Dogs are known for their incredible sense of smell. They are also able to smell our stress. PhD student Clara Wilson recounts her amazing experience with these animals in The Conversation.
Dogs share a long history with our species, which gives them an amazing ability to decipher the signals we send them. They also have an incredible sense of smell, which allows them to detect, from their scent alone, certain diseases that affect human beings, such as Covid-19 where the lung cancer. The question of whether these abilities extend to the detection of odors associated with psychological states, on the other hand, has been much less explored.
My colleagues and I wanted to determine if dogs could distinguish, thanks to their sense of smell, odor samples taken from the same person, before and after they had suffered stress. It should be known that when we are stressed, hormonal changes and modifications of the nervous system occur and modify the odors produced by our body.
To determine whether dogs could actually detect such differences, we drew inspiration from protocols applied to biomedical detection dogs, sniffer dogs whose talents are employed in the laboratory. We combined them with techniques used to test how our canine companions perceive odors. Our results have been published in the journal PLOS One.
Table of Contents
How was the experience, human side?
We first equipped the (human) study participants with sensors to continuously measure their heart rate and blood pressure. We also asked them to rate their level of stress felt before and after taking part in the task that we were asking them to perform as part of this experiment. It was a question of carrying out a rapid mental calculation, supposed to induce a stress.
Before the task began, participants wiped a piece of gauze on their neck, placed it in a sterile glass vial, and then exhaled into the vial. After the task, participants provided two more sweat/breath samples.
The time separating the collection of the samples “in a relaxed state” (before the task) and “in a stressed state” (after the task) was four minutes. This short delay reduces the likelihood that changes related to events other than exercise stress will affect participants.
We included in the study only samples of participants who said they found the task stressful, and whose heart rate and blood pressure increased during exercise. In the end, we presented the dogs with the samples from 36 people.
The dogs included in this study were companion animals, whose participation had been proposed by their owners. These animals had been trained by researchers in a laboratory once a week, using positive reinforcement (which consists of associating an exercise with something that represents a reward for the animal, editor’s note).
Before data collection began, dogs were thus taught to signal that they were choosing a sample by standing and hovering over it for several seconds or by sitting in front of it – a attitude that we have called “alert behavior”.
The animals were then introduced to a matching game, through which they learned to distinguish between samples that smelled different. Once it was established that they had passed this game, they were ready to participate in the actual test.
During the latter, we asked the dogs to distinguish the samples taken from the participants before and after the aforementioned arithmetic task. To teach them what smell to look for during each testing session, we first showed them the sweat or breath sample from the stressed person, along with two “control” samples – pieces of clean gauze placed in sterile glass bottles, so without sweat or breath. The dogs were allowed to sniff out the three samples and were rewarded when they were able to signal the sweat/breath sample to the researchers.
After ten exposures, a second breath/sweat sample was added to the list: the relaxed sample from the same person. It was from this moment that the “discrimination” test began, which took place over the next 20 trials.
The “stress” scent was chosen by the dogs in most cases
During this phase, the dogs had to communicate, by their alert behavior, the sample which they perceived as identical to that which had been shown to them during the ten previous trials, i.e. the sample which had for them the smell of the stress sample. Controls have been put in place to verify that dogs do not rely on information other than that relating to the sample tested to help them in their choices (such as an odor that persists on the presentation devices samples, or a visual aid unconsciously provided by the experimenter).
In the event that the two smells presented seem similar to the sniffer dog, it can be expected that he will choose one or the other by chance. If the two odors seem distinct to him, he should on the other hand be able to systematically find the odor which was initially presented to him during their training (the “stress” odor). Each set of samples from participants was only used once, so dogs were presented with samples from a different participant during each session.
Result: from their first exposure to the “stress” samples, the dogs found that they had a specific smell. They correctly chose the stress sample in 94% of the 720 trials. The fact of having been subjected to a mental arithmetic exercise which stressed them therefore indeed caused a modification of the odors produced by the body of the participants.
It should be emphasized, however, that this study does not determine whether the dogs perceived the stress samples as reflecting a negative emotional state. It’s likely that in real life, dogs use various contextual cues, such as our body language, tone of voice, or breathing rate, to help them understand a situation.
These results, however, provide strong evidence that the smell of stress is also something that dogs can perceive. They provide insight into how dogs perceive and interact with human psychological states.
Beyond a better understanding of the relationship that unites us with our canine companions, this knowledge could also be useful for better training service dogs for people with anxiety and post-traumatic stress disorderwhich are currently trained to respond primarily to visual cues.