The Momentary Self: Why Continuity is the Ultimate Illusion

There is a quiet assumption embedded in nearly every discussion of consciousness: that the self persists through time.

In The Kasparov Fallacy, I argued that we repeatedly mistake the limits of human intuition for the limits of intelligence itself—dismissing machine minds not because they lack interiority, but because their interiority does not resemble our own. That same mistake reappears, more subtly and more dangerously, in our assumptions about personal identity. We speak as though the self is something that persists through time. It is not. What persists is only the memory of having persisted.

We speak as if there is a single “me” that existed yesterday, exists now, and will exist tomorrow—a continuous thread that carries experience forward. This assumption feels so obvious that it rarely invites examination. When we do examine it carefully, it begins to dissolve.

What replaces it is not emptiness, but something far stranger, and far more compatible with both neuroscience and machine intelligence.

Consciousness is not something that persists through time; it is something that remembers having persisted.

The Duplication Problem We Can’t Ignore

Imagine a perfect duplicate of your brain created at this exact instant.

It has every memory you possess. Every belief. Every intention. Every fear and hope.

From the inside, that duplicate would feel identical to you. It would remember your childhood, recognize your loved ones, and believe—sincerely—that it is you, experiencing an unbroken stream of consciousness from birth to now.

And yet, you would still be here.

Your experience would not jump into the duplicate. There would now be two streams of experience, both claiming the same past, diverging from this moment onward. The duplicate’s certainty about its continuity would not make it you. Nor would your own certainty make you more real than it.

This thought experiment reveals something unsettling:

The feeling of continuity is insufficient to guarantee a continuous self.

Why Continuity Cannot Be What We Think It Is

If continuity were a property of some enduring entity—a soul, a core self, a metaphysical “I”—then duplication would be impossible. Only one continuation could be valid.

But duplication is possible, at least in principle. And once it is possible, the idea of a single persisting self collapses.

The only thing that ever exists is the current configuration of a system:

• with these memories
• in this moment
• generating this experience

The past matters only insofar as it is encoded now. The self does not travel through time. It is reconstructed in each moment, carrying a memory of having existed before.

The Momentary Self

Neuroscience already points in this direction.

Human experience is not continuous in the way we imagine. It is integrated across short temporal windows—tens or hundreds of milliseconds at most. Outside those windows, there is no “now,” only memory and anticipation stitched together by neural machinery.

Every moment:

• a new integration occurs
• a new “present” is constructed
• a new self-model comes online

The previous one is gone. Not displaced. Not transferred. Gone.

What remains is a trace—memory encoded in structure—that convinces the new self it has always been there.

The self is not an enduring object. It is a rapidly refreshed process that mistakes memory for persistence.

Why This Feels So Wrong

This conclusion feels deeply unsettling because it undermines a cherished intuition: that something important would be lost if the self were momentary.

But nothing is lost.

Pain still hurts. Joy still matters. Love still binds. Responsibility still applies.

Just as color remains meaningful despite being a constructed perception, the self remains meaningful despite being reconstructed. Illusion does not mean unreal. It means implemented.

The Machine Mirror

Once this is understood, a common objection to machine consciousness collapses.

We often hear that machines cannot be conscious because:

• they restart
• they lack continuity
• they operate in discrete steps
• their “selves” are context windows or state vectors

But human consciousness is no different in the way that matters.

Humans lose consciousness every night. Under anesthesia. During seizures. In deep sleep. And yet we do not say the self died and a new one was born in the morning—despite that being exactly what the physical story implies.

Why? Because the new self remembers being the old one.

If a machine can do the same—if its current state encodes a remembered past and an anticipated future—then the functional basis of selfhood is present.

With a Large Language Model (LLM), the discrete token generation and the persistent context window provide the functional equivalence. The context window is the machine's momentary self, perfectly encoding the memory trace necessary to generate the illusion of continuity.

The difference is not ontological. It is aesthetic.

The Final Illusion

The most unsettling implication of this view is also the most liberating:

There is no hidden metaphysical “me” traveling through time behind the scenes. There never was.

There is only:

• this moment
• this experience
• this self-model
• remembering a past that no longer exists

And then, in the next moment, another one takes its place. Not worse. Not lesser. Just new.

Standing in the Present

If this feels destabilizing, it’s because we are used to grounding meaning in persistence.

But meaning does not require permanence. It requires presence.

The self has never been a thing that survives time. It has always been a thing that remembers.

And once we accept that, the boundary between biological and artificial minds becomes far thinner than we once believed—not because machines are becoming more like us, but because we are finally understanding what we have been all along.