The standard model of the psyche is an onion: peel away enough defensive layers and you arrive at a single core self at the center. The metaphor predicts the therapeutic goal — keep peeling, keep getting more central, eventually arrive at the undefended original.
IFS says the psyche is structured like garlic. Peel away the outer skin and you find not a single core but a cluster of cloves — each one distinct, each one complete in itself, each one a subsystem with its own internal structure. And when you open a clove, you find that it too has its own layers, its own organization, its own nested interior.1
This is not a metaphor deployed for pedagogical convenience. It is Schwartz's description of what practitioners actually encounter when working deeply into an internal system: parts that have their own parts. Managers that contain their own Managers and Exiles. Exiles that have their own protective sub-parts that must be addressed before the exile itself can be approached. The nested structure goes down as far as the work goes.
The theoretical name Schwartz gives this structure is the fractal Self — a system that maintains the same organizational pattern at every level of scale you examine it.1 What you find inside a part looks like what you find inside the whole system. The three-group ecology (Managers, Firefighters, Exiles, organized by a Self) is not a feature of the top level of the psyche — it is the organizational logic of the psyche at every level.
The fractal-nested model makes two distinct structural claims that should be held separately before being synthesized.
Claim One: Parts have their own Self. Every sub-personality in the system carries its own access to Self energy — its own version of the calm, curious, connective capacity that Schwartz identifies as the undamaged core of every person. This means the work of connecting part to Self does not only happen at the systemic level; it happens within parts too. A Manager part may have access to its own Self and be able to use that connection to regulate its own sub-parts. The therapeutic resources available to the system are not only top-down (the main Self contacting the parts) but distributed throughout the structure.1
Claim Two: Parts have their own parts. The organizational structure of the whole — the ecology in which a vulnerable element (Exile) is surrounded by protective elements (Managers and Firefighters) organized by a central capacity (Self) — is reproduced inside individual parts. An exile may have its own protectors, parts within the exile that prevent the exile itself from fully expressing its wound. A Manager may have its own internal exile, a suppressed element within the Manager that the Manager is managing with the same logic it applies externally. The layering is real; practitioners who work to the outer edge of the system often find themselves opening into new systems within.1
This is not theoretical. Schwartz reports in No Bad Parts that he has worked clinically with subparts of subparts — with the internal structure of a part's internal structure, one additional level down from the standard IFS map.1
What this means in practice: after successfully unburdening a part and working through the part's relationship with its own sub-parts, those sub-parts may themselves contain sub-sub-parts that were not visible from the previous level. The work can continue downward. There is, in principle, no floor — the fractal structure is, from the clinical evidence, as deep as the work is capable of going.
The practical limit is not conceptual but clinical. Most clients do not require work at this level of depth. Most presenting concerns are addressed at the first or second level of the standard model. But for clients with the most complex trauma presentations — severe early developmental disruption, extensive dissociation, multi-generational family burden — practitioners may find themselves needing to navigate a system that has more internal architecture than the standard model's one-level-down view can map.
Schwartz's garlic/fractal metaphor is partly designed to correct a misreading that a simpler nested model would invite — the Russian doll (matryoshka) image.1
If the psyche were structured like nested Russian dolls, you would open the outer doll to find a smaller identical doll inside, and open that to find a smaller one still, and continue until you reached the smallest doll, which would be the final center — the irreducible self at the bottom of all the nesting.
The Russian doll model implies: (1) the "real self" is the smallest, most central layer, and (2) getting to it requires systematically opening and setting aside all the enclosing layers. This is a more sophisticated version of the onion model — still organized around the arrival at a single irreducible core.
IFS's fractal model diverges in both implications: (1) the Self is not found at the bottom of any nesting — it is present at every level simultaneously, distributed throughout the structure; and (2) the work is not about opening layers to reach a center but about meeting what is at each level and enabling it to access its own Self-connection. The center is not at the bottom. The center is everywhere, at every scale, already present but sometimes obscured.
The fractal logic extends beyond the individual psyche into the relationship between individuals and their social contexts — a claim that links the nested-Self model to IFS's broader social and cultural application.
The structure of the individual inner system (Self + parts in three-group ecology) is isomorphic with the structure of the family system (a functional Self-presence in the family + family parts in analogous roles). It is isomorphic with the structure of organizations and cultures. The same organizational pattern — a central capacity for curious, non-reactive leadership organizing a set of differentiated sub-systems, some of which carry unprocessed burdens — appears at every scale from the individual psyche to the nation.1
The isomorphism is not merely metaphorical. Schwartz's claim is structural: the same organizational dynamics (exile suppression, protector management, firefighter eruption, Self-leadership) operate at every level. Working with an individual's inner system produces changes that ripple into their family system. Changes in a family system ripple into the culture it produces. The individual is not a microcosm of the culture — they are the same system at a different scale.
This recursive structural claim is among the most ambitious IFS makes, and among the most difficult to verify. But it produces a specific therapeutic implication: healing the individual's relationship to their own parts will change how they interact with parts in other people. Self-led individuals create conditions in which other people's parts are less likely to need to be extreme. The therapeutic value of individual IFS work propagates outward — not through will or intention but through the structural effect of Self-energy in a system.
The fractal-nested model has direct clinical implications that change what practitioners do and how they understand impasses.
The protector within the exile. When a practitioner approaches an exile and finds something blocking access — a part within the exile that refuses to allow the exile's experience to be witnessed — this is not a failure to reach the exile. It is the structure doing what the structure does. There is a part inside the exile that is protecting the exile. That part needs to be addressed with the same care as the Managers and Firefighters in the outer system. The approach is the same: curiosity, no blending, genuine interest in what the inner-exile protector is afraid will happen if it steps aside.
Parts that seem like Self. One of the most clinically subtle features of deeply nested systems is that a part — usually a Manager — can mimic the qualities of Self so convincingly that neither the practitioner nor the client detects the substitution. This is sometimes called a Self-like Manager or "Self-lite." (See IFS: Self and Self-Leadership for the diagnostic criteria.) In deeply nested systems, these Self-like Managers may operate at multiple levels — with each level's Self-impersonator blocking access to the actual Self-energy distributed through the structure.
Clinical sequencing. Standard IFS work addresses the outer protectors before going to the exile. The nested model adds a second level of the same logic: when working with an exile's own internal sub-parts, address that level's protectors before going to that level's exile. The sequencing principle is fractal — it applies at whatever level you are working.
The depth of the work. Knowing that parts have their own parts prevents the practitioner from declaring work "done" too quickly. A significant unburdening at one level may relieve presenting symptoms while leaving deeper levels of the system intact. The client may improve substantially. But if the practitioner or client treats that improvement as a completion rather than a partial resolution, they may be surprised when new symptoms emerge from the next layer down — not as failure but as the natural next presentation of the fractal system unfolding.
The fractal-nested Self is primarily a 2021 elaboration.1 The 1995 textbook describes the three-group ecology and works with protectors and exiles at one level of the system, but does not explicitly develop the claim that parts have their own Self and their own parts. The structural logic of nested systems is implied but not made central.
No Bad Parts treats the fractal structure as one of the model's more important theoretical refinements. Schwartz introduces it with enough specific clinical detail — reporting that he has worked with subparts of subparts — to establish that it is not a theoretical extrapolation from the main model but an observation from clinical practice that the theory then accounts for.
The significance of the shift: the 1995 model could be read as describing the psyche as having one layer of internal structure (a surface ecology of parts organized by a Self). The 2021 model describes a structure with indefinite depth. This is a more complex model, but it is also a more satisfying account of why therapy on apparently resolved issues sometimes reveals new layers, and why the most complex presentations do not resolve through standard one-layer work.
The two texts converge entirely on the practical clinical approach — curiosity toward all parts at all levels, no part is bad, Self-leadership as the goal. They diverge on depth: the 2021 model requires practitioners to hold open the possibility of indefinite structural complexity, while the 1995 model's one-level-down view could lead practitioners to prematurely declare the system mapped.
Psychology — IFS: Parts as Innate: The fractal-nested model extends the innate-nature claim downward. If each part has its own Self and its own parts, then the structure that exists at the top level of the system — undamaged Self organizing a community of sub-personalities — is reproduced inside every element of that community. The innate-nature claim is not just that parts have natural functions before trauma distorts them; it is that the whole organizational template (Self + parts) is innate and constitutive of the system at every scale. Trauma doesn't damage the template — it produces extreme configurations at specific levels while leaving the template intact at others.
Psychology — IFS: Cultural and Societal Application: The fractal structure's extension across system levels — individual → family → culture — is the bridge between IFS as individual therapy and IFS as social theory. If the same organizational pattern (Self + Manager/Firefighter/Exile ecology) operates at every scale, then the individual work and the cultural/political work are not analogies of each other — they are the same work at different scales. The fractal model makes the social application logically necessary rather than metaphorically appealing: if the structure is genuinely isomorphic, working at one level has effects at the others.
Eastern Spirituality — Indra's Net: The Mahayana Buddhist image of Indra's Net — every jewel in the net reflecting every other jewel, which reflects every other, in infinite recursive depth — is a structural parallel to the fractal-Self model's claim that Self-energy is distributed throughout the system at every level, each part containing within itself the reflection of the whole. The convergence points to something neither framework states directly: the possibility that the recursive self-similarity of inner systems is not just a clinical finding or a metaphysical claim but reflects something about how organized wholes maintain coherence across scales. Where the frameworks diverge: IFS locates this pattern empirically in clinical observation; Indra's Net is a meditative revelation of interdependence. The gap between clinical report and meditative recognition is the live edge.
The Sharpest Implication
If the psyche is fractal all the way down — if every part has its own parts, and those parts have their own parts — then "getting to the bottom of it" is not a meaningful therapeutic goal. There is no bottom. The Russian doll model's promise (open enough layers, reach the inviolate center, be finished) turns out to be a comforting fiction. What replaces it is something more demanding and, eventually, more interesting: the work does not conclude, it deepens. The measure of progress is not whether you have resolved all the layers but whether the layers you can currently access are in better relationship. The Self is not at the bottom waiting to be found — it is distributed throughout a system that keeps revealing more of itself the more carefully you look. This is not discouraging news. It is the structural reason why people who do years of IFS work report that it keeps producing something new — not because the method changes but because the system has more architecture than any single course of therapy can fully map.
Generative Questions