What can David Letterman’s Stupid Animal Tricks Teach us About Forensic Science?

One of the “types” of forensic science that is used in modern policing today is the use of specially trained canines.  Canines, proponents say, can be trained for various targeted goals including cadaver search, explosive detection, drug interdiction, accelerant detection and general protection/patrol just to name a few.

You can find some limiting language in the Scientific Working Group on Dog and Orthogonal Detector ( SWGDOG) http://swgdog.org/ Guidelines themselves.

The list is very very long into why there are a lot of limitations to this idea that well-trained domesticated pet can validly and with an acceptable rate of error detect controlled substances, accelerants, human cadavars or explosives.

• For some reason canine training tends to attract profiteers.  As such, it is very prone to some really bad original training.  Just Google Russell Lee Ebersole to see the allegations made against him by others and what he plead guilty to and was sentenced to in Court.
• There is also the issue of bad handling in over-working them, over-rewarding them, cuing them, the handler not using sunglasses, the handler using a leash lead with tapping instead of free roam, lack of practice using “the box drill”, incorrect type of alert presented at examination but the handler calling it “a hit” (passive versus active alert), and on and on and on.
• There is also the BS justification of the “rental car” approach or the “residual odor” defense where the handler seeks to justify that the dog was not wrong when a search revealed no drugs and that it must have been due to “residual odor” in the air because my pet could never be wrong.  (See “You have cocaine on you guaranteed!  Cocaine on Currency“)

The main scientific issue with the canine itself is not so much the discriminating power of their nose (the selectivity) and their lack of specificity, but rather in their ability to communicate with us their findings free of bias, cuing or error.  Most dogs are not distraction trained.  They are over-worked and even when “properly” trained lose their training over time.

Here is a NSW review that includes the following language which is instructive:  http://www.ombo.nsw.gov.au/news-and-publications/publications/reports/legislative-reviews/review-of-the-police-powers-drug-detection-dogs-act-2001

No drugs were located in almost three-quarters of searches following indications, raising questions about the accuracy of drug detection dogs. This in turn casts doubt on the legitimacy of police relying on the dogs to determine whether they may reasonably suspect that a person is in possession of a prohibited drug.

Accuracy rate

During the review period almost three-quarters of persons searched in public as a result of an indication by a trained drug detection dog were not found in possession of prohibited drugs.

NSW Police have suggested that the dogs’ accuracy is 70%. This takes into account the 26% of searches where drugs were located, and incidents in which no drugs were located but the person made some admission of prior drug contact. Admissions of drug contact included persons admitting to previous drug use (usually cannabis), or indicating that they might have been around others who were using drugs (usually cannabis).

There are a number of difficulties with using the 70% figure as a measure of accuracy. First, it takes into account admissions of drug contact involving cannabis smoke, which a person may have been inadvertently exposed to. In addition, NSW Police say drug detection dogs are not trained to indicate this scent. Second, some admissions of personal drug use were so remote in time they did not provide plausible explanations for the indication. For example, a number of admissions recorded by police in information reports involved drug use dating back days, weeks, months, and in some cases, years. Third, although some admissions may support the accuracy of drug detection dogs in picking up the scent of prohibited drugs, this should not be confused with the accuracy of the dogs detecting persons currently in possession of prohibited drugs, which is the purpose for their use.

The central issue, if you believe in the use of trained canines and its validation, seems to surround what precisely is the dog alerting to?  What is the calibration compound?  In the case of cocaine, it is not actually cocaine that they are alerting to when there is an indication.  They are not trained to detect cocaine in terms of IUPAC Standard InChI:

* InChI=1S/C17H21NO4/c1-18-12-8-9-13(18)15(17(20)21-2)14(10-12)22-16(19)11-6-4-3-5-7-11/h3-7,12-15H,8-10H2,1-2H3/t12?,13?,14-,15+/m0/s1

They are universally trained to detect methyl benzoate.

Why do they use this compound, one may ask. Methyl Benzoate has a chemical formula of C₆H₅COOCH₃.  It is a colorless liquid that is poorly soluble in water.  It is not cocaine.  However, it is claimed that cocaine hydorchloride hydrolyzes IN MOIST AIR to Methyl Benzoate.

The bottom line:

Dogs are man’s best friends.  They should be celebrated as such.  However, their use in the forensic arena to form even the low burden of probable cause based upon the above and other arguments should be limited to the truth of it meaning that there are extreme limitations to the accuracy and the validity of the conclusions derived from their use.