Emergence Lab

Emergence LabTheory

How new ideas arise

The theoretical foundations of the lab.

Emergence Lab is an applied research lab in computational creativity. Its work draws on three accounts of how new ideas arise — Margaret Boden’s three forms of creativity, Arthur Koestler’s bisociation, and Gilles Fauconnier and Mark Turner’s conceptual blending — and on one working hypothesis: that emergence can be studied, provoked and traced.

I·Computational creativity

Computational creativity (CC) is a scientific discipline at the intersection of artificial intelligence, cognitive science, and art. It studies how computational systems produce — alone or alongside humans — ideas, artworks, and solutions deemed creative.

Its best-known definition comes from Margaret Boden: creativity is the ability to come up with ideas or artefacts that are new, surprising and valuable. Boden also distinguished three ways of arriving at them.

But something has shifted. For thirty years, CC studied AI as an object: can a machine be creative? Today, AI is no longer just a subject of study — it has become a full-fledged co-creation partner. The real frontier is no longer machine creativity, but the creativity that emerges from encounter: how do humans and machines create together?

Margaret Boden

“Creativity is the ability to come up with ideas or artefacts that are new, surprising and valuable.”

— Margaret A. Boden, The Creative Mind, 2nd ed., 2004
  • Novelty
  • Surprise
  • Value

The three forms of creativity

Boden, 1990

  1. Type 01

    Combinational creativity

    Producing new associations between existing ideas. The most computationally accessible form — exploring new combinations in a known conceptual space. Metaphor, analogy, and collage are examples.

  2. Type 02

    Exploratory creativity

    Systematically exploring the boundaries of a defined conceptual space. The system creates no new rules — it pushes existing rules to their limits. Constraint-based generative music, or variations on a style, are typical examples.

  3. Type 03

    Transformational creativity

    Redefining the very rules of the conceptual space — the most radical form. The system generates a new way of thinking about a domain, making possible what was previously inconceivable. The rarest, most disputed, and most powerful of the three.

Reference Margaret A. Boden, The Creative Mind: Myths and Mechanisms (1990; 2nd ed. 2004).

II·Bisociation

In The Act of Creation (1964), the writer Arthur Koestler proposed that the creative act — in science, in art and in humour — is a bisociation: the meeting of two frames of reference that had previously been unrelated.

Ordinary association works within a single frame. Bisociation connects two, and the new idea comes from their meeting.

Collider Lab’s collision descends from bisociation: it specifies a distant destination and requires relational transfer from it.

III·Conceptual blending

In The Way We Think (2002), the linguist Gilles Fauconnier and the cognitive scientist Mark Turner describe conceptual blending: two mental spaces are partially matched and projected into a third, blended space. What they share is held in a generic space; in the blend, a new structure emerges that neither input contained.

They treat this operation as basic to everyday thought, not as a privilege of exceptional moments of invention.

Collider Lab’s approach descends from conceptual blending as well: in a collision, a mechanism used in another field is mapped role by role onto the target problem.

IV·Emergence

Something none of its parts contained.

The lab’s working hypothesis is that emergence — the appearance of something none of its parts contained — is not an accident but a process that can be studied, provoked and traced.

When two distant concepts collide in a controlled way, something new appears that existed in neither. The same holds for the encounter between a person and a machine: from it emerges what neither could have produced alone.

  1. 01

    Studied

    The lab measures what a collision changes in the ideas it produces, and publishes each result with its limits.

  2. 02

    Provoked

    Collision brings together two things that nothing connects, makes them collide, then selects among what emerges.

  3. 03

    Traced

    Every step stays traceable, from the starting concept to the idea that emerged.

V·The lab’s position

Precisely at this intersection.

That’s where — at the exact intersection of human and machine — Emergence Lab works. The lab harnesses the power of AI, combined with human intuition, to reinvent the interfaces between the two.

And it does so by using this partnership as its very method: a mise en abîme where the tool, the approach, and the object of study become one.

  1. V.1

    AI is a partner, not a tool

    The lab’s projects are not “AI-assisted”: they are co-designed with AI, a full partner to think with. Every project in the lab is born from this dialogue.

  2. V.2

    The machine steers, the human selects

    The machine steers the search towards a distant field; selection stays human. The claim is about populations of ideas, never about any single one: an automatic selector built to detect good ideas found none, so candidate ideas still require human judgement and external validation.

  3. V.3

    Glass-box traceability

    Most AI ideation is a black box: one prompt in, one idea out, and nothing in between. Here every collision leaves a complete record — the starting concept, where it breaks, what it collided with, the mechanism transferred, and the idea that emerged. An opaque system returns a verdict; a glass-box one returns a file you can argue with.

What is not yet established

So far, the evidence shows that collisions move AI-generated ideas away from the usual answers and repeat earlier ideas far less often. It does not yet show that this amounts to creativity. That question is still under test.

Traceable does not mean verified: nothing yet checks automatically that an idea actually resolves the difficulty it was aimed at. The measurement itself is still under test.

This remains a working theory: moving away from the usual answers is measurable, giving that movement a direction is testable, but neither is yet a synonym for creativity.