Oxytocin increases anxiety to unpredictable threat

Abstract

The nonapeptide oxytocin (OT), dubbed by the media as the ‘moral’ or ‘love’ molecule, has a variety of pro-social effects across species. A relatively simple explanation for these complex effects is that OT alleviates anxiety,1 thereby indirectly promoting trust, cooperation and other affiliative behaviors.2,3 Indeed, OT can reduce anxiety-like behavior in animals, via its neuromodulatory influence on the amygdala, a core hub mediating fear and anxiety.4 Surprisingly, OT also promotes territoriality, aggression and other defensive behaviors toward out-group members,2 complicating any straightforward interpretation of OT’s socio-emotional effects. Here, we show that OT does not reduce but rather increases defensive responding to unpredictable threat in humans, suggesting that OT enhances anxiety. A critical distinction has been recognized between two kinds of defensive states, sustained anxiety and phasic fear.5 Although the former is elicited by ambiguity and uncertainty (that is, unpredictable threat), the latter is evoked by explicit threat cues (that is, predictable threat).5 The hypothesis that OT differentially modulates defensive responses to unpredictable versus predictable threats directly follows evidence that OT released from the paraventricular nucleus of the hypothalamus facilitates excitatory signaling in the bed nucleus of the stria terminalis (BNST),7,8 a critical structure mediating defensive responses to unpredictable threats.5 In addition, OT receptor density varies by subdivision within the central nucleus of the rodent amygdala (CeA)4,6 and these subdivisions are differentially involved in defensive responses to predictable (fear) and unpredictable threats (anxiety). The medial CeA (mCeA) and lateral CeA (lCeA) are involved in fear and anxiety,5 respectively, but only the lCeA is rich in OT receptors.6 Finally, while in rodents endogenous OT reduces freezing to a fear conditioned cue,8 systemic OT administration does not affect fear-potentiated startle to a threat cue.9 Accordingly, we tested the possibility that OT may selectively increase defensive responding to unpredictable but not predictable threats using startle as a measure of fear and anxiety.5 Methods are fully described in Supplementary Information (see also Schmitz and Grillon10). We assayed startle reactivity, a well-established index of aversive state,5 in three conditions (see Figure 1) no shock (N), 2) predictable shocks (P) administered only during threat cues (red circles in the figure), and unpredictable shocks (U) administered randomly. Using a double-blind randomized within-subjects design, we administered placebo, OT, the structurally-related neuropeptide arginine vasopressin (AVP), and placebo (PLC) to 43 subjects in three sessions. Figure 1 (a) There were three conditions: (1) no shock (N), (2) predictable shocks (P) and unpredictable shocks (U). Acoustic startle stimuli were administered in the presence (thick colored arrow pointing up) and in the absence (that is, during inter-trial intervals ... Results are shown in the figure (see also Supplementary Information). Defensive responses to unpredictable shocks were significantly increased by OT compared with both PLC (F(1,42) = 5.1, P<0.03) and AVP (F(1,42) = 6.6, P<0.01), demonstrating selectivity in OT effects on anxiety. Defensive responses to predictable shocks were not affected by treatment (F(2,84) = 0.02, NS), suggesting that threat unpredictability is a critical factor underlying OT’s effects. This is the first evidence that OT can be anxiogenic in humans. The partial dissociation in the effect of OT on defensive responses to predictable and unpredictable shocks can be explained by evidence in rodents that OT facilitates excitatory signaling in a structure that underlies sustained anxiety-like behavior to uncertainty, the BNST,7,8 by directly activating corticotrophin releasing factor neurons. Several reasons might explain the failure to detect any AVP effect. The AVP dose may have been too low. Alternatively, OT effects may be mediated by the OT receptor as opposed to AVP receptors (for example, V1a and V1b). Finally, OT and AVP may affect differentially the brainstem pathways that mediate the various components of anxiety,4 suggesting that startle potentiation is selectively sensitive to OT but not AVP. Although the finding of increased anxiety following OT administration does not directly inform on OT’s pro-social effects, it may explain why OT affects responses to in-group (familiar) and out-group (unfamiliar) members differently.2,3 Anxiety is an adaptive defensive response to unpredictable threat. Our data suggest that OT sensitizes such responses in the face of ambiguity/uncertainty, which, when extended to the social domain, would include unfamiliar individuals whose behavior is inherently unpredictable. The hypothesis that OT enhances defensive responses in uncertain social contexts is consistent with findings that OT boosts anti-social behaviors towards unfamiliar individuals.2,3 Although in humans OT may strengthen existing social bonds (for example, between relatives/friends), perhaps reducing anxiety secondarily, it does not create bonds de novo (for example, between strangers), implying that it is less of a ‘love’ or ‘moral’ molecule than an ‘us and them’ molecule2 (see also Supplemental Information). Given the complexity of OT effects, it is too early to speculate about its therapeutic use for treating socio-emotional disorders (for example, autism and social anxiety) before we have a better understanding of the non-social effects of OT.