About A Life Between the Layers

A Life Between the Layers is an autobiographical work by John Brown. It is being assembled from lived experience, memory, surviving records, correspondence and other evidence accumulated across many years.

The title reflects one of the central problems running through the book: the difference between what was experienced, what was remembered, what was recorded by institutions, and what can later be established from surviving evidence.

The book is being published progressively rather than waiting for the entire work to be completed. The first block of chapters is now available, while a second block is being researched and written.

This also means that the work is not intended to be frozen in its first published form. Further records, correspondence and other evidence may clarify dates, names or events and may result in corrections or additions. Where significant changes are made, they can be recorded on the Updates page.

Supporting material and information about the kinds of records used in constructing the account can be found on the Sources page.

AI, Memory and the Writing of This Book

I have used artificial intelligence extensively in developing A Life Between the Layers.

That requires some explanation.

There is understandable concern that AI can become a substitute for thinking. A person can ask for an answer, accept what appears on the screen and avoid much of the intellectual work which might otherwise have been required.

That has not been my experience of using it here.

For me, the process has often worked in the opposite direction.

I bring memories, documents, questions, arguments I made years ago, contradictions, things I know, things I think I know, things which have returned unexpectedly, and things for which I still have no satisfactory explanation.

The conversation frequently sends me back into those layers. It prompts another memory. A forgotten name returns. An assumption becomes a question. Something I once regarded as certain becomes less certain. Something which had remained disconnected for years suddenly acquires context.

I find myself asking more questions, not fewer.

Finding the Thought Inside the Sentence

My natural writing style can become complicated.

When I am close to emotionally significant material, past, present and future can move through the same paragraph — sometimes through the same sentence.

I can write a sentence of extraordinary length while apparently regarding commas as an unnecessary luxury.

That is not simply a grammatical habit. My style changes according to my relationship with what I am remembering.

Sometimes I am describing an event. Sometimes I am inside it. Sometimes I am examining what I believed then. Sometimes I am comparing it with what I understand now.

Those positions can change very quickly.

For years I have sometimes had difficulty taking the considerable amount of material in my head and expressing the central thought with the precision I wanted. That difficulty has itself been a subject of conversations with my psychologist, Dr Julian Lim.

AI has proved unexpectedly useful there.

I can put down the untidy version. The long sentence. The half-memory. The argument mixed with the emotion which produced it.

Then we can separate the pieces.

What happened?

What did I believe at the time?

What does the record say?

What am I inferring now?

What remains uncertain?

What am I actually trying to say?

Very often, somewhere inside several hundred words, there is a much shorter proposition.

Finding it does not remove my thought. It reveals it.

Testing What I Think I Know

During the years in which I was trying to understand the psychiatric diagnosis that had been placed upon me, and later exploring Dissociative Amnesia with Dr Julian Lim, I developed several ways of testing my own thinking.

One was deliberately uncomfortable.

I would take a position for several days.

Then I would take the opposite position for several days.

Not merely pretend that the opposing view existed, but try to inhabit it seriously enough to understand what supported it.

After spending time on both sides, I could ask a different question:

Which position actually survives best when I have tried to argue against myself?

Sometimes the answer was the position I had started with. Sometimes it was not. Sometimes neither survived intact.

That was useful too.

The purpose was not to win an argument with myself. It was to discover what I could reasonably continue to believe.

Everything I Know — Or Think I Know

Another exercise usually began with words something like:

This is everything I know, or think I know, about this subject.

Then I would write.

The resulting block of material was deliberately provisional.

Some of it might be factual. Some might be memory. Some might be inference. Some might be fragile. Some might simply be wrong.

I was not declaring it true by writing it down.

I was putting it somewhere that it could be examined.

That distinction became extremely important to me.

I wanted the material to be factual. But wanting something to be factual does not make it factual.

So the first task was to expose what I thought I knew.

The second was to find out what kind of knowledge each part actually was.

A Working Block, Not a Declaration

That way of working eventually became something close to the method used throughout this book.

I can begin with a block of recollection or argument without requiring every part of it to have the same status.

One sentence may come from direct memory. Another from a medical record. Another from something somebody told me. Another may be an interpretation I formed years later. Another may simply be a question.

Putting them together does not make them equivalent.

It gives me something to work on.

The block is open to correction. It can be divided, reordered, challenged, checked against records, compared with research, or rejected altogether.

The important thing is that I do not have to choose between saying nothing until I possess certainty and treating every recollection as established fact.

There is a middle position:

put down what I know or think I know, identify what kind of claim each part is, and then test it.

AI Entered an Existing Method

That is one reason artificial intelligence has fitted this work so naturally.

It did not introduce the habit of questioning my own conclusions.

That habit was already there.

What it provided was another place in which the unfinished block could be placed and worked upon.

I can say: this is what I remember; this is what I think it means; this part troubles me; this part may be wrong; argue the other side; separate the facts from the interpretation; tell me what needs checking.

The result is often not an answer.

It is a better question.

And sometimes the greatest value has been discovering that something I felt certain about needs qualification, while something I had treated tentatively has much stronger evidence behind it than I realised.

Clarification Rather Than Substitution

I do not want AI to decide what I believe.

I want it to help me see more clearly what I am attempting to examine.

I remain responsible for the memory, for accepting or rejecting an interpretation, for deciding whether wording represents me, for identifying when something has become too certain, and for returning to evidence.

I remain responsible for saying: no, that isn't what I meant.

Or occasionally:

yes — that is exactly what I have been trying to say for years.

At its best, the process has not narrowed my thinking. It has widened the number of directions from which I can examine it.

A Style in Transition

I have sometimes tried to describe my writing “style.”

I am increasingly unsure that I have one fixed style.

It is more like a state of transition.

The language changes with the emotional distance between me and the event.

Some passages need to remain raw. Some need chronology. Some require evidence. Some require uncertainty. Some become almost forensic. Others become philosophical.

The important thing is not to make all of them sound alike.

It is to make each one clear enough that another person can follow where the thought came from.

There are moments when a complicated idea which has occupied pages of my earlier writing can finally be expressed in a few sentences.

Sometimes I recognise in that economy something I have admired in people such as Richard Feynman or Robert Sapolsky: not imitation of their language, but the satisfaction of getting past the clutter and reaching the proposition underneath.

An External Workspace for Thought

Because I live with Dissociative Amnesia, the process has another dimension.

The conversation can function as an external workspace.

An idea placed here does not have to depend entirely upon my ability to hold every part of it simultaneously.

Earlier material can be brought back beside later material. Contradictions become visible. Questions can remain open. A memory can be recorded without immediately being declared fact. A recovered detail can be marked for checking rather than forced into certainty.

In that sense, the technology has not simply been writing assistance.

It has helped me construct a place in which I can think across the layers.

The Desire Is Not to Be Right

There is a difference between wanting to be right and wanting what I write to be right.

I increasingly care about the second.

If evidence destroys one of my explanations, then the explanation should go.

If another interpretation accounts for the evidence better, I want to know that.

If uncertainty remains, I would rather preserve the uncertainty than manufacture resolution.

That is not always emotionally comfortable when the material concerns my own life.

But it is probably the only way I know to write this book.

The Thinking Still Has to Be Mine

None of this makes AI inherently beneficial.

Used differently, it could undoubtedly save somebody the trouble of thinking.

That is precisely why the method matters.

I do not regard an AI-generated statement as true because an AI generated it.

Sources still have to be checked.

Claims still have to survive evidence.

My memories remain memories.

Documents remain documents.

Interpretations remain interpretations.

And sometimes both the AI and I are wrong.

The useful relationship is not:

Ask the machine for the answer.

It is:

Bring the unfinished thought into a space where it can be questioned.

That has been one of the unexpected parts of writing this book.

Rather than taking me away from my memories and thinking, the process has repeatedly taken me further into them.