In 2014, a neuroscientist named Gregory Berns trained a group of dogs to lie completely still inside a functional MRI scanner and measured what happened in their brains when they smelled their owners — and the reward centers of the dogs' brains lit up more brightly for their owners than for any other scent, including the scent of other dogs in their own households

In March 2014, three American researchers published a paper in Behavioural Processes titled “Scent of the Familiar: An fMRI Study of Canine Brain Responses to Familiar and Unfamiliar Human and Dog Odors.” The authors were the Emory University neuroscientist Gregory Berns, his postdoctoral colleague Andrew Brooks, and the professional dog trainer Mark Spivak of Atlanta’s Comprehensive Pet Therapy. Over the preceding two years, the Berns lab had trained twelve domestic dogs to lie completely motionless — unsedated, unrestrained, and voluntarily — inside a Siemens 3 Tesla Trio fMRI scanner for periods of up to thirty seconds at a time. During each scan, the research team presented each dog with gauze pads carrying five different biological scents: the dog’s own scent, an unfamiliar dog’s scent, a familiar dog from the same household, an unfamiliar human, and a familiar human from the same household. They measured which scent produced the largest activation in the caudate nucleus, the reward-processing centre of the mammalian brain. The answer was the familiar human.

The first substantial systematic investigation of the neural correlates of the dog-human bond was conducted by a neuroscientist whose original academic training had been in the comparative-neuroscience of financial decision-making rather than in canine cognition, and whose personal motivation for the specific research programme had originated in the specific 2011 death of his pug, Newton, at the age of fifteen. Gregory Berns, who at the specific 2011 date of Newton’s death was serving as the director of Emory University’s Center for Neuropolicy and had established a substantial reputation across the preceding decade as one of the founding figures of the field of neuroeconomics, encountered in the specific psychological aftermath of losing his dog the substantive question of whether Newton had, in essential respects, been experiencing something during their thirteen-year relationship that could reasonably be described in the same emotional vocabulary humans use to describe their own attachments to other humans. The comparative-cognition literature had, at that point, essentially no direct neural evidence with which to answer the question — every prior fMRI study of the brains of non-human animals had required substantial anaesthetisation of the research subjects, which produced fundamentally unusable data for research into questions about conscious emotional experience. Berns’s specific insight, developed jointly with Mark Spivak across the 2011-2012 period, was that dogs could, in principle, be trained through positive-reinforcement techniques to voluntarily lie still in the fMRI scanner in the awake state, permitting the specific neural imaging protocols that had transformed human cognitive neuroscience across the preceding two decades to be extended to the specific research questions the domestic dog raised about the neural basis of interspecies emotional attachment.

According to the specific peer-reviewed 2014 Berns-Brooks-Spivak paper indexed on the National Institutes of Health’s PubMed database, the experimental protocol required approximately 26 to 113 days of individual training per participating dog (an average of approximately 67 days between the original fMRI training session and the actual scent experiment). Each dog was progressively acclimatised to the specific auditory environment of the MRI scanner (a substantial engineering challenge, since the specific electromagnetic gradient switching of an operational fMRI scanner produces sound levels of approximately 110 decibels — comparable to a rock concert — that required the specific development of noise-cancelling ear protection for the dog subjects, marketed under the trade name “Mutt Muffs”), to the specific physical geometry of the scanner’s head coil (custom-fitted to canine cranial anatomy), and to the specific behavioural requirement that the dog voluntarily hold its head motionless within the coil for the specific 30-second intervals required for BOLD signal acquisition. The specific research design used cotton swabs rather than direct scent exposure to eliminate confounds from air-currents within the scanner bore. The specific “familiar human” scent samples were, importantly, not collected from the dog’s own handler — the specific research design deliberately used a different household member (typically the handler’s spouse) to isolate the specific effect of the biological scent from the confounding effects of the person’s physical presence.

What the caudate did

The specific neural response the twelve dogs produced across the substantial repeated experimental trials was, per Emory Health Sciences’ institutional summary of the specific 2014 Berns-Brooks-Spivak research findings, substantially unambiguous. All five scents produced approximately equivalent activation of the dog’s olfactory bulb and olfactory peduncle — the specific neural structures that process the initial reception of chemical scent signals from the nasal epithelium. This baseline olfactory response indicated that the dogs were, in fact, perceiving all five scents at approximately equivalent sensory magnitudes and that the specific differences the experiment eventually documented could not be attributed to differential sensitivity to the different scent sources. The specific caudate nucleus responses, however, were dramatically different. The scent of the familiar human produced a caudate activation that was substantially larger than the caudate activation produced by any of the other four scents — larger than the response to the dog’s own scent, larger than the response to the unfamiliar human, larger than the response to the unfamiliar dog, and, most substantively, larger than the response to the specific familiar dog that the subject animal actually lived with. The specific interpretive weight of that final comparison is worth pausing on. Dogs are, in essentially every non-neuroscientific measure of canine social behaviour, substantially oriented toward other dogs — they prefer canine playmates to human ones, they respond to canine communication signals more fluently than to human ones, and their basic social vocabulary was developed across approximately 40,000 years of wolf-and-early-dog pack behaviour rather than across the substantially more recent 15,000-40,000-year period of domestication. The specific caudate result the Berns lab documented in 2014 indicated that the specific reward-processing region of the domestic dog’s brain nevertheless assigns the highest positive-expectation weight to the specific scent of the individual human being with whom the dog lives.

Why the finding actually matters

The specific substantive interpretive question that the 2014 caudate finding raised — namely, whether the specific neural activation the twelve dogs produced could reasonably be described in the specific emotional vocabulary of human interpersonal attachment — has, in the substantial subsequent decade of Berns-lab research and comparative-cognition analysis, remained substantially unresolved. As reported in National Geographic’s contemporary coverage of the Berns lab’s 2014 canine olfaction research and its specific comparative-neuroscience implications, Berns himself has consistently declined to make the strong claim that the caudate activation his research documented constitutes direct neural evidence that dogs “love” their humans in the specific emotional sense that humans love other humans — the specific comparative-neuroscience framework requires substantial additional theoretical work to translate between the specific neural signals fMRI can measure and the specific subjective experiential states that emotional vocabulary describes. As acknowledged in the Companion Animal Psychology institutional review of the specific 2014 Berns findings and their interpretive limits, the specific alternative interpretation that Berns himself has publicly acknowledged is that the familiar human’s scent may simply have been associated, across the preceding several years of the dog’s life, with substantial repeated positive experiences (food, play, affection) — meaning that the caudate activation could represent a substantial conditioned response rather than a substantive emotional attachment. What Berns has consistently maintained, across the subsequent decade of his ongoing Emory University Dog Project research (including his 2013 New York Times bestseller “How Dogs Love Us” and his 2017 follow-up book “What It’s Like to Be a Dog”), is that the specific 2014 finding establishes, at minimum, that the substantial physical presence of humans in a dog’s life produces measurable, persistent, and specifically-directed neural signatures in the dog’s brain — that the specific individual human being with whom the dog lives is, in some substantively real neural sense, the substantially most important single element of the specific reward landscape that the dog’s brain is progressively organising its own experience around.