The results of the study led by Barbara Ferry of the Centre of Research in Neuroscience in Lyon were recently published in PLoS One. “Our collaboration with the Division of the Technical and Scientific Police in Ecully started in 2010, when Sophie Marchal, a chemical engineer and first author of our publication, started working there. First, we had to enter all the data, which had been collected by the Division since 2003 and had been written down on piles of paper, in digital spreadsheets. Then we had to find the best way statistically to evaluate them. This took a lot of time,” Ferry observed.
By analysing data obtained with twelve German and Belgian Shepherd dogs, the researchers found that the animals could be trained accurately to recognise the smell of an individual in 84 out of 100 cases. The dogs did not make any false attributions. They matched samples which were collected in the same way - either 'directly' from the hands or 'indirectly' from objects and clothes - with higher sensitivity than samples which were collected in different ways. In France, suspects are not obliged to allow the collection of scent samples. However, if they refuse, they are condemned to a prison sentence. This enables a police officer to collect any scent traces from their clothes and belongings. In court case identifications with several detection dogs, successful identifications were more frequent with fresh scent traces, not older than 24 hours, than with older samples. Positive identifications made by the dogs prompted confessions and helped to solve 120 out of 435 cases.
The unique scent of a human being, which is to be detected by the dogs, stems from a mixture of volatile compounds, produced by the skin with its millions of secretory glands. Human axillary sweat, for example, contains several C6-C10 straight chains, branched and unsaturated acids. On the fingers alone, over 300 compounds were found. When we move, we leave behind shed skin with our sweat and bacteria attached. Canine ability to distinguish human scent was documented as early as 1887, when George J. Romanes described that his setter bitch could track him by the scent of his boots and by the scent emanating from his head - even when she could not see him and his scent was masked by aniseed oil or the presence of the scents of many other people. Interestingly, compared to his boots, the scent of his bare feet or stockings presented the dog less clear leads.
Olfaction begins with sniffing. When air enters a dog's nose, the airstream is divided. Whereas most of it reaches the lungs, a proportion flows directly into the olfactory system to be analysed. With their vomeronasal organ, dogs can also detect pheromones. These ‘creatures of the nose’ surpass our olfactory abilities because they have 50 times more olfactory receptors for odorant molecules in their noses: as many as 300 million, which connect to their large olfactory bulbs. “From studies in canines and other species, we know that the olfactory bulbs are involved in discriminating between odours and in enhancing the sensitivity of odour detection. Olfactory bulb neurons transmit signals to the olfactory cortex. The pathway then projects to the hippocampal formation and the thalamus, which relays information to neocortical areas such as the medial and orbitofrontal cortex, where the olfactory signal is interpreted,” Ferry explained. The neuroscientist studies brain structures involved in learning and memory. Her research focuses on the circuits involved in the acquisition, consolidation and retrieval of memories with the aim of understanding the basis of memory disorders.
For their career as police detection dogs, the animals were trained for 18 to 20 months before entering the judicial case programme. “The police force preferentially works with German and Belgian Shepherds because of their high aptitudes for training, their high motivation and working capacity. It seems that hounds are very good at following particular scent traces, but this breed does not enjoy working indoors and doing identification tasks repetitively,” the neuroscientist commented.
The training comprised six to eight daily matching exercises, in the course of which the dogs first learned to recognise human scent in general and then to identify individual human scents. If they successfully matched a sample to the corresponding sample in a line-up of five samples in glass jars, and lay down in front of the matching sample, they were rewarded with a sausage or allowed to play with a ball. The dogs received additional training for eight to ten months throughout their lives. In the matching tasks, sensitivity and specificity improved with increasing rounds of training. In future experiments, the scientists want to improve the hit rates in matching trials with differently collected samples. “This is very important, since during official judicial cases, ‘indirect’ trace scents collected at the crime scene are usually compared to the ‘direct’ body scent of the suspect in custody,” Ferry noted. “We also want to analyse, whether our conservation procedure of human scents is good enough to preserve olfactory characteristics over the years and thus influences the identification performance of the dogs. Another interesting question is, whether the scent of persons changes during their life.”
So far in France, for some magistrates and judges, identification based on human scent remains a controversial form of legal evidence. Compared to forensic DNA analysis, there are arguments in favour of forensic odorology. “Our study in PloS One showed that rigorous training of dogs leads to accurate and reproducible results in human scent identification, which does not require much time,” Ferry said. She added that whereas a criminal could use strategies to avoid DNA traces, he could not move without leaving his scent. “When DNA analysis sometimes cannot be used to prove a suspect was present at a crime scene, human scent identification does allow us to show if he or she was there.”
Bettina Dupont
Photo: Police dog Cisko during a scent detection test © DGPN-SICOP