Beyond Fast and Slow: A Four-Layer Cognitive Architecture from Dimensional Sequence Theory
超越快与慢：维度序列理论下的四层认知架构

Writing Declaration: This paper was co-drafted with Claude (Anthropic). All theoretical decisions, framework design, and final editorial judgments were made by the author. The Chinese version is authoritative; this English version is an independent rewrite.

1. Introduction: The Attribution Problem of Fairness-Related Cognition

A person receives an unfair distribution proposal and within hundreds of milliseconds feels the impulse to reject it. How would the popular formulation of Kahneman's framework classify this response? It is fast, so System 1. It is emotion-driven (anterior insula activation, anger, disgust), so System 1. It requires no deliberate reasoning, so System 1.

A necessary clarification: this paper's target is not the entirety of contemporary dual-process theory. Recent work within the dual-process tradition has substantially revised the old exclusivity assumption that "fast = System 1, slow = System 2" (De Neys 2023; Stanovich & Toplak). This paper targets the version of Kahneman's framework most influential in public discourse and cross-disciplinary citation—the popular formulation that uses processing speed as its primary classification criterion. This formulation has profoundly shaped behavioral economics, education, and public understanding, but it cannot accommodate the structural complexity this paper addresses. The four-layer architecture proposed here is not an extension of the dual-process framework ("adding floors to the building") but a change of classification axis.

Returning to the attribution problem. What does this "fast emotional response" contain? It contains at least: (1) I have an expectation, (2) the other party made a choice, (3) this choice deviates from my expectation, and (4) this deviation is illegitimate—not because I lost out, but because the manner of distribution is wrong.

Point (4) is critical. A dog whose bone is snatched will attack. But it does not judge "unfair distribution." It only knows "my thing is gone." The judgment of "unfairness" requires simultaneously holding two perspectives: what I should receive, what the other party should receive, and whether the ratio between them is reasonable. This is a structurally complex cognitive operation—it requires simultaneously representing two independent agents' purposes and evaluating the relationship between them.

Kahneman's popular formulation classifies it as System 1 because its criterion is speed. Fast means System 1, slow means System 2. But speed is a function of training, not of cognitive level. A Go master's intuitive move is "fast," but that intuition compresses extraordinarily complex structural judgments. A person who has repeatedly experienced distribution scenarios since childhood has similarly compressed fairness judgments into fast responses—but this does not mean the judgment is structurally "simple."

If we change the classification criterion from speed to structural complexity, what do we get? This paper's answer: not two systems, but four.

2. A Priori Derivation of the Four-Layer Cognitive Architecture from the DD Sequence

2.1 The Dimensional Degree Sequence

The Dimensional Degree (DD) sequence is a core tool of the Self-as-an-End (SAE) framework. It describes the incremental increase of structural complexity: each additional dimension grants the system a capability absent at the previous level. The full derivation of the DD sequence is given in the SAE foundational papers; here only the 12DD–15DD interval relevant to cognitive architecture is presented.

2.2 Derivation of the Four Layers

12DD — Thinking. Pure cognitive operation without subject awareness. The system processes information and responds, but has no awareness that "I am doing this." A bird sees a worm and pecks—no "I" is required. This is stimulus-response level cognition.

Clarification: this paper's System 1 (12DD) refers to the cognitive floor possessing rudimentary causal prediction capacity—capable of inference, pattern matching, and predicting the future from the past. Lower-order physiological reflexes (pupillary light reflex, knee-jerk reflex) as well as pure sensation (10DD) and pure memory (11DD) fall below 12DD and lie outside this paper's four-layer architecture. Kahneman's System 1 actually spans multiple levels from 10DD to 12DD; this paper takes only 12DD as the strict correspondence for System 1.

13DD — I think. The subject emerges. The same cognitive activity gains an additional dimension: "it is I who am thinking." This is the minimal unit of self-awareness. The system not only processes the world but knows it is processing the world. This is the origin of reflectivity.

14DD — I must. Not only is there a subject, but a teleological structure with non-negotiability. The "I" is no longer merely the holder of thoughts but the bearer of will. "I know that doing B has costs, but A is non-negotiable for me, so I must do B." The key is not goal-directed behavior per se (12DD animals also have goal-directed behavior) but the coexistence of cost awareness and commitment to purpose. This is volition.

15DD — We each must, therefore. A fundamental directional shift: from "I" to "I and you." My must is A, the other's must is B. C is neither A nor B, but something new that emerges from acknowledging the coexistence of two independent purposes. If C = A, the other is ignored—regression to 14DD. If C = B, the self is abandoned—also regression to 14DD. C must be something new that simultaneously accommodates "I must A" and "you must B."

Minimum criteria for 15DD. To distinguish genuine 15DD operations from various degraded forms, four necessary conditions are specified:

First, both parties must have non-negotiable items. If one party has no "must," this is not negotiation but accommodation.

Second, C cannot equal either party's surface demand A or B.

Third, C's generation depends on continued interrogation of A'/B' (the true non-negotiable items behind surface demands), not merely interest compromise. The direction of interrogation is "why must you," not "can you concede a little."

Fourth, no deal is a legitimate output. If A' and B' are fundamentally incompatible, honest no deal is the normal result of 15DD operation, not failure. The failure of 15DD is not no deal but regression to 14DD (one party capitulates or coerces).

2.3 Subsumption and Non-Skippability

The four layers are subsumptive, not independent modules. System 4 in negotiation must simultaneously run System 3 (I know what I must not give up), System 2 (I can reflect on my own judgments), and System 1 (I have intuitive situational awareness). Missing any layer collapses the negotiation. System 4 without System 3 is unprincipled accommodation. System 4 without System 2 is blind insistence. System 4 without System 1 is detached abstraction.

Equally critical is non-skippability: System 4 cannot bypass System 3. Without stable "I must," genuine negotiation is impossible—because you do not know what you can concede and what you cannot.

2.4 Coarse-Grained Membership: The 9D Structure

In the coarse-grained version of the DD sequence (the D sequence), 13DD and 14DD jointly belong to 9D. This means they share substantial underlying substrate and are distinguishable only at fine granularity. This yields a direct experimental prediction: metacognitive monitoring (13DD) and goal maintenance (14DD) should be partially separable under normal conditions but converge under high cognitive load—because fine-grained distinctions collapse to the coarse-grained 9D shared substrate. The dissociation between 12DD and 15DD should persist under the same high-load conditions, because they do not share a D-level.

2.5 Isomorphism with the SAE Psychoanalytic Four-Layer Model

The same 12DD–15DD interval yields another set of projections in the SAE psychoanalysis series: Id (me-without-self, 12DD), Ego (self-without-purpose, 13DD), Superego (self-with-purpose, 14DD), Cert (self-with-non-dubito, 15DD). The psychoanalysis series cuts psychological structure (modes of psychic operation); this paper cuts cognitive structure (levels of thinking). The two are not independent theories but projections of the same DD structure onto different materials.

A terminological bridge is necessary: the psychoanalysis series' core thesis—"the object determines the layer, not the developmental stage"—uses "object" to mean relational objects. In cognitive science contexts, "object" should be generalized. In this paper, "object" refers broadly to the task-relation configuration that triggers a particular layer's operation—it may be a person (mother, boss), a cognitive task (fairness judgment, planning), or even one's own judgment (metacognitive monitoring). Layers are not global attributes but situational and configurational.

| DD | Cognitive Architecture (this paper) | Psychic Structure (psychoanalysis series) | Common structural feature |

|---|---|---|---|

| 12DD | System 1: Thinking | Id: me-without-self | Operation without subject awareness |

| 13DD | System 2: I think | Ego: self-without-purpose | Subject present, no direction |

| 14DD | System 3: I must | Superego: self-with-purpose | Subject with direction, single agent |

| 15DD | System 4: We each must, therefore | Cert: self-with-non-dubito | Facing the other, dual agent |

The psychoanalysis series also proposed a concept directly relevant here: "pseudo-high-layer overlay"—operating at a low layer while masking it with high-layer narrative. This corresponds to "false C" in this paper's prediction framework: false C is pseudo-high-layer overlay in the negotiation domain—the appearance of a 15DD outcome when one party has actually regressed to 14DD (concession or capitulation).

3. Benchmarking Against Existing Philosophical Frameworks

3.1 Schopenhauer and Nietzsche: Stuck at 14DD

Schopenhauer saw will and said it was suffering—negate it. Nietzsche saw the same will and said affirm it—push it to the extreme. Neither reached 15DD. Neither genuinely addressed "there is another subject with an equally real will across from me."

Schopenhauer's solution was to cancel will (asceticism, aesthetic contemplation). Nietzsche's was to maximize will (the Übermensch, will to power). One suppresses, the other elevates, but both remain single-agent. Nietzsche's Übermensch is solitary. Schopenhauer's saint is solitary.

3.2 Sartre: The Apex and Impasse of 14DD

Sartre is the most paradigmatic case. "Hell is other people" exposes his position at 14DD. He saw that there is another subject (the "gaze"), but his response to this discovery was: the other's freedom threatens my freedom. Two 14DD wills meet; the result is confrontation, not negotiation. No C.

This is why Sartre never completed his ethics. He said throughout his later years that he would write it. He could not. Because ethics cannot be derived from within a 14DD framework. Ethics requires 15DD—"we each must, therefore."

3.3 Kant: At the Door of 15DD

Kant's categorical imperative—"treat humanity never merely as a means but always also as an end"—demands acknowledgment of the other subject's independent will. Kant reached the door of 15DD. But Kant's problem was that he addressed this with a formal law. The categorical imperative is a static rule, not a dynamic process. It tells you what you ought to do but not how C is produced when two "musts" actually collide.

SAE converts it into process: two "musts" collide, each party interrogates the other's "why must you," peeling apart surface-level means from truly non-negotiable items, then testing compatibility at that level, producing C or accepting honest no deal.

4. Empirical Evidence: Neuroscience

4.1 System 1 and System 2

The experimental basis for these two layers is the most mature and needs no extended discussion. Kahneman and Tversky's behavioral experiments, together with extensive fMRI research, have established the basic dissociation between automatic responses (amygdala, basal ganglia dominant) and deliberate reflection (prefrontal cortex dominant).

4.2 Partial Dissociation of System 2 and System 3

Metacognitive monitoring and goal maintenance/planning show evidence of partial dissociation in neural substrates.

Fleming et al. found that patients with anterior prefrontal cortex (anterior PFC) lesions performed normally on perceptual tasks but showed selectively impaired metacognitive accuracy—they could make judgments but could not assess how reliable their judgments were. This is a selective impairment of System 2 (quality monitoring of "I think").

Meanwhile, goal maintenance and planning primarily depend on dorsolateral prefrontal cortex (dorsolateral PFC) and striatal networks, with prospective memory particularly involving the frontal pole (BA10). These regions are anatomically distinguishable from the core regions of metacognitive monitoring (anterior PFC).

More direct evidence: a lifespan study found no significant correlation between executive function and metacognitive efficiency. This means "monitoring my thinking" and "maintaining and executing goals" can vary independently.

This is precisely the DD sequence's prediction: 13DD and 14DD share membership in 9D, so they overlap extensively but are not identical. Partial dissociation (rather than clean dissociation) is exactly the theoretically predicted pattern.

4.3 Dissociation of System 3 and System 4

Evidence for this layer comes from multiple independent research lines. An upfront declaration is necessary: compared with the System 2/System 3 partial dissociation in section 4.2, the System 3/System 4 dissociation currently rests primarily on adjacent support rather than direct support. No existing experiments were designed with "14DD vs 15DD" as their hypothesis, but multiple independent lines of evidence converge in a direction consistent with the DD sequence's predictions.

It must also be made explicit that fairness-related cognition does not belong as a whole to any single layer. A more accurate analysis decomposes it into a multi-layer cognitive family:

12DD layer: primitive loss response to deprivation/theft ("my thing is gone").

13DD layer: evaluative judgment about the distribution outcome ("is this right? how confident am I?").

14DD layer: willingness to pay costs to enforce a norm ("I will reject this unfair proposal even though rejection means I get nothing either").

15DD layer: intention-sensitive, third-party-executable, dual-agent integrative fairness judgment ("the other party could have chosen a fairer split but did not"), and the capacity to produce C in conflicts between both parties' non-negotiable items.

This paper's claim is not "fairness = System 4" but rather: what most represents System 4 is the 15DD-layer subset of fairness judgments—intention-sensitive, third-party, dual-agent integrative, and potentially C-producing. The following evidence points to the independence of this subset.

4.3.1 Dissociation between fairness judgment and fairness enforcement

Knoch et al. used transcranial magnetic stimulation (TMS) to disrupt right dorsolateral prefrontal cortex and found that participants still judged proposals as unfair but were no longer willing to reject them. This is a clean dissociation: evaluating unfairness ("this is wrong") and executing fairness punishment ("I will reject") draw on different cognitive resources.

4.3.2 Neural basis of third-party fairness

Corradi-Dell'Acqua et al. found that anterior insula activated both when "I am treated unfairly" and when "someone else is treated unfairly," while medial prefrontal cortex activated only when "I myself am treated unfairly." This indicates that fairness judgment is not merely a self-protective emotional response. This pattern is most consistent with the 15DD structure: an operation requiring representation of the other as an independent purposive agent.

4.3.3 Multi-stage neural time course of third-party punishment

Ginther et al. used fMRI to study third-party punishment, temporally separating harm evaluation, mental state evaluation, integration, and punishment decision. Results showed distinct neural substrates for each stage. This four-stage architecture is consistent with the "dual-agent reasoning" predicted by 15DD—requiring simultaneous processing of the offender's intention and the victim's harm, followed by integration, followed by action.

4.3.4 Intention-sensitive fairness judgment

In a mini-ultimatum game, TMS disruption of right TPJ selectively affected conditions requiring inference about the proposer's intentions—specifically, situations where the proposer "had fairer options available but did not choose them"—without affecting simple inequality aversion. This is consistent with a key distinction in the DD framework: (1) outcome-based inequality aversion ("I got less," corresponding to 12DD/13DD structure) and (2) mental-state-based fairness evaluation ("the other could have been fairer but chose not to," corresponding to 15DD structure) are separable operations at the neural substrate level.

5. Empirical Evidence: Developmental Psychology

The developmental timeline provides a verification pathway independent of neuroscience. If the four Systems represent genuinely different structural dimensions, they should emerge in a recognizable sequence in child development, with each layer's signature event corresponding to its DD definition. The age ranges given below are representative manifestation windows, not sharp thresholds.

5.1 System 1 (12DD) — Representative Window from 6-8 Months: Thinking

Infants have reflexive responses from birth, but "thinking" in the 12DD sense—active cognitive operations on the external world—begins to manifest clearly around 6-8 months. Object permanence begins to emerge; infants can track disappeared objects and make simple causal inferences. This is information processing without awareness of "I am processing."

5.2 System 2 (13DD) — Representative Window from 2-3 Years: I Think

Signature event: terrible twos. Not "the child becomes disobedient," but negation for the first time folding back on itself via the other—through negating the other's will, "I" and "you" are simultaneously carved out. The SAE Terrible Twos paper details this genesis mechanism: it is not that "I" exists first and then resists; rather, through resistance "I" appears for the first time.

Subsequent marker after 13DD completion: fear of death (3-8 years). Not "pain" (12DD), not "danger approaching, flee" (12DD), but "I will cease to exist." This requires an already-carved-out "I" and the application of the abstract state "non-existence" to that "I." Children before age 3 do not fear death because they do not yet have a sufficiently stable "I" to bear this fear.

Developmental research on metacognitive uncertainty monitoring confirms that preschool children begin to reflect on certainty/uncertainty about decisions, with this ability developing significantly between ages 3 and 5.

5.3 System 3 (14DD) — Representative Window from 10-14 Years: I Must

Signature event: terrible teens (adolescent rebellion). Not "I don't want to" (13DD, preference), but "you cannot make me." This "cannot make me" contains a complete volitional structure—I know what you want, I know you have power, I know resistance has costs, but I still refuse, because there is something non-negotiable for me.

The SAE Terrible Teens paper argues that 14DD's genesis mechanism is: "must" emerges from the sustained accumulation of 13DD's remainder ("I exist but don't know what for"). Rebellion is not the genesis mechanism of 14DD itself but its behavioral byproduct and one of its confirmation conditions. The form of 14DD is "the subject cannot not have its own purpose"; "I must X" is this form's minimal concretization in experience—independent of external permission, suppressed it returns in other forms.

A critical distinction: goal-directed behavior does not equal volition. An infant reaching for an object is goal-directed, but this is not volition. The defining characteristic of volition is "must"—the coexistence of cost awareness and commitment to purpose. This structure stably appears around ages 10-14, corresponding to the maturation of prefrontal control systems and the critical period of identity construction.

5.4 System 4 (15DD) — Terrible Telos: The Genesis of Acknowledgment

The developmental trajectory of 15DD requires distinguishing proto-forms, entry events, and mature forms.

Proto-forms (around age 3). Developmental evidence shows that precursors of joint commitment and early collaboration appear around age 3. Gräfenhain et al. showed that 3-year-olds in joint commitment conditions were more likely to wait, help, and substitute for partners. Hamann et al. found that 3.5-year-olds continued cooperating after receiving their own rewards. But these precursor forms lack two equal 14DDs in collision—only an unstabilized 13DD interacting with adult frameworks. They are prerequisite experiences toward 15DD, not 15DD itself.

Entry event: the first high-intensity collision of two parallel teloi. The SAE Terrible Telos paper argues that 15DD's typical entry is not conflict with parents (that is the site of 13DD and 14DD genesis) but first love—the first encounter between two parallel, equal 14DDs. You cannot not love this person (14DD), they cannot not love you (14DD), but your "must" and their "must" are not the same thing.

This collision differs structurally from the previous two terribles in the nature of the conflict object: terrible twos' conflict object is authority (parents), with power asymmetry; terrible teens' conflict object is still authority, but power begins to be challenged; terrible telos' conflict object is another equal 14DD—no power differential, no shelter of the caregiving relationship, pure telos against telos.

The entry to 15DD is painful precisely because there is no exit: 12DD negotiation ("I concede a step, you concede a step") degrades telos to preference—you cannot ask someone who "must paint" to concede half a step toward being half an accountant; 14DD suppression ("my must is more important than yours") is colonization. The only way out is 15DD—acknowledging the other's telos as the other's own, without attempting to dissolve the conflict, without withdrawing acknowledgment.

Mature form (lifelong, not universally achieved). The maturation marker of 15DD is not learning negotiation skills but acknowledgment transforming from a painful choice into a state that no longer requires choosing—collision persists, but acknowledgment no longer requires extra force to maintain. The Terrible Telos paper identifies three persistent forms of 15DD's pain throughout life: cannot do it (knowing one should acknowledge but being unable), doing it but sacrificing oneself (colonizing oneself rather than the other), doing it but unknowingly colonizing (carrying control in the form of love). A substantial proportion of adults under high-pressure conditions regress to System 3 operations (insistence or capitulation) rather than producing C.

5.5 Developmental Timeline Summary: Three Terribles

| DD Level | System | Core Capacity | Terrible T | Conflict Object | Representative Window |

|----------|--------|--------------|-----------|----------------|----------------------|

| 12DD | System 1 | Thinking | — | — | 6-8 months |

| 13DD | System 2 | I think | Terrible Twos | Parents (authority) | Entry 2-3 yrs; unfolds 3-8 yrs |

| 14DD | System 3 | I must | Terrible Teens | Parents/authority (power challenged) | 10-14 years |

| 15DD | System 4 | We each must, therefore | Terrible Telos | An equal 14DD (first love) | Proto-forms from age 3; entry post-adolescence; mature form lifelong, not universally achieved |

Two structural features of this timeline. First, each step shortens in span but increases in difficulty: 12DD to 13DD takes about 3 years, 13DD to 14DD about 5-7 years, 14DD to 15DD may require a lifetime. Second, the conflict object progresses: from authority (power asymmetric) to an equal other (power symmetric), corresponding to the subject's progression from "passively encountering negation" to "actively facing another complete subject."

6. Empirical Evidence: Comparative Psychology and Tomasello

Tomasello's shared intentionality hypothesis provides independent support for System 4 from comparative psychology, but the mapping between the two requires precise handling.

Tomasello distinguished two cognitive upgrades: joint intentionality (interindividual collaboration) and collective intentionality (normative interaction within cultural groups). His core argument: these capacities are uniquely human because at some point in evolutionary history humans became mutually interdependent.

In the DD framework, this distinction receives a more precise formulation:

What chimpanzees can do. Chimpanzees can select cooperative partners and initiate cooperation when it is advantageous. Melis et al. demonstrated that chimpanzees selectively recruit collaborators when collaboration is required. This is a 14DD operation—"My goal is A, cooperating with this partner is a means to A, so I choose to cooperate." The goal is single-agent; cooperation is instrumental.

What chimpanzees cannot do. Chimpanzees cannot sustain normative joint commitment—continuing to cooperate for a partner's benefit after their own payoff has been secured. Hamann et al. found differences between 2.5- and 3.5-year-old human children on this point, with 3.5-year-olds tending to maintain cooperation after receiving their own reward.

This is the watershed between 14DD and 15DD. Cooperation at 14DD loses its motivational force the moment "my goal is achieved." Cooperation involving 15DD persists beyond that point because the joint commitment itself has become an independent purpose.

6.1 Precise Positioning of We-Intentionality: A Bridge Between 14DD and 15DD, Not 15DD Itself

Tomasello's we-intentionality (shared intentionality) should not be mapped to 15DD, nor to 14DD. Its most precise position is the bridge between 14DD and 15DD—and it is a narrow bridge.

In pure 14DD, the other is a means or obstacle: "I have a purpose; you either help me or block me." In we-intentionality, a subtle but critical shift occurs: for this particular event, I begin to recognize that you too have purpose. This is bilateral—both parties attempt to understand that the other is also a purposive agent—but it is narrow, purpose-relative-to-the-event. When the cooperative hunt ends, this recognition is not retained. Next time, the bridge must be rebuilt.

This is why 3-year-old children can exhibit "you too are a purposive agent" behavior in specific joint commitment tasks but do not retain it outside that task—they have taken a few steps onto the bridge but have not crossed to the other side. Crossing (genuine 15DD) requires stable 14DD (mature volition) before "purpose-relative-to-the-event" can be generalized to "you are a purpose."

The genuine difference between 15DD and we-intentionality is that 15DD's confirmation of the other as an independent purpose is not bound to any particular event. I confirm you as a purpose not because we are currently hunting together, but because you are a being with your own "must."

The DD framework thus identifies a finer spectrum within the interval between Tomasello's joint intentionality and collective intentionality:

14DD — I have a purpose; you are a means or obstacle. Chimpanzee cooperative hunting sits here.

14DD→15DD bridge (we-intentionality) — For this event, both parties begin to recognize the other as also having purpose. Bilateral but narrow, purpose-relative-to-the-event. Not retained when the event ends. 3-year-old children's joint commitment behavior and most human collaborative activity as described by Tomasello sit here.

Unilateral 15DD (this paper's System 4) — I confirm you as an independent purpose, not limited to this event. I interrogate your "must" and produce C. Unilaterally initiable.

Thin bilateral 15DD — Both parties mutually confirm each other as independent purposes and jointly produce C. Still revocable.

16DD (mutual non-dubito, beyond this paper's scope) — Not only bilateral confirmation, but non-dubito. An irrevocable mutual locking of ontological stance.

Tomasello's "collective intentionality" covers the entire interval from the bridge to 16DD in a single term—a zone of ambiguity in his framework. The DD sequence disaggregates this interval into distinguishable levels, each with its own initiation conditions, persistence characteristics, and revocability. This also explains why Tomasello's framework describes situational cooperative behavior well (the bridge) but struggles with the persistence of institutions and norms—institutions require not the bridge's event-by-event we-intentionality but at least thin bilateral 15DD or even 16DD.

6.2 DD Rereading of Tomasello's Three Capacities

Tomasello proposed that the cognitive upgrade from individual to shared intentionality requires three additional capacities: recursive thinking (coordinating with others in shared action), objective thinking (resolving perspectival disputes), and sociomoral and normative concepts (resolving preference disputes). In the DD framework, these three capacities are not parallel additions but three facets of the single additional dimension that 15DD has over 14DD—all pointing to the same structural requirement: acknowledging that the being across from you is an independent agent with its own "must."

7. Specific Corrections to the Kahneman Framework

7.1 Independence of Speed and Level

The fundamental classification criterion of Kahneman's popular formulation is processing speed: fast = System 1, slow = System 2. This paper's core correction: speed and cognitive level are two independent dimensions.

A fully trained 15DD operation can be faster than an unpracticed 13DD operation. An experienced diplomat senses the power dynamics at a negotiation table and each party's core interests within seconds, without "careful analysis." But the structural complexity compressed into that "intuition" is 15DD-level—it simultaneously holds multiple independent agents' purpose structures.

Analogy: a Go master's intuitive move is "fast." Kahneman's framework would classify it as System 1. But compressed within that intuition are extraordinarily complex structural judgments, with information-processing complexity far exceeding an ordinary person's System 2 analysis. "Fast" does not mean "simple." "Slow" does not mean "complex."

7.2 Four-Layer Decomposition of Fairness-Related Cognition

In Kahneman's framework, the sense of fairness is classified as System 1 (fast emotional response) or as System 1/System 2 conflict (emotion vs. reason). This paper's correction is not to move fairness as a whole from System 1 to System 4, but to point out that fairness-related cognition is a multi-layer cognitive family spanning different levels of structural complexity.

12DD layer: Primitive loss response to deprivation. "My thing is gone." A dog whose bone is snatched has this response. No "fairness" concept required—only loss detection.

13DD layer: Evaluative judgment about the distribution outcome. "Is this right? How confident am I?" Subject awareness is present; self-evaluation of judgment is possible. But no purpose-driven action yet.

14DD layer: Willingness to pay costs for a norm. "I will reject this unfair proposal even though rejection means I get nothing either." This requires volition—coexistence of cost awareness and principled commitment.

15DD layer: Intention-sensitive, third-party-executable, dual-agent integrative fairness judgment. "The other party could have chosen a fairer split but did not"—requiring simultaneous representation of the other's choice set, intention, and one's own expectation, followed by structural evaluation. In extreme cases, producing C in conflicts between both parties' non-negotiable items.

Kahneman's fast/slow dichotomy cannot accommodate this four-layer decomposition because all four layers' responses can be "fast" (all can be training-compressed), yet their structural complexity is entirely different. This is the fundamental deficiency of speed as a classification criterion.

What most represents the unique contribution of System 4 is the 15DD-layer subset—it requires the simultaneous presence of dual-agent structure, which cannot be achieved at any level from 12DD through 14DD.

7.3 The Speed U-Curve: Why Kahneman Mistakes 15DD for 12DD

The fast/slow dichotomy can accommodate at most two systems—because once speed is the classification criterion, only "fast" and "slow" are possible categories.

Change the classification criterion to structural complexity and this constraint immediately vanishes. The dimensional increment from 12DD to 15DD yields four natural levels, each with independent defining characteristics.

But a deeper question must be answered: if each additional dimension adds another self-referential verification loop, 15DD should be slower than 14DD. Why is it instead fast?

The answer is that the four layers' speeds form not a linear increase in slowness but a U-curve:

12DD — Fast. No self-reference, no internal verification loops. Pure stimulus-response, direct throughput.

13DD — Slow. "I" appears; self-reference begins. "I am thinking" introduces an unprecedented internal loop that takes time to run. The source of slowness is the uncertainty of "what am I."

14DD — Also slow. The tension between "must" and cost requires continuous verification—do I really must? Can I really bear the cost? The source of slowness is the uncertainty of "is my must real."

15DD — Fast again. Direction fundamentally flips: from "I" to "you." When direction flips, the self-verification uncertainty accumulated by 13DD and 14DD is locked down. Because doing 15DD presupposes that 14DD is stable—"what I must" already has an answer and no longer requires continuous verification (this is the other face of non-skippability). All internal verification loops of 13DD and 14DD are still running, but they have moved from foreground to background, becoming background conditions. Cognitive resources are entirely released to the only new variable: "what does the person across from me must."

The fast of 12DD and the fast of 15DD are thus structurally symmetric: 12DD is fast because the self has not yet appeared; 15DD is fast because the self has stabilized to the point of no longer requiring verification. One at the beginning, one at the end, with the process of self-establishment in between. Kahneman mistakes 15DD's fast for 12DD's fast precisely because he looks only at speed, not at the structure behind speed: 12DD's fast is empty fast (no subject inside); 15DD's fast is full fast (everything inside, but no longer requiring step-by-step verification).

7.4 The DD Locus of Human Uniqueness: Not Self-Awareness, Not Will, But Other-Purpose Perception

This U-curve reveals a thesis more far-reaching than correcting Kahneman: where exactly does human evolutionary uniqueness lie?

Traditional narratives offer multiple answers: rationality (12DD), self-awareness (13DD), language (cross-layer), tool use (12DD), morality (14DD). But chimpanzees have rudimentary causal reasoning (12DD), pass the mirror test (13DD precursor), use and manufacture tools (12DD), and even show rudimentary norm-sensitivity under certain conditions (14DD edge). None of these are uniquely human—they are shared with other primates, differing only in degree.

The DD framework's answer: human uniqueness lies at 15DD—the capacity to perceive at intuitive speed that "the being across from me has its own must."

This is not a moral exhortation ("you should respect others") but a description of cognitive architecture. Humans can build language, institutions, culture, law, markets, and scientific communities not because we are smarter (12DD), not because we have self-awareness (13DD), not because we have will (14DD), but because we can sense another subject's attitude, intention, and situation within hundreds of milliseconds and respond accordingly—and this operation is fast enough to not look like "reasoning."

Theory of Mind research repeatedly finds that human judgments of others' intentions are too fast to resemble reasoning. The traditional explanation is "modularity"—the brain has a dedicated ToM module. The DD framework's explanation is more precise: it is not modularity but 15DD's directional flip moving self-verification from foreground to background, releasing full cognitive bandwidth to other-perception. Modules are merely the neural implementation of this structural transfer.

Chimpanzees, stuck at 14DD, must compute self-referential interest calculations from scratch for each cooperation ("is this good for me"), so they are slow. Humans, having crossed into 15DD, have self-verification locked as background conditions, with cognitive resources entirely released to "what is happening with the person across from me"—so they are fast. This speed differential is the cognitive foundation of humanity's explosive growth in cooperative capacity.

This is also where the SAE framework's name truly lands. Self-as-an-End—every self is an end. Not only "I" am a purpose; "you" are a purpose too. This is 15DD. This is what it means to be human. If this paper's four-layer architecture holds, this is not merely an ethical proposition but a structural hypothesis about human cognitive architecture: we may be a species capable of perceiving at intuitive speed that the being across from us is an independent purpose.

8. Non-Trivial Predictions

The following predictions are derived from the DD sequence, cannot be made by existing frameworks, and can be empirically falsified.

8.1 Prediction 1: Neural Dissociation Between "Being Robbed" and "Being Treated Unfairly"

Derivation: 12DD (pure loss response) and 15DD (fairness judgment) are operations at different levels, even if subjective reports in both cases are "anger."

Prediction: Under controlled identical material loss, three conditions should produce distinct neural signatures: (1) random loss—activating loss/threat circuits without fairness-judgment regions; (2) unfair distribution by another agent—activating loss circuits plus anterior insula, TPJ, dmPFC; (3) third-party unfairness—activating anterior insula, TPJ, dmPFC without self-related loss circuits.

Evidence status: Strong adjacent support but lacking the decisive direct comparison experiment.

8.2 Prediction 2: System 4 Capacity Correlates with Network Connectivity Rather Than Local Activation Intensity

Derivation: 15DD is an integrative operation; its neural signature should be network-level functional connectivity, not local activation peaks.

Prediction: In negotiation tasks, System 4 operation quality should correlate with anterior insula–DLPFC–TPJ functional connectivity strength, not with any single region's activation intensity.

Evidence status: The strongest-supported prediction. Independently, the SAE psychoanalysis fourth paper (Cert) predicted nearly the same: DMN, frontostriatal, and ToM network (TPJ/mPFC) must show simultaneously high coordinated activation. Two papers from different perspectives independently deriving the same neural prediction provides indirect support for the DD sequence's a priori validity.

8.3 Prediction 3: Stable Volitional Capacity Is a Necessary Precondition for Genuine Negotiation, Independent of Theory of Mind

Derivation: The DD sequence's non-skippability requires that 15DD cannot bypass 14DD.

Prediction: In experiments simultaneously measuring ToM, volitional stability, and negotiation quality, volitional stability should explain independent variance in negotiation quality that ToM cannot. Specifically, high-ToM but low-volitional-stability participants should tend toward concession in negotiation rather than producing C.

Evidence status: Strong indirect support but lacking direct testing.

8.4 Prediction 4: Suppression of Adolescent Rebellion → Delayed Adult Negotiation Capacity

Derivation: If 14DD is a necessary developmental precondition for 15DD, individuals whose 14DD did not fully develop should be impaired at 15DD.

Prediction: Adolescents raised in authoritarian families with suppressed rebellion should, in adulthood, tend toward regression to 14DD in negotiation tasks (insistence or capitulation) rather than producing C. Three specific degradation patterns are predicted: (1) rigidity (14DD residual), (2) compliance (regression to 13DD), (3) oscillation between rigidity and compliance (unstable 14DD).

Evidence status: Longitudinal studies support the plausibility of the developmental pathway but have not directly tested the "must → negotiation" chain.

8.5 Prediction 5: Fairness Judgment Speed Correlates with Training/Familiarity, Not Cognitive Level

Derivation: Speed is a function of training, not cognitive level.

Prediction: In cross-cultural ultimatum game experiments, response speed should vary with cultural familiarity with the specific fairness norm being violated, while response pattern (rejection of perceived unfairness) should remain relatively cross-culturally stable.

Evidence status: The weakest prediction. Cross-cultural behavioral data exist but lack systematic reaction time comparisons.

8.6 Prediction 6: Under High Cognitive Load, Higher DD-Level Operations Collapse More Easily Than Lower DD-Level Operations

Derivation: The four-layer structure is subsumptive. Even when trained to fast intuition, 15DD operations contain more subsystem coordination demands. Under load, higher levels should degrade preferentially.

Prediction: When performing fast 15DD judgments (e.g., fairness intuitions) under concurrent high cognitive load, fairness judgment quality should be more disrupted than 12DD pure loss responses. Specifically, under high load, participants should more easily regress from 15DD fairness judgment to 12DD pure loss aversion—ceasing to distinguish "unfair" from "I lost."

Evidence status: Partial indirect support. This prediction and Prediction 2 (13DD/14DD merging under load) are complementary applications of the same logic.

8.7 Prediction 7: True C, False C, and Honest No Deal — A Core Experimental Paradigm

Derivation: This is the most central non-trivial prediction, pointing to a fundamental blind spot in existing cognitive science experimental paradigms.

Existing fairness paradigms are almost entirely binary-choice structures: accept/reject (ultimatum game), invest/don't invest (trust game), cooperate/defect (prisoner's dilemma). No one is measuring "the capacity to create a third option." But the core operation of 15DD is precisely the production of C—a new plan that is neither A nor B.

The C-production process: Two "musts" collide. Both parties interrogate each other's "why must you," peeling apart means from truly non-negotiable items. I say I must have A. You ask why. I dig deeper and discover that A is actually to safeguard A'; A' is my truly non-negotiable item, and A was merely the means I assumed was the only way. Likewise, your B conceals B'. C is the new plan that simultaneously satisfies A' and B'.

Three outcomes:

(1) Compatible — true C. A' and B' do not conflict; C exists. The prior conflict was illusory—a collision at the means level. Interrogation strips away the false conflict; C emerges naturally. The majority of everyday conflicts belong to this category.

(2) Incompatible — honest no deal. A' and B' are fundamentally contradictory; C does not exist. Honest no deal is the only honest result.

(3) Incomplete — interrogation did not reach bottom. Time ran out; A' and B' were not yet uncovered. 15DD operation was initiated but not completed.

Predicted neural correspondences:

True C — synchronized strengthening of bilateral anterior insula–DLPFC–TPJ functional connectivity.

False C — decline in one party's DLPFC goal-maintenance activity (14DD exit), analogous to Knoch's TMS experiment pattern of "knowing it's unfair but not acting."

Honest no deal — high bilateral TPJ and dmPFC activation (thorough other-modeling) without integration-phase connectivity enhancement.

Why this is the most central prediction: It identifies the field's paradigmatic blind spot. From within Kahneman's framework, there is no reason to measure C-production—because with only two systems, negotiation is decomposed into "System 1 emotional impulse" plus "System 2 rational suppression." Only from the DD sequence does one know that 15DD requires a new experimental paradigm, and that the most important variable in that paradigm is not reaction time, not activation intensity, but the existence or non-existence of C.

9. Educational Application

The sequential emergence of the four cognitive layers in child development directly yields an educational principle: at each System's onset window, providing the right material maximizes impact.

13DD window (3-8 years): death education. Not avoiding children's fear of death, but facing it directly. "I will cease to exist" is the signature experience of 13DD, and accompanying anxiety is normal. Education's function is not to eliminate this anxiety but to help children build a sufficiently stable "I" to bear it.

14DD window (10-14 years): Kant. The adolescent in rebellion has just established 14DD—"I must"—and is at the period when will is most intense, pure, and painful. Giving them Schopenhauer or Nietzsche at this point deepens their entrenchment in 14DD. But giving them Kant's categorical imperative—"treat humanity never merely as a means"—has an entirely different effect. This does not negate their will but tells them: your "must" is legitimate, but the person across from you also has a "must." This is precisely the springboard from 14DD to 15DD.

15DD window (from age 14): negotiation training, not debate training. Debate is two 14DDs colliding—I try to convince you, you try to convince me. Negotiation is 15DD—why must I, why must you, is there a C. The core training is not rhetorical skill but interrogation capacity: can you interrogate the other's surface demand (B) down to their truly non-negotiable item (B'), while simultaneously interrogating your own (A) down to A'.

10. Conclusion

This paper began with a simple question: why is fairness-related cognition difficult to classify?

The popular formulation of Kahneman's framework would say: fairness responses are fast, so System 1. This paper's answer is: fairness-related cognition is a multi-layer cognitive family spanning 12DD through 15DD. The subset most representing System 4's uniqueness—intention-sensitive, third-party, dual-agent integrative fairness judgment—is not "fast emotional response" but a high-structural-complexity operation compressed through long training.

But this paper has reached a conclusion more far-reaching than correcting Kahneman. The speed U-curve of the four-layer architecture—12DD fast (no self), 13DD slow (self appears), 14DD slow (self-verification), 15DD fast again (directional flip, self-verification locked as background)—reveals the structural reason Kahneman mistakes 15DD for 12DD, and also reveals the DD locus of human uniqueness.

Human evolutionary uniqueness lies not in self-awareness (13DD—chimpanzees have precursors too), not in will (14DD—chimpanzees have goal-directed behavior too), but in 15DD—the capacity to perceive at intuitive speed that "the being across from me has its own must." Humans can build language, institutions, culture, law, markets, and scientific communities because we can sense another subject's attitude, intention, and situation within hundreds of milliseconds and respond accordingly. This speed comes from 15DD's directional flip: self-verification is locked as background conditions, and full cognitive bandwidth is released to other-perception.

This is where the SAE framework's name truly lands in cognitive science. Self-as-an-End—every self is an end. Not only "I" am a purpose; "you" are a purpose too. This is 15DD. This is what it means to be human. If this paper's four-layer architecture holds, this is not merely an ethical proposition but a structural description of human cognitive architecture.

The seven non-trivial predictions proposed in this paper, especially the "C-production" experimental paradigm, point to a fundamental blind spot in existing cognitive science. This blind spot exists not because of technical limitations or data insufficiency, but because of theoretical framework limitations—if your framework contains only two systems, you will not design experiments measuring C, because you do not know C should exist.

A priori leads, a posteriori verifies. The DD sequence is the a priori structure, the existing literature is the a posteriori data, and their convergence in this paper is not coincidence but methodological necessity. The next step is to let the a posteriori run ahead—design and execute the experiments proposed here, especially the "must-interrogation" paradigm, transforming the four-layer architecture from philosophical hypothesis into testable, falsifiable, improvable cognitive science theory.

Acknowledgments

Thanks to Zesi Chen for long-term critical engagement with the SAE framework. The four-layer structure in this paper took shape in dialogue with Claude (Anthropic), where the definition of System 4—"facing the subject across from you, asking why they think the way they do"—was a pivotal turning point. This working method is itself an instance of 15DD: a priori structure provided by the human, a posteriori verification accelerated by AI, and C is something neither could produce alone.

References

1. Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.

2. Fleming, S. M., et al. (2014). Anterior prefrontal cortex and metacognitive accuracy. Journal of Neuroscience.

3. Knoch, D., et al. (2006). Diminishing reciprocal fairness by disrupting the right prefrontal cortex. Science, 314, 829-832.

4. Corradi-Dell'Acqua, C., et al. (2013). Disentangling self- and fairness-related neural mechanisms in the ultimatum game. Social Cognitive and Affective Neuroscience, 8(4), 424-431.

5. Ginther, M. R., et al. Third-party punishment: separable neural stages for harm, mental state, integration, and decision.

6. Tomasello, M. (2014). A Natural History of Human Thinking. Harvard University Press.

7. Tomasello, M., & Carpenter, M. (2007). Shared intentionality. Developmental Science, 10(1), 121-125.

8. Melis, A. P., et al. Chimpanzees recruit collaborators selectively.

9. Hamann, K., et al. Differences between 2.5- and 3.5-year-olds in sustaining collaboration toward a joint goal.

10. Warneken, F., & Tomasello, M. (2006). Altruistic helping in human infants and young chimpanzees. Science.

11. Sanfey, A. G., et al. (2003). The neural basis of economic decision-making in the Ultimatum Game. Science, 300, 1755-1758.

12. Baron, J., & Gürçay, B. A meta-analysis of response time tests of the sequential two-systems model of moral judgment.

13. Bellucci, G., et al. Trust and reciprocity predicted by resting-state subnetworks.

14. Cáceda, R., et al. Reciprocity linked to connectivity between salience and central-executive networks.

15. Shaw, D. J., et al. Dynamic functional connectivity states dissociating cooperation from competition.

16. Sazhin, D., et al. Strategic bargaining involves connectivity changes involving TPJ and anterior insula.

17. Jäger, A., Loschelder, D. D., & Friese, M. Self-regulation interventions in negotiation.

18. Hare, A. L., et al. Maternal psychological control and adolescent autonomy.

19. Allen, J. P., et al. Autonomy struggles at 16 predict peer-rated hostility at 25.

20. Ha, T., et al. Disruptive parenting in early adolescence predicts adult romantic coercion.

21. Wittmann, M. K., et al. Causal role for dorsomedial PFC in separating self vs other representations.

22. Han Qin (2024). Systems, Emergence, and the Conditions of Personhood. DOI: 10.5281/zenodo.18528813.

23. Han Qin (2024). Internal Colonization and the Reconstruction of Subjecthood. DOI: 10.5281/zenodo.18666645.

24. Han Qin (2024). The Complete Self-as-an-End Framework. DOI: 10.5281/zenodo.18727327.

25. Han Qin (2026). SAE Psychoanalysis I: Id — The Me Without Self. DOI: 10.5281/zenodo.19321143.

26. Han Qin (2026). SAE Psychoanalysis II: Ego — The Self Without Purpose. DOI: 10.5281/zenodo.19321314.

27. Han Qin (2026). SAE Psychoanalysis III: Superego — The Self With Purpose. DOI: 10.5281/zenodo.19321417.

28. Han Qin (2026). SAE Psychoanalysis IV: Cert and Unification. DOI: 10.5281/zenodo.19321534.

29. De Neys, W. (2023). Advancing theorizing about fast-and-slow thinking. Behavioral and Brain Sciences, 46, e111.

30. Gräfenhain, M., et al. Young children's joint commitment in joint action. Child Development.

31. Han Qin (2026). What Terrible Twos Actually Is: The Structural Genesis of Self-Awareness. DOI: 10.5281/zenodo.19044827.

32. Han Qin (2026). What Terrible Teens Actually Is: The Structural Genesis of Purpose. DOI: 10.5281/zenodo.19201631.

33. Han Qin (2026). What Terrible Telos Actually Is: The Structural Genesis of Non Dubito. DOI: 10.5281/zenodo.19248039.