Internal Working Model

Transfers

Bowlby chose the term “working model” deliberately, borrowing from engineering and manufacturing where a working model is a functional prototype used to predict how a full-scale system will behave. The “working” is load-bearing: it means the model is dynamic, revisable, and defined by its function (prediction) rather than its fidelity.

Key structural parallels:

• Prediction before commitment — a factory builds a working model to test whether a production process will succeed before investing in full-scale machinery. Bowlby argued that infants build mental models of their caregivers to predict responses before investing the metabolically expensive effort of signaling distress. The child who has learned that crying produces comfort will cry; the child whose model predicts indifference will not. The metaphor frames attachment behavior as rational prediction, not instinct or habit.
• The model is built from operating experience — manufacturing models are refined through iterative testing on the factory floor, not designed from first principles. Similarly, internal working models are constructed from repeated interaction patterns: thousands of micro-episodes of bid, response, and repair (or non-repair). The metaphor captures why early experience is so influential — the model is calibrated during the period of most intensive data collection.
• “Working” means revisable in principle — a working model is not a final design. It is meant to be updated as conditions change. Bowlby intended this to be optimistic: pathological attachment patterns are not permanent because the model that generates them can be revised. Therapy, new relationships, and corrective emotional experience are model-revision processes.
• Multiple models for multiple domains — a factory may maintain different working models for different product lines. Bowlby proposed that individuals maintain multiple working models — of self, of specific others, of relationships in general — that may be inconsistent with each other. A person can have a model of “relationships are safe” alongside a contradictory model of “I am unworthy of care,” inherited from different relational histories.

Limits

• Working models are inspectable; internal models are not — the defining advantage of a factory model is that engineers can examine it, measure it, compare it against specifications. Internal working models resist introspection. People act from their models without being able to articulate what those models contain. This is not a minor limitation — it means the core therapeutic challenge (making the model visible so it can be revised) has no analogue in the source domain, where visibility is the model’s whole point.
• Manufacturing models are updated deliberately — when production conditions change, engineers revise the model. Internal working models update slowly, unevenly, and often not at all. A person whose childhood model predicts rejection may maintain that model through decades of contradictory evidence from loving relationships. The source domain’s assumption of rational updating is precisely what fails in the target domain, and the metaphor can obscure this failure by suggesting that “enough new data” should suffice.
• Promiscuous generalization — a factory model for automobile assembly is not applied to semiconductor fabrication. But internal working models generalize recklessly across domains: a model built from interactions with a neglectful parent gets projected onto romantic partners, employers, friends, and therapists. This over-generalization is the primary mechanism of attachment pathology, and it has no structural equivalent in manufacturing, where models are domain-specific tools.
• The metaphor individualizes a relational phenomenon — “internal” working model places the locus of attachment entirely inside one person’s head. But attachment patterns are co-constructed and maintained by both parties in a relationship. The manufacturing metaphor, with its single-engineer-and-model structure, cannot represent the dyadic, mutually regulating character of real attachment.

Expressions

• “Updating your working model” — therapeutic language for revising attachment expectations through corrective experience
• “His internal model of relationships” — clinical shorthand in attachment-informed psychotherapy
• “Mental models” — the term migrated from Bowlby through cognitive science (Johnson-Laird, 1983) into business strategy (Senge’s Fifth Discipline), losing its attachment-specific meaning
• “Running the model” — computational metaphor layered on top of the manufacturing one, treating the working model as a simulation engine
• “Schema” — the Piagetian and cognitive-therapy term for the same concept, often used interchangeably with “working model” in clinical practice, though schema theory and attachment theory have distinct theoretical lineages

Origin Story

Bowlby borrowed the term from Kenneth Craik’s The Nature of Explanation (1943), which proposed that organisms carry “small-scale models” of external reality that they use to predict events. Craik was himself drawing on the engineering practice of building functional prototypes. Bowlby’s innovation was to apply this to representations of relationships specifically — not just models of the physical world but models of how particular people will respond to one’s bids for proximity and care. The term entered developmental psychology through Bowlby’s Attachment and Loss trilogy (1969-1980) and was refined by Mary Main, who distinguished between models that are coherent and integrated versus models that are contradictory and dissociated — a distinction that maps imperfectly but suggestively onto the manufacturing source domain (a factory whose working models contradict each other will produce inconsistent output).

References

• Bowlby, J. Attachment and Loss, Vol. 1: Attachment (1969)
• Craik, K.J.W. The Nature of Explanation (1943)
• Main, M., Kaplan, N., and Cassidy, J. “Security in Infancy, Childhood, and Adulthood: A Move to the Level of Representation,” Monographs of the Society for Research in Child Development 50 (1985)
• Johnson-Laird, P.N. Mental Models (1983)
• Bretherton, I. and Munholland, K.A. “Internal Working Models in Attachment Relationships: Elaborating a Central Construct in Attachment Theory,” in Cassidy and Shaver (eds.) Handbook of Attachment (2008)