The Shape of Seeing the City

Topology of Mind explores the geometric structure of neural representations through high-dimensional data analysis. This project reveals how the brain encodes visual information by transforming neural population activity into interpretable geometric shapes.

1. Starting Point: A Tool That Should Eventually Exist

This project begins with a technological premise.

It imagines a system that generates architectural space in response to a specified emotional state. As affective computing, brain signal interpretation, and generative AI continue to advance, such a system is not a distant fantasy but an emerging technical possibility.

However, the project is not primarily concerned with building this tool. Instead, the tool is used as a lens.

If emotional and spatial mapping becomes automatable, what does this reveal about how our era understands emotion, knowledge, and design?

This project uses a speculative design artifact to ask epistemological and cultural questions rather than to solve a technical one.

2. What Does It Mean to Know?

Human knowledge has often been described through two complementary processes: deduction and induction. Deduction expands relationships within an existing logical structure. It strengthens internal consistency but does not introduce fundamentally new information. Induction, by contrast, extracts relational patterns from repeated experience. It filters noise and builds models of the world from observed regularities.

From this perspective, knowing is not the accumulation of isolated facts, but the understanding of relational structure. Objects do not carry meaning on their own; they gain meaning through their position within a network of relationships.

3. Knowledge in the Brain and in Artificial Intelligence

Both biological and artificial neural systems operate in this relational manner. They do not store "things" as fixed entities. Instead, they encode patterns of relationships among stimuli. A concept such as "sky" is not a static image but a configuration of color gradients, spatial orientation, boundaries, and contextual associations.

Through inductive learning, these systems construct relational structures. Deductive reasoning then operates on top of these structures, allowing inference in new situations. Human cognition and machine intelligence thus share a structural logic: learning builds the relational field; reasoning navigates it.

4. The Crisis of Comprehension

The difficulty arises because these relational structures exceed the scale of human intuition. Human perception evolved to interpret low-dimensional sensory environments. By contrast, neural data and AI representations exist in high-dimensional relational spaces. Their internal operations cannot be grasped through ordinary perceptual or intuitive means.

This project treats this gap as a design problem: how can complex relational systems be translated into forms that remain cognitively accessible?

Topological Data Analysis (TDA) is explored here as one such translation strategy. By transforming high-dimensional data into geometric and topological forms, TDA offers not merely visualization but an interface between machine-scale structure and human-scale perception.

5. A Shift in the Structure of Design

Design and art can be understood as practices that shape structures capable of eliciting emotional responses. Traditionally, designers develop this capacity through long cycles of experience, feedback, and refinement—an inductive learning process. In architecture, this process is especially slow because feedback emerges only after construction and occupation.

If emotional effects can be computationally modeled, this inductive training process is partially externalized. The designer's role begins to shift. Rather than discovering emotional effects through prolonged experience, the designer operates at the level of selecting, configuring, and interpreting relational structures.

Design thus moves from experiential calibration toward structural authorship.

6. Project Position

This project does not propose a finished solution. Instead, it builds a speculative system as a conceptual probe. Through this artifact, it asks:

• What becomes of "understanding" when emotional responses can be algorithmically anticipated?
• How do we maintain human interpretability within increasingly non-human representational systems?
• What remains the role of the designer when affective responses become computational variables?
• If design becomes a technology for optimizing emotion, what remains of art?

The project is therefore less about automation than about revealing the underlying structures that shape how humans, machines, and environments relate.

Next Steps

To explore the neural data and recording methodology that forms the foundation of this research, visit the Dataset section. To understand how this data is transformed into interpretable geometric structures, explore the Topology section where we detail the complete analytical pipeline.