To be an expectant mother, or the anxious partner of one, is to be keenly, even agonizingly aware of how chemicals affect a developing life. The basic advice is well known, and obsessively followed: Alcohol in strict moderation, and no nicotine at all. Don’t mess with mercury. Folic acid is your friend. More protein and less caffeine. Stay away from BPA, PBCs and PFA, and generally make an enemy of the unpronounceable.

But, if we take the results of a provocative recent paper seriously, there may be another important, and deeply underappreciated chemical influence at work: a man’s odor. The research, by a team headed by Noam Sobel of the Weizmann Institute of Science, suggests that there is a relationship between women’s response to “social odors” contained in male sweat and the heartbreaking condition of unexplained repeated pregnancy loss (uRPL). Specifically, in blind smell-tests, these scientists observed that women who had experienced uRPL were significantly better at identifying their spouse’s odor than age-matched controls. Additionally, their brains responded differently to nonspouse odors and they displayed unique olfactory neuroanatomy. Taken in the context of a large body of literature on chemosignaling in nonhuman animals, these results make it conceivable that the human nose could also communicate with the womb and may even influence a pregnancy.

So far, the results are strictly correlative, and in no way point to male odor as some kind of pheromonal smoking gun that explains pregnancy loss. Hypothetically, it could also be true that women experiencing uRPL have, on average, larger middle toes, larger whites of their eyes, thinner wrists and a proclivity for wearing purple socks. None of these would give one pause or prompt a serious search for some kind of causal link to pregnancy loss. Yet this particular link between smell and pregnancy loss is intriguing because of how prevalent and robust it is in other mammals, including primates. Many miscarriages still have unexplained causes, which makes any lead, correlative or not, a particularly interesting and worthwhile area of research.

The phenomenon of pregnancy block was first documented in mice by Hilda Bruce, in the 1950s. After the first signs of pregnancy, Bruce housed female mice with either the original “stud male” (the father), or a “strange male” that had not been encountered before. All the females housed with stud males carried their pregnancies to term, but a striking 30 percent of pregnancies were blocked in females housed with the strange male. The factor producing the block was likely related to the hormonal state of the males, since the incidence of block decreased by about 50 percent once the strange males were castrated. In later studies that carefully controlled and isolated the potential ways that males could be influencing a pregnancy (presenting physical contact or damaging the brain’s olfactory centers) it was found that odor was both necessary and sufficient for what is now known as “the Bruce effect.” The most likely reason for this mysterious effect is that it’s an efficient, if merciless, mechanism for maximizing reproductive fitness: there’s no point investing in a resource-intensive pregnancy if the father is not around and a new male mouse is only going to kill the offspring.

Similar controlled experiments can’t, of course, be ethically replicated in humans, so instead Sobel’s group set out to investigate changes in olfactory identification of mates for women who had already experienced pregnancy loss. If this Bruce-type effect existed in humans, we might expect it to show up as a difference in the detection or recognition of mate odor that could persist after pregnancy loss. In the main experiment, the uRPL group was asked to identify their spouse’s soiled T-shirt by smell from a lineup including a stranger’s soiled T-shirt and a “blank” unsoiled shirt. The uRPL group performed significantly above chance in the task, and the experimenters noted, anecdotally, that many participants spontaneously offered remarks like “Oh, this is my spouse.” In contrast, an age-matched control group that had not experienced rUPL performed at only chance levels on this task.

Although this result could in principle be explained by a more acute sense of smell, without any mate-specific olfactory alchemy, additional experiments seemed to rule out this possibility. When tested with a battery of general odor detection and identification tasks, the rUPL women performed no differently from the controls, indicating that the group’s heightened olfactory identification was mate-specific.

If there is a link between smell and pregnancy loss, then the neural events that mediate this are of considerable interest. In mice, the brain’s sorting of “studs” from “strangers” involves the vomeronasal system, an olfactory subsystem that was once called “the social nose” but which is now seen as even richer and more complex. Scientists have proposed a wide range of mechanisms to account for the “Bruce effect” in mice, including changes in neurogenesis (the birth and incorporation of new neurons). But, the most likely theory is the formation of a template memory in the vomeronasal system. In the same way that your brain learns to “subtract out” your own smell (effectively recognizing the self as a nonthreatening or uninteresting stimulus), the pregnant mouse brain could do the same for a mate’s smell.

Naturally, the neural mechanisms of mate identification cannot be studied in humans with anything close to the control, precision and granularity that are possible in mice. Still, Sobel’s group has made some interesting observations about social odor processing using fMRI that may hint at a neurological basis for the observed phenomenon. At a gross structural level, uRPL subjects tended to have smaller olfactory bulbs than their control counterparts (the olfactory bulb is an early and critical brain structure that receives direct sensory inputs from the nose). In another experiment designed to monitor real-time neural responses to social odors, subjects were placed in a scanner while being shown arousing videos and smelling nonspousal odors presented at concentrations well below conscious detection. Interestingly, the scientists didn’t just observe a difference in the amount of stimulus-evoked brain activity between the two groups. Rather, they found a difference in the direction of the groups’ responses, with activity in the hypothalamus—a key regulator of basic physiological functions including sexual behavior and menstruation—decreasing in the control group and increasing in the uRPL group.

Sobel’s studies show that something is noticeably different about social odor perception in women with repeated miscarriages, and that these differences influence how their brains process social odors. While this could potentially hint at a “Bruce-like effect” in humans, there are still many questions left unanswered. The main one is that because of the experimental design (driven by the ethical constraints of working with humans) we don’t know in what direction the arrow of causality points. It could be that pregnancy loss has nothing to do with smell, and caused downstream changes in olfactory processing, in a sort of “backwards” Bruce effect, for example. And, even if odor does prove to be causal, it may also turn out that the effect owes more to the uniqueness of certain male odors than a difference in certain female brains. Regardless of the end result, we are sure to learn something new and important about how we communicate using our most underappreciated sense.