I. The Problem

For thirty years, Belk’s extended self framework has shaped how researchers and practitioners understand identity in consumer societies. The logic is intuitive: we are what we own. Our possessions, relationships, and experiences extend outward from a core self, forming layers of identity like rings in a tree. When social media arrived, the framework adapted seamlessly. Digital possessions—profiles, posts, avatars—became new categories of extension. The model held.

Except it didn’t.

The extended self model assumes a direction: from core to periphery, from being to having, from self to symbol. But something has shifted in the seven years since we first published our analysis of the digital self and customer loyalty (Bischoff, Berezan & Scardicchio, 2019). The evidence now points toward an inversion. Digital platforms don’t merely provide new surfaces for self-expression—they have become the primary site where self is constructed. The teenager doesn’t choose a brand to express identity; the algorithm surfaces the brand that constructs it. The professional doesn’t curate a LinkedIn profile to reflect their career; the profile’s engagement metrics reshape their career decisions. The direction has reversed, and every product, metric, and marketing strategy built on the extension model is calibrated to a reality that no longer exists.

This paper proposes a different model. Not the onion—concentric rings extending from a stable core—but the puzzle. A jigsaw where each piece’s shape is defined by the pieces surrounding it. Where the “self” at the center has no inherent form, only the form its connections create. Where removing or rearranging a single piece changes the shape of everything it touches.

II. The Self, Revisited

Belk’s (1988) original formulation was elegant: the self incorporates possessions through passion, authority, and personalization. A motorcyclist builds identity through the act of riding, the ownership of the bike, and the customization of the machine. The three mechanisms—wanting it, controlling it, making it yours—transform objects into extensions of who you are. Goffman (1959) had already established that we present ourselves through deliberate symbolic performance; Belk showed that the symbols we choose become us.

The framework translated to digital spaces more easily than anyone expected. Belk himself (2013) demonstrated that virtual possessions carry equivalent psychological weight: losing a digital inventory triggers grief; sharing a playlist constitutes generosity; customizing an avatar feels like self-expression. Our 2019 study confirmed this—passion, authority, and personalization operate identically whether the object is physical or digital (Bischoff et al., 2019).

But here is the instability that neither framework addressed: in digital environments, the three mechanisms can operate without the self initiating them. An algorithm can generate passion through addictive content loops. A platform can manufacture authority through gamified metrics. Personalization can be automated through recommendation engines that construct a taste profile the user then internalizes. When the mechanisms of self-extension operate independently of conscious choice, we are no longer describing extension. We are describing construction.

III. The Puzzle of Self

The visualization above encodes a structural claim: each concept’s meaning depends on its connections. “Need for autonomy” means something different when it connects to “looking glass self”—social comparison driving self-determination—than when it connects to “ideal self”—aspiration driving self-governance. The piece’s shape, its tabs and blanks, is literally defined by its neighbors.

Three tiers organize the puzzle: the core self (darkest), the extended self (medium), and the online self (lightest). But unlike the onion model, the boundary between tiers is porous. A piece from the outermost ring—say, “online narrative”—can reshape the meaning of the core. This is the fundamental difference: in extension, the core determines the periphery. In construction, influence flows in all directions.

Consider the chain from Purpose through Need for Autonomy to Looking Glass Self and finally Ideal Self. In the extension model, this reads left to right: a person’s purpose drives their autonomy, which shapes how they see themselves reflected in others, which produces an aspirational identity. Clean, causal, outward. But trace it in reverse and the construction model emerges: the ideal self—algorithmically curated through feeds of aspirational content—reshapes the looking glass, which redefines what autonomy means, which reconstructs purpose itself. The direction of causality is no longer fixed. It is emergent.

IV. The Motivation Engine, Compromised

Self-determination theory identifies three psychological needs—autonomy, competence, and relatedness—as the engines of human motivation (Ryan & Deci, 2000). Our 2019 paper demonstrated how social media satisfies these needs through different platform affordances (Bischoff et al., 2019). But the satisfaction has become compromised in ways that deepen the instability of the extension model.

Autonomy—the need to control one’s own life—is the most degraded. The defining feature of digital identity construction is that it feels autonomous while being architecturally constrained. Users “choose” to post, follow, and engage, but these choices occur within attention-optimized environments designed to maximize platform outcomes eg., engagement and time spent on platform, not user self-determination. The puzzle captures this: the autonomy piece connects to “looking glass self” and “ideal self,” suggesting that what feels like self-directed identity construction is often a response to perceived social judgment.

Competence—the need to deal effectively with one’s environment—has been gamified. Follower counts, likes, views, engagement metrics, and algorithmic visibility scores have replaced the organic development of mastery with quantified performance indicators. The competence need is technically satisfied, but by metrics that measure platform value, not personal growth.

Relatedness—the need for close, affectionate relationships—remains the most intact, yet paradoxically the most exploited. Dunbar’s research on network layers demonstrates that meaningful relationships require time-intensive social grooming (Dunbar, 2012). Digital platforms substitute fast, low-cost signaling—likes, shares, emoji reactions—for this grooming, creating an illusion of relational satisfaction while the quality of social bonds declines (Roberts & Dunbar, 2011).

V. The Network Paradox

Roberts and Dunbar (2011) established that personal networks follow a predictable hierarchy: approximately 5 intimate relationships, 15 close friends, 50 regular contacts, and 150 meaningful acquaintances. Social media has not altered these cognitive limits—but it has created an additional layer of 500–1,500 weak ties that users experience as part of their social world (Sutcliffe et al., 2012).

The paradox: expanding the network’s outer ring pressures users to increase social investment without providing the means to do so meaningfully. Virtual “grooming”—likes, pokes, shares—does not trigger the neurochemical responses that sustain real bonds (Lehmann & Dunbar, 2007). The result is a network that feels expansive but functions poorly: more connections, weaker bonds, and a growing gap between social capital and social satisfaction.

The puzzle model positions relationships at the intersection of multiple chains, reflecting this complexity. “Personal relationships” connects simultaneously to “community involvement,” “need for competence,” and ultimately “belonging”—but each connection operates under different conditions and at different Dunbar layers. Belonging, at the puzzle’s edge, depends on every piece between it and the core. Remove one—community involvement, for instance—and belonging doesn’t diminish gradually. It becomes structurally impossible.

VI. Global Architectures of Self

The construction model operates differently across cultures—a dimension largely absent from Western-centric identity frameworks. China’s guanxi, the system of personal connections, referral trust, and reciprocity that structures business and social relationships, represents a fundamentally relational approach to self-construction that predates digital platforms by centuries (Gold, Guthrie & Wank, 2002).

Platforms like WeChat do not merely facilitate guanxi; they formalize it. Payment systems, trust scores, and social commerce create a measurable infrastructure for relational identity. The self is quite literally constructed through transactional relationships—who you pay, who vouches for you, whose referral you accept (Peng & Luo, 2000). Western platforms monetize attention; Chinese platforms monetize trust. Both construct identity, but through different mechanisms.

This cross-cultural comparison strengthens the puzzle model: the same pieces exist in different cultures, but they connect differently. “Trust” in a guanxi context links to commerce and reciprocity. “Trust” in a Western social media context links to self-disclosure and authenticity performance. The piece is the same; the connections—and therefore the self—are different.

VII. Three Propositions for 2026

VIII. The Self That Builds Itself

The self has always been assembled from its connections. What has changed is who does the assembling. In the pre-digital era, individuals constructed identity through deliberate acts of possession, relationship, and expression. In the extension era, digital tools amplified this process. In the construction era—where we now find ourselves—the tools have become co-authors.

The puzzle model acknowledges this shift. Unlike the onion, which implies a fixed core radiating outward, the puzzle allows for an identity that is genuinely emergent—shaped by its pieces, transformed by their arrangement, and fundamentally incomplete without them. The tabs and blanks of each piece are not decorative. They are structural. They define what fits, what connects, and what the whole becomes.

We began, in 2018, by observing that social media had transformed the expression of self. We demonstrated, in 2019, that digital possessions function psychologically like physical ones. We propose now, in 2026, something more unsettling: that the distinction between tool and self has collapsed. The puzzle is not a metaphor for identity. It is identity—contingent, interlocking, and perpetually under construction.