Generative AI (Computational Archive) │ │ │ Psychoanalysis ─────── Dementia (Unconscious Processes) (Embodied Disintegration) │ │ Phantom Mirror (AI Phototherapy Method) │ │ Psychic Snapshots (Symbolic Images from Lived Experience)
The conceptual structure of Phantom Mirror emerges at the intersection of three domains: lived experience of dementia, psychoanalytic theory, and generative artificial intelligence. Dementia destabilises embodied perception and recognition, exposing the fragility of the structures through which the world normally appears coherent. Psychoanalysis provides a framework for understanding how experiences that resist immediate articulation can surface symbolically through images and affect. Generative AI introduces a third element: a vast computational archive capable of producing images through statistical pattern recognition without possessing lived experience or worldhood. The practice-led methodology developed in this research, termed AI phototherapy, operates at the centre of this triangle. Through dialogic prompting and iterative image selection, lived emotional events are translated into symbolic images that I describe as psychic snapshots. These images capture moments where experience becomes visible before it can be fully articulated in language
Chapter One
Phantom Mirror: Origins, Methodology, and Research Context
1. Introduction: Lived Experience and the Origins of the Project
This research began not as a theoretical investigation but as a practical response to lived experience. The onset of my stepmother’s dementia introduced a series of destabilising encounters within everyday family life. Moments of confusion, misrecognition, and emotional distress created an urgent need to find a symbolic way of processing what was unfolding.
My previous artistic practice had already explored themes of grief and memory through phototherapy and reenactment photography, drawing on the feminist photographic methodologies developed by Jo Spence and Rosy Martin. These practices use staged imagery to externalise internal emotional states and to revisit unresolved personal histories.
When dementia entered my family life, I began experimenting with generative AI image systems as a way of extending this process. Initially this experimentation was informal. It was not conceived as research but simply as a method for translating emotional experience into visual form.
One of the earliest images emerged following a visit in which my stepmother became distressed because she could not find the correct bed sheets. The image generated from this encounter was unexpectedly powerful. It condensed a complex emotional situation into a symbolic visual scene.
This experience was surprising. It suggested that generative AI might function as a reflective tool capable of externalising lived experience in ways that traditional photography could not.
From this moment onward, the practice developed into what I now call AI phototherapy.
2. Development of the Method: AI Phototherapy
AI phototherapy is a process in which lived emotional experiences are translated into symbolic images through dialogue with generative AI systems.
The method typically unfolds in several stages:
A moment of emotional intensity or confusion occurs within lived experience. The event is described in conversation with an AI language model. The resulting prompts are used to generate images through diffusion models. Multiple iterations are produced and reviewed. Images that resonate emotionally are selected.
This process often occurs rapidly, sometimes within minutes of the initial event. The images therefore emerge very close to the emotional moment that triggered them.
Over time it became clear that these images were not functioning as simple illustrations of events. Instead they appeared to capture what I later began calling psychic snapshots.
These snapshots seemed to register emotional states that had not yet been fully articulated in language.
The discovery that such images could emerge through AI interaction became the central methodological question of this research.
3. Emerging Research Questions
The development of AI phototherapy gradually led to several key questions:
How can AI-generated imagery externalise emotional experience? Why do certain images resonate with such unexpected symbolic force? What is the relationship between lived experience and machine-generated imagery? What does dementia reveal about perception, recognition, and identity? What does the encounter between human experience and generative systems reveal about the nature of subjectivity?
These questions situate the project within a broader inquiry into human perception, memory, and technological mediation.
4. Dementia and the Question of Recognition
Although dementia was not chosen as the subject of the project, it became a central lens through which these questions emerged.
Dementia destabilises fundamental aspects of human experience:
memory recognition spatial orientation identity sense of home
Moments in which my stepmother failed to recognise familiar people or places forced me to confront the fragility of what philosophers call worldhood.
These encounters led me toward phenomenology, particularly the work of Maurice Merleau-Ponty, who argues that perception is fundamentally embodied.
Merleau-Ponty writes:
“The body is our general medium for having a world.”
From this perspective, dementia can be understood not simply as memory loss but as a disruption of the embodied structures that allow the world to appear coherent.
5. Theoretical Framework
As the project developed, several theoretical traditions proved helpful in articulating what the practice was doing. These theories were not imposed on the work in advance. Rather, they helped explain patterns that had already emerged within the image-making process.
Key thinkers informing this research include:
Walter Benjamin Sigmund Freud Jacques Lacan Christopher Bollas Donald Winnicott Martin Heidegger Maurice Merleau-Ponty Roland Barthes Jorge Luis Borges
These thinkers provide philosophical, psychoanalytic, and phenomenological frameworks through which the images and processes within Phantom Mirror can be understood.
6. Structure of the Thesis
The remainder of the thesis develops these ideas across several chapters:
Chapter 2 – AI Phototherapy as Method
Chapter 3 – Dementia, Perception, and Dual Unreality
Chapter 4 – Psychic Snapshots and the Structure of Mourning
Chapter 5 – AI, Archive, and the Aleph
Chapter 6 – Relational Consciousness and the Human–Machine Encounter
Working Concept Map for the Thesis
(Reference list of key concepts discussed in this project)
Core Methodological Concepts
AI phototherapy
Dialogic prompting
Image iteration
Refusal and selection
Psychic snapshot
Embodied prompting
Generative symbolic translation
Phenomenology and Perception
Embodiment (Merleau-Ponty)
Worldhood
Recognition
Misrecognition
Sense of home
Dual unreality
Embodied orientation
Psychoanalytic Framework
Freud – Mourning and Melancholia
Lacan – The Real
Bollas – The Unthought Known
Winnicott – Holding Environment
Winnicott – Transitional Space
Transformational Object
Symbolic condensation
Psychic rupture
Philosophy of Technology
Heidegger – Enframing (Gestell)
Heidegger – The Clearing
Heidegger – Dwelling
Poetic revelation
Technological standing-reserve
AI and Archive
Generative datasets
Hallucination
Pattern cognition
Machine perception vs human perception
Borges – The Aleph
Infinite archive
Prompt as extraction
Integrative Concepts Emerging from the Project
Relational consciousness
Human–machine reflection
AI as phantom mirror
Symbolic externalisation of emotion
Practice-led knowledge production
(This document functions as a working reference for Chapter One and as a conceptual map for the wider thesis.)