Functional Neuro-Anatomical Models of Human Attachment (NAMA & NAMDA)

Functional Neuro-Anatomical Models of Human Attachment:

1. General Considerations

Attachment theorists postulate the existence of an attachment behavioural system that orchestrates proximity seeking behaviour, particularly in times of distress and/or need.

In 2012, Prof Patrik Vuilleumier and Dr Pascal Vrticka proposed a first social neuroscience account of the attachment behavioural system by summarising the to date available experimental evidence and suggesting a “functional neuro-anatomical model of the influence of adult attachment style on social processing” (published Open Access in the Journal Frontiers in Human Neuroscience – see here)

After providing an updated version of the above-mentioned first model as part of a book chapter on “The Social Neuroscience of Attachment“ in 2017, a new review paper is now available in the journal Cortex. It provides an extended and refined “Functional Neuro-Anatomical Model of Human Attachment (NAMA)” – Section 2 below -, and explains its biological and neural components / modules by a newly formulated set of prototypical attachment pathways – Section 3 below.

More recently, a first account of an extension of NAMA to also include Disrupted and Disorganised Attachment (NAMDA) has been published Open Access in Frontiers in Psychiatry – Social Cognition (as part of Research “Topic Social Interaction in Neuropsychiatry”). This new account is summarised in Section 4 below.

2. Description of NAMA

Functonal neuro-anatomical model of human attachment (NAMA) – Figure 2 in Long et al., 2020. For more information, see here.

NAMA aims at providing a model for the putative biological and neural underpinnings of the attachment behavioural system in humans. To do so, NAMA differentiates between an affective evaluation (left) and a cognitive control (right) system that are in a dynamic balance (“push-pull”).

The affective evaluation system is thought to process incoming information rather quickly, automatically and sometimes even non-consciously (i.e., bottom-up). Conversely, the cognitive control system likely represents more voluntary and thus somewhat slower and conscious regulatory processes that can up- or downregulate affective evaluation mechanisms (i.e., top-down). However, it is important to note that such dissociation of emotional versus cognitive processing should not be understood to correspond to an old/lower versus new/higher brain anatomical organisation – for an excellent recent cautionary tale on why “Your Brain Is Not an Onion With a Tiny Reptile Inside” can be found here. Both emotional and cognitive (neural) processes are vital for and integral to attachment and caregiving, as also reflected in attachment internal working models (IWMs) – because IMWs can be described in terms of both cognitive schemas as well as associations learned by means of classical and operant conditioning (for more information, see here).

NAMA further differentiates the affective evaluation system into an aversion (red) and an approach (green) module (that also are in a dynamic balance; “push-pull”), and the cognitive control system into an emotion regulation (blue) and a mental state representation (orange) module. Finally, NAMA suggests a (non-exhaustive) list of involved brain circuits for each module & system, and a (non-exhaustive) list of involved neurotransmitters / -peptides. For an explanation of abbreviations, please see here.

Please note that there is no dedicated “attachment neural area” or “attachment neural network” within the human brain. Attachment draws upon neural computations in a vast network of brain regions that are involved in many different processes – also including attachment. Accordingly, the above-mentioned aversion, approach, emotion regulation and mental state representation modules should be regarded as roughly corresponding to extended brain networks that are also described as part of other social neuroscience accounts. The rough correspondences are as follows: aversion module = salience network for perceiving and responding to domain-general homeostatic demands (e.g., Seeley, 2019); approach module = reward circuit processing motivation and reward (e.g., Haber & Knutson, 2010); emotion regulation module = central executive network involved in attention allocation and cognitive regulation (e.g., Menon, 2011); mental state representation module = default mode network activated during mentalisation about the self and others – i.e., inward-focused processes (e.g., Alves et al., 2019).

The so far available data (mainly from adults and to a lesser degree from adolescents) suggest that the organised attachment orientations / classifications of insecure-avoidant and insecure-anxious (as compared to secure / autonomous) differentially associate with affective evaluation and cognitive control mechanisms. While both insecure attachment orientations appear to (at least partially) interfere with emotion (self-)regulation mechanisms as part of the cognitive control system (see also here), attachment avoidance and anxiety seem to have opposite effects on affective evaluation processes. Attachment avoidance may be mainly associated with a general blunting of affective evaluation processes, particularly in the case of positive social information (neurally represented in the approach module). In turn, attachment anxiety may be mainly linked to a general intensification of affective evaluation processes, especially in the case of negative social information (neurally represented in the aversion module). The above said, interesting additional and more complex interactions are starting to emerge. For a more comprehensive picture, please refer to Figure 3 in Long et al. (2020).

As NAMA is only concerned with the three organised attachment orientations of security, avoidance and anxiety, we recently also suggested a first attempt to extend NAMA to attachment disorganisation and trauma (“A Functional Neuro-Anatomical Model of Disorganised AttachmentNAMDA“). More information on NAMDA can be found in Section 4 below.

3. The Underlying Prototypical Attachment Pathways of NAMA

Prototypical attachment pathways – Figure 1 in Long et al., 2020. For more information, see here.

The two components / four modules of NAMA explained in Section 2 are conceptually derived from an initial prototypical attachment pathway – see the light orange path (A) in the Figure above.

The attachment behavioural system is thought to become activated most prominently and strongly by an (external or internal) event (E) that is appraised as threatening (and thus deemed salient / relevant for the organism at a given moment in time). Consequently, the event (E) triggers an appropriate fear response that entails a deviation from the bodily and mental balance or homeostasis. Threat detection and the fear response are likely biologically and neurally maintained by the aversion module of NAMA.

As a core element of attachment, the innate response to threat / fear is physical proximity seeking behaviour to establish closeness to a significant other. Such social approach behaviour is likely biologically and neurally maintained by the approach module of NAMA.

If proximity seeking is successful (and the source of threat / danger removed), the next step in the attachment behavioural system can occur: social co-regulation of emotion through social allostasis. Although initially mainly externally driven, this step becomes increasingly autonomous (i.e., emerging self-regulation) and is likely biologically and neurally maintained by the emotion (self-)regulation module of NAMA.

If co-regulation (and later increasingly self-regulation) is successful, the organism can return to its bodily and mental balance or homeostasis. Likely biologically and neurally encoded as a subjectively positive experience, this positive experience of the return to homeostasis is accompanied by a felt sense of security conveyed through the soothing and comforting social interaction as part of social co-regulation (or later on also imagined / remembered social support). Such socially rewarding experience is likely once more biologically and neurally encoded by the approach module of NAMA.

Finally, through repeated cycles of the above-described prototypical initial attachment pathway, predictions about the own capacity to elicit help and the availability and responsiveness of others in times of need start emerging. These are the fundamental building blocks of internal working models (IWMs) of attachment. IWMs are likely neurally maintained in part by the mental state representation module of NAMA – although also importantly mediated through safety learning principles that occur more automatically and non-consciously in the two affective evaluation (i.e., approach and aversion) modules.

As derivatives of the prototypical initial attachment pathway (A), NAMA comprises suggestions of alterations associated with attachment security (B), avoidance (C) and anxiety (D). These derivatives reflect the primary attachment strategy of (physical and psychological) proximity seeking in the case of security (B), the secondary attachment strategy of de-activation in the case of avoidance (C), and the secondary attachment strategy of hyper-activation in the case of anxiety (D).

More information is available here.

4. NAMDA: Extension to Disrupted and Disorganised Attachment

Figure 1 taken from White et al., 2020 (

As indicated above, NAMA is only concerned with the organised attachment orientations secure, insecure-avoidant and insecure-anxious. We therefore recently proposed an extension of NAMA to also include considerations of attachment disruption and disorganisation – NAMDA.

NAMDA proposes to conceptualise attachment disruption and disorganisation in terms of a principal distinction of caregiver behaviour in terms of abuse (i.e., threatening) versus neglect (i.e., insufficient availability). On a neurobiological level, this distinction is then suggested to manifest itself in the presence of hypo-arousal (i.e., approach-avoidance conflict and over-compliance) leading to harm-avoidance and rigid control in the case of abuse, or hypo-arousal (i.e., no overt / covert distress, passivity, and resignation) leading to rigidly independent self-regulation in the case of neglect.

Importantly, we acknowledge that attachment disorganisation should by no means be equated to maltreatment (abuse / threat and abandonment / neglect). Furthermore, several sub-categories of attachment disorganisation (e.g., classification of disorganised with secondary classification of secure, avoidant and / or anxious) may exist, making this a very inhomogeneous construct. Please also note that disrupted and disorganised attachment using the above denominations is conceptualised in terms of a representational model amalgamated from the history of caregiving experiences and not a singular or set of singular event(s). It is furthermore our hope that such conceptualisation may aide us in understanding the many (initially) adaptive behaviours children and adults show in the face of adversity. For more detailed information, please refer to the Attachment Disruption and Disorganisation section of this website.

In the remaining parts of the NAMDA paper, we selectively review the to date available evidence from direct social neuroscience research on disorganised attachment, combine the latter with indirect evidence from maltreatment-related research, and summarise the so far available evidence in relation to NAMA. This is still very much work in progress, because social neuroscience research on disrupted and disorganised attachment is sparse so that more data is needed for replication of extant patterns and the addition of new insights.

For more detailed considerations, please refer to the the NAMDA paper (now published in Frontiers in Psychiatry – Social Cognition as part of Research Topic Social Interaction in Neuropsychiatry).

More information on disrupted / disorganised attachment and particularly our own research efforts aimed at dissociating its contributing factors (especially abuse versus neglect) can be found here.

5. Additional Considerations

5.1 Different Neuro-Biological Accounts of Human Attachment

It should be mentioned here that there are other neurobiological accounts of human attachment available in the literature. A nice summary can be found in a book chapter by Gillath, Karantzas, & Fraley (2016): “What Can Neuroscience, Genetics, and Physiology Tell Us About Attachment?“.

In our view, one prominent account by Helen Fisher (see, for example, here) suggests that three principal human social behaviours can be dissociated by their underlying neurotransmitter / -peptide systems: (1) the sex drive / lust mediated by androgens, (2) attraction / mate choice / romantic love mediated by catecholamines, and (3) attachment (understood as a long-term non-sexual bond between individuals; child-parent, but also romantic partners, etc.) mediated by oxytocin.

A second prominent account by Ruth Feldman (for a recent review, see here) suggests a differentiation of human attachment behaviours by means of four principal mammalian bonds, namely: (1) parent-infant bonds, (2) pair-bonds, (3) peer bonds, and (4) bonds between conspecifics. Furthermore, Ruth Feldman proposes that the strength of these bonds can be categorically distinguished by their degree of bio-behavioural synchrony expressed within the bond – with parent-infant bonds having the strongest degree of bio-behavioural synchrony. Bio-behavioural refers to the temporal alignment of behavioural (e.g. touch, eye gaze), physiological (e.g. heart rate), endocrine (e.g. secretion of cortisol, oxytocin), and neural (e.g. brain activation) patterns. As suggested by Shir Atzil et al. (see here), bio-behavioural synchrony in association with external allostasis co-regulation is vital for social emotional development, as social interactions provide key wiring instructions that determine brain development.

In our above-described functional neuro-anatomical model of human attachment (NAMA) we propose, the considerations by Fisher, Feldman and Atzil are also considered. These considerations, however, are extended by the notion put forward by attachment theory that the very same mammalian bond / attachment  – mediated by the very same neurotransmitter / -peptide systems – can be subjected to strong interindividual differences depending on the quality of the bond and thus attachment. For example, parent-infant bonds may be generally characterised by high bio-behavioural synchrony and prominently mediated by oxytocin, but strong inter-individual differences in the bio-behavioural signature are to be expected depending on the quality of the bond – the latter being reflected in secure versus insecure, anxious and/or avoidant attachment orientations / classifications (not only on the individual but also the dyadic and family level). Inter-individual differences therefore are key in NAMA and we are particularly interested in describing their influence on the psychological, behavioural, biological, physiological, and neural basis of human attachment.