On Souls

February 1, 2025

I’ve spent months building tools that extract signals from images—faces detected, expressions measured, poses mapped. The technology works. It finds what it looks for. And yet, the more precisely I can measure the visible, the more I’m haunted by what remains invisible.

What is this thing we call the soul?

The subtraction problem

If I photograph you, I can extract:

The geometry of your face (478 landmarks, if using MediaPipe)
Your expression (52 blendshapes measuring every muscle movement)
Your pose (33 body keypoints in three-dimensional space)
The scene around you (indoor, outdoor, day, night)
The metadata (when, where, what device)

I can aggregate these across thousands of images and find patterns. I can track how your expressions cluster, how your posture shifts across contexts, who appears beside you and how often.

But if I subtract all of this—every measurable signal—something remains. The part that decides what the smile means. The part that knows what you were thinking when the photo was taken. The observer behind the observation.

Is that the soul?

What the ancients said

The concept appears everywhere, wearing different names:

The Hebrew nephesh—the breath that animates
The Greek psyche—that which flutters away at death
The Sanskrit atman—the self that persists
The Chinese hun and po—the ethereal and corporeal souls, plural

Aristotle thought the soul was the “form” of the body—not a separate thing but the organizing principle that makes matter alive. Descartes split us in two: extended substance (the body) and thinking substance (the mind/soul). The Buddhists took another path entirely: anatta, no-self, the idea that what we call soul is an illusion, a story we tell about a process.

What strikes me is how every tradition grapples with the same observation: there’s the body we can see, and there’s… something else. Something that experiences.

The question of multiplicity

This brings us to a strange case: dissociative identity disorder, once called multiple personality disorder.

In DID, a single body hosts what appear to be distinct identities—different names, different memories, different mannerisms, sometimes different handedness or allergies. Brain imaging studies show different patterns of activation for different alters. The body is one; the selves are many.

If we believe in souls, this poses a genuine puzzle: does such a person have multiple souls?

The theological answers vary:

Some religious traditions would say no—one body, one soul, and the apparent multiplicity is a disorder of the mind, not a fracturing of the essential self. The soul remains singular even if the personality fragments.

Others might say the concept doesn’t apply—that DID is precisely the kind of case that reveals the soul as metaphor rather than substance. There is no ghost in the machine; there is only the machine, running different programs.

The clinical perspective:

Psychiatry treats DID as a response to severe trauma, usually in early childhood. The mind, overwhelmed, partitions itself. The “alters” are understood as dissociated parts of a single consciousness, not separate beings. Treatment often aims at integration—helping the parts recognize themselves as aspects of one whole.

But here’s what’s interesting: many people with DID report that their alters experience themselves as genuinely distinct. They have their own preferences, their own histories, their own sense of being. From the inside, multiplicity feels real.

Who are we to say it isn’t?

What signals cannot capture

I built an image analysis tool that reports what it sees without interpretation. It will tell you that a face shows “mouth corners raised” but refuses to conclude “happy.” It measures head angle but won’t infer intent. This was a deliberate choice—humility encoded in software.

But watching this tool work has taught me something: even if we could measure everything, we wouldn’t capture the soul.

Not because the soul is supernatural (it might or might not be), but because the soul—whatever it is—is the experiencer, not the experienced. It’s the subject, not the object. Every measurement we make is of the body, the face, the behavior. The thing that witnesses these measurements remains outside the frame.

William James wrote about this in The Principles of Psychology (1890). He called it the “I” versus the “Me”—the self that observes versus the self that is observed. The Me can be studied: your body, your memories, your social roles. The I cannot. It’s always the one doing the studying.

The multiplicity in all of us

Perhaps DID is an extreme case of something universal.

Don’t we all contain multitudes? The self I am at work differs from the self I am with my family. The me who writes these words is not quite the me who will read them tomorrow. We speak of “being of two minds” about a decision. We surprise ourselves—who was that, we wonder, who said that cruel thing, who acted so bravely, who made that choice?

Walt Whitman wrote: “I am large, I contain multitudes.”

Maybe the soul isn’t a single point but a process. Not a thing but a happening. And maybe what we see in DID isn’t a soul shattered but a soul whose multiplicity became visible.

The limits of detection

I keep returning to this: some things can only be known from the inside.

Guilt and regret, as I wrote recently, leave no reliable trace on the face. They must be confessed, not detected. The same is true for the soul. No camera will capture it. No algorithm will extract it. Not because our technology is insufficient, but because the soul—by definition—is not an object to be observed.

It is the observer.

This isn’t mysticism. It’s epistemology. There are things about human experience that are first-person by nature. Pain, consciousness, the felt sense of being. These are real (more real to us than anything), yet they resist third-person capture.

An unanswerable question

Does a person with DID have multiple souls?

I don’t know. I suspect the question itself might be malformed—that “soul” is a word we use for something we don’t understand, and asking how many souls are present is like asking what color a song is.

But I find the question valuable precisely because it reveals the limits of our categories. It shows us where our tools break down, where measurement gives way to mystery.

The body can be photographed. The face can be analyzed. The behavior can be catalogued. But the one who lives behind the eyes—singular or plural, permanent or passing—remains beyond the reach of any instrument I can build.

And perhaps that’s as it should be.

This is part of an ongoing exploration of what signals can and cannot tell us about human nature. The image analysis tool on this site is an experiment in honest observation—capturing geometry without claiming to capture the soul.