How Do We Know What to Believe?

On epistemology in an age of infinite information

We are living through a strange inversion. For most of human history, the problem was scarcity: too little information, too few books, too few witnesses to any given event. Now the problem has flipped entirely. We drown in claims. A question that once required a trip to a library now returns ten million answers in half a second, several of them contradicting one another, all delivered with the same calm, authoritative tone, whether generated by a Nobel laureate, an anonymous forum poster, or a language model synthesising the two without distinction.

This is not, at its root, a new problem. It is the oldest problem in philosophy wearing a new coat. Epistemology, the branch of philosophy concerned with the nature and limits of knowledge, has always asked some version of the question we now ask of our phones: how do I know that what I believe is true, and not merely what I have been told, or what is convenient, or what confirms what I already suspected?

How do we know what to believe? Critical thinking is key!

The classical answer, and where it runs thin

Plato gave us one of the oldest working definitions: knowledge is justified true belief. You know something if you believe it, if it is in fact true, and if you have good reason for believing it. The definition has taken a beating in the centuries since, most famously from Edmund Gettier, who showed that a person can have a justified true belief that still doesn't feel like knowledge, because the justification and the truth are connected only by accident.

But the definition remains useful as a diagnostic. When you encounter a claim today, whether from a newspaper, a preacher, a politician, or an AI system quoting all three at once, you can ask the same three questions.

1. Do I actually believe this, or am I merely repeating it?

2. Is it true, and how would I find out independently?

3. And is my reason for accepting it a real reason, or is it simply that the claim arrived wearing the costume of authority: a confident tone, a polished interface, a familiar voice?

That last question matters more now than at almost any point before. Artificial intelligence systems are extraordinary at synthesis. They can summarise a great deal of what has been written on a subject with impressive fluency. But fluency is not truth, and synthesis is not verification. An AI model trained on the internet inherits the internet's errors, biases, and confident nonsense along with its genuine insight, and it delivers all of it in the same even, unhesitating register. The danger is not that these tools lie to us. It is that they never sound uncertain even when the underlying material is deeply uncertain, and human beings are wired to read confidence as competence.

Testimony, trust, and the limits of your own senses

Here philosophy offers something genuinely useful: the field of testimony, which asks when it is rational to believe something simply because someone told you, rather than because you verified it yourself. The honest answer is that almost everything we believe rests on testimony. You believe the earth orbits the sun not because you have personally tracked its path but because you trust a chain of people, going back to Copernicus and Galileo, who did the observing and the mathematics on your behalf. Trust in testimony is not naive. It is how any finite mind, unable to verify everything from first principles, manages to function at all.

The question, then, is not whether to trust testimony, because you have no real alternative. The question is which testimony deserves trust, and on what grounds. Philosophers who work on this, such as C.A.J. Coady, point to a few reliable markers:

1. Does the source have relevant expertise,

2. Does the source have anything to gain from your belief,

3. Can the source be corroborated independently,

4. And would the source be falsifiable if it were wrong.

These are not exotic tools. They are questions any careful reader can learn to ask as a matter of habit.

Where the shamanic perspective agrees with the philosophical one

It may surprise some readers that shamanic practice and rigorous epistemology arrive at a similar place, though by different roads. In shamanic tradition, direct experience is treated as a primary source of knowledge, not a secondary one to be discounted in favour of what an authority has said. You are taught to notice what you actually perceive, in the body and in the world, before accepting an interpretation handed to you by someone else. That instinct, distrust of secondhand certainty, insistence on checking a claim against your own encounter with reality wherever that is possible, is not so far from what a good epistemologist asks of us. Both traditions, in their own vocabulary, are suspicious of belief that has never been tested against experience.

A working discipline of what to believe, not a final answer

There is no formula that will sort every true claim from every false one, and any philosopher who promises you one is selling something. But there is a discipline that helps:

1. Ask who benefits from your believing this,

2. Ask what would have to be true for it to be false,

3. Seek at least one source with no stake in the outcome,

4. And notice when a claim's confidence exceeds its evidence.

Do this consistently, and you will not become infallible. You will become considerably harder to mislead, which, in an age when misleading has become both easy and profitable, is no small achievement.

The question "how do we know what to believe" is not one we answer once and set aside. It is a practice, taken up again with every article, every algorithm's output, every confident voice asking for our trust. That, perhaps, is the most honest thing philosophy has to offer: not certainty, but a better way of living without it.

Kathy Postelle Rixon is Chair of The Philosophical Society: Oxford, a researcher of philosophy of mind and physics, and shamanic practitioner. You can contact her at kathy@magicinharmony.com or visit her website at www.magicinharmony.com.