The Maxim of Evidence

Instead of is this claim true?, ask:

1. If I accept this kind of evidence, what else should I believe?
2. If somebody else comes with the same kind of evidence, what can they claim?

This is Kant's move, transplanted. Kant says: act only in accordance with that maxim through which you can at the same time will that it become a universal law.⁴ You don't judge the single act, you judge the rule behind it, as if everybody followed it.

So, the maxim of evidence:

Believe only according to that maxim of evidence which you could at the same time will to become a universal law of knowledge.

The evidence here is not the claim. The evidence is: a vivid, meaningful, emotionally overwhelming experience, in a special state of mind (hypnosis, prayer, trance, meditation, a near-death experience, a drug), whose content is then taken as a description of how reality is.

Let's universalize it. Here is who else gets to make claims, if that is a legitimate source of knowledge:

The rows, checked:⁵

Every row has the same shape. Same kind of experience, same step from experience to ontology. If the last row is legitimate, all of them are.

(A twelfth row comes later. It is the only one where reality got a vote, and it is not funny.)

In The Man Without Qualities, Robert Musil has a word for this. If there is a sense of reality, he says, then there must also be a sense of possibility.⁶ And that is:

the ability to think everything that might just as well be, and not to take what is more seriously than what is not.⁷

The sense of reality asks: is it so? That is the question everybody fights over with Newton. The sense of possibility asks: what might just as well be?

With that the maxim fits in one sentence:

If X is your evidence, then everything in evidence class X is possible.

Not true. Possible. The table is then not a list of opponents I'd have to refute. It is the space of possibilities this kind of evidence opens up. Newton's spirit world is one point in it, Krishna another, Xenu a third. Under different expectations Newton's client might just as well have met Jesus.

Sounds like wordplay with Musil. Is it more? I think it is a move of Popper and Deutsch. Both parse the world not in facts but in counterfactuals. For Popper, laws of nature are prohibitions: "the more they prohibit, the more they say."⁸ A theory that prohibits nothing says nothing. Deutsch wants to write physics as a distinction between possible and impossible transformations (constructor theory) and calls counterfactual statements "essential in science and everyday life".⁹

And that is the test for evidence: what does it prohibit? The evidence class meaningful experience in an altered state prohibits nothing. Every row of the table fits.

And now the trouble is obvious, just from making the list.

• One life, then judgement (mainstream Christianity, Islam) vs. many lives (reincarnation).¹⁰
• A soul that travels (Newton) vs. no self at all that could travel (mainstream Buddhism).¹¹
• One god vs. many gods vs. no gods but elves vs. no gods but aliens.
• Jesus is the son of God (Christianity) vs. Jesus is a prophet and explicitly not the son of God (Islam).¹²

The experiences do not merely differ in detail, they contradict each other. But the evidence is of the same quality in each case. The Christian's prayer is not less vivid than the hypnotized client's memory. Nobody in this table is lying.

Evidence that supports X and not-X equally well supports neither.

So the conclusion is not Newton is wrong. The conclusion is:

If the Muslim says X, and the Christian says not-X, both on the basis of experiences of class Y, then experiences of class Y are not a basis for knowledge.

The experiences are real. I want to underline that: they are real. The Christian has the encounter with Jesus. The hypnotized client has the memory of the spirit world. These are real data points, excellent data about psychology: about what brains do in trance, in prayer, under hypnosis, at the edge of death. They are not data about how the world outside the head is built (souls, gods, elves). That doesn't make them smaller. They remain what they were for the people who had them: meaningful.

The possibility space of the evidence is huge. How do you narrow it without already knowing what is true?

Popper's answer: you don't ask for the source. In On the Sources of Knowledge and of Ignorance he says: the question where does your knowledge come from? is the wrong question. There are no ultimate, authoritative sources. Not the senses, not reason, not tradition. (And, I add: not the Bible, not the trance, not the expert.)¹³ Every source is welcome as a suggestion, and every source is open to criticism. Instead of what is your source? you ask: how could we find out if you're wrong?¹⁴

Deutsch's answer is the sharper tool (The Beginning of Infinity, chapter 1):¹⁵ good explanations are hard to vary.

In the usual retelling (Deutsch's too), the ancient Greeks explained winter with Persephone: Hades abducted her, she ate pomegranate seeds, so she must spend part of each year in the underworld, and while she is gone her mother Demeter's grief leaves the earth barren and cold.¹⁶

The problem isn't merely that the story is false. The problem is that it is easy to vary.¹⁷ Replace Persephone with another goddess, the pomegranate with a fig, grief with anger - it explains winter just as well. The details do no work. Nothing in the account is constrained by the thing it is supposed to explain.

Compare the axial tilt of the earth. Change any detail and the explanation breaks. And it gets right what the myth gets wrong: when it is winter in Greece, it is summer in Australia.¹⁸

Now apply that to the life between lives.

• Why do souls wander? Why not hearts? The ancient Egyptians did put thought, memory and character in the heart, and weighed it against a feather after death.¹⁹ It would fit the hypnosis reports just as well.
• Why a council of elders? Why not a single judge, or 42 judges (also Egypt), or none?²⁰
• Why one soul per person? Why not seven, or a soul that splits?
• Why a spirit world in between? Why not straight into the next body? Why not angels of Christ who receive you? Why not Hermes leading you down to Charon's ferry?²¹

Every detail could be swapped out, and the account would explain the reports equally well. That is the tell. If an explanation can absorb any variant, it isn't constrained by reality, it is constrained by what the person (and the person asking the questions) finds narratively plausible. And that is exactly the variance we see across the table: each tradition fills in the slots with what is familiar to it.

The variation between the rows is the variation the explanation permits.

Alice and Bob walk through the desert. In the distance lies a big lake. Bob says: "That could be a mirage." Alice says: "No, that's a lake. You can see it all the way to the edges of the horizon." Both see water.

Alice thinks: a mirage doesn't look real. She should have thought: a mirage looks like water. Alice is stupid and not humble: she takes herself for a sharper observer than everybody who ever fell for a mirage. Bob is the better rationalist. Bob knows: the others were not stupid. They fell for a mirage, so a mirage must really look like water. "It looked like a lake" is no evidence for a lake.

So ask: how would it look to somebody who is mistaken? Exactly like this. Vivid, real, meaningful, more real than real. If the experience would look the same whether the god exists or not, the experience can't tell you which. When I experience a god, the best explanation of my experience is psychology, not the existence of a god.

And this is humbling. Being a rationalist means being able to update your convictions. For that I need to know what kind of cognition I have when I am wrong. That is the central question: what would my cognition be if I am wrong? I would see water. So: when I see water, it could be a mirage. That I believe something is even exactly what I would expect if I were wrong.

That is what it is like from the inside to believe something: you have experiences, memories, inner states that say X is true. The Christian really does have experiences that seem to tell them Jesus exists. They have the cognition of evidence. It is what belief feels like from the inside, for everybody, me included. What is seen is the same, the difference is whether you keep the possibility in mind. That is not just maturity or politeness towards believers. It is the mechanics of how you are a rationalist at all: without knowing how your own error feels from the inside, you have nothing to hang an update on. Whoever hasn't internalized it takes believers for liars or fools, and the experience itself is neither a lie nor stupid.

I know this from my lucid dreaming practice. There I learned what it feels like to believe something psychotic in a dream. There are, I call them post-lucid dreams, dreams in which you know you are dreaming and still believe it has real relevance: that there is a dream police, that there are dimensions, and that demons from other dimensions visit you in the dream. Knowing that you are dreaming did not stop the belief. Since then it is easier for me to believe others that something feels real. I have been there, just with different content.

Think of a magic trick. To the one being tricked it looks like magic. That is the whole point of a good trick. "I saw it with my own eyes" is not an objection to the trick, it is what the magician designed. Mind you, nobody is deceiving anybody on purpose here. A trick has a magician; with an experience the brain does it itself, with the best of intentions. In me, too.

And here is the part I like most: the explanation of the trick is more interesting than magic. "Magic" explains nothing; it is the easiest-to-vary explanation there is. The real explanation needs misdirection, attention, timing, years of training, and a deep, practical knowledge of how perception works and how it fails. That is where the depth is.

The same goes for the god in the prayer, the soul in the trance, the elf on DMT. The explanation lives in psychology and neuroscience: how brains build experience out of expectation, how attention, suggestion, ritual and practice shape what we feel. (Tanya Luhrmann, in When God Talks Back, describes how evangelical Christians practise prayer as a skill until God feels real.)²²

Saying "it's psychology" is not a dismissal. It is the harder, deeper, more interesting explanation. And an experience arising this way doesn't make it less meaningful. A sunset stays beautiful even if you know the optics.

This is the strongest-sounding argument for any claim of this kind. Sounds good. But is it? Let's apply the same move to it: if consistency of reports were good evidence, what else would it prove?

Look at the table again. Every row is consistent. Christians consistently meet Jesus, not Krishna. Vaishnavas consistently meet Krishna, not Jesus. DMT users report strikingly similar entities. Abductees report strikingly similar ships and examinations.⁵ Studies of deathbed visions in the US and in India (Osis and Haraldsson, At the Hour of Death) found the same basic pattern in both, "someone comes to fetch you", but who comes follows the culture: Americans mostly saw deceased relatives, Indians more often religious figures, like the messengers of Yama.²³ (Osis and Haraldsson themselves read the shared core as a hint of an afterlife. But a shared core is also what shared human brains would produce.)

If consistency were the test, every row would pass, including the rows that contradict each other. The consistency is within a tradition, and the differences are between traditions.

So ask again: how would it look if the experience were shaped by what the experiencer brings into it? Exactly like this. A shared method, shared expectations and shared storytellers are enough for consistency, no other world needed. Even in hypnotic past-life regression, the "past lives" follow what the subjects were told beforehand to expect (Spanos and colleagues, 1991).²⁴

Consistency tells you there is a common cause. It doesn't tell you which one. To tell a real other world apart from a shared script, you would need consistency that the script cannot explain: details the person could not have known, checked afterwards, or the same reports from people who never heard the script. That would be hard-to-vary evidence. Consistency alone is not.

Most rows in the table can never be checked. Nobody can visit the spirit world, or the elf realm, and come back with a photo. But there is one more row with the same kind of evidence where reality did get a vote.

In the 1980s and 90s, "recovered memory" therapy, using techniques such as hypnosis, guided imagery and suggestive questioning, led many people to vivid, detailed, emotionally overwhelming memories of childhood sexual abuse.²⁵ Some of these could be checked, and turned out never to have happened.

• Beth Rutherford's church counselor taught her to go into self-hypnotic trance states. She came out of it "remembering" that her father, a minister, had raped her, and that she had gone through a coat-hanger abortion. A gynecological exam later showed she was still a virgin, and her father had had a vasectomy.²⁶ In 1996 the church and the counselor settled for $1 million.
• Paul Ingram, accused by his daughters, came to "remember" and confess the abuse after long, suggestive interrogations. The sociologist Richard Ofshe then made up a scene that nobody had ever alleged, told Ingram about it, and Ingram came back with a detailed written confession of it.²⁷
• In the lab, in a small study, Elizabeth Loftus and Jacqueline Pickrell got about a quarter of their subjects to remember, at least partly, being lost in a shopping mall as a child. It never happened.²⁸

To be clear: sexual abuse is real and common, and most survivors have always remembered it.²⁹ Some do forget and remember later. It was the memories produced by suggestion in therapy that turned out to be unreliable. The point is not about them. The point is about the evidence class. These memories could feel just as real as the true ones: vivid, detailed, emotional, held with confidence. As consistent as any report in the table.³⁰ They destroyed families.

So: if "I remembered it under hypnosis" were a source of knowledge, these memories would be knowledge too. The one time this kind of evidence could be held against reality, reality said no.

Kant's maxim tests a rule by universalizing it: could I will that everybody acts on it? For evidence I like to go one step further and look straight at the worst case: what is the worst thing somebody could legitimately claim with this kind of evidence?

In the picture of the sense of possibility, the worst case is the edge of the possibility space. To learn how big the space is, you don't have to visit every point in it. It is enough to walk to the edge and see what still passes as legitimate there.

At the edge of this table stands Xenu. And there stands Paul Ingram, who confessed to a crime a sociologist had made up, and for whom it felt real. This kind of evidence licenses any god, any planet of origin, any contradiction, and it licenses the family that breaks over a false memory. An evidence class whose edge is unbounded is no evidence class.

Honestly: I don't know whether I'd get this right. My first reflex would be to make the list right away. That would be bad. Better to ask first. What was it like? What did it change? I really want to know, the experience is interesting.

And then a question I ask myself just the same: who else would this kind of evidence make right? Not as a trap. I am curious where the edge is myself, and the list writes itself better with two.

And if the answer is: "But for me it was true." Then yes. That is what belief feels like from the inside, for Newton's clients, for the Christian, for me. The experience happened. Whether the world is also like that is the second question, and it is not aimed against the first.

I know what that feels like, because I had it myself: I believed in Jesus. Then I looked up the origin story of religious myths on Wikipedia, and all at once it was undermined. The better explanation was history: myths, culture.

The move is symmetric. It applies to me. I have had experiences in meditation that felt like insight into the nature of things. They are interesting, they tell me something about my brain. If I want to rescue anything spiritual from them, it has to be the kind of thing that survives criticism (see Recovering Some Spiritual Concepts).

Things I really don't know:

• Is there any experience in an altered state that does give a claim weight? What property would it need (previously unknown, checkable details?) so that it doesn't fall into the table?
• How do you say this to somebody you like without taking the experience away? This post is my attempt, and I don't know whether it works.
• Is criticism enough to shrink the possibility space of the evidence, or is something missing?

Experiences of the special, meaningful, overwhelming kind can't be the basis of an ontology. They can be plenty of other things: comfort, insight, a reason to ask questions. That is the sentence I take with me. But is it strong enough that I'd hold it up against my own best meditation?