Operant Treatment Of Separation-Related Problem Behaviors



Erica N. Feuerbacher, MS, CPDT-KA
University of Florida
Department of Psychology
945 Center Drive
P.O. Box 112250
Gainesville, FL 32611-2250
Doctoral Advisor: Clive D. L. Wynne

Kristy Muir
Kristy Muir MS, CPDT-KSA
Animal Behavior Training Solutions, Inc.

Separation related problem behaviors, such as excessive vocalization, defecation/urination, and destruction are a common problem in owned dogs and is a common cause of relinquishment of dogs to shelters (Bailey, 1991). We propose a novel approach to treating separation related behavior problems. First, we propose an operant approach to treating separation disorders. Traditional behavioral techniques use a counter conditioning and desensitization treatment. We hypothesize that owner return is a reinforcer for the dog's behavior and can be used to shape behavior.  We also hypothesize that, by making owner return (or possibly food delivery) contingent on desirable behavior, training will be facilitated and the dog's behavior will improve faster than with more traditional techniques. We will video record the dog's behavior in baseline and treatment phases. Training will entail one of two conditions. Owner return: the owner will enter the room contingent on the dog's calm response that is incompatible with behaviors observed in baseline. Remote food: a remote dispenser will deliver food as a reinforcer contingent on calm behavior. Using these potential reinforcers, we will attempt to shape an incompatible behavior over longer durations of owner absence. We will video all sessions and will use the dog's behavior as a direct measure of treatment efficacy, rather than the traditional indirect method of owner perception of behavioral improvement. If successful, our research will offer practitioners a novel treatment as well as greater understanding of behavioral change during treatment through direct measurement of the dog's behavior.

Separation related problem behaviors in dogs is a severe behavioral disorder often resulting in referral to a behaviorist, administration of medication, or even surrender of the dog to a shelter (Bailey, 1991). After aggression, it is the most frequently reported behavioral problem of dogs that have been referred to a behavioral clinic (Takeuchi, Houpt, & Scarlet, 2000). Estimates of prevalence of separation-related problem behaviors were assumed to be 15% of the owned dog population (Guthrie, 1999), but prevalence reached over 18% in dogs adopted from a New York shelter (Spain, Scarlet, & Houpt, 2004). While high rates in shelter dogs might not be surprising, the prevalence of this problem was above 20% in a longitudinal study of border collie and labrador retrievers which were reared by breeders and did not pass through a shelter (Bradshaw, McPherson, Casey, & Larter, 2002), indicating this is not restricted to shelter dogs.

Common problem behaviors reported as separation related are excessive barking or whining, excessive salivation, property destruction and possible concomitant injury to the dog, and improper elimination when the dog is separated from its owner. To date, most research on separation related problem behavior has focused on pharmacological interventions paired with behavioral interventions, for example clomipramine (King, et al., 2000), fluoxetine (Ibáñez & Anzola, 2009; Landsberg, et al. 2008), and diazepam (Ibáñez & Anzola). The emphasis in these studies, however, has typically been assessing the efficacy of the administered drug, rather than the behavioral intervention. In each study, either all dogs were supposed to receive behavioral therapy while receiving the drug or all dogs did not receive behavioral therapy while receiving the dog. In either case, the independent variable between groups was the administration of the drug, not the behavioral therapy. With medication, most dogs showed improvement, but not all dogs and not complete behavioral improvement. Thus, the need for effective behavioral treatments remains. Even given the utility of medications, some owners are averse to using pharmacological interventions for treatment or potentially as long-term treatment. Additionally, an effective behavioral treatment would augment any benefits from pharmacological treatment as well.

Only two studies have addressed strictly behavioral treatments for separation related problem behaviors and focused on the efficacy of these treatments without the confound of simultaneous drug administration. Hothersall and Tuber (1979) described their procedure as desensitization and counter conditioning in which the dog was gradually exposed to increasing levels of the anxiety producing stimulus (owner departure and absence). However, the description of their procedures more closely approaches an operant procedure in which calm behavior is reinforced by owner presence, "our aim is to establish a strong positive association between the owner's departure, contrived good behavior by the dog, and the owner's return which culminates in an extraordinary reward period for the dog" (p.244). More recently, Butler, Sargisson, and Elliffe (2011) investigated a combination of counterconditioning and systematic desensitization for treating separation related problem behaviors. In the systematic desensitization component, owners were to leave the dogs for 5 min and then return; if the dog showed no signs of problem behavior, owners could increase the length of absence by an additional 5 min, but had to return to the last duration of no occurrence of problem behavior if the dog showed signs of separation distress. How the owners proceeded (the rate of duration increase) was unmoderated, however, and might have accounted for differential outcomes in treatment efficacy. The counter conditioning involved giving the dogs treats as the owner departed, as well as a punishment component in which the owners ignored the dogs for 30 min upon owner return if there were signs of separation related problem behaviors.  The authors concluded that systematic desensitization was the critical component for successful treatment; counter conditioning and other behavioral advice did not influence the behavioral outcome. 

The part of the treatment that was useful in Butler et al.'s treatment (2011) was the systematic desensitization in which the owner would return after an increasingly long duration.  This is also suggestive of an operant procedure in which the owner's return could function as a reinforcer and reinforce the dog's behavior—if the dog was not engaging in problem behaviors this would inadvertently reinforce more desirable behaviors. If this procedure were made more explicitly operant (i.e., owner return is contingent on desirable behavior, could we produce faster and more stable behavioral improvement? Snider (2007) demonstrated that treatment of aggression through an explicitly operant procedure (shaping a calm, alternative behavior) was faster (hours to effective treatment) than typical desensitization and counter conditioning treatments (weeks or months to effective treatment).
We propose to investigate the efficacy of an explicitly operant treatment procedure for separation-related problem behavior.  In our procedure, we will shape desirable behavior (e.g., sitting or lying calmly) and deliver a reinforcer (owner return or remote delivery of food) contingent on that behavior, whereas typical treatments have not made consequences contingent on behavior (e.g., owner gives treats to dogs as she leaves regardless of dog's behavior and owner returns and enters the house after a set amount of time, regardless of the dog's behavior). We predict that this will facilitate the dog's acquisition of desirable behavior and make behavior change faster and more complete than traditional desensitization and counter conditioning procedures. We also predict that an operant approach will be effective for all dogs as long as an appropriate reinforcer is identified and delivered. We will investigate this procedure using two different reinforcers: owner return and food.

First, we hypothesize that owner return functions as a reinforcer for dog behavior and, in fact, might be the reinforcer that has inadvertently shaped up and maintained separation related problem behavior. Thus we will first explore the possibility of shaping calm, appropriate behavior by having the owner re-enter the house or room contingent on calm behavior. If this is successful, we will gradually increase the duration of separation; that is, the dog must engage in an alternative behavior for longer amounts of time before the reinforcer (owner return) is delivered.

Second, we propose to investigate the possibility of using food as a reinforcer to shape incompatible behavior during owner-dog separations.  Although food delivery is not likely the reinforcer that has produced or maintained the problem behaviors, it is a potent reinforcer for dogs (Feuerbacher & Wynne, 2012) and can readily bring about behavior change in dogs. Additionally, food can be easily delivered via remote device (e.g., MannersMinder©) potentially allowing for owners to remotely reinforce their dog's desirable behavior throughout the day, even when the owner is not actually returning home. For example, the owner could program a KONG© Time device to release a food filled KONG© in the owner's absence to reinforce calm behavior. If successful it might also be a way for trainers to improve a dog's separation behavior even when the owner is unavailable for training sessions. In our study, after training the dog to retrieve food from a MannersMinder©, we will begin shaping a calm, alternative behavior using the MannersMinder© to deliver the food consequence.  As with owner return as the reinforcer, we will gradually shape the dog to engage in an alternative behavior for longer and longer durations.  We predict that this might be an effective treatment for some, but not all dogs, as some dogs with severe separation problems will not eat when the owner is absent. 

Our research will investigate a novel treatment for separation related problem behaviors and, if successful, will give a more explicit procedure for practitioners to implement in their behavioral practices. Additionally, if we are successful in shaping alternative behaviors using owner return as the reinforcer, this might offer insight into how separation problems arise (poorly timed owner return or attention, such as returning to the house if the owner hears the dog scratching).  Additionally, it might offer a preventative treatment plan for owners newly acquiring a dog. By implementing this procedure from the beginning, owners might prevent separation related problem behaviors from arising in the first place.

To make our reinforcer delivery contingent on appropriate behavior we will use a webcam to monitor the dog's behavior. This addresses two problems characteristic of the previously reported studies on separation related problem behavior: fidelity to the treatment plan and how successful treatment is assessed. First, the behavioral therapy in these studies consisted of instructions given to owners but there was no independent, direct observation of the fidelity of the owner's implementation of the behavioral therapy. At most, owners were asked about whether they were implementing the behavioral treatment, but no independent assessment was made as to whether the owners were correctly implementing the behavioral treatment.  Improper implementation, or faulty self-reporting of the fidelity in implementing the procedure might have produced the variable outcomes reported in the published studies. When assessing the utility of a behavioral treatment, it would be useful to have direct measures that the owners implemented the plan as detailed, otherwise the actual efficacy of the treatment cannot be determined. Lack of success might not be a fault with the procedure but the fidelity with which it was implemented. Some studies (e.g., Butler, Sargisson, & White, 2011) did have owners report durations of absences, delivery of treats but owners might improve reports to appear to be implementing the plan as detailed, or owners might be able to correctly verbally indicate the procedure but unknowingly change critical features when implementing the plan (e.g., when to deliver treats). In our study, one of the authors will be present to coach the owner to implement the procedure correctly and we will be able to detect and correct any errors.

The second measure of concern in the previous studies on separation related problem behavior is the outcome measure for successful treatment: owner reports on improvement was the main dependent measure (Butler, Sargisson, & White, 2011, made a slight improvement by also having an independent observer make occasioanal spot checks for destruction produced by the dog). However, given the problems with accuracy of owner reports to begin with, and because some separation related problem behaviors might occur without the owner perceiving them (e.g., excessive whining or barking) relying solely owner reports might inherently bias the data to report a lower incidence of problem behavior (Palestrini, et al. 2010). In their longitudinal study, Bradshaw, McPherson, Casey and Larter (2002) reported that excessive vocalizations were the first separation related problem behavior to appear in the lifespan of the dog and only later did destruction begin to appear as a problem behavior. Thus, correctly identifying and treating vocalization might be a crucial component of successful treatment. Additionally, owners might report improvement in the dog's behavior regardless of actual improvement as humans are prone to over-report improvement during treatment (Lilienfeld, Lynn, Namy, & Woolf, 2008). Again, a direct measure (i.e., direct observation of the dog's behavior during separation) seems warranted to correctly determine the efficacy of treatment and to learn about the trajectory of behavioral improvement. Our study will directly observe and measure the dog's behavior and we can thus directly assess the efficacy of our treatment.  Additionally, how the behavior of a dog that shows separation related problem behavior changes with treatment (both within and across sessions) has not been documented. Understanding how quickly behavior changes, whether there are any hurdles (e.g., specific durations) that are particularly problematic during treatment will allow practitioners to be better prepared to set their clients up for success during treatment. 

Thus, given the state of the art for treating separation related problem behavior, there is space for more precise and effective behavioral interventions, as well as a lack of direct data on the implementation and efficacy of treatment. We are proposing to fill these gaps by investigating a new treatment plan for separation related problem behaviors and in so doing take direct measures of the implementation of treatment, and direct measures of the dog's behavior to determine efficacy of treatment.

General Procedures


We will first collect baseline data on each dog's separation related problem behavior. We will use a webcam set up in a discreet spot in the owner's house where the owner leaves the dog during separations. We will first record the dog's behavior during separation from the owner for 30 min. Palestrini et al. (2010) found that most dogs displayed problem behaviors within 10 min of owner absence. This will allow us to determine the exact topographies of behavior that the dog emits (e.g., excessive vocalization, destruction), including duration and frequency, and the latency from owner departure to initiation of problem behavior. It will also allow us to determine appropriate alternative behaviors to train in the shaping phase. After collecting this 30 min. After assessing the baseline behavior, we will begin implementing the treatment phase.

Treatment Phase:

In both treatment phases, the owner will always give the dog a novel safety cue (a discriminative stimulus that predicts training) before leaving the room and will not use this cue outside of training until training is complete. This cue is intended to help the dog discriminate between the training sessions in which new behavior is being trained, and when the owner leaves during the normal course of the day to prevent long owner absences from impede training.  Once training is complete, the owner will use this cue during normal absences and this cue should control the dog's appropriate behavior.

Owner Return Condition

In the owner return condition, we will begin by having the owner give the dog the safety cue and then exit the room. The experimenter will monitor the dog's behavior via webcam and, as soon as the dog emits a calm behavior incompatible with problem behavior, the owner will re-enter the room and sit down. The owner will interact normally with the dog if the dog is within approximately 1 m or remain neutral if the dog is farther away than 1 m. The owner will remain in the room for 1 min before a subsequent trial is started. Once the dog emits no problem behavior for three consecutive trials, the owner will delay returning to the room for an additional 5 s after the dog emits an incompatible behavior. This criterion will continue as follows: three consecutive successful trials and the owner will add another 5 s to the delay before return. This will continue until the duration is 1 min, at which point 15 s will be added to each step in duration. When the dog can remain alone successfully for 5 min durations, 30 s will be added per step.  Finally, at 10 min of successful duration, we will add 1 min per step. These are general guidelines only, and we will adjust them as necessary if the dog fails too often on those step sizes.

If, at any duration the dog shows problem behavior, the duration of owner absence will be reduced to the last successful duration step.

If the dog begins to show problem behavior before the owner leaves the room, we will begin shaping by having the owner walk towards the door, but return to the chair and interact as detailed above for 1 min contingent on calm behavior. After 1 min, the owner walk again towards the door. If the owner can walk to the same place three times in a row without the dog showing problem behavior (and has been reinforced for emitting incompatible behaviors) the owner will walk 5 feet closer to the door in the next trial.  This will continue until the owner can leave the room without the dog emitting problem behavior.
If the owner entry seems too delayed compared to when the dog emits the desirable behavior, we might add a tone that we can remotely play for the dog at the moment the dog emits the desirable behavior and this tone will predict owner return (i.e., it will function as a "clicker" to mark the behavior and predict reinforcer delivery).

Remote Food Delivery

We will begin by training the dog to retrieve food from a MannersMinder©.  Once the dog has learned to access food from the MannersMinder©, we will begin shaping the dog's behavior. We will have the owner give the safety cue and leave the room. The experimenter will monitor the dog's behavior via webcam and, as soon as the dog emits a calm behavior incompatible with problem behavior, the experimenter will trigger the remote feeder to deliver food. Once the dog has finished eating and emits an incompatible behavior, the experimenter will again trigger the remote feeder.  Once the dog emits no problem behavior for three consecutive trials, the experimenter will begin to delay food delivery. The duration increases will parallel those in the owner return condition.

If the dog begins to show problem behavior before the owner leaves the room, we will begin shaping by having the owner walk towards the door and we will deliver food remotely for the dog's calm behavior. We will follow the same criteria for decreasing proximity to the door as in the owner return condition.


Our proposed research has direct implications for providing new tools to practitioners to treat separation related problem behaviors. First, if our treatment plan is successful, we will provide practitioners with an operantly based treatment for separation disorders. Potentially, two different reinforcers (owner return and food) might be effective, in which case this would allow practitioners to opt for the reinforcer most potent to the dog and/or most convenient for training. If training with food is possible, this might allow trainers to improve the behavior dogs with separation disorders even while the owner is not present, potentially making training for flexible. We predict that an operant procedure will expedite and facilitate training as the behavior change is explicitly reinforced; additionally, many trainers are more comfortable with delivering reinforcers contingent on appropriate behavior reinforcers and thus might feel more comfortable using this technique, compared to noncontingent techniques (desensitization and counter conditioning). We hope that such a treatment might be easily implemented by both well-versed practitioners and naïve owners. If successful, in future studies we can study the relative ease of implementation of an operant versus systematic desensitization procedure for owners.

Additionally, because we will video record the dogs' behavior, we will be able to accurately detect behavioral changes and be in a position to indicate how quickly behavior change occurs. This will be useful to practitioners who could then tell their clients an estimate of treatment time, potential behavioral pitfalls they might have to navigate around, and the likely success of the treatment. Treatment success will not be a measure simply of owner perception. Although client satisfaction is necessary, practitioners are largely concerned with effective treatment for the dog itself and, likely, client satisfaction will come along with effective and humane treatments.

We will disseminate the results of our works by producing a manuscript for hopeful publication, by attending and presenting the results and academic and practitioner conferences (e.g., APDT, Association of Behavior Analysis International, and Florida Association of Behavior Analysis). We will also, whenever possible, present our work at workshops.


We will begin recruiting dogs in Spring 2013 through the behavioral practice of the second investigator. We plan to treat a total of 12 dogs: six dogs in each condition (owner return condition and remote food delivery condition). Although it will depend on individual dogs, we estimate that each dog will require a 2 hour baseline session in which we collect dog data from the owner and run the baseline session. Following this session, we estimate conducting two 90 min training sessions with each dog, with a maximum of 4 sessions if criterion is not met in the first two sessions. We will end sessions after 90 min to prevent behavior change being due to fatigue.

We expect to run approximately 2 sessions per week such that we would complete data collection within approximately 30 weeks. We will have research assistants that help with research for course credit code the video sessions within one week of the session being run, such that we should have the videos coded shortly after collection. The remaining time can be spent on data analysis and allows a small buffer for extra data collection if a dog drops out of the study midway through and we need to work with an additional dog. We expect the project to be completed with the data analyzed and the results available within 1 year.

We will assess progress in terms of several factors: Short-term progress: 1) number of sessions we run per week with our target average 2 sessions per week and 2) the amount of video coded on a weekly basis. Long-term progress: number of dogs that are currently undergoing or have completed treatment.

We will assess our short-term progress on a weekly basis, and our long-term progress on a monthly basis. 


We will use the funds to purchase equipment to film the dogs during treatment, deliver treats remotely, and the treats themselves. We will use the remaining money to pay the experimenters salary during treatment sessions.

4 camera webcam DVR with harddrive $280 1 $280
MannerMinder $98 2 $196
Treats (Bil Jac Little Jacs) $8 10 $80
  $15 60 $900
TOTAL     $1456


We will use the video analysis of each individual dog's behavior to determine the success of the treatment.  We conclude that the treatment is successful if 1) the frequency and duration of separation related problem behaviors decreases across sessions or is completely eliminated while the dog is separated from the owner and 2) the dog can remain alone without emitting a problem behavior for a duration equal to 3 times the time until the dog first showed problem behavior in baseline (e.g., if the dog first showed problem behavior after a 6 min absence, this duration would be 18 min) or 15 min, whichever is longer. We will assess treatment efficacy on by assessing the outcome of each individual dog rather than pooling data. We will report number of dogs successfully treated as well as individual dog's trends in frequency and duration of problem behavior, as well as maximum duration each dog can remain alone without emitting a problem behavior. If not all dogs show improvement in their behavior or reach criterion, we will conduct a post hoc analysis to identify possible factors that predict successful treatment.


Erica Feuerbacher is a doctoral candidate at the University of Florida. She is in the Canine Cognition and Behavior Lab under the advisorship of Dr. Clive D. L. Wynne.  She earned her Master's of Science in Behavior Analysis from the University of North Texas, where she studied concept formation in dogs (thesis), as well as effects of the click:food ratio on dog behavior. At UF she studies effects of human social interactions on dog behavior, and has published papers on dogs. She has taught Principles of Behavior Analysis at UF and UNT, and is adjunct faculty at Santa Fe College, teaching General Psychology. Prior to graduate school, she worked as an animal behaviorist at Peninsula Humane Society & SPCA in San Mateo, CA, Training and Programs Manager at A DOG'S LIFE Doggy Daycare in Palo Alto, CA, and ran a private training and behavior practice teaching obedience and agility, and treating behavioral problems.

Feuerbacher, E. N. & Wynne, C. D. L. (2012). Relative efficacy of human social interaction and food as reinforcers for domestic dogs and hand-reared wolves. Journal of the Experimental Analysis of Behavior, 98(1), 105-129.

Feuerbacher E. N. & Wynne, C. D. L. (2011).  A history of dogs as subjects in North American experimental psychological research. Comparative Cognition & Behavior Reviews, 6, 46-71.


Bailey, G. (1991). Parting with a Pet Survey. Burford, UK: Blue Cross Publication.

Bradshaw, J.W.S., McPherson, J.A., Casey, R.A., & Larter, I.S., (2002). Aetiology of separation-related behaviour in domestic dogs. Veterinary Record, 151, 43–46.

Feuerbacher, E. N. & Wynne, C. D. L. (2012). Relative efficacy of human social interaction and food as reinforcers for domestic dogs and hand-reared wolves. Journal of the Experimental Analysis of Behavior, 98(1), 105-129.

Guthrie, A. (1999). Dogs behaving badly—canine separation disorder research. Veterinary Practice, 31, 12-13.

Hothersall, D.  & Tuber, D. S. (1979). Fears in companion dogs: Characteristics and treatment. Psychopathology in Animals: Research and Clinical Implications. New York: Academic.

Ibáñez, M., & Anzola, B. (2009). Use of fluoxetine, diazepam, and behavior modification as therapy for treatment of anxiety-related disorders in dogs. Journal of Veterinary Behavior: Clinical Applications and Research, 4(6), 223-229.

King, J.N., Simpson, B.S., Overall, K.L., Appleby, P., Pageat, C.R., Chaurand, J.P., Heath, S., Beata, C., Weiss, A.B., Muller, G., Paris, T., Bataille, B.G., Parker, J., Petit, S., & Wren, J. (2000). Treatment of separation anxiety in dogs with clomipramine: results from a prospective, randomized, double-blind, placebo-controlled, parallel-group, multicenter clinical trial. Applied Animal Behaviour Science, 67, 255–275.

Landsberg, G., Melese, P., Sherman, B., Neilson, J., Zimmerman, A., & Clarke, T. (2008). Effectiveness of fluoxetine chewable tables in the treatment of canine separation anxiety. Journal of Veterinary Behavior:  Clinical Applications and Research, 3, 12-19.

Lilienfeld, S. O., Lynn, S. J., Namy, L. L, & Woolf, N. J. (2008). Psychology: From Inquiry to Understanding. Upper Saddle River, NJ: Pearson.

Palestrini, C., Minero, M., Cannas, S., Rossi, E., & Frank, D. (2010). Video analysis of dogs with separation-related behaviors. Applied Animal Behaviour Science, 124(1), 61-67.

Snider, K. S. (2007). A constructional canine aggression treatment: using a negative reinforcement shaping procedure with dogs in home and community settings. Unpublished thesis. (Master's thesis) Retrieved from http://digital.library.unt.edu/ark:/67531/metadc5120/m2/1/high_res_d/thesis.pdf

Spain, C. V., Scarlett, J. M., & Houpt, K. A. (2004). Long-term risks and benefits of early-age gonadectomy in dogs. Journal of Applied Veterinary Medicine, 224, 380–387.

Takeuchi, Y., Houpt, K. H., & Scarlett, J. N. (2000). Evaluation of treatments for separation anxiety in dogs. Journal of Applied Veterinary Medicine, 217, 342–345.

< Return to Awarded Grants