The caregiving style of parents influences the attachment style their offspring are likely to form. This project will produce a simulation that models this phenomenon. The simulation will include a proposed model of the causal relations that explain how attachment style is transferred from adult to infant. Starting from the information processing description of attachment phenomena currently used by Attachment Theorists, this work proposes architectural explanations of these phenomena at a greater level of precision and detail. The relevant individual differences between infants are explained by mechanisms involved with goals and sub-goal production. The individual differences between adults are explained by mechanisms involved in attentional filters. This thesis proposes that perturbances to meta-management may be caused by biases in attentional filtering. These perturbances are what give rise to the empirically observed individual differences in adult attachment behaviour. This research uses a variant of the design based methodology used in Autonomous Agent research in AI which may allow a deeper architectural design and a closer mapping between empirical data and simulation to be produced.