Dopamine in the medial amygdala network mediates human bonding

Abstract

Research in humans and nonhuman animals indicates that social affiliation, and particularly maternal bonding, depends on reward circuitry. Although numerous mechanistic studies in rodents demonstrated that maternal bonding depends on striatal dopamine transmission, the neurochemistry supporting maternal behavior in humans has not been described so far. In this study, we tested the role of central dopamine in human bonding. We applied a combined functional MRI-PET scanner to simultaneously probe mothers' dopamine responses to their infants and the connectivity between the nucleus accumbens (NAcc), the amygdala, and the medial prefrontal cortex (mPFC), which form an intrinsic network (referred to as the "medial amygdala network") that supports social functioning. We also measured the mothers' behavioral synchrony with their infants and plasma oxytocin. The results of this study suggest that synchronous maternal behavior is associated with increased dopamine responses to the mother's infant and stronger intrinsic connectivity within the medial amygdala network. Moreover, stronger network connectivity is associated with increased dopamine responses within the network and decreased plasma oxytocin. Together, these data indicate that dopamine is involved in human bonding. Compared with other mammals, humans have an unusually complex social life. The complexity of human bonding cannot be fully captured in nonhuman animal models, particularly in pathological bonding, such as that in autistic spectrum disorder or postpartum depression. Thus, investigations of the neurochemistry of social bonding in humans, for which this study provides initial evidence, are warranted.

Keywords: dopamine; humans; maternal behavior; network connectivity; social affiliation.

Fig. 1.

The medial amygdala network (in red) includes the medial sector of the amygdala along with a set of connected regions in the NAcc, rostral hypothalamus, vmPFC, PCC, and sgACC (25). The map is based on 150 healthy adults and reflects a whole-brain intrinsic connectivity analysis seeded in the right medial amygdala. Previous research shows that stronger intrinsic connectivity within this network is associated with larger and more complex social networks (25). The connectivity and dopamine response within the network were tested for association with maternal bonding behavior and plasma oxytocin levels. We hypothesized that dopamine secretion within the medial amygdala network recruits the network to support human bonding (green arrowheads). We predicted that synchronous mothering would be associated with an enhanced dopamine response within the network that specifically signifies the own-infant over the unfamiliar-infant condition.

Fig. 2.

Evidence that behavioral synchrony between mothers and infants is associated with maternal striatal dopamine responses to the infant. (A) A general linear model analysis demonstrates that high-synchrony mothers have higher relative dopamine responses than low-synchrony mothers in the own- vs. unfamiliar-infant comparison (indexed by percent [¹¹C]raclopride BPnd change). Regions of interest are presented according to their effect size (see the full list of regions in Table S1). Error bars represent the SEM. Mothers were assigned to a high- or low-synchrony group based on the vocalization synchrony median (2.36%). (B and C) Individual differences in vocalization synchrony in mothers are positively correlated with dopamine responses (indexed by the percent [¹¹C]raclopride BPnd change) in the right pallidum (one-tailed, n = 19, r = 0.517, P < 0.012) (B) and right NAcc (one-tailed, n = 19, r = 0.378, P < 0.055, trending) (C).

Fig. S1.

Individual [¹¹C]raclopride BPnd data showing that more high-synchrony mothers than low-synchrony mothers have a dopamine preference to their own infants. Lower [¹¹c]raclopride BPnd represents increased endogenous dopamine. High-synchrony mothers are depicted in green. Low-synchrony mothers are depicted in blue. (A) Right amygdala: Six of 10 mothers in the high-synchrony group have lower raclopride BPnd in the own-infant condition (indexing increased endogenous dopamine) compared with one of nine mothers in the low-synchrony group. (B) Right NAcc: Six of 10 mothers in the high-synchrony group have lower raclopride BPnd in the own-infant condition (indexing increased endogenous dopamine) compared with two of nine mothers in the low-synchrony group. (C) Right sgACC: Six of 10 mothers in the high-synchrony group have lower raclopride BPnd in the own-infant condition (indexing increased endogenous dopamine) compared with three of nine mothers in the low-synchrony group. (D) Right PCC: Six of 10 mothers in the high-synchrony group have lower raclopride BPnd in the own-infant condition (indexing increased endogenous dopamine) compared with one of nine mothers in the low-synchrony group. (E) Right pallidum: Six of 10 mothers in the high-synchrony group have lower raclopride BPnd in the own-infant condition (indexing increased endogenous dopamine) compared with zero of nine mothers in the low-synchrony group. (F) Right caudate: Six of 10 mothers in the high-synchrony group have lower raclopride BPnd in the own-infant condition (indexing increased endogenous dopamine) compared with two of nine mothers in the low-synchrony group.

Fig. S2.

Differences in synchrony by endogenous dopamine preference. Mothers whose right pallidum response showed an endogenous dopaminergic preference for their own infant (green bar, n = 6) had significantly higher vocalization synchrony scores than mothers whose right pallidum dopaminergic response was higher for the unfamiliar infant (red bar, n = 13) (F = 4.3, P < 0.05).

Fig. 3.

Mothers with a stronger medial amygdala network are more attuned to their infants (n = 19, r = 0.46, P < 0.02, one-tailed). Medial amygdala network connectivity is represented as Fisher’s r-to-z transformed Pearson correlation coefficients between the right medial amygdala seed and the rest of the network's nodes. Maternal attunement was measured as the percent of time during a 2-min interaction in which mothers provided positive vocal stimulation to their infants while the infants were content and socially engaged.

Fig. 4.

Stronger intrinsic connectivity in the medial amygdala network is predicted by increased in-network dopamine responses during the own-infant condition. (A–C, Upper) Intrinsic connectivity maps of the medial amygdala network (in red), overlaid with regions of interest for PET analysis [manually illustrated in green, according to FreeSurfer segmentation atlases (34)] in which [¹¹C]raclopride BPnd is correlated with the network connectivity. (A–C, Lower) The Pearson one-tailed correlation graphs (n = 19). (A) Right sgACC (r = 0.45, P < 0.03). (B) Right amygdala (r = 0.455, P < 0.02). (C) Right NAcc (r = 0.38, P < 0.05). In the x axes, an increase in dopamine responses during the own-infant condition is indexed by a decrease in [¹¹C]raclopride BPnd. In the y axis, medial amygdala network connectivity is represented as Fisher’s r-to-z transformed Pearson correlation coefficients between the right medial amygdala seed and the rest of the network's nodes.

Fig. 5.

Plasma oxytocin negatively predicts connectivity in the medial amygdala network (two-tailed, n = 17, r = −0.415, P < 0.049). In the x axis, medial amygdala network connectivity is represented as Fisher’s r-to-z transformed Pearson correlation coefficient values between the right medial amygdala seed and the rest of the network's nodes.