medical school but refuse to take the entrance exam, your reasons for that goal should also push you toward the means. Structural rationality is different. It doesn’t care whether each attitude is a good response to reasons. It cares about how your attitudes combine.
Imagine Ravi. Ravi believes that all taxes are illegitimate. He also believes that income taxes are legitimate. A left-leaning friend might think his first belief is wrong; an anarcho-capitalist might think his second is wrong. Yet both can agree that Ravi’s combination of beliefs is structurally irrational. Their judgment doesn’t depend on knowing which belief has the best reasons. The failure is the clash itself.
This distinction has four features that make structural rationality stand out. First, it’s about combinations, not single attitudes. Believing p and believing not-p is structurally irrational, even if each belief, by itself, might seem okay to some observer. Second, we can judge structural irrationality without knowing which attitude is substantively right — just the pattern matters. Third, we don’t need to know the agent’s situation or evidence. If someone intends to go shopping, believes they must bike to get there, but lacks an intention to bike, that’s structurally off regardless of whether the shop is actually nearby. Fourth, structural requirements are often formal. The rule “don’t believe both p and not-p” applies to any proposition, no matter its content. It’s not about what you believe; it’s about the shape of your believing.
Some philosophers, like Scanlon and Broome, think that all genuine requirements of rationality are structural. They argue that rationality supervenes on the mind — you can’t become irrational without a change in your mental states. Since reasons often depend on facts in the world (like poisonous fish you don’t know about), they say responsiveness to those reasons isn’t a requirement of rationality but something else, like doing what you ought. Others push back, saying substantive rationality matters too, because you can be irrational for ignoring the evidence you do have. This debate still simmers.
The Akrasia Tangle: When You Believe You Should but Don’t Intend To
Have you ever believed you should study, yet didn’t intend to do it? That’s a mild form of akrasia — acting against your own best judgment. Many philosophers say this is a structural failing. There seems to be an enkratic requirement: you shouldn’t believe that you ought to do something and fail to intend to do it. But how exactly should we state that rule?
Imagine Ava, an environmental activist. Somehow she comes to believe that she ought to extinguish all human life. It’s a terrible, unjustified belief. Must she now form an intention to carry out mass murder to be rational? Many would answer no. This is where the scope of the requirement matters.
On a narrow-scope view, the requirement says: If you believe you ought to φ, then rationality requires you to intend to φ. With this reading, Ava’s crazy belief forces a requirement to form a heinous intention. Worse, if rational requirements are normatively important — if you should do what rationality demands — then Ava has a reason to intend genocide. That’s a classic bootstrapping problem: a rotten belief creates a reason for a rotten intention. Most philosophers want to avoid that.
The wide-scope view, defended by Broome and others, puts the whole conditional inside the requirement: Rationality requires that (if you believe you ought to φ, then you intend to φ). This is symmetric. Ava can satisfy the requirement either by forming the intention or by dropping her belief that she ought to kill everyone. Since the belief is monstrous, dropping it is fine. Wide-scopers say this is the right result — you should not be trapped by your worst beliefs.
But narrow-scopers reply that there’s an important difference between forming an intention based on your normative belief and abandoning a belief just because you lack the intention. It feels more rational, they say, to move from a belief that you ought to act to the intention to act, than to give up a belief because you can’t bring yourself to follow through. So the debate isn’t settled.
Similar puzzles arise for doxastic attitudes. Suppose Darren believes that his evidence supports the claim that the earth is flat, but he doesn’t believe the earth is flat. That’s epistemic akrasia. A narrow-scope requirement might say he must believe the earth is flat. The wide-scope version lets him instead drop the belief about what his evidence says. Again, wide-scopers argue that we shouldn’t be forced by a mistaken higher-order belief to adopt a crazy first-order belief.
The scope debate shapes how we think about the very rules our minds follow. It asks: can rationality let you walk away from your own declared standards, or does it chain you to them?
Why Should We Care About Coherence?
Now a deeper question: even if there are structural requirements, do we have any good reason to follow them? Niko Kolodny (b. 1972) famously asked, “Why be rational?” Usually, reasons point to something valuable: we have reasons to avoid pain, to seek truth, to keep promises. But what’s so valuable about mere coherence? If your contradictory beliefs are both false, fixing the clash doesn’t make you closer to the truth. If you’re akratic and never act on your best judgment, that might be bad for you — but is the badness just that you’re being incoherent?
Some philosophers try to give a derivative answer: being coherent helps you achieve other things that matter. Michael Bratman (b. 1945) argues that structural rationality is important for self-governance — it helps you act like a unified agent rather than a collection of conflicting impulses. Others say that coherence makes you interpretable to others, which is needed for cooperation. The problem is that it’s not clear that coherence always promotes these goods. A coherent villain who intends to conquer the world and takes all the necessary means may be extremely self-governed, but that doesn’t seem to give him a reason to be that way.
Other philosophers, like H. A. Prichard with morality, take a non-derivative route: there are just basic reasons to be rational, period. But critics call this fetishistic. Why care about coherence for its own sake? And any such reason might be a wrong-kind reason — like someone offering you money to believe a contradiction. That’s a reason, but it’s not the kind of reason that makes the belief itself reasonable. It’s hard to see why structural rationality would generate the right kind of reasons.
A different strategy is to reduce structural irrationality to substantive irrationality. The philosopher Benjamin Kiesewetter suggests that whenever you have an incoherent set of attitudes, at least one of them must fail to be substantively rational, no matter what your situation is. For instance, you can never have good reasons that simultaneously support believing p and believing not-p. So the incoherence guarantees a failure to respond to reasons properly. This view doesn’t add extra “reasons of rationality”; it just says structural irrationality is an especially strong form of messing up with your actual reasons. However, critics point to cases where each attitude in the incoherent set is individually permitted. You might be allowed (but not required) to intend to go to the supermarket, allowed to believe you must bike to get there, and allowed not to intend to bike. Individually, no failure. Yet the combination feels off. So the reduction may not be complete.
The question of why coherence matters remains a live, difficult puzzle. It forces us to ask: is having a tidy mind a real value, or just a side effect of other valuable things?
What This Means for You
Most of us don’t walk around believing we’re Superman. But we do hold beliefs that don’t fit together neatly. Maybe you believe that lying is always wrong, but also think a little white lie to save a friend’s feelings is okay. Maybe you fully intend to ace your history test and know you must study now, yet you don’t form an intention to open the book. These everyday clashes are cases of structural irrationality — the same kind philosophers debate.
Understanding structural rationality can sharpen the way you think. It reminds you that being reasonable isn’t just about having the right facts. It’s about how you arrange those facts in your head. If you catch yourself believing two contradictory things, the rational pressure isn’t about which one is true — it’s about the conflict itself. You might resolve it by dropping one belief, or by refining both until they fit. The choice is yours, but the pressure is real.
This also matters for how we judge others. When a politician says one thing and does another, we might call them a hypocrite. That’s a structural failing: their stated belief doesn’t match their actions. But if we think rationality includes only responding to reasons, we might excuse them by saying they have good reasons for both. The structural view captures something the purely substantive view might miss — that a mismatch between word and deed can be irrational even if each part has some justification.
The debate between wide and narrow scope, and over whether coherence has built-in reasons, may seem abstract. But it shapes what we expect from ourselves and others. Should you always act on your belief about what you ought to do? Or can you sensibly give up that belief? Next time you face a stubborn disagreement inside your own head, you’re living the very puzzle that philosophers from Darwall to Kolodny have wrestled with. And they still haven’t found a final answer.
Think about it
- Suppose you believe that lying is always wrong, but you also think it’s okay to lie to protect a friend’s feelings. Are you being irrational? What kind of mistake is that, if any?
- If you had a friend like Ali who believes contradictory superhero facts, how would you help them become more rational? Would you target one belief or the whole pattern?
- Could a computer program ever be structurally irrational? What would it need to have — or lack — for the charge of irrationality to make sense?
