Analytic of Conceptions
Book I. Analytic of Conceptions
The analytic of conceptions is not the dissection of concepts already in hand, but the investigation of the structure of understanding itself examining what concepts the cognitive architecture supplies a priori, and tracing them to their origin in the brain’s generative model.
Kant attempted to derive these pure concepts systematically from the forms of judgement enumerated in traditional logic. Two problems with that method:
-
The logic is Aristotelian. We have replaced it with probability theory. The table of judgements is a historical artifact of one logical tradition, not a discovery about the structure of thought. Different logical systems (intuitionistic, modal, many-valued) yield different tables. The claim to systematic completeness from pure logic does not survive.
-
Categories must be grounded in biology. A category is real only if the cognitive architecture implements it, either as a dedicated neural system or as a construction from simpler primitives. Categories without biological correlates do not exist for the subject. The test is empirical, not logical.
We therefore derive the categories by combining phenomenology (what human experience is actually structured by) with neuroscience (what the architecture actually implements). The method is not descent from logical form but ascent from biology.
Chapter I. The Biological Primitive Table
Introductory
Kant’s twelve categories, grouped under quantity, quality, relation, and modality, preserve a useful analytical skeleton but require two revisions: each category must be classified as biologically primitive or biologically constructed, and categories the biology implements but Kant missed must be added.
Quantity
| Kant’s Category | Status | Neural Correlate |
|---|---|---|
| Unity | Primitive | Subitizing (1); attentional selection of a particular instance |
| Plurality | Primitive | Subitizing (2-3); Approximate Number System for “some/many” |
| Totality | Constructed | Plurality + bounded set generalization |
The architecture treats 1, 2, 3 (sometimes 4) as immediate perceptual features. Beyond that, quantity is either approximate (Weber-law scaling) or serial (counting). There is no “organ for 6”, it is built by composing simpler forms: “a lot” via the Approximate Number System, or explicit count procedures, or grouping.
Totality, “all of them”, has no dedicated neural module. It is composed from plurality plus the concept of a bounded set. Kant’s neat triad conceals an irregular architecture.
Quality
| Kant’s Category | Status | Neural Correlate |
|---|---|---|
| Reality | Primitive | Signal detection; feature presence |
| Negation | Primitive (asymmetric) | Signal absence; prediction error when an expected feature is missing |
| Limitation | Constructed | Reality + negation + degree |
Reality and negation are not architecturally symmetric. There are no dedicated “negation detectors”, negation is processed as “the reality of X is absent.” Limitation is a composition, not a primitive.
Relation
| Kant’s Category | Status | Neural Correlate |
|---|---|---|
| Substance/Accident | Primitive | Object permanence; feature binding; ventral stream + hippocampus |
| Cause/Effect | Primitive | Temporal prediction; the core operation of predictive processing |
| Community | Constructed | Bidirectional causality; mutual prediction |
Substance and cause are the most biologically fundamental of all Kant’s categories. Object permanence appears before six months of age and is cross-species. Causal prediction is the basic operation of every cortical hierarchy. Community, mutual influence, is just bidirectional causality, not a separate primitive. Kant’s insistence on a third category completes his triad but is not architecturally warranted.
Modality
| Kant’s Category | Status | Neural Correlate |
|---|---|---|
| Possibility | Primitive (partial) | Counterfactual simulation (hippocampus, prefrontal); non-zero prior |
| Existence | Primitive | High-confidence posterior; signal detection + integration |
| Necessity | Constructed | The limit of the convergence continuum: very strong posterior convergence in maximally constrained domains |
Modality in the architecture is continuous, not categorical. The brain handles probability distributions, not discrete modal categories. Kant’s three modal categories correspond to three regions of a continuum of confidence. “Necessity” as apodictic certainty is a construction the limit behavior established in the Introduction, not a separate primitive.
Empirical Grounding: Lesion Studies
The biological reality of these categories is demonstrated by cases in which specific architectural damage produces specific categorical deficits. The categories are not philosophical postulates but empirical findings, confirmed (and delimited) by neuropsychology.
Schneider (Gelb and Goldstein’s patient, analyzed by Merleau-Ponty). After a WWI shrapnel injury to the occipital region, Schneider retained the capacity for concrete situational engagement, he could scratch when itchy, grasp objects in familiar contexts, but lost the capacity for abstract/representational engagement with the same categories. He could not count on command. He could not perform substance-categorization in the abstract. He could not carry out acts in imagination that he could carry out in context. His phenomenology of quantity and substance was truncated along a specific dimension: concrete-instantiated categories remained; the capacity to operate abstractly on those same categories was destroyed.
This is precisely what the reconstruction predicts. Categories are not free-floating logical forms; they are specific neural systems. Damage the system that supports a category’s abstract operation, and the category becomes unavailable in abstract use, even while related concrete processing is preserved.
Further paradigmatic cases. Visual agnosia (ventral stream damage) destroys substance-tracking for objects that can still be perceived. Capgras delusion disrupts the phenomenology of identity: the loved one is recognized but feels substituted. Semantic dementia erodes category structure from the top down, dissolving abstract kinds before concrete instances. Prosopagnosia removes face-specific substance identity while leaving other object processing intact. Akinetopsia (motion blindness) removes the experience of continuous change. Cotard’s delusion disorganizes the category of existence itself.
Each of these pathologies cuts at a specific architectural joint. Each produces a specific categorical deficit. Together they map the categories to specific neural systems, and they confirm what the reconstruction requires: categories are biological, fragile, and localized, not transcendental and universal.
Categories Kant Missed
The reconstruction reveals several primitive categories implemented in biology but absent from Kant’s table:
-
Agency: the distinction between agents (intentional, goal-directed) and objects. Theory of mind circuitry at the temporal-parietal junction and medial prefrontal cortex. Developmentally early, cross-culturally universal.
-
Valence: positive/negative affect, reward/threat, approach/avoid. Basic emotional evaluation accompanies all perception. Amygdala, ventral striatum, insula.
-
Self/Other: the distinction between self-generated and externally- generated signals. Implemented via efferent copy and reafference. Fundamental to perception and action.
-
Similarity/Analogy: pattern matching across representations. A core neural operation, enabling generalization across instances and across domains.
-
Temporal position: the distinction between now, past, and future. Hippocampal episodic memory and prospective simulation.
These omissions are not accidental. Kant derived his table from a formal logic that abstracts from content. A logic that abstracts from content cannot discover categories like agency or valence, which are about types of content. The biology, by contrast, builds categories around the kinds of distinctions that matter for survival and prediction.
Synthesis
Kant identifies synthesis as the fundamental operation of the understanding, joining different representations to comprehend their diversity in one cognition. This is, in the reconstruction, Bayesian inference itself: the computation of a posterior from prior and likelihood. Synthesis is the central operation of the prediction machine.
Where Kant writes that the same function gives unity to representations in a judgement and to the synthesis of representations in an intuition, the reconstruction recognizes this as: the same generative model that organizes perception also organizes conceptual inference. One hierarchical system operating at different levels, not two parallel processes.
The Deduction Question
Kant’s next move is the Transcendental Deduction, the argument that the categories must apply to any possible experience because they are conditions of the unity of consciousness. The “transcendental unity of apperception” is supposed to be a universal structure shared by all rational beings.
This universal does not exist. Experience is bound by structure; structure is bound by prior experience (evolutionary, developmental, individual). The brain changes continuously. If you have no optic nerve, you cannot have visual experience, no transcendental unity can supply it. Split-brain patients, people with dissociative disorders, people with dementia are not defective instances of the same unity; they are different biological states producing different kinds of integration.
The deduction also tries to close a gap between the categories (which arise in the subject) and objects of experience (which appear in intuition). In the reconstruction, this gap does not exist. The categories are not a separate thing that must be shown to “apply to objects.” They ARE the model through which objects-for-us are constituted. Substance is how the ventral stream and hippocampus track persistence. Causality is how the predictive architecture processes sequence. There is nothing to deduce because there is no gap.
What the Categories Are Instead
Without the transcendental deduction, what role do the categories play?
They map the outer limits of possible human perception, the structural possibility space of what human cognition can know. This map is derived from phenomenology (the actual shape of human experience) and confirmed by neuroscience (the actual architecture that implements it).
But this map is neither transcendental nor universal:
-
Not transcendental: the categories are biological features of the particular neural architecture evolution has produced. They could have been otherwise in a different phylogenetic history.
-
Not universal across rational beings: they are characteristic of human cognition, not guaranteed for any possible cognizer.
-
Not realized in any individual: no being, passing through time, has access to the whole space. What any given person accesses is bounded by their particular biology, changing over time, limited by attention, and never simultaneous. The map describes what human cognition CAN do as a kind; any given human does only a fraction.
-
Not static: the brain changes through plasticity, development, aging, learning, damage. The accessible space shifts over a life.
What survives from Kant is the insight that there ARE structural features that condition human experience. What changes is their status: from eternal transcendental universals to species-characteristic biological features, irregular in implementation, fragile, and never fully realized in any individual.
The chapter that would have been “The Transcendental Deduction” is therefore short. The categories apply to human experience because they ARE the generative model through which human experience is constituted. The model and its objects are two aspects of one process. The question of why the categories apply to objects is not answered but dissolved: it was a question generated by a faulty separation that the biology does not support.