Earned Security
Earned security was first identified by Mary Main and her colleagues in the development of the Adult Attachment Interview. The critical finding was that some adults who reported difficult or disrupted early caregiving nevertheless demonstrated the linguistic and cognitive markers of secure attachment: coherent narratives, balanced perspectives, the capacity to reflect on painful experiences without being overwhelmed. They had not been lucky in childhood — they had been lucky, or deliberate, later. Through therapy, through a transformative partnership, through some process of sustained reflective engagement with their own history, they had revised their internal working models. The past was still there, fully acknowledged — but it no longer controlled the present. For the AI transition, the finding is decisive: the professional whose fishbowl has cracked is not condemned to detachment. Revision is possible, but only under specific conditions that cannot be manufactured by policy alone.
In the AI Story
The finding emerged from Mary Main's Adult Attachment Interview research in the 1980s and 1990s. Subsequent work by Erik Hesse, Miriam Steele, Peter Fonagy, and others has documented the cognitive and relational conditions under which earned security develops. Three consistent findings: the process requires reflective function, it develops in specific kinds of relationships rather than in isolation, and it integrates rather than erases the original insecurity — producing a more complex and flexible model rather than a fresh start.
The conditions for earned security map directly onto what the AI transition requires and what most institutions are failing to provide. The disrupted professional cannot revise her working model alone; she needs a relationship (therapist, mentor, community, partner) that provides the attachment functions absent in her original formative experiences — or that have been destabilized by the disruption. The relationship must be consistent enough to disconfirm expected unreliability, responsive enough to disconfirm expected rejection, and patient enough to withstand the testing that an insecure model will impose on any new relational offer.
The Segal framework's orange pill moment — the crack in the fishbowl, the visibility of the previously invisible — is the activation event for earned security. Bowlby's clinical insight is that the activation is necessary but insufficient. Without relational support, the crack produces not earned security but disorganization: a state in which the person cannot integrate the new information and oscillates between incompatible coping strategies. The orange pill cannot be swallowed alone.
The research carries a warning. Earned security takes time — measured in years, not weeks. The insistence on speed that dominates contemporary change management is, in attachment terms, a form of institutional insensitivity that actively undermines the conditions for revision. The good-enough caregiver provides conditions and waits; she does not pull the child to her feet before the child is ready to stand. Organizations that want their workforces to achieve earned security during the AI transition must learn the same patience — a patience most have neither the temperament nor the incentive structures to sustain.
Origin
Mary Main, Erik Hesse, and colleagues at UC Berkeley identified and operationalized earned security through the Adult Attachment Interview in the 1980s and 1990s. The concept was consolidated in the research literature by Daniel Siegel's work on the mindsight framework and by the mentalization-based treatment work of Peter Fonagy and Anthony Bateman.
Subsequent longitudinal research — particularly the Minnesota Longitudinal Study of Parents and Children — has documented that earned security in adulthood is associated with parenting behaviors and child outcomes comparable to those of continuously secure adults.
Key Ideas
Revision is possible. Internal working models formed in childhood are not destiny; they can be revised through sustained relational experience.
Requires reflective function. The capacity to think about one's own mental states and those of others — mentalization — is both a prerequisite for and a product of the revision process.
Relational, not solitary. Earned security develops within relationships that provide the attachment functions absent or inadequate in the original caregiving relationship.
Integration, not erasure. The earned secure adult does not forget her history. She integrates it into a more flexible and realistic working model that includes the knowledge of what loss and disruption feel like.
Takes time. The process is measured in years of sustained relational experience, not in weeks of training — a timeline that institutional change-management programs systematically fail to respect.
Debates & Critiques
Researchers debate whether earned security is fully equivalent to continuous security or whether it retains vulnerabilities that emerge under extreme stress. The Berkeley research suggests strong equivalence on most measures; other work suggests that earned-secure adults may be more vulnerable to relapse under conditions that closely mirror their original attachment disruptions. For the AI context, the implication is that earned security achieved during one technological transition may require active maintenance to persist through subsequent transitions.
Appears in the Orange Pill Cycle
Further reading
- Mary Main et al., 'Adult Attachment Classification System' (unpublished manuscript, 2002)
- Daniel Siegel, The Developing Mind (Guilford, 1999; 3rd ed. 2020)
- Peter Fonagy and Anthony Bateman, Mentalization-Based Treatment (Oxford, 2006)
- L. Alan Sroufe et al., The Development of the Person (Guilford, 2005)