Three dog teams were selected to participate in the study. The dogs were trained in human remains detection (HRD), but not cross-trained on live human odor, and performed a passive alert, either a sit or a down. Teams were certified as cadaver dogs by the California Emergency Management Agency. Handlers had to be comfortable working on-lead. Two of the dogs were German shepherds, one was a Labrador.
The experiments were conducted in six square plots, 10 meters by 10 meters, in an area with Jeffrey pines, mountain mahogany shrubs, sage brush, and native grasses. Human teeth were soaked in distilled water in a sterilized glass jar with a metal lid for 24 hours, followed by air drying in sunshine. Teeth were not placed in preservative or otherwise chemically treated. Teeth were stored together in a single sterilized container. The reason for the combined storage may have been to provide some common “bouquet” for all the teeth so the dogs would be smelling roughly the same thing with each tooth.
Ten teeth were placed in each plot on October 29, 2008, a relatively high number of target hides for a single location. Each plot was subdivided into one meter square subgrids and teeth were randombly placed in subgrid locations, with no more than one tooth per subgrid. Teeth were partially buried with one end flush with the surface or placed on the surface under pine litter. Wind on the day of the trials averaged 15.7 miles per hour, which could be expected to reduce the dogs’ effectiveness in finding such small items as teeth. Results might be expected to be better on a calmer day.
Trials were double blind, and neither the handler nor the data collector knew the number or locations of placements in the plots. Trials were conducted two days after the teeth were placed. Handlers stated out loud when a dog alerted. Handlers covered the grids with a two-pass strategy, first crossing the grid up and down three times (a total of six passes) to go from one side of the grid to the other, then three back and forth in a crosswise direction (again a total of six passes) until the grid was fully covered a second time. The maximum leash distance meant that a dog would twice have access to all the ground in a grid. The graphic shows an aerial view of what covering a portion of the grid could look like. Handlers were given no time limit. Mean search times of the three teams were 27.5, 45, and 47.5 minutes. The team that was quickest, Team C at 27.5 minutes, had the worst performance. (Some handlers would argue that a search under strong wind conditions should involve making passes downwind of the search area, but it is not clear if this would have been useful here.) When a dog alerted, as stated by the handler, the time and location were recorded, and the handler placed a flag numbered sequentially at the location of the alert. If the handler saw a tooth when placing the flag, she could reward the dog with play, provide verbal praise or petting, or if she chose, give no reward. Handlers could also place a flag based on a dog’s change of behavior. (In much of police detection dog work, a failure to give a clear alert would not justify further police action.) To be considered a “hit,” the handler had to have placed a flag no further than 0.45 meters (c. 18 inches) from a tooth. False positives were calculated as the proportion of alerts greater than 0.45 meters from a target, divided by all alerts.
The highest recovery rate was 7 of 9 teeth in a plot, while the lowest was 0 of 10 teeth. Of the 29 teeth located by the teams, 24 were pinpointed precisely, while 5 were within 0.45 meters of a tooth. Nine flags were more than 0.45 meters from a tooth. Team A found 7 of 9 teeth in both plots searched, with one false alert. Team B found 5 of 9 in one plot and 6 of 9 in another, with 4 false positives. Team C found 4 of 10 teeth in one plot and none of ten in a second plot, with three false alerts.
The researchers concluded that the “dog teams were shown to be able to locate individual human teeth in a field environment.” The poor performance of Team C shows that “significant variability exists in dog teams, even among those that have met the same minimum criteria for certification, alert type, work style, and trained target odors.” Nevertheless, dogs can clearly be useful in finding teeth. The researchers stated:
“Of interest is that in the natural setting teeth may or may not have associated blood or decomposition of other tissues, depending on the circumstances surrounding death, postmortem processes, and time. Despite the relatively clean state of the teeth used in this study, dog teams were able to find and in many instances pinpoint the location of individual teeth to which they had not been previously exposed.”
The researchers placed considerable emphasis on good recordkeeping:
“Maintaining accurate and quantifiable training records such as those kept in this study would provide for on-going assessment and the ability for handlers to more accurately present probability of detection (POD) to search managers, detectives, and investigators.”
The fact that some flags were planted more than 0.45 meters from a tooth did not necessarily mean that the dog had not detected something, but it did mean that it had not gotten close enough to make finding the tooth easy:
“In reality, the relationship between the distances of an actual tooth from the dog’s alert, or handler’s flag, translates to processing effort. The farther away from a target the dog alerts and a flag is planted, the more time and effort required to find that target. This holds true for any target. Processing effort translates into time and cost. Most of the flags were planted very close to the actual teeth, <0.45 m (18”), which would make locating the tooth relatively easy with a sieve.”
The researchers emphasize that the “idea of minimum level of proficiency, or baseline specification, is also important for actual crime scene deployments where the legal defensibility tied to any team’s qualifications is a critical factor.” (As handlers sometimes say: “Train like you work; work like you train.”) They say that their findings reinforce “the need for handlers to conduct training that mimics actual search assignments and conditions and to train toward maximizing recovery through the use of blind problems.” It must be noted, however, that having ten target odor finds in such a small area will not often mimic an actual search assignment.
The research provides support for using cadaver dogs even when an area is not expected to yield an intact body, or even large body parts. This is a significant study for the cadaver dog world.
This blog was written by John Ensminger and L.E. Papet.
Source: M.E. Cablk and J.C. Sagebiel (2011) Field Capability of Dogs to Locate Individual Human Teeth, Journal of Forensic Sciences (doi: 10.1111/j.1556-4029.2011.01785.x). See also A.J. Rebmann, Cadaver Dog Handbook (Taylor & Francis/CRC Press 2000); J.J. Ensminger, Police and Military Dogs (Taylor & Francis, CRC Press 2011) (Chapter 19: Cadaver Dogs, by J.J. Ensminger & L.E. Papet).
0 comments:
Post a Comment