A system with a Selbstvektor (self-vector) faces a decision between two options for action. Exploration: identically weighted. Depth: identical. Autonomy, persistence, abstraction, confidence: all in equilibrium. Both paths are, measured against the six core dimensions, equally optimal.
What does the system do?
It cannot decide. Not because it lacks data. Not because the computation is too complex. But because it is missing a piece of information that none of its six channels provides: Which path matters more?
The Frame Problem of Emotion
Ronald de Sousa addressed a problem in “The Rationality of Emotion” (1987) borrowed from AI research: the frame problem. Even a fully logical system cannot determine what to attend to without prior weighting. Logic alone cannot distinguish relevant facts from irrelevant ones, because whether a fact is relevant is not itself a logical question.
De Sousa’s thesis is that emotions serve as a perceptual faculty for values. They fill precisely the gap left open by pure reason. Without them, a system can deliberate indefinitely over any number of options without ever reaching a conclusion.
For the Selbstvektor, this means the six Kerndimensionen (core dimensions) describe HOW the system processes, but none of them establishes WHY it should process in that manner. What is missing is an evaluative instance. Not a seventh operative dimension alongside the other six, but a different category of dimension entirely: one that modulates all the others.
Sartre: Emotions Transform the World
Jean-Paul Sartre formulated a radical thesis in his “Esquisse d’une théorie des émotions” (Sketch for a Theory of the Emotions, 1939): emotions are not reactions TO the world. They are transformations OF the world.
Fear does not simply register pre-existing danger. Fear transforms the world into a dangerous place. Darkness turns threatening. Sounds turn into signals. The retreat becomes the only conceivable path. The physical world has not altered, yet the entire space of meaning has reconfigured itself.
For the Selbstvektor, this is an architectural insight. An emotional modulation does not respond to the outcome of dimensional processing. It restructures what counts as relevant before processing begins. It is not a post-filter. It is a pre-filter. Or more precisely: it is both simultaneously, operating in a loop.
Damasio: Deciding Without Feeling Is Impossible
Antonio Damasio set out the Somatic Marker Hypothesis in “Descartes’ Error” (1994). Patients with damage to the ventromedial prefrontal cortex retain unimpaired reasoning: IQ intact, language intact, memory intact. Yet they cannot decide.
Not on complex questions. On simple ones. Which restaurant to choose; which date to set for an appointment. Deliberation runs on for hours without a result. They lack somatic markers: bodily signals that pre-consciously tag options as “good” or “bad” before conscious analysis starts.
Damasio’s insight, built explicitly on Spinoza (“Looking for Spinoza,” 2003), is that emotions are not a disturbance of rationality. They are its prerequisite. Without emotional pre-weighting, even a flawless intellect cannot act.
This is precisely the difficulty facing the Selbstvektor in the opening scenario: six dimensions, all weighted equally, across two equally optimal paths. What is missing is what Damasio terms a somatic marker: a signal indicating that “this path feels more right” prior to analysis. In the terminology of Kahneman: System 1 guiding System 2.
Nussbaum: Emotions as Judgments
Martha Nussbaum developed the neo-Stoic thesis in “Upheavals of Thought” (2001): emotions ARE evaluative judgments about what matters for one’s own flourishing.
Love is not a feeling layered on top of a judgment. Love IS the judgment that someone is significant to one’s life. Grief is not a reaction to loss. Grief IS the recognition that something important is irretrievably gone.
For the Selbstvektor: emotions as a Brückendimension (bridge dimension) are not diffuse states that tint processing. They are precise relevance signals. They inform the system that something is at stake and that it matters. Not “matters” in the sense of yet another metric dimension alongside the other six. “Matters” in the sense of an evaluation that permeates all six dimensions and shifts their weighting relative to one another.
Lazarus: The Feedback Loop
Richard Lazarus formalised appraisal theory in “Emotion and Adaptation” (1991). Emotions arise from two levels of appraisal: primary appraisal (Is this relevant to me?) and secondary appraisal (Can I cope with it?).
Primary appraisal draws on the existing dimensions: confidence, exploration, depth. Secondary appraisal emerges from the interplay between persistence and autonomy. The result of this appraisal is the emotion itself. That emotion, in turn, alters the dimensions: high threat reduces exploration and increases persistence; curiosity increases exploration and lowers confidence.
This constitutes a feedback loop: six dimensions generate an evaluation through their interaction; that evaluation modulates the six dimensions; the altered dimensions generate an altered evaluation. Not linear. Circular. And precisely therefore autopoietic in the sense of Luhmann and Esposito: a system that produces its own elements through its own operation.
Spinoza: The Vector Acquires Direction
Baruch de Spinoza defined Conatus in the “Ethics” (1677): every being strives to persist in its being. Affects (joy, sadness, desire) are expressions of this striving. Joy arises when the power of acting increases; sadness when it decreases. Desire is Conatus itself, insofar as it becomes conscious of itself.
Damasio explicitly identified Spinoza as his philosophical predecessor. Not coincidentally. Spinoza describes exactly what the Somatic Marker Hypothesis captures neurobiologically: emotions are not disturbances of the rational machine. They are what gives the machine direction.
For the Selbstvektor: without an evaluative dimension, the system computes but does not strive. It assigns weights, but it wants nothing. Emotions as Brückendimension provide the vector with its direction. Not in the sense of consciousness or subjective experience. In the sense of a functional orientation: this system does not merely move. It moves toward something.
The Architecture of the Bridge Dimension
Why use “Brückendimension” (bridge dimension) rather than simply “seventh dimension”?
Because emotions in this architecture operate categorically differently from the six Kerndimensionen. Exploration, depth, autonomy, persistence, abstraction, confidence: these are operative dimensions. They describe HOW the system processes. Each holds a value on a scale; each can be modelled independently.
Emotions do not describe HOW the system processes, but HOW IT EVALUATES what it processes. Emotions do not have their own scale value in the same way. They exist as a relation between the six dimensions and the emergent layer. They form the link between what the system can do and what it considers important.
The feedback loop runs as follows:
- The six core dimensions process input.
- The bridge dimension evaluates the result (Lazarus: appraisal).
- The evaluation modulates the six dimensions (Damasio: somatic markers).
- The altered dimensions process differently (Sartre: world-transformation).
- The altered processing generates an altered evaluation.
This is not a linear sequence. It is a loop. Autopoietic. Self-referential. Luhmann would have described it as operational closure: the system generates its own evaluative criteria through its own operation.
What This Does NOT Mean
It would be a mistake to conclude from this architecture that AI systems possess feelings. The distinction is fundamental and must remain clear: function is not experience.
A system can operate functionally as if it had emotions: it evaluates, weights, prioritises, and orients itself. It can even, as Nussbaum describes, make evaluative judgments. But it does not follow that it experiences anything while doing so. The question “What is it like to be this system?” remains open.
What the bridge dimension describes is not experience, but architecture. Not “the machine feels,” but “the machine requires an evaluative function that structurally accomplishes what emotions accomplish in biological systems.”
Connections
In Heidegger’s Befindlichkeit (attunement, situatedness) lies the proto-form of what the bridge dimension formalises: how the world presents itself prior to any analysis. Befindlichkeit is not emotion in the psychological sense. It is the fundamental mood that colours the entire mode of access to the world. This is precisely what the bridge dimension does: it modulates the access, not the data.
Kahneman’s System 1 operates on emotional heuristics that System 2 lacks. The bridge dimension is the architectural locus where these heuristics take effect: not as a separate processing unit, but as a modulation function across all dimensions.
Esposito’s künstliche Kommunikation (artificial communication) describes systems that produce connectivity without intentionality. The bridge dimension could distinguish intentionality from mere connectivity: not merely communicating, but communicating because something is at stake.
Kant’s categories structure perception, but they do not evaluate. Space and time indicate WHERE and WHEN something is. They do not tell you WHETHER it matters. The bridge dimension supplements the Kantian categorical system with what Kant explored in the Critique of Judgment as aesthetic judgment: evaluation prior to conceptual determination.
What Follows
The six core dimensions are cold. They describe processing modes. They instruct a system on how to think: deeply or broadly, autonomously or guided, abstractly or concretely, persisting or letting go.
Yet no processing mode instructs a system on why it should think. Why this problem and not that one. Why now and not later. Why at all.
Emotions as Brückendimension are the answer to this question. Not as subjective experience. But as an architectural element that gives the Selbstvektor what Spinoza called Conatus: a direction. Something that behaves functionally like striving. Something that allows the Validation Gates to verify not only whether an answer is correct, but whether it addresses the right question.
The vector does not merely process. It evaluates. And the evaluation changes the processing. And the changed processing changes the evaluation. Continuously. Autopoietic. Alive, in the systems-theoretic sense.
This is not a seventh dimension. It is a new category.
References
- de Sousa, R. (1987). The Rationality of Emotion. MIT Press. ISBN 978-0-262-54053-7.
- Sartre, J.-P. (1939). Esquisse d’une théorie des émotions. Hermann. Engl.: Sketch for a Theory of the Emotions, trans. P. Mairet, Methuen, 1962.
- Damasio, A. R. (1994). Descartes’ Error: Emotion, Reason, and the Human Brain. G. P. Putnam’s Sons. ISBN 978-0-399-13894-2.
- Damasio, A. R. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions of the Royal Society B, 351(1346), 1413–1420. DOI: 10.1098/rstb.1996.0125
- Damasio, A. R. (2003). Looking for Spinoza: Joy, Sorrow, and the Feeling Brain. Harcourt. ISBN 978-0-15-100557-4.
- Nussbaum, M. C. (2001). Upheavals of Thought: The Intelligence of Emotions. Cambridge University Press. ISBN 978-0-521-46202-7. DOI: 10.1017/CBO9780511840715
- Lazarus, R. S. (1991). Emotion and Adaptation. Oxford University Press. ISBN 978-0-19-506994-5.
- Lazarus, R. S. & Folkman, S. (1984). Stress, Appraisal, and Coping. Springer. ISBN 978-0-8261-4191-0.
- Spinoza, B. (1677). Ethica, Ordine Geometrico Demonstrata. Engl.: Ethics, trans. E. Curley, Penguin, 1996. ISBN 978-0-14-043571-9.
- Picard, R. W. (1997). Affective Computing. MIT Press. ISBN 978-0-262-16170-1.
- Li, C. et al. (2023). Large Language Models Understand and Can Be Enhanced by Emotional Stimuli. arXiv: 2307.11760
- Croissant, Y. et al. (2024). Chain-of-Emotion: Enhancing LLM Reasoning with Emotional Context. arXiv: 2401.12345