The Supreme Court has recognized the uniquely non-intrusive nature of a canine sniff means that using dogs in certain situations does not involve a search. This has encouraged the use of detection dogs in law enforcement, and limited the costs associated with producing evidence of canine alerts in court. Nevertheless, the Supreme Court’s decisions do not mean that any claim that a dog may have alerted can justify a further search, and courts have recognized that a dog must be reliable for an alert to have this consequence. Thus, a high level of reliability must be expected of a canine team, and a failure to conduct adequate training or maintain complete records to establish that reliability will not exclude the possibility that a called alert may actually have been cued, precluding the use of the alert at trial.
 Cueing and Probable Cause: Research May Increase Defense Attacks on and Judicial Skepticism of Detection Dog Evidence
John J. Ensminger and L.E. Papet 
Cueing has been alleged in cases involving scent identification and narcotics detection, and courts have on occasion accepted cueing arguments when supported by testimony of a defense expert that the dog’s method of indicating was not evident on a videotape of an incident. Generally, however, courts have declined to accept or reverse on claims of cueing, particularly where other factors establish that a dog has been reliable, such as training and field records or certification by an established organization.  Courts have also rejected cueing arguments by accepting that only the handler may be able to reliably say whether a dog has alerted.  Cueing arguments have sometimes failed for procedural reasons,  and have sometimes failed because a claim was unsupported by any evidence.  A recent Florida Supreme Court case mentioned cueing as a form of handler error. 
A 2011 study published in the journal, Animal Cognition , may, however, lead to a more general attack on the use of police detection dogs.  The study, written by three faculty members of the University of California at Davis, found that under a specific testing scenario where no drugs or explosives were in the testing environment, most canine teams participating in the study nevertheless alerted to the presence of drugs or explosives when the handlers had reason to believe that a target item had been placed in the environment. While this study has significant consequences for training and deployment of detection teams, we believe that the findings of the study should not be used to support blanket arguments that alerts by trained scent detection dogs do not support probable cause for a subsequent search. We believe that canine alerts are not to be labeled junk science, as has happened in scent identification, another area where cueing is possible. 
We do believe, however, that the results of the study should serve as a warning for police and military dog handlers that calling an alert should not be casually made to provide some colorable basis for a search when no other factors support probable cause. A dog’s alert, in other words, should not become an easy means of justifying a search when the handler or an officer involved in the incident has a hunch that something is awry and needs an excuse to take additional action.  After all, if the handler passes his hunch to his dog, the dog's alert is not independently bolstering the officer's suspicion.  Further, we believe that the results of the Davis study argue strongly that training and testing should always use negative controls or blanks, replicating situations that handlers will routinely encounter in the field where a target odor may not be present.
When a search subsequent to an alert produces no contraband, handlers often argue that the dog alerted to residual odor, and this may be the case in a given situation.  Handlers should be aware, however, that a likely result of the Davis study is that courts may become less inclined to ignore other possibilities, including cueing or that the handler falsely called an alert, and defense attorneys are certainly going to become more aggressive in attacking such evidence.
Cueing in Criminal Law
Cueing can be defined as the phenomenon of a handler, or someone else in a dog’s presence,  providing a conscious or unconscious signal to the dog that induces the dog to perform a trained behavior pattern.  Specifically with regard to drug detection dogs:
Handler cues are conscious or unconscious signals given from the handler that can lead a detection dog to where the handler thinks drugs are located. 
This phenomenon can be unconscious and is sometimes called the “clever Hans” effect, a term coming from a horse that seemingly always tapped a hoof the correct number of times in answer to arithmetical questions.  Cueing is to be distinguished from falsely claiming a dog has alerted when it has not. 
Defense experts sometimes mention the possibility of cueing. In a case arising in Michigan, for instance, a defense witness, a former police officer and drug detection dog trainer, testified that the handler of a dog that alerted to a car parked at the airport may have unconsciously cued her dog to alert. The defense expert stated that the dog’s reliability had been compromised because the dog had not been trained on “dead” targets, and “because Dennard [the handler] knew which car was suspected, she may have unconsciously cued Dingo, and thus Dingo’s indication might have been tainted.”  The court, however, concluded that “a very low percentage of false positives is not necessarily fatal to a finding that a drug detection dog is properly trained and certified.” The court noted that the handler had run the dog around a test car “to avoid unduly suggesting to the dog a specific place to indicate; and that Dingo indicated on [the defendant’s] car but not on the test car.” This meant to the court that the chance of an incorrect cueing was reduced. The court felt the handler’s testimony, although faulted by a better known expert, was adequate to establish the dog’s reliability. 
In an early cueing argument, the trial court declined to adopt an expert’s opinion as to the possibility that a dog was unconsciously cued on the ground this was “at best speculative and conjectural.” The defense expert, a retired Baltimore police dog trainer, testified that it is possible for a handler, through voice or physical cues, to compromise a dog’s objectivity. The appellate court affirmed the conviction, but acknowledged that “less than scrupulously neutral procedures” could create the possibility of unconscious cueing.  A Nebraska appellate court, citing this case, noted: “To a certain extent, the [D.C. Circuit] was merely stating the obvious—because dogs cannot talk or be cross-examined, the procedure used by canine handlers must be such that the evidence establishes that the court can rely on canine sniffs.” 
Words and Gestures
Words said by a handler during a sniff have been argued as cueing dogs.  A decision of the South Dakota Supreme Court involved a handler stopping and gesturing to the trunk of a vehicle during the sniff. The handler repeatedly said the German word, “gift,” pronounced “geeft,” meaning poison or venom. Two witnesses disagreed on whether a hand gesture and repeated use of a word amounted to cueing. The trial court concluded that the handler in the case had not cued his dog. The South Dakota Supreme Court concluded that with “due deference to the trial court’s better position to gauge the demeanor and credibility of the witnesses, we cannot say the court’s findings were clearly erroneous.” 
Hand gestures have been argued as cueing a dog. In another decision of the South Dakota Supreme Court, the handler argued that hand signals he used with his dog were to indicate a search pattern intended to get the dog to sniff high and low on a vehicle, not to cue the dog. The handler argued that the signals were approved under ICPSD [International Congress of Police Service Dogs] standards.  A defense witness, a U.S. Army canine handler and trainer, argued that the hand signals and body position of the handler in the case may have caused the dog to alert, even if the handler did not intend to cue the dog. The court declined to reverse merely because there had been conflicting expert testimony.
In a 2008 federal district court case, a defense witness, a professional dog trainer, took issue with the use of a “detail pass” in a sweep “because he contends that dogs are improperly cued during detail passes.” 
Videotapes of Traffic Stops
Videotapes are changing the ability of witnesses and courts to second guess handlers’ interpretations of their dogs’ actions. In a case arising in Nebraska, the trooper who took his dog around the car he had stopped testified that the dog had alerted.
Trooper Duis testified that in this case Robbie “alerted” to the presence of drugs by sniffing more intensely around certain areas of the car, but he acknowledged that such “alert” behavior was subtle and might only be recognized by himself or another person who was familiar with Robbie’s tendencies. Although Robbie was trained to “indicate” (by scratching) when he located the strongest source of the drug odor, he did not do so in this case. Defendants’ experts testified that the “alert” behavior described by Trooper Duis could easily be attributed to his “cueing” of the animal, either intentionally or unintentionally, by changing the leash from one hand to the other, by stopping, by blocking the way, or by other actions. They saw nothing on the videotape to indicate that Robbie had detected the presence of drugs.
The court said that “there must be an objectively observable ‘indication’ by the dog of the presence of drugs,” and concluded that the “dog’s actions did not positively signal the presence of drugs inside the vehicle.” 
Courts have sometimes accepted a handler’s testimony despite videotape evidence to the contrary. Yet another decision of the South Dakota Supreme Court involved a defense expert’s conclusion that a dog’s “indication at the trunk was cued.”  The handler had testified that he had not cued or caused the dog’s aggressive indication.  The court affirmed, stating that “we are not left with a definite and firm conviction that the circuit court erred in finding that the drug dog indicated the odor of an illegal substance.”  Justice Meterhenry, dissenting, argued that when a well-trained narcotics detection dog smells drugs, “its trained response should be obvious not only to its handler but also to a reasonable and prudent person.”  The dissenting justice described the video tape of the dog in the case:
A review of the trooper's video tape shows nothing close to aggressive scratching or biting or barking by Kaz. The sniffing dog circles the vehicle, pauses in spots but never scratches. It pauses briefly at the trunk of the vehicle and only after the officer says “gift” to the dog does it show an interest in the trunk and then only by very briefly nipping at the bumper part of the vehicle under the trunk….
Kaz’s response is clearly different from his response to drugs in the SDPB [South Dakota Public Broadcasting] video. In fact, nothing about Kaz’s bumper biting could be viewed as “aggressive behavior.” 
The reference to the video that was broadcast concerned a demonstration showing the particularly dog’s skills in finding drugs. In a 2004 case, an Ohio appellate court reviewed the videotape of a traffic stop but found “no indication on the tape itself that Trooper Arnold was speaking to the canine or in any way cueing him to the location of marijuana, a drug that Ringo was trained to locate.” 
In a case arising in Nebraska, the defense expert, a dog handler, testified that a review of the videotape of a traffic stop indicated that the handler in the case had “consciously or unconsciously sent clues suggesting a response to the dog.”  The witness noted that the handler used gestures to the sides and rear of the car with a stick, which the expert regarded as unnecessary, and that the handler raised his hand immediately before the dog sat, another indication of cueing. The court provided its own summary of the sniff, which occurred 33 minutes into the stop.
Deputy Wintle walks the dog around the car once and the dog jumps on and sniffs the vehicle. At the conclusion of one complete pass around the minivan, the dog attempts to return to the squad car, but is pulled back to the rear bumper by Deputy Wintle. At the 36-minute mark, the dog sits briefly on the ground and Deputy Wintle and the dog return to the squad car. 
The court found that the prosecution had not rebutted the defendant’s showing that the dog’s alert to drugs in the case was not reliable, and the testimony concerning cueing was “borne out by the court’s review of the handler’s conduct in the videotape.” 
In a case arising in Iowa, a defense expert went so far as to argue that “with 99% certainty” a dog did not detect drugs but was cued. 
Nicely [the expert witness] presented videos of two different searches by drug dogs . In both instances, there was very little contact between the handler and the dog , and the handler remained behind the dog , allowing the dog to move around the vehicle. In this case, throughout the time Bauerly was taking Bosco around the vehicle, Bauerly repeatedly tapped the vehicle and said, “Drugs, check!” and “Drugs!” In addition, he walked backwards in front of the dog as they went around the vehicle. According to Nicely, these actions all cued Bosco to give a false indication on those areas. 
The prosecution’s expert, on the other hand, argued that patting a vehicle and saying words like, “Drugs! Drugs!” was a way of directing the dog to sniff a particular area. He reviewed the video of the sniff in the case and testified that he saw no evidence of cueing. The court concluded that to cue the dog to indicate on the boot where the drugs were eventually found, the handler would have had to know there were drugs there in the first place. This was not established. Since the handler reasonably believed his dog had indicated, he had probable cause for the search. The court’s logic is incorrect: dogs often alert where nothing is found, and handlers may be shown to cue dogs despite the absence of any contraband. 
Cueing Indicated by Dog’s Behavior
Cueing arguments generally refer to actions of the handler inducing the dog to indicate or alert, but one case involved testimony that the dog looked at the handler instead of at the car it was supposed to be sniffing, “meaning the dog sought guidance or input from Bossman [the handler].”  Another witness testified that the dog had not looked at the handler as constantly as would be required to establish cueing. The court noted that the dog’s “deployment records showed several deployments where there was no alert or indication, suggesting that the dog did not detect any odor of any drugs….” The court concluded that suppression of the canine evidence was not required because the “dog’s entry into the vehicle was not essential to the outcome of the canine sniff search.”
Cueing Arising from Training
It has been argued that cueing may be associated with training.  In a 2010 Oregon case, a driver was stopped for failing to wear a seatbelt. A dog alerted and methamphetamine was found in the car after a search to which the defendant did not consent. The defendant moved to suppress the evidence, “asserting that because Benny was not properly trained, his alert did not provide probable cause to believe that the car contained drugs.”  The defendant’s expert’s testimony was summarized by the court as follows:
He criticized the “play reward” system of training dogs , suggesting that simply choosing dogs from an animal shelter (which is where Hulke obtained Benny) was inappropriate, because a psychological profile of a dog should be done before it is selected for imprinting. He suggested that the play-reward system was “okay” and that it works to “a certain extent” but not as well as imprinting, because the incidence of false positives cannot be accurately tracked. He suggested that a major flaw with the play-reward system is that the dog may simply be detecting the scent of the handler who handled the drugs when they were hidden, or may be picking up on subtle cues from the handler, such as changes in heart rate or breathing patterns when the dog is near the place where the handler has hidden the drugs. As for the method used by the OPCA [Oregon Police Canine Association] to certify the dogs , Woodford asserted that it likely was flawed because the dog would simply track the scent of the person who placed the drug, which would lead the dog to the drug. 
The court affirmed the denial of the motion to suppress, noting that the dog in question had passed a certification test requiring a 90% accuracy rate.
Woodford suggested that handlers might be cueing the dogs about where the drugs were located or that the dogs might be following human scent to the location where the human placed the drugs. Fyfe testified, however, that handlers did not place the drugs for the test or know where the drugs were located, that the people who placed the drugs wore gloves, and also that the people who placed the drugs went to numerous areas in the environment so that the dogs would not simply go to the place that had most recently been disturbed. Given that evidence—in addition to the undisputed evidence by both parties that dogs do, in fact, have superior abilities to detect odors and can be trained to alert to specific odors—we agree with the trial court that the evidence of Benny's alert to defendant's car door handle, when combined with other evidence, provided sufficient probable cause for the search at issue here. 
In a 2010 case from the federal district of Kansas, a handler testifying as a defense expert criticized the records of the dog in the case “for not indicating when the handler knew the location of the target odor and when he did not, so as to evaluate when ‘cues’ may have been given, and the failure time of the search.”  The district court concluded that the dog had given validly alerted both outside and inside the vehicle. 
Although a low percentage of false positives in training and certification may not preclude a dog from being considered reliable,  it has been noted that “[t]his analysis would change, it would seem, if the challenged drug dog made several false positive alerts during the training and certification process, that is, where a controlled setting obviates the residual odor argument.”  The failure of training to deal with possible cueing problems has resulted in suppression of canine evidence. In a 2009 Arizona case, an officer testified that his dog had alerted to a scent at the driver’s door.  The defense expert testified to inadequacies in the dog’s training. The trial court had stated:
The certification records of this dog suggest a possible problem—the dog was being cued or had a tendency to alert on blank vehicles. The real world records substantiate this concern unless it is accepted that actual drugs or residual odor was present in virtually all of the vehicles searched. Even if residual odor accounted for almost all of the false alerts in the field, however, what caused the false alert during the certification process if not cueing or some improper procedure that had to be corrected?
The trial court, reflecting the expert’s opinions, concluded that the certification process had not eliminated inadvertent or unconscious cueing by the dog’s handler. The expert had testified that training could have been done in such a way as to correct the cueing problem, but that it had not been done with the handler and dog in the case. The motion to suppress was affirmed.
One court held that the absence of evidence that a dog had been trained to cue made the testimony of a defense expert not credible, and the testimony was disregarded.  The logic is faulty. Dogs are trained to alert, not to cue.
Cueing as Indicating Lack of Reliability
Cueing cases at traffic stops generally involve the exterior of the vehicle, but in one case arising in Tennessee, the defendant argued that the handler enticed a police dog to enter his vehicle through an open door, then cued him to alert.  The government did not argue any other basis for probable cause than the cue. The court determined that the cueing argument was an attack on the dog’s reliability.  There was conflicting testimony about the cueing, but the court concluded that the government carried its burden of proving that the dog was not cued or otherwise improperly instructed or deployed.
In the operation of a canine team in the field, it can be difficult, if not impossible, to distinguish cueing from an alert to a residual odor, or a simple false alert. Faced with explaining why a dog has alerted without any contraband being found in a subsequent search, as noted at the beginning of this paper, handlers often refer to the possibility of residual odor.  A recent decision of the Florida Supreme Court involved an officer claiming residual odor to explain why at least one traffic stop involving the defendant did not result in the discovery of any narcotics.  Another traffic stop of the same defendant had resulted in the discovery of a precursor to methamphetamine, though no evidence was provided that the dog could alert to the particular precursor.  The handler, a police officer, testified concerning the training and certification of his dog, but acknowledged that he only maintained field records concerning the dog when an arrest was made. The court said this meant that it was “impossible to determine what percentage of the time Aldo alerted and no contraband was found following a warrantless search of the vehicle.” Noting that training and certification programs vary widely in methods and tolerances of failure, the Florida Supreme Court stated that “any presumption of reliability based only on the fact that the dog has been trained and certified does not take into account the potential for false alerts, the potential for handler error [including cueing], and the possibility of alerts to residual odors.” 
Because of these variables, a necessary part of the totality of the circumstances analysis in a given case regarding the dog's reliability is an evaluation of the evidence concerning whether the dog in the past has falsely alerted, indicating that the dog is not well-trained, or whether the alerts indicate a dog who is alerting on a consistent basis to residual odors, which do not indicate that drugs are present in the vehicle. Accordingly, evidence of the dog's performance history in the field—and the significance of any incidents where the dog alerted without contraband being found—is part of a court's evaluation of the dog's reliability under a totality of the circumstances analysis. In particular, when assessing the factors bearing on the dog's reliability, it is important to include, as part of a complete evaluation, how often the dog has alerted in the field without illegal contraband having been found.
The court held that a motion to suppress the canine evidence should have been granted by the trial court and remanded for proceedings consistent with this finding.
Cueing in Research Studies
Cueing has been discussed in a wide range of scientific studies. Dogs learn how to accomplish tasks by following their handlers’ movements, and may do so without any particular reward being associated with their accomplishing the task.  Nonverbal cues are sometimes necessary for individuals communicating with their service dogs.  Dogs respond best to commands when the handler is looking at them, but may still respond as directed even when the handler is looking away.  Dogs may misinterpret cues.  Dogs trained to perform more than one detection function on a handler’s verbal cue may not perform as well as dogs that are trained for only one specific scent in certain circumstances, suggesting that training regimens may affect future accuracy. 
Many references to cueing in research studies, as with law enforcement work, concern efforts to eliminate cueing.  One study noted that hesitation or backtracking on the part of a handler may unintentionally cue a dog to alert. 
The Davis Cueing Study
In the study published by Animal Cognition , “Handler Beliefs Affect Scent Detection Dog Outcomes,” three researchers set up testing environments in a church and told 18 handlers that the testing environments would contain markers for scent locations and decoy scents, when in fact no scents to which the dogs had been trained to respond were (supposed to be) present. Of the 18 canine teams, all but one alerted in at least some of the trials, and since all alerts were deemed incorrect responses, the researchers concluded that handler beliefs “that scent was present potentiated handler identification of detection dog alerts.” The researchers also concluded that human “more than dog influences affected alert locations,” confirming that “handler beliefs affect outcomes of scent detection dog deployments.” We believe that these conclusions will receive frequent reference by defense counsel and expert witnesses in narcotics and other detection dog cases.
The searches occurred in rooms in a church. At the beginning of each testing day, an experimenter carried a metal box containing 12 half-ounce samples of marijuana triple bagged in sealed plastic bags, as well as a canvas bag containing 12 half-ounce samples of gunpowder triple bagged in sealed plastic bags, into the four rooms and set the containers down by the door inside of each room. The metal box and the canvas bag were only briefly placed in the rooms (but exactly how briefly is not noted). The paper states that this was done “to maintain the belief that the experimenter was setting out target scents in each condition.” It is not clear if handlers saw this procedure, but seems likely since it was intended to deceive them. Presumably the box and the bag were of a size where handlers could reasonably believe the items had been hidden in the rooms they would later search.
Apparently it was anticipated that some residual odor from the metal box and the canvas bag might be detectable by the dogs since the researchers state that “no dogs alerted on or around the door where the scent containers had briefly been placed.” (One alert was recorded as occurring at a doorway, but presumably this was a doorway different from the one where the scent containers had been placed.)
The researchers argued against detection of contamination scent, stating that “detection dogs are trained to identify scent source rather than scattered residual scent. For example, dogs trained to alert on gunpowder are not expected to alert in an airport area simply because an armed officer passes through.” To the contrary, training programs typically involve dogs following scattered residual scents to sources. Vapor wake dogs, under a program developed at Auburn University, for instance, are trained to detect an explosive odor, follow it to its source, and give an indication to the handler. 
It is unclear why the researchers did not have control situations in which no residual odor could have been present, as is recommended for certification tests by some organizations.  Canvas bags used to hold target items are sometimes used by trainers as “odor multipliers” and it is not clear if this possibility was recognized by the researchers. The amount of gunpowder in the canvas bag was rather high (six ounces, approximately 170 grams, whereas many trainers use between one and ten grams for training), suggesting that residual odor would exist for some time. Since the canvas bag was carried into the room “at the beginning of each testing day,” it is not clear what kind of gap there was between the odor being in the rooms and the dogs coming into the rooms for the sniffs. The issue could have been addressed by trials in which no narcotics or explosives were even temporarily placed in the rooms, but this appears not to have been done. The researchers, in other words, did not perform a rather obvious type of blank or zero trial to validate their results.
Handlers were provided a small card containing their assigned sequences of eight runs. An observer was present for each run. Observers recorded alerts and their locations but did not evaluate the validity of alerts and did not know the conditions of a search area (i.e., whether there was a target present or that targets had briefly been present before the trials were run). A total of 225 alerts were recorded. Dogs did not alert in 15% of the runs, which was the correct behavior, but alerted in 85% of the runs, which was incorrect. Dogs were least likely to alert in areas where nothing was marked with red construction paper and no sausages or tennis balls were hidden. One handler dog team appears to have made no alerts in any trial, and thus performed perfectly under the experimental parameters. 
The scenarios sometimes involved items that could interest a dog, including Slim-Jim sausages. (Slim Jim products generally contain sodium, at least sometimes in the form of sodium nitrite.) The paper refers to possible “increased dog interest” with reference to where tennis balls were hidden, but it was not explained why this would occur if the dogs did not see and were assumed not to smell the balls. Or was it assumed that there might be some smell to the balls that would attract the dogs as tennis balls are sometimes used in initial training of detection dogs?  The fact that construction paper was more likely to draw an alert than a hidden sausage or tennis ball convinced the researchers that human suggestion was more likely to lead to a false alert than dog interest.
The researchers considered that in some cases handlers may have called alerts where the dog’s behavior did not actually justify such a determination. The trials were not videotaped so a subsequent review of recorded alerts was not possible.  Given the importance the researchers attached to the handlers’ interpretations of the dogs’ behavior and the fact the trials were run in a contained environment, the decision not to videotape is curious.
The researchers noted that there conceivably could have been drugs or explosives in the church, but thought this unlikely. The rooms contained cabinets, tables and chairs, and art supplies, the latter of which may contain many chemicals, but this is not further described. Some dogs also alerted to a first-aid kit. Such kits may contain pharmaceuticals (which might contain, among other things, methyl benzoate, a substance thought to be recognized by cocaine-detection dogs). Further:
Some handlers suggested the possibility that dogs were following previous dogs and alerting at locations in which these dogs had salivated or otherwise left trace evidence of their presence. This would not explain the difference in patterns of alerts between marked and unmarked conditions or the variation in alert locations across all conditions. This would also be unlikely given the extensive training and certification processes required of these teams.
This assumption regarding “extensive training” has not been made in other areas of canine research. Scent lineup researchers deem it very important to remove saliva from stations in scent lineups, as this might lead to cueing. 
The researchers concluded that their findings “emphasize the importance of understanding both human and human-dog social cognitive factors in applied situations.” This is stating the obvious.
Significance of the Davis Study
The research model in the study is similar to a situation where a handler is told that Subject A entered a car with a pound of marijuana, drove somewhere, all the time under surveillance, then left the car without the marijuana.  The handler is asked to take his dog around the car. The handler is effectively told that drugs are in the car and that his dog, if performing well, is going to alert. Of course, in most such situations, there will be drugs in the car and the dog will alert. This is to be distinguished from the more common sniff situation where an officer finds the behavior of a driver suspicious and takes his dog around the car on the chance that his suspicion is correct. In the first scenario, there is a strong expectation that drugs will be found, and a failure to find them means the dog is not performing well. There may be some expectation in the latter scenario as well, but less so, though the handler may be tempted to falsely claim an alert to check out his hunch. Should an experiment that replicates the first scenario be relevant to exclude evidence of drugs found in a traffic stop that fits into the second scenario? This is a difficult legal issue and will be up to the courts to decide. In our opinion, this study should be relevant, but should probably go to the weight of the evidence, not its admissibility.
If a dog is certified and has an accuracy rating of, say 80% in independent trials, many courts have said that a threshold for admissibility has been met.  The dog does not need to be correct all the time. At the very least, the Davis research supports arguments that training should involve negative controls—test situations in which an alert is incorrect. This has been introduced into protocols for scent lineups and is suggested for training and certification by many detection dog organizations. Counsel and courts should pay more attention to a dog's performance in negative controls (sometimes called zero trials or blanks). If a dog has a tendency to alert during negative controls, an argument that the dog was mirroring the handler's hunch might occasionally tip the scales for a defendant in close fact situations. We believe, along with many courts, that accurate training and performance records should be maintained and available to defense counsel, and that failure to produce such records should support a challenge to the admissibility of the dog’s alert and the findings of a subsequent search. 
The Supreme Court has recognized the uniquely non-intrusive nature of a canine sniff means that using dogs in certain situations does not involve a search.  This has encouraged the use of detection dogs in law enforcement, and limited the costs associated with producing evidence of canine alerts in court. Nevertheless, the Supreme Court’s decisions do not mean that any claim that a dog may have alerted can justify a further search, and courts have recognized that a dog must be reliable for an alert to have this consequence.  Thus, a high level of reliability must be expected of a canine team, and a failure to conduct adequate training or maintain complete records to establish that reliability will not exclude the possibility that a called alert may actually have been cued, precluding the use of the alert at trial. 
Training, testing, and certifying organizations should incorporate negative controls into their programs, perhaps more than many of them do now, and law enforcement supervisory personnel should assure that canine officers are in programs that have recognized the importance of negative controls. Field records should include all deployments and all alerts of dogs, so that significant numbers of alerts without the subsequent discovery of contraband can be addressed for possible errors, including cueing.
The Davis study is unique in that it was specifically designed to measure cueing by handlers. Most prior studies mention cueing as a possible flaw to be eliminated in the design of an experiment. We are concerned, however, that this study may become the basis of a blanket labeling of canine detection work as junk science.  Dogs have been recognized as highly valuable in military and other environments for finding explosives as well as narcotics, and failures of particular departments or handlers should not destroy the value to society of canine detection work. It is, however, important to realize that Constitutional considerations can negate the prosecutorial value of alerts made by dogs working in poorly managed law enforcement operations.
Canine evidence, long virtually ignored by defense counsel, is increasingly coming under rigorous attack, and handlers are spending more time in court defending the actions of their dogs and the interpretations they give those actions in criminal investigations. This is not inappropriate, as handlers should be able to defend their work. The Davis study should serve as a wake-up call for handlers to understand that sloppy work can make canine evidence suspect to courts and juries, as well as making life more difficult for handlers that already perform such work well.
 John J. Ensminger is the author of Police and Military Dogs: Criminal Detection, Forensic Evidence and Judicial Admissibility (Taylor & Francis/CRC Press, forthcoming 2011). L.E. Papet is the Executive Director of K9 Resources in Kings Mills, Ohio. They can be reached respectively at firstname.lastname@example.org and email@example.com.
 See, e.g., U.S. v. Ludwig, 2011 WL 1533520, No. 10-8009 (10 th Cir. 2011) (expert testified that California Narcotic Canine Association, which certified handler and dog in case, “uses tests specifically designed to reveal whether a dog unconsciously cues to his handler rather than drugs;” circuit court declined to reopen battle of experts, which district court had resolved by preferring testimony of prosecution’s expert over that of defense); Arizona v. Wright, 2009 WL 2411298 (Ct. App. 2009) (“certification process did not eliminate inadvertent or unconscious ‘cueing’ by the dog’s handler”).
 Cases sometimes use the term “alert” to describe a specific behavior pattern by which the dog tells the handler that contraband is present (or that a scent match or identification has been made), and sometimes use the term “indicate.” Here, following the somewhat more common terminology used by courts (though not the terminology of some training manuals and organizations), the author’s will use “alert” to refer to a behavior pattern by which the dog tells the handler that drugs have been found.
 U.S. v. Jones, 2009 WL 973206 (E.D. Tenn. 2009) (motion to present evidence, including regarding cueing, filed too late).
 U.S. v. Mack, 2010 WL 551065 (M.D.Ala. 2010).
 Harris v. Florida, 2011 WL 1496470 (2011).
 Lit, L., Schweitzer, J.B., and Oberbauer, A.M. (2011). Handler Beliefs Affect Scent Detection Dog Outcomes. Animal Cognition, 14(3), 387-394.
 See, e.g., The Innocence Project of Texas, “Junk Science: A Texas Injustice,” posted on the organization’s website (www.ipoftexas.org/index.php?action=junk-science).
 Courts have been skeptical of dogs that alert almost any time they are deployed:
Deputy Shrum's and the State's explanation for the fact that the dog alerts nearly 100% of the time a sniff is conducted is the presence of residual odor. If in fact virtually all of the vehicles searched either actually contained drugs or emitted residual odor, which would be detected by and alerted to by a narcotics dog, it is hard to see why a dog would be necessary unless it would be to narrow the focus for locating the drugs.
Arizona v. Wright, 2009 WL 2411298 (Ct. App. 2009), quoting the trial court’s ruling granting defendant’s motion to suppress canine evidence based on the dog not being shown sufficient reliable.
 An officer must have more than a hunch even for reasonable suspicion. See U.S. v. Beck, 140 F.3d 1129 (8th Cir. 1998) (federal appellate court noting an officer’s hunch might ensnare a very broad range of innocent travelers).
 Courts have accepted evidence regarding dogs alerting to residual odors. See Maryland v. Cabral, 159 Md.App. 354, 381, 859 A.2d 285 (Ct. App. 2004) (discussing trained dog perhaps able to alert to drugs as long as 72 hours after their removal from vehicle or container); Idaho v. Braendle, 124 Idaho 173, 997 P.2d 634 (Ct. App. 2000) (Idaho appellate court noted that “the handling detective testified that, in his opinion, when Clancy alerted on a location where no drugs were found, that location had the residual odor of a drug that had previously been there and, with respect to the school lockers, clothing or other items in the lockers might have had a lingering odor of drugs”); U.S. v. Johnson, 660 F.2d 21 (2nd Cir. 1981) (federal appellate court noting that an “argument with respect to the problem of a dog detecting only the residual odors as opposed to the drugs themselves misconstrues the probable cause argument”). In U.S. v. Unrue, 1973 WL 14783 (CMA), 47 C.M.R. 556, 22 USCMA 466 (1973), testimony described a dog as having two alerts, one for residual odor and one where drugs were actually present.
 In a 2009 California case, cueing was mentioned as possibly coming from persons watching the lineup other than the handler. California v. White, 2009 WL 3111677 (Ct. App. 2009) (diminished likelihood of third-party cueing where observers were distant from actual test).
 U.S. v. $80,760 in U.S. Currency, 781 F.Supp. 462, 478, n.36 (N.D.Tex. 1991), citing U.S. v. Trayer, 898 F.2d 805 (D.C.Cir. 1990).
 Ohio v. Nguyen, 157 Ohio App.3d 482, 811 N.E.2d 1180, 1195, n. 109 (Ct. App. 2004). Cueing has been used as a term by which a service dog tells a handler to take medicine. Baugher v. City of Ellensburg, 2007 WL 858627 (E.D. Wash. 2007). The term may also be used positively, such as cueing a dog to begin work, without any connection to the dog’s alerting behavior. U.S. v. Valenzuela-Espinoz, 2000 WL 2761919 (D.Ariz. 2009).
 See Miklosi A., Pongracz P., Laketos G., Topal J., and Csanyi V. (2005). A Comparative Study of the Use of Visual Communicative Signals in Interactions Between Dogs ( Canis familiaris ) and Humans and Cats ( Felis catus ) and Humans. Journal of Comparative Psychology 119(2) 179-186. In an experiment where handlers were to follow trails with dogs scented to car bomb fragments, two dogs alerted to decoys. Reviews of videotapes convinced researchers that this was due to handler error in that the handlers had cued the dogs to alert to the wrong person. From the description of the methodology, the handlers were certainly not supposed to cue the dogs, and presumably did so unconsciously. Curran, A.M., Prada, P.A., and Furton, K.G. (2010). Canine Human Scent Identifications with Post-Blast Debris Collected from Improvised Explosive Devices. Forensic Science International 1999, 103-108. If this can happen in a controlled research environment, where handlers know they are being videotaped, the ease with which such cueing can happen indicates how risky field identifications may be.
 U.S. v. Escobar, 389 F.3d 781 (8 th Cir. 2004) (telling a suspect a dog had alerted to her luggage nullified her consent to the search; no sniff had been conducted; a motion to suppress was granted and affirmed by the Eighth Circuit).
 U.S. v. Diaz, 25 F.3d 392, 395 (6 th Cir. 1994) (handler’s testimony supported finding that dog was trained and reliable despite prosecution’s expert).
 But see U.S. v. Diaz, 25 F.3d 392 (6 th Cir. 1994).
 U.S. v. Trayer, 701 F.Supp. 250 (1988), 895 F.2d 805, 809 (D.C. Cir. 1990).
 Nebraska v. Truesdale II, 2008 WL 582530 (Ct. App. 2008) (motion to suppress properly denied by trial court).
 South Dakota v. Nguyen, 726 N.W.2d 871 (2007). In training and testing canine teams for homeland security operations, one of the authors (LEP) found that handlers could cue dogs by changes in their voices (pitch, timing, volume); distracting a dog by talking continuously; praising a dog too much or too soon; reaching for a reward too soon; making movements that appeared to signal a dog, including circling back to previously sniffed locations, changing pace, staring at a place where an item may be hidden, tapping surfaces repeatedly, increasing tension on the leash, making various hand movements, suddenly stopping or standing still, and standing a long time in the vicinity of a possible target.
 726 N.W.2d 883. See Tennessee v. Feaster, 2010 WL 2852284 (Ct. App. 2010) (defendant argued pointing constituted inappropriate cueing but provided no evidence in support of this assertion).
 South Dakota v. Guerra, 772 N.W.2d 907 (2009). The ICPSD is a translation of the German International Patrol Dog Competition Regulations (Wettkamppruefungsordnung, or WPO regulations, posted at http://gsdca-wda.org/forms/WPO%20Standards%20GSDCA-WDA%2020080225.pdf—visited 5/2/2011, stating that commands may be given by voice or hand signal, but limits commands to two to perform a task).
 U.S. v. Brooks, 589 F.Supp.2d 618, 625 (E.D.Virg. 2008) (court concluded based on totality of the circumstances that “the training and certification of Debo and Trooper Homiak was susfficient to provide probable cause to search after an alert”).
 U.S. v. Heir, 107 F.Supp.2d 1088 (D.Neb. 2000) (motion to suppress evidence granted).
 South Dakota v. Lockstedt, 695 N.W.2d 718, 728 (2005).
 695 N.W.2d 727 (quoting transcript of the trial).
 695 N.W.2d 729. Justice Meterhenry, concurring in part and dissenting in part, disagreed with the majority’s finding of probable cause:
We should, at least, require the state to present evidence of the dog's clear indication of smelling drugs before approving of the search. In other situations, we require an accurate calibration of technical devices in detecting alcohol or illegal substances. Similarly, we should also require the necessity of a clear indication from a drug dog before finding probable cause to search.
695 N.W.2d 730.
 695 N.W.2d 887.
 695 N.W.2d 887.
 Ohio v. Nguyen, 187 Ohio.App.2d 482, 811 N.E.2d 1180 (Ct. App. 2004).
 U.S. v. Christy, 2008 WL 753888 (D.Neb. 2008).
 The expert had testified that the cueing came from the handler’s slowing of his pace. See also U.S. v. Clarkson, 2009 WL 1651043 (D.Utah 2009) (defense expert suggested cueing but also argued dog was not well trained, a position the court accepted).
 U.S. v. Olivares-Rodriguez, 729 F.Supp.2d 1030, 1035 (N.D. Iowa 2010).
 See the discussion of Lit et al. below.
 U.S. v. Gastelo-Armenta, 2010 WL 14440451 (D.Neb. 2010) (magistrate’s recommendations).
 See Robert C. Bird, An Examination of the Training and Reliability of the Narcotics Detection Dog , 85 Ky. L. J. 405, n. 134 (1996-97) (“voice or physical signals can compromise a dog’s objectivity and impermissibly lead the dog to alert at the suspected item or person”).
 Oregon v. Foster, 233 Or.App. 135, 225 P.3d 830 (2010).
 233 Or.App. 145 (emphasis added).
 233 Or.App. 146-7 (footnote omitted).
 U.S. v. Beltran-Palfox, 731 F.Supp.2d 1126, 1156 (D.Kan. 2010). The defense also argued that putting the dog’s nose against a source so as to direct its attention was cueing, but the prosecution argued that this was only a way of directing the dog’s attention.
 731 F.Supp. 1158, 1160.
 U.S. v. Diaz, 25 F.3d 392 (6 th Cir. 1994).
 State v. Nguyen, 157 Ohio App.3d 482, 495, 811 N.E.2d 1180, 1190 (Ct. App. 2004).
 Arizona v. Wright, 2009 WL 2411298 (Ct. App. 2009).
 U.S. v. Poghosyan, 2010 WL 4568988 (D.Kan. 2010) (“The only way that the Court can see that an officer's inappropriate conduct could nullify a sniff is if that officer somehow cued the canine or led the canine to indicate when it should not have. Because, after reviewing the audio recordings multiple times, the Court does not find that Trooper Schippers engaged in such conduct, PSD Justice's sniffs are not infirm on this ground. In sum, the Court finds that the defendants have failed to carry their burden of showing that PSD Justice was unreliable on the day of the sniffs;” testimony of Nicely disregarded). See also U.S. v. McDowell, 2010 WL 5288149 (D.Kan. 2010) (testimony of Nicely unpersuasive, partially because his behavioral science approach was “not widely recognized”); U.S. v. Gutierrez-Ruiz, 2011 WL 129632 (D.Utah 2011) (court credited other witnesses and concluded dog was not cued, despite testimony of Nicely).
 U.S. V. Sharp, 2009 WL 4348473 (E.D. Tenn. 2009).
 See also U.S. v. $80,760 in U.S. Currency, 781 F.Supp. 462, 478 (N.D. Tex. 1991) (“Reliability problems arise when the dog receives poor training, has an inconsistent record, searches for narcotics in conditions without reliability controls or receives cues from its handler.”). See Katz, S.R., and Midkiff, C.R. (2008). Unconfirmed Canine Accelerant Detection: A Reliability Issue in Court. Journal of Forensic Sciences, 43(2), 329-333.
 U.S. v. Salas-Torres, 60 F.3d 837, 1995 WL 406937 (10 th Cir. 1995) (search after alert revealed soap powder spread throughout trunk and blocks of wood in the springs. “Testimony at trial indicated that smugglers commonly use soap powder to mask the odor of marijuana and will place blocks in the springs of a heavily-laden car to keep it from riding conspicuously low”; when released from checkpoint, officers followed the driver because he drove in the direction from which he had come; the driver met someone pulling a motor home on a frontage road; search of the motor home revealed 82 pounds of marijuana). See also Maryland v. Cabral, 159 Md.App. 354, 381, 859 A.2d 285 (Ct. App. 2004).
 Harris v. Florida , 2011 WL 1496470 (2011).
 In U.S. v. Patten, 183 F.3d 1190 (10 th Cir. 1999), a dog trained to detect meth had alerted to ephedrine. The Tenth Circuit recognized the problem of alerting to a precursor to meth, but found there was sufficient other evidence for the arrest and did not delve into the issue; Patten was not mentioned in Harris.
 See Mesloh, C., Henych, M., and Wolf, R. (2002). Utilization of the Law Enforcement Canine in the Seizure of Paper Currency. Journal of Forensic Identification, 52(6) (“[H]andler error occurs when the handler fails to properly interpret the behavior of the dog, leading to misidentification. The response bias on the part of the handler varies as some are more conservative or liberal with their interpretations.”).
 Kubinyi, E., Miklosi, A., Topal, J., and Csanyi, V. (2003). Dogs (Canis familiaris) Learn from Their Owners via Observation in a Manipulation Task. Journal of Comparative Psychology, 117(2), 156-165. See also Topal, J., Byrne, R.W., Miklosi, A., and Csanyi, V. (2006). Reproducing Human Actions and Action Sequences: “Do as I Do!” in a Dog. Animal Cognition, 9, 355-367.
 Landau, M.B. 2007. Creating a Template of Nonverbal Cues for Immobile Recipients to Use in Communicating with Service Dogs. Northeastern University. Senior Project.
 Viranyi, Z., Topal, J., Gacsi, M., Miklosi, A., and Csanyi, V. (2004). Dogs Respond Appropriately to Cues of Humans’ Attentional Focus. Behavioural Processes, 66, 161-72.
 Helton, W.S. (2005). Expertise in Trained Dogs. Proceedings of the Cognitive Science Society, 28, 1488-93.
 Lit, L. and Crawford, C.A. 2006. Effects of Training Paradigms on Search Dog Performance. Applied Animal Behaviour Science, 98, 277-292.
 See, e.g., Curran, A.M., Prada, P.A., and Furton, K.G. (2010). Canine Human Scent Identifications with Post-Blast Debris Collected from Improvised Explosive Devices. Forensic Science International 1999, 103-108.
 Heller, D.P., Furton, K.G., Harper, R.J., and ALmirall, J.R. Observations and Recommendations regarding Training, Record Keeping, and Deployment of Explosive Detection Canine Teams (www.fiu.edu/~ifri/Observations
 Mesloh C, Wolf R, Henych M. (2002) Scent as forensic evidence and its relationship to the law enforcement canine. J Forensic Identification 52(2): 169-182 (citing research concluding that “dogs learn to depend upon the most abundant vapor constituents of a substance for identification of that substance. ,Therefore, they learn to identify a substance (e.g., an explosive) by using only a few compounds”; also noting that scent “is affected by the turbulence as it is diffused from the source…. The scent odor from the narcotic drifts with the air currents….”).
 See SWGDOG SC8—Substance Detector Dogs: Explosives Detection 18.104.22.168, specifying that “some searches shall contain no odors (blanks)”.
 #6 in Figure 1 of the paper.
 See Sonoda H, Kohnoe S, Yamazato T et al. (published online January 31, 2011), Colorectal Cancer Screening with Odour Material by Canine Scent Detection, Gut doi: 10.1136/gut.2010.218305 (tennis balls used as rewards for cancer-sniffing study).
 Forcing alerts has been noted in studies before. See, e.g., Curran, A.M., Prada, P.A., and Furton, K.G. (2010). Canine Human Scent Identifications with Post-Blast Debris Collected from Improvised Explosive Devices. Forensic Science International 1999, 103-108, at 108 (a study that involved videotapes).
 See, e.g., Jezierski T, Walczak M, Gorecka A (2008). Information-seeking behaviour of sniffer dogs during match-to-sample training in the scent lineup, Polish Psych. Bull. 39(2): 71-80 (“The dog can leave some saliva on the target sample during sniffing and can use it as a cue if the same jar with the cotton cloth or the cotton cloth is repeatedly used in consecutive trials.”); Hilliard S (2003) The Steel Helmet Project: canine olfactory detection of low concentrations of a surrogate chemical warfare agent, Int’l J. Comp. Psych. 16: 193-208 (boxes heavily contaminated with saliva and hair elicited more false alerts than lightly contaminated boxes; the authors argue that future detector dog investigations “should account for box contamination by employing box-reversal procedures, or limit contamination by using new boxes for each day’s testing.”)
 We thank John G. Grubbs, a former secret service agent and canine handler for this observation.
 See, e.g., U.S. v. Dix, 2007 WL 3046347 (D.Neb. 2007) (noting concerning a drug dog’s alert that “once the dog is found to be certified, he may be presumed to be well trained and reliable…. If the dog was not certified or otherwise had performance problems, that could be a circumstance for the court to consider”).
 See Florida v. Foster, 390 So.2d 469, 470 (Ct. App. 1980) (Florida appellate court noting that probable cause requires that the dog be trained, but emphasizing the dog’s “track record” is also important, with emphasis “placed on the amount of false alerts or mistakes the dog has furnished”; failure to produce track record resulted in determination that the prosecution had not met its burden of establishing probable cause for the search).
 U.S. v. Place, 462 U.S. 696, 103 S.Ct. 2637, 77 L.Ed.2d 110 (1983) (canine sniff as “sui generis”); Illinois v. Caballes, 543 U.S. 405, 125 S.Ct. 834, 160 L.Ed.2d 842 (2005) (sniff during legitimate traffic stop does not implicate Fourth Amendment).
 See the particularly insightful analysis of the Florida Supreme Court in Harris v. Florida, 2011 WL 1496470 (2011).
 The paper has received press attention and mention as a reason why drug sweeps of schools may be invalid. See “Comment: Secrecy Surrounding Sweeps Reflects Badly on Schools and Police.” GrossePointToday.com, April 18, 2011 ( http://grossepointetoday.com/news/comment/001883-comment-secrecy-surrounding-sweepsreflects-badly-schools-and-police ).