As you sit down to dinner at your favourite restaurant the waiter comes over and asks you what you would like to drink. You don’t find this a difficult question. A gin and tonic is (say) what you want and you know it’s what you want. So you place your order. Your irritating companion asks you how you know want a gin and tonic. A very strange question, no doubt, and probably a conversation stopper. Anxious to drink your gin and tonic you say you just know, and that is all there is to it.

Although your impatience with your companion’s question is perfectly understandable it raises an interesting philosophical question. On the face of it, any assertion is open to the challenge “How do you know?” For example, if you assert that it is raining in Mombasa then you can be asked how you know. So if “I want a gin and tonic” is a genuine assertion then it, too, is exposed to the question “How do you know?” You might not know the answer but there must be an answer. Equally, if you know you believe it’s raining in Mombasa there must be an answer to the question how you know that that is what you believe. Being asked for the answer over a gin and tonic might be a bit much, but if there is such a thing as knowledge of one’s own desires, beliefs, hopes, fears, and intentions it should be possible in principle to explain how such self-knowledge is possible.

In philosophy, rationalism is much impressed by the role of reasons in our mental lives and its account of self-knowledge is constructed on this basis. So if you are a rationalist you might be tempted to suggest that our beliefs and desires are normally determined by our reasons and so are knowable by reflecting on our reasons. For example, if your belief that it is raining in Mombasa is formed in response to the reasons in favour of believing this then you can know that you believe that it is raining in Mombasa by consideration of these reasons. By the same token, you can know that you want a gin and tonic by consideration of the reasons in favour of wanting one, as long as your desires are determined by your reasons. To put the point more simply, you can answer the question whether you actually want a gin and tonic by answering the question whether you ought rationally to want one.

Unfortunately, many of our desires are not determined by our reasons. If your doctor has told you to cut down on your drinking then you have a good reason not to want a gin and tonic but that doesn’t alter the fact that you want one. So consideration of what you ought rationally to want won’t be a good guide to what you actually want unless you are the kind of being whose desires are rationally determined. No doubt reason plays a part in the formation of our desires and beliefs but as human beings we are also influenced by a wide range of non-rational factors, including environmental and biological factors, character traits and biases. If your desire for a gin and tonic is impervious to the reasons for not wanting one then reflecting on those reasons won’t tell you what you want: you ought not to want one but you do. Other attitudes are no different. You shouldn’t fear the spider in your bathroom but you do, and it would be remarkable if every one of your beliefs is one that you ought rationally to have.

There is a parallel with economics. Behavioural economists argue that economics goes wrong when it operates with the conception of the economic agent as an unswervingly rational homo economicus. They argue that the study of economic behaviour must be the study of the economic behaviour of real human beings, most of whom are far from unswervingly rational. Homo philosophicus is the philosophical cousin of homo economicus and no less mythical. By stipulation, homo philosophicus’ beliefs and desires are determined exclusively by its reasons and so are knowable by reflection on its reasons. To the extent that we depart from the ideal of homo philosophicus rationalism doesn’t account for our self-knowledge. Rationalism is the philosophy of homo philosophicus rather than the philosophy of homo sapiens.

In that case, how do we know our own desires and beliefs? Before tackling this question it is worth noting the peculiarity of the idea that knowledge of your desire for a gin and tonic is a form of “self-knowledge”. If it is self-knowledge then it is a strikingly mundane piece of self-knowledge, and not self-knowledge as we ordinarily think of it. Self-knowledge in the ordinary sense is substantial self-knowledge. This includes knowledge of such things as one’s character traits, values, complex emotions and fundamental desires. This is the self-knowledge we tend to value and that is presumably the subject of the ancient injunction to “Know thyself”. As it happens, substantial self-knowledge is also somewhat easier to explain than mundane self-knowledge so it makes sense to start at the substantial end of the spectrum and see whether mundane self-knowledge can be accounted for along similar lines.

Krista Lawlor of Stanford University gives the example of a woman called Katherine trying to work out whether she wants another child. As she folds her son’s now-too-small clothes she “catches herself imagining, remembering, and feeling certain things”. From these and other such “internal promptings” Katherine infers that she wants another child. When she says to herself “I want another child” her self-attribution feels right to her. It feels like, and is, an expression of a piece of substantial self-knowledge. The implication is that self-knowledge is based on evidence and the product of inference. Katherine infers from her passing thoughts, imaginings, memories, feelings and daydreams that she wants another child, and her knowledge that she wants another child is based on this evidence. Her evidence is psychological rather than behavioural but it is evidence nonetheless. Although she reasons her way to the conclusion that she wants another child she doesn’t answer the question whether she wants another child by asking herself whether she ought rationally to want another child. She knows that she wants another child but might be hard pushed to say whether she ought rationally to want one.

Standard philosophical accounts of self-knowledge start by emphasising the peculiarity of self-knowledge. Self-knowledge, they say, is fundamentally different from other kinds of knowledge. Other kinds of knowledge are based on evidence but self-knowledge isn’t normally based on evidence, behavioural or otherwise. You might need to infer what someone else wants or believes but you don’t have to infer what you want or believe. Self-knowledge, on this view, is normally direct and not subject to the failings of conclusions based on evidence. However, Katherine’s knowledge that she wants another child is indirect, and what she thinks she wants might not be what she really wants. She doesn’t “just know” what she wants. She has to work it out.

Proust gives another example along similar lines. Marcel assumes that he no longer loves Albertine but then he hears that announcement “Mademoiselle Albertine has gone”. The anguish these words produce in Marcel reveal to him how wrong he was about his feelings. He now knows that he still loves Albertine and his route to self-knowledge is suffering. But how can suffering or anguish be a route to self-knowledge? By providing Marcel with evidence of his continuing love for Albertine. He is now in a position to infer that he still loves Albertine, and the inference works because he takes his suffering on hearing news of her departure to be caused by his love for Albertine. His anguish reveals his underlying emotion, but it only does that on the assumption that Marcel understands the significance of his anguish.

How does this help with mundane self-knowledge? You might think that while substantial self-knowledge is indirect, mundane self-knowledge is not. Katherine relies on evidence to know that she wants another child but you don’t need evidence to know that you want a gin and tonic. On reflection, however, the difference between mundane and substantial self-knowledge is a difference in degree rather than in kind, and examples like Katherine and Albertine should make us receptive to the possibility that mundane self-knowledge is also based on psychological evidence and the product of inference or reasoning.

How can that be? When the waiter asks you what you would like to drink you might perhaps visualise a gin and tonic or imagine its taste or remember how much you enjoyed drinking one last night. The effect that these visualisations, imaginings and memories have on you, including the feelings they produce, are the evidence you have to go on. You infer from this evidence that you want a gin a tonic but your inference may be so rapid as to be barely conscious. You don’t have to think about whether you want a gin and tonic in the explicit way that Katherine has to think about whether she wants another child but you don’t “just know”.

This explains why we are sometimes wrong about what we want. If knowledge of our own desires is based on evidence then there is the possibility of error because evidence can almost always be misinterpreted. The evidence on the basis of which Katherine concludes that she wants another child might in fact be evidence of something else. In principle you can also be wrong about what you want to drink. You think you want a gin and tonic but the moment your drink arrives you realise that it wasn’t what you really wanted.

The emerging picture of self-knowledge is very different from the standard philosophical picture of self-knowledge. Self-knowledge, even mundane self-knowledge, is indirect. It is mediated by the evidence on which it is based and by our understanding of the significance of that evidence. It should come as no surprise that some people are better than others at knowing what they want since some people are better at reading their own minds. Self-knowledge is the product of self-interpretation and self-interpretation is something that can be done more or less well, with more or less psychological insight.

Knowing what you want is one thing but what about knowing what you hope or intend or fear or believe? Can the evidential picture account for these varieties of self-knowledge? The case of belief is the trickiest since it has seemed to many philosophers that you can’t possibly be said to know your own beliefs on the basis of evidence. What could possibly be more direct than knowledge of your own beliefs?

Consider this case: imagine that you are reading Capital in the Twenty-First Century by Thomas Piketty. You have heard a lot about the book and are curious to know what it says. A few pages into the book you encounter this sentence: “Economists are all too often preoccupied with petty mathematical problems of interest only to themselves”. As you read the sentence you realise that what Piketty is saying here is exactly what you think and perhaps have always thought. This is self-knowledge. It is knowledge of what you do and have believed.

Even in this relatively straightforward example there are multiple possible pathways to self-knowledge. Piketty’s words might produce in you a feeling of conviction, but your agreement with him needn’t be experienced by you as the reaffirmation of what you already think. Your sense of agreement might instead be experienced by you as an indication of a new belief. There is also the case in which you have inchoately believed that economists are all too often preoccupied with petty mathematical problems and in which reading Piketty’s words brings your inchoate belief to light by presenting you with a compelling formulation of what you have at some level always felt.

In each case, you come to know your newly formed or pre-existing belief on the basis of psychological evidence. The feelings of conviction that Piketty’s words produce in you are psychological evidence that you have the belief those words express. In contrast, consciously disagreeing with his statement is an indication that you don’t share his view of economists. However, your immediate psychological reaction to someone else’s words can be misleading. Listening to a particularly charismatic speaker you might feel that you agree with them even though, in the cold light of day, you don’t share their beliefs. But this is not an objection to the evidential picture. It only serves to bring out again the extent to which we can be mistaken about our own beliefs. Human beings aren’t immune to self-ignorance. You might think you believe that, say, men and women are equal but your behaviour suggests otherwise. Our desire to think well of ourselves is one among many obstacles to self-knowledge.

Should we be worried? That depends on why self-knowledge matters. What is so good about self-knowledge and so bad about self-ignorance? What should we say to someone whose reaction to “Know thyself” is to ask “Why bother?” It has been suggested that the true value of self-knowledge can only be understood by reference to higher ideals like authenticity. To be authentic is to be true to yourself but how can you be true to yourself if you don’t know yourself? The answer to this question is that being true to yourself is a matter of being true to your actual character and values, and it’s far from obvious that this requires knowledge of your actual character and values. The fastidious person who lives fastidiously is being true to himself even if he has no idea that he is fastidious. Anyway, what’s so great about being true to yourself? Doesn’t it depend on the kind of person you are?

In reality, the value of self-knowledge is practical. Imagine the consequences of having to make choices, whether trivial or serious, in the absence of a capacity to know one’s own desires. You can tell the waiter what you want because you know what you want. When it comes to substantial self-knowledge the fundamental issue is whether having more of it rather than less of it makes a positive difference to overall happiness or well-being. No doubt self-knowledge can be a mixed blessing. Perhaps there are truths about oneself one is better off not knowing and there is evidence that mild self-ignorance can increase levels of well-being. Self-illusions can motivate self-improvement and thereby make one’s life go better. However, this line of thinking can only be pushed so far. Research by Timothy Wilson and Elizabeth Dunn supports the intuitive view that extreme self-illusions can be harmful and that a degree of self-knowledge is essential to well-being. There is no better explanation of the value of self-knowledge, and no better case for making the effort to know yourself.

Quassim Cassam is Professor of Philosophy at the University of Warwick and the author of Self-knowledge for Humans. He also writes about self-knowledge on his website.