This paper addresses the formation of infant attachment types within the context of active inference: a holistic account of action, perception and learning in the brain. We show how the organised forms of attachment (secure, avoidant and ambivalent) might arise in (Bayesian) infants. Specifically, we show that these distinct forms of attachment emerge from a minimisation of free energy—over interoceptive states relating to internal stress levels—when seeking proximity to caregivers who have a varying impact on these interoceptive states. In line with empirical findings in disrupted patterns of affective communication, we then demonstrate how exteroceptive cues (in the form of caregiver-mediated AMBIANCE affective communication errors, ACE) can result in disorganised forms of attachment in infants of caregivers who consistently increase stress when the infant seeks proximity, but can have an organising (towards ambivalence) effect in infants of inconsistent caregivers. In particular, we differentiate disorganised attachment from avoidance in terms of the high epistemic value of proximity seeking behaviours (resulting from the caregiver’s misleading exteroceptive cues) that preclude the emergence of coherent and organised behavioural policies. Our work, the first to formulate infant attachment in terms of active inference, makes a new testable prediction with regards to the types of affective communication errors that engender ambivalent attachment.

We build on a neuroanatomical model of how empathic states can motivate caregiving behavior, via empathy circuit-driven activation of regions in the hypothalamus and amygdala, which in turn stimulate a mesolimbic–ventral pallidum pathway, by integrating findings related to the perception of pain in self and others. On this basis, we propose a network to capture states of personal distress and (weak and strong forms of) empathic concern, which are particularly relevant for psychotherapists conducting attachment-based interventions. This model is then extended for the case of self-attachment therapy, in which conceptualized components of the self serve as both the source of and target for empathic resonance. In particular, we consider how states of empathic concern involving an other that is perceived as being closely related to the self might enhance the motivation for self-directed bonding (which in turn is proposed to lead the individual toward more compassionate states) in terms of medial prefrontal cortex–mediated activation of these caregiving pathways. We simulate our model computationally and discuss the interplay between the bonding and empathy protocols of the therapy.

We build on a previous neural model of bonding circuitry, in which the orbitofrontal cortex mediates between facilitative and stress reactivity to social stimuli, via the dorsomedial and paraventricular nucleus of the hypothalamus. We integrate recent neuroscientific findings, and consider how the introduction of additional reward could drive a further, counter-conditioning based re-balancing mechanism between activation of these networks, via increasing prefrontal-driven inhibition of the central nucleus of the amygdala. We simulate our model computationally, and hypothesise that such a process may be involved in a particular phase of self-bonding in the newly introduced self-attachment psychotherapy.

We propose a new conceptual multi-agent framework which, given a game with an undesirable Nash equilibrium, will almost surely generate a new Nash equilibrium at some predetermined, more desirable pure action profile. The agent(s) targeted for reinforcement learn independently according to a standard model-free algorithm, using internally-generated states corresponding to high-level preference rankings over outcomes. We focus in particular on the case in which the additional reward can be considered as resulting from an internal (re-)appraisal, such that the new equilibrium is stable independent of the continued application of the procedure.

We develop an arousal-based neural model of infant attachment using a deep learning architecture. We show how our model can differentiate between attachment classifications during strange situation-like separation and reunion episodes, in terms of both signalling behaviour and patterns of autonomic arousal, according to the sensitivity of previous interaction.