Analytic of Principles: Schematism

Book II. Analytic of Principles

The analytic of conceptions identified the primitive categories of human cognition. The analytic of principles asks how these categories apply to experience, how abstract structural features of the model meet concrete sensory content.

Kant poses the problem sharply: the pure conceptions of the understanding seem “heterogeneous” with sensory intuitions. You cannot find causality directly in a phenomenon. There must, he argues, be some mediating representation, a “third thing”, that is homogeneous with both the category and the phenomenon. He calls this the schema, and the procedure of constructing schemata he calls schematism.

Kant calls the schematism “an art, hidden in the depths of the human soul.” The reconstruction removes the mystery. The schema is not a mysterious mediator between static abstract categories and static concrete phenomena. It is the dynamic, physical, temporally-extended process of neural computation itself, shaped by both architectural constraint and individual experience.

Chapter I. The Schematism: Neural Tissue Computing in Time

The Dissolution of the Heterogeneity Problem

Kant’s problem assumes that categories are abstract things (conceptual, logical, non-sensuous) that somehow need to be applied to concrete things (sensory, temporal, intuited). On that assumption, a mediator is required.

Under the reconstruction, there is no such gap. Categories are not abstract entities stored somewhere separate from sensory processing. They are specific neural circuits instantiated in specific cerebral locations, processing sensory content in time. Substance is not an abstract concept awaiting application, it is the predictive tracking of persistence carried out by the ventral stream and hippocampus. Causality is not a logical category applied to events, it is the temporal prediction carried out by the cortical hierarchy. The “categories” and the “phenomena” are not two things separated by an ontological gap. They are aspects of one physical process: neural tissue computing on sensory signals in time.

The heterogeneity problem dissolves because the premise, that categories and phenomena are heterogeneous, is false. They are homogeneous: both are activity patterns in biological tissue.

Time and Cerebral Localization

Kant identifies time as the universal mediator, every schema is a temporal determination. He is correct that time is fundamental. What he does not see, because he lacks the neuroscience, is that time is fundamental because neural computation unfolds in time: spike trains, oscillations, conduction delays, sequential activation. Every cognitive process is realized temporally because neurons process signals temporally.

To this the reconstruction adds a second fundamental fact Kant could not have known: cerebral localization. Categories are instantiated in specific places. Grid cells in entorhinal cortex. Object recognition in the inferior temporal cortex. Causal prediction distributed across cortical hierarchies with specific laminar organization. Emotional valence in amygdala and insula. Agency detection in medial prefrontal and temporal-parietal regions.

Time and cerebral localization are the two fundamental physical facts that the schematism must be understood through. The schema is specific neural tissue in a specific location computing in time.

Constrained Plasticity

If the schematism were only this, specific tissue computing in time it would still be too static to capture what the biology actually does. The brain is a plastic organ. It reshapes itself physically in response to experience:

Hebbian learning: neurons that fire together wire together. Processing is restructuring.
Synaptic plasticity: long-term potentiation and depression modify connection strengths.
Dendritic growth and pruning: the physical shape of neurons changes.
Myelination: conduction velocities change with use.
In some regions, adult neurogenesis: new neurons are added throughout life.

The tissue that imposes categories is literally reshaped by the experience it categorizes. London taxi drivers develop enlarged posterior hippocampi from years of spatial learning. Congenitally blind individuals recruit visual cortex for language and touch processing. Musicians have enlarged auditory and motor cortices. Children raised in different linguistic environments develop different phonemic categories physically instantiated in different cortical organization.

The world reshapes the brain. The brain reshapes the world. These are two aspects of one continuous process. Structure and content co-constitute each other through time.

But plasticity is not unbounded. It operates within:

Genetic scaffolding: cortical architecture is channeled by genetic programs. Grid cells form hexagonal patterns, not random topologies. V1 has specific laminar structure. Critical periods constrain when certain learning can occur.
Physical constraints: conduction velocities, metabolic limits, the chemistry of neurotransmission, the geometry of tissue.
Evolutionary shaping: the space of possible transformations has been selected for species-level survival. Plasticity cannot turn neural tissue into something non-neural.

The schematism is therefore constrained plasticity: the continuous mutual shaping of brain and world, operating within evolutionarily-given architectural bounds. Kant’s “rule of the imagination” becomes: the trajectory of architectural change under experience, constrained by biology, realized in time in specific locations.

Shared Phenomenology Through Convergence

If every brain is reshaped differently by different experiences, how is shared phenomenology possible? How do two people come to understand one another, inhabit a common world, agree on what they see?

The answer is convergence under shared constraint, and it parallels exactly the Bayesian solution to the induction problem.

In Bayesian inference, the induction problem is solved not by a special faculty or a transcendental guarantee but by the fact that different priors, updated against strong shared evidence, converge on similar posteriors. De Finetti’s exchangeability theorem formalizes this: under sufficient shared evidence, rational agents with different priors arrive at the same posterior. Agreement is forced by the structure of the evidence, not by prior sameness.

The same structure applies to cognitive architecture. Different genetic starting points, subjected to:

Shared survival pressures (nutrition, threat, mating, social coordination),
Shared physical constraints (the same universe, the same physics, the same chemistry, the same embodiment),
Shared environmental structure (objects that persist, events that have causes, agents that act with intent),

produce convergent cognitive architectures. Humans share phenomenology not because we all have identical brains, not because we share a transcendental faculty, but because we are cost-optimizing on similar survival functions within shared physical constraints. The convergence is forced by the problem, not granted from above.

This is not relativism. It is convergence from below rather than universality from above. Objectivity is what convergence looks like when the constraints are strong enough to force it, the same story, whether in Bayesian updating or in evolutionary cognition.

The Specific Schemata Reconsidered

Kant identifies specific schemata for each category. Under the reconstruction, these become specific claims about neural computation:

Schema of quantity (number; successive synthesis): subitizing for small numbers; serial counting for larger ones; ANS for approximate quantities. All implemented in parietal circuits operating in time.
Schema of reality (filling of time with sensation): signal detection. The presence of a feature corresponds to cortical activity above baseline, filling a temporal window with predictive-matching content.
Schema of negation (empty time): prediction error when an expected signal fails to arrive. A specific form of cortical processing distinct from reality but parasitic on it.
Schema of substance (permanence of the real in time): the hippocampal-ventral-stream system tracking object identity across occlusion, feature change, and temporal gaps. Predictive persistence is the schema.
Schema of cause (succession subjected to a rule): hierarchical temporal prediction. Rule-governed succession is exactly what the cortical hierarchy computes.
Schema of community (coexistence according to a rule): bidirectional prediction between coupled systems. A composition, not a primitive.
Schema of possibility: counterfactual simulation. The hippocampus and prefrontal cortex generate predictions for non-actual states.
Schema of existence: the current posterior, high-confidence representation of what is the case.
Schema of necessity: the limit case of strong convergence under rigid constraint. Not a separate schema but the extreme end of the modal continuum.

The Hidden Art Unveiled

Kant called the schematism “an art, hidden in the depths of the human soul, whose true modes of action we shall only with difficulty discover and unveil.” The depth is real and the difficulty is real. But the art is not hidden in a soul. It is hidden in neural tissue.

The schematism is:

Specific neural circuits (cerebral localization)
Computing in time (all neural computation is temporal)
Reshaping themselves through experience (plasticity)
Within evolutionary and physical constraints (bounded plasticity)
Producing convergent phenomenology across individuals through shared optimization pressure (objectivity as convergence)

The “third thing” Kant sought between category and phenomenon does not exist because the two were never separate. What exists is one physical process, the brain’s continuous, constrained, localized, temporal restructuring of itself, and the schematism is that process under description.