For me, The Self-Evidencing Agent is probably Jakob Hohwy’s most ambitious work to date. It is not merely another book about predictive processing. It is an attempt to make predictive processing into a first philosophy—a philosophical foundation for understanding existence, selfhood, consciousness, agency, value, and even wisdom.
The central thesis can be summarized in one sentence:
To exist as an agent is to continually generate evidence for one’s own existence.
This is what Hohwy calls self-evidencing.
1. From predictive processing to self-evidencing
Earlier predictive-processing theories mainly said:
- the brain predicts sensory input;
- prediction errors are minimized;
- perception is Bayesian inference.
Hohwy argues that this is not yet fundamental enough.
The deeper question is:
Why is prediction occurring at all?
His answer is:
because organisms must maintain themselves as the kind of things they are.
Living organisms therefore seek evidence that confirms
“I am still the organism I model myself to be.”
This is not psychological reassurance.
It is an existential requirement.
A living system survives only by continually producing evidence that it still exists as that kind of system.
2. Existence becomes statistical
The book begins with an unusual move.
Instead of asking
What is consciousness?
or
What is mind?
it asks
What does it mean for something to exist?
Drawing on Karl Friston’s Free Energy Principle, Hohwy treats existence statistically.
A living thing exists because
- it occupies a relatively stable region of possible states;
- it continually resists entropy;
- it actively keeps itself within those viable states.
Existence therefore becomes
an ongoing achievement,
not a static fact.
3. The self is a model
This is perhaps the most important implication.
The self is not
- a soul,
- an ego,
- a Cartesian substance.
Instead,
the self is
the model that best enables an organism to continue self-evidencing.
A self-model helps predict
- bodily states,
- environmental changes,
- social interactions,
- future outcomes.
Its function is not to discover metaphysical truth.
Its function is to keep the organism alive.
This is why Hohwy devotes a whole chapter to Self-Models and Selves.
4. Action is evidence gathering
Traditional philosophy often says
perception receives information,
while
action changes the world.
Hohwy blurs this distinction.
Action itself becomes
an experiment.
Whenever I move,
I generate sensory evidence.
If the evidence matches my predictions,
my model gains support.
Thus
action is not merely doing.
It is actively sampling reality to test the model of myself and my world.
This extends predictive processing into what is often called active inference.
5. Consciousness
One of the later chapters argues that consciousness is deeply connected with self-evidencing.
Conscious experience is not simply information processing.
Rather,
consciousness helps an organism
- regulate uncertainty,
- evaluate competing hypotheses,
- coordinate action over time.
Earlier papers by Hohwy suggest that consciousness may be a particularly sophisticated form of self-evidencing rather than something added on top of prediction.
6. Value and meaning
Perhaps surprisingly,
the book eventually reaches ethics.
If existence itself requires self-evidencing,
then
- value,
- purpose,
- meaning,
are not mysterious extras.
They arise because an organism must continually determine
which futures best sustain its existence.
Hence the final chapters discuss
- volition,
- emotion,
- wisdom,
- meaning,
- mindfulness,
as increasingly sophisticated ways of managing self-evidencing.
Why this book is philosophically important
I think the book marks a shift.
Hohwy’s earlier work (The Predictive Mind) mainly asked
How does the brain infer the world?
The new book asks
Why must an agent infer anything at all?
That is a much deeper philosophical question.
Predictive processing becomes
ontology,
rather than merely cognitive science.
Relation to your recent thinking
Reading your recent conversations, I suspect what will interest you most is not predictive processing itself, but where Hohwy may differ from your own direction.
You have repeatedly said things like:
- “跟著碎片走 (follow the fragments).”
- “增加世界的陌生性.”
- “尋找有趣的生命型態.”
- “生成大於存有.”
- “讓物歸位.”
These suggest a philosophy in which the world continually exceeds our models. Novelty, surprise, and the “otherness of the other” are not merely errors to be reduced; they are what call us beyond ourselves.
Hohwy, by contrast, gives primacy to the organism’s need to preserve itself through increasingly successful self-models. Even when he discusses openness, uncertainty, and exploration, they remain in service of maintaining the agent’s continued existence through self-evidencing.
That is where an interesting dialogue emerges.
One could say:
- Hohwy: We live by continually confirming ourselves.
- Your emerging “Kikujiro Sixteen Forms”: We live by repeatedly allowing ourselves to be interrupted by fragments that we could not have predicted.
Or, in your own vocabulary:
Hohwy asks how an agent maintains its world model.
You increasingly ask how an agent can step outside its attractor basin and let another world begin.
That difference may turn out to be a fruitful conversation between predictive processing and what you have been developing through Heidegger’s Ereignis, Deleuze’s plateau, Winnicott, and your idea of “following the fragments.” Hohwy offers a remarkably sophisticated theory of how a world is maintained; your recent work seems to be searching for a philosophy of how another world becomes possible.
