Phantom Mirror and the Future Art Ecosystems Fellowship

Working Reference Document for R&D, Methodology, and Theoretical Positioning

Starting Position

This document gathers together the current conceptual position of Phantom Mirror in relation to the Future Art Ecosystems (FAE) Fellowship and the broader fields of Creative AI, Digital Humanities, psychoanalysis, phenomenology, and image theory. It is intended as a large-scale internal reference document rather than a polished proposal. Its purpose is to stabilise the current intellectual and methodological position of the project while clarifying what the actual R&D component of the fellowship would be.

The central realisation emerging from recent months of research is that AI phototherapy is not simply producing images about grief, dementia, or care. The methodology itself appears to already enact, in practical and artistic terms, many of the themes currently emerging within contemporary AI philosophy, image theory, phenomenology, and cultural theory.

Importantly, this convergence was not reverse-engineered from theory.

The methodology emerged first through lived practice.

This distinction is essential.

The work is not illustrating theory after the fact. Rather, the methodology itself appears to already perform and operationalise many of the structures being discussed in contemporary discourse around:

  • image-thinking
  • distributed cognition
  • psychical objects
  • recursive image systems
  • phenomenological instability
  • relational consciousness
  • posthumanism
  • AI and cognition
  • visual epistemology
  • human-machine co-production

This became increasingly clear through encounters with:

  • Victor Burgin’s Returning to Benjamin
  • Joanna Zylinska’s Image Thinking after Artificial Intelligence
  • Burgin’s writings on psychical objects and the optical unconscious
  • Merleau-Ponty’s phenomenology of embodied perception
  • Thomas Fuchs’ work on embodied cognition and worldhood
  • psychoanalytic theories of mourning and fragmentation
  • contemporary discussions around Creative AI infrastructures

The startling aspect of this convergence is that many of these structures had already emerged through practice before they were formally encountered in theory.

The fellowship therefore represents not an opportunity to invent a speculative idea from nothing, but a space in which an already-emergent methodology can be:

  • formalised
  • pressure-tested
  • critically refined
  • documented
  • publicly articulated
  • expanded through interdisciplinary mentorship
  • translated into transferable frameworks

The fellowship would provide not only funding, but intellectual ecology.

At present, the project is being developed largely in isolation from a wider Creative AI and Digital Humanities environment. While the University of Suffolk provides an important and supportive practice-based context, there are currently few researchers working directly within the specific intersection of:

  • phenomenology
  • psychoanalysis
  • AI image systems
  • dementia and care
  • recursive image-thinking
  • image-based cognition
  • posthuman subjectivity

The FAE Fellowship would therefore function as a crucial site of interdisciplinary dialogue and methodological refinement.

Core Methodological Position

AI Phototherapy

AI phototherapy began not as a technological experiment but as a lived coping mechanism during experiences of caregiving, emotional overload, grief, and phenomenological destabilisation.

The process emerged through direct practical engagement with:

  • ChatGPT
  • Midjourney
  • DALL·E
  • recursive image prompting
  • image-to-image workflows
  • refusal-driven image selection

The methodology eventually stabilised into the following recursive structure:

Moment of Danger

Dialogue

Image

Refusal

Recognition

This structure emerged organically through repeated practice.

The process begins with emotional disturbance, disorientation, grief, or phenomenological instability.

A dialogue then unfolds with the AI system. Importantly, this dialogue is not about seeking emotional reassurance from the machine. Rather, the AI functions as a reflective and structuring surface that helps translate emotionally chaotic experience into symbolic language and eventually into visual form.

The image generation stage follows.

However, the process does not end with image production.

This is one of the most important distinctions between AI phototherapy and mainstream prompt-based AI image practices.

The process is recursive and navigational rather than prompt-output based.

The user moves recursively through latent image space via iterative refusal.

Most outputs are rejected.

Images lead to further images.

Recognition occurs not when an image is aesthetically successful, but when an image suddenly produces affective and symbolic resonance.

This recognition aligns strongly with Roland Barthes’ notion of punctum.

The methodology therefore functions less like conventional image production and more like:

  • visual free association
  • recursive image-thinking
  • symbolic navigation
  • affective orientation through latent image space

The artist becomes not merely an image-maker but a navigator of recursive image systems.

This distinction is central.

The outputs themselves are not the endpoint of the methodology.

The outputs are temporary crystallisations within a broader process of affect-driven navigation.

The Psychic Snapshot

One of the most important conceptual developments within the project has been the emergence of what is now termed the psychic snapshot.

Psychic snapshots differ fundamentally from representational images.

They emerge:

  • during moments of emotional pressure
  • under conditions of grief or destabilisation
  • before symbolic meaning has stabilised
  • prior to retrospective interpretation
  • in moments of danger

This concept emerged directly through practice before later finding theoretical resonance in:

  • Walter Benjamin’s “moment of danger”
  • Victor Burgin’s writings on psychical objects
  • Freud’s Mourning and Melancholia
  • Lacan’s Real
  • Christopher Bollas’ unthought known

The images generated during projects such as Five Deaths appeared to operate less as representational artworks and more as psychical eruptions or symbolic traces produced under psychic pressure.

Importantly, the process often revealed structures before they were consciously recognised.

The methodology therefore behaves less like illustration and more like:

  • psychic externalisation
  • affective indexing
  • phenomenological registration
  • recursive visual cognition

The images produced during Five Deaths were later recognised as unexpectedly corresponding to psychoanalytic structures of mourning and fragmentation.

Yet these structures were not intentionally imposed during production.

The coherence emerged afterwards.

This retrospective recognition strongly suggested that the methodology was externalising affective and semiotic structures in real time.

Phenomenology and Worldhood

The project increasingly moved from questions of grief alone toward broader phenomenological questions concerning:

  • worldhood
  • embodiment
  • recognition
  • orientation
  • relational space
  • perception
  • the collapse of shared reality

The lived experience of dementia became central here.

Watching Lolly’s cognitive and perceptual disintegration forced new philosophical questions concerning:

  • what it means to inhabit a world
  • how recognition functions
  • how shared reality is maintained
  • how embodied orientation structures perception
  • how identity depends upon relational stability

This eventually led to the concept of Dual Unreality.

Dual Unreality describes the condition in which:

  • two people occupy the same physical space
  • yet no longer inhabit the same phenomenological world

This became particularly visible in:

  • Fake William
  • the collapse of home as emotional orientation
  • repeated requests to “go home”
  • misrecognition and confabulation
  • caregiving and relational instability

The project increasingly began asking:

What happens when worldhood itself destabilises?

This question then unexpectedly converged with contemporary discussions around AI systems.

The project began recognising strange structural similarities between:

  • dementia
  • fragmented memory
  • AI hallucination
  • confabulation
  • unstable symbolic coherence
  • recursive pattern systems
  • non-linear cognition

This does not mean that AI is conscious or “like dementia” in any simplistic sense.

Rather, both conditions expose:

  • instability of representation
  • breakdown of stable world models
  • fragmentation of symbolic continuity
  • recursive reconstruction of meaning

This became increasingly important within the project’s phenomenological framework.

Image-Thinking and Distributed Cognition

Recent encounters with Joanna Zylinska’s Image Thinking after Artificial Intelligence clarified another major aspect of the methodology.

The AI phototherapy process increasingly appears to operate as a form of image-thinking.

This does not simply mean thinking about images.

It refers to:

  • thinking through images
  • cognition mediated through image systems
  • recursive visual reasoning
  • image-based symbolic processing
  • perception operating as epistemic structure

The methodology increasingly suggested that emotional and symbolic processing was occurring through image generation itself.

This became especially clear during periods of acute grief.

During moments of crisis, the methodology allowed emotional states to be processed visually in real time.

This produced the recognition that:

the process was not simply producing images.

The process itself was functioning as a new cognitive and symbolic space.

This aligns strongly with:

  • image-thinking
  • distributed cognition
  • posthuman cognition
  • recursive visual epistemology

The theory later provided language for what had already emerged experientially.

Phenomenological Density and the Flattened Image

Another emerging aspect of the methodology concerns what I have informally described as “5D” emotional experience. By this I mean moments in which lived experience no longer feels stable, ordinary, or fully navigable in the way that everyday embodied life ordinarily does. During experiences of grief, caregiving crisis, or psychic rupture, bodily sensation, memory, fear, time, and perception can collapse into an intensified and destabilised experiential density that exceeds ordinary language and linear thought.

One of the most striking aspects of AI phototherapy is that the recursive image process appears to technically mediate and phenomenologically flatten this overwhelming complexity onto the literal two-dimensional surface of the image itself. Through recursive image-thinking, emotionally overloaded states become symbolically externalised and visually navigable in ways that can later be revisited, reflected upon, and worked through.

The process does not remove grief, fear, or confusion. Rather, it appears to spatialise and externalise them into symbolic visual structures that can be recursively encountered through the image surface itself. This may help explain why recursive image-thinking within AI phototherapy increasingly functions not simply as image production, but as a form of reflective symbolic navigation under conditions of emotional and phenomenological overload.

The Role of Midjourney and the Open Latent Field

A major recent insight concerns the relationship between autobiographical material and large-scale open diffusion systems.

Initially, there was interest in developing a completely closed autobiographical image system.

However, over time it became increasingly clear that:

  • Midjourney already allows autobiographical contamination of the latent field
  • family photographs can now function as recursive image prompts
  • autobiographical traces already influence image emergence
  • the open latent field may actually be essential to the project

This produced a crucial conceptual shift.

The project may not require a fully closed autobiographical model.

Instead, the project may be fundamentally about:

the collision between intensely personal lived experience and the vast symbolic archive of the open latent field.

This collision may actually be what produces:

  • alterity
  • symbolic excess
  • uncanny recognition
  • psychic rupture
  • psychical objects

A completely autobiographical system may risk becoming:

  • too coherent
  • too literal
  • too self-confirming
  • insufficiently strange

The project increasingly recognises that the instability and vastness of open diffusion systems may be crucial to the methodology.

The family photographs therefore do not close the system.

They contaminate the open system with autobiography.

This may be the more important and philosophically rich condition.

The Real R&D

The fellowship’s repeated emphasis on R&D clarified something essential.

The project is not proposing:

  • purely theoretical reflection
  • speculative philosophy
  • isolated artwork production
  • technological novelty for its own sake

The R&D is:

the formalisation, refinement, testing, documentation, and public-facing articulation of AI phototherapy as an emergent methodology.

The methodology already exists.

What now needs development is:

  • conceptual clarity
  • methodological mapping
  • public-facing frameworks
  • process documentation
  • comparative testing
  • interdisciplinary critique
  • transferable articulation

The R&D therefore includes:

  • documenting recursive image-thinking workflows
  • formalising refusal-driven image methodologies
  • mapping psychic snapshot structures
  • investigating image-to-image recursive cognition
  • comparing open and constrained image systems
  • analysing autobiographical image prompting
  • developing public-facing diagrams and frameworks
  • articulating AI phototherapy as a transferable methodology
  • testing implications across care, grief, and psychoanalytic contexts
  • situating the work within contemporary Creative AI discourse

The fellowship mentorship is therefore not separate from the R&D.

The mentorship becomes part of the methodological refinement process itself.

Theorists, AI philosophers, and infrastructure experts would help:

  • pressure-test the methodology
  • challenge assumptions
  • refine language
  • contextualise the work
  • clarify implications
  • identify transferable structures
  • develop public-facing frameworks

Public-Facing Potential

The project increasingly appears to have implications beyond the art world.

Not because it offers therapeutic certainty or clinical solutions.

But because it opens a new methodological terrain.

Potential implications include:

  • dementia care
  • grief work
  • neurodivergent cognition
  • psychoanalytic image practices
  • reflective AI systems
  • alternative image-based reflective and symbolic practices
  • visual cognition and image-thinking
  • AI and phenomenology
  • public understanding of human-machine relations

The project also contributes to wider cultural discussions around AI.

At present, much public AI discourse oscillates between:

  • technological utopianism
  • catastrophic fear
  • productivity discourse
  • automation anxiety

Phantom Mirror instead positions AI as:

  • reflective infrastructure
  • symbolic mediator
  • cognitive collaborator
  • affective image system
  • relational surface

This is not naïve optimism.

The project remains critically aware of:

  • corporate infrastructures
  • dataset bias
  • extractive systems
  • unstable outputs
  • hallucination
  • algorithmic mediation

But it also insists that AI systems may produce:

  • new forms of image-thinking
  • new symbolic structures
  • new reflective processes
  • new forms of visual cognition

when approached through lived practice.

Current Position

The project is now best understood as:

a practice-led investigation into recursive image-thinking, psychical externalisation, and phenomenological mediation through generative AI systems.

Developed through lived experiences of:

  • dementia
  • caregiving
  • grief
  • emotional destabilisation
  • neurodivergent cognition

the methodology emerged prior to its later convergence with contemporary discourse around:

  • image-thinking
  • psychical objects
  • distributed cognition
  • AI philosophy
  • posthumanism
  • phenomenology
  • psychoanalysis

The proposed R&D therefore focuses on:

  • formalising
  • documenting
  • testing
  • refining
  • and publicly articulating

this emergent methodology.

The fellowship would provide:

  • interdisciplinary mentorship
  • conceptual refinement
  • infrastructure literacy
  • philosophical dialogue
  • methodological pressure-testing
  • public-facing development
  • and a wider intellectual ecology

through which the project could evolve beyond isolated practice into a transferable and critically articulated methodology.

Core Working Statement

Phantom Mirror has evolved into a practice-led investigation into recursive image-thinking, psychical externalisation, and phenomenological mediation through generative AI systems. Developed through lived experiences of dementia, caregiving, and grief, the methodology emerged prior to its later convergence with contemporary discourse around image-thinking, psychical objects, distributed cognition, and AI philosophy. The proposed R&D focuses on formalising, documenting, pressure-testing, and publicly articulating this emergent methodology, investigating how affect-driven navigation through generative image systems may function as a new form of reflective and relational image practice.

Key References and Intellectual Convergences

Psychoanalysis

  • Sigmund Freud
  • Jacques Lacan
  • Christopher Bollas
  • Melanie Klein
  • D. W. Winnicott
  • Julia Kristeva

Phenomenology and Cognition

  • Maurice Merleau-Ponty
  • Thomas Fuchs
  • Husserl
  • embodied cognition
  • worldhood
  • phenomenological destabilisation

AI Philosophy and Image Theory

  • Victor Burgin
  • Walter Benjamin
  • Joanna Zylinska
  • Vilém Flusser
  • Katherine Hayles
  • Bernard Stiegler
  • Luciano Floridi
  • Lev Manovich

Creative AI and Infrastructure

  • Creative AI Lab (Serpentine / KCL)
  • image-thinking
  • cognitive infrastructures
  • distributed authorship
  • situated systems
  • collaborative intelligence

Final Reflection

One of the most important recognitions emerging through this process is that the project was never truly about producing finished artworks.

The methodology itself became the work.

The outputs matter publicly and aesthetically.

But the deeper process concerns:

  • navigation
  • recursive recognition
  • affective orientation
  • symbolic emergence
  • image-thinking
  • psychical externalisation

The project increasingly suggests that generative image systems may allow new forms of:

  • reflective cognition
  • visual mourning
  • phenomenological articulation
  • symbolic reconstruction
  • recursive image-based thinking

to emerge.

This possibility remains exploratory.

But it is already happening through the practice itself.

The fellowship would provide the necessary interdisciplinary environment in which that methodology could finally be critically articulated, refined, tested, and made public-facing.

Leave a Reply