Training of Castaway
Contacting “Castaway”: Training a Wild, Deaf, Offshore Atlantic bottlenose Dolphin (Tursiops truncatus)
Sarah C. Sayre1, Nancy Cooper1 , Stacey J. Anderson2, Jill Richardson1,2
1. Dolphins Plus Inc., 31 Corinne Place, Key Largo, FL 33037 (education@dolphinsplus.com)
2. Marine Mammal Conservancy, P.O. Box 1625, Key Largo, FL 33037 (rgl@marinemammalconservancy.org)
History
Castaway, a deaf, visually impaired, neurologically challenged, offshore Atlantic Bottlenose dolphin (Tursiops truncatus) was found stranded on Castaway Cove beach, Vero Beach (Florida, USA) on 11, November 2006. Harbor Branch Oceanographic Institute (HBOI) responded in situ, and she was subsequently transported to Mote Marine Laboratory (Mote) in Sarasota, Florida. After 79 days of rehabilitation, four unsuccessful release attempts on 30, January 2007 culminated in her transport to Marine Mammal Conservancy (MMC) in Key Largo, Florida. Castaway remained at MMC for approximately six months, where she was housed in various Florida Bay, natural lagoon settings (depth and size ranges: 3 – 6 m depth; 165 m2 – 3300 m2 enclosed area) and isolated from conspecifics.
Two days after Castaway arrived at MMC, a remarkable lack of response to auditory stimuli (e.g. “pingers” and tapping a partially submerged stainless steel bucket) was noted. Suspicion of her deafness was confirmed via two passive auditory tests conducted by Dr. David Mann (University of South Florida) on 18, February 2007. In addition to her hearing deficits, Castaway’s left eye displayed mild to moderate impairment, and possible neurological deficiencies were noted. Castaway was remarkably maladroit, with weak gross motor coordination, lacked interest in or focus on environmental stimuli, and appeared unable to recognize patterns. Though her capacity for response improved slowly over the course of six months at MMC, her anomalous mannerisms and limitations presented many training and enrichment challenges.
Castaway was transported to Dolphins Plus on 25 July 2007, where she currently resides with two adult female conspecifics in a natural lagoon setting (approximate depth and area: 4.5 m and 388 m2).
Training
While at MMC, Castaway’s inability to hear, compounded by her solitary confinement, inertness, and deficient attention span, had a large impact on her training in both formal and informal sessions. At the commencement of her training regimen, primary reinforcement was established (i.e. herring) as the predominant training tool. In general, her training sessions at MMC were short, and the criteria for success were low. After she moved to Dolphins Plus, her focus improved dramatically, and thus the training criteria were much more stringent.
Targeting
In order to acquire Castaway to station for feeds at MMC, fist targeting was implemented immediately. Presentations of a human fist were rigorous to ensure detection, and general orienting was immediately rewarding with an Atlantic herring (Clupea harengus) underwater. Within three to five sessions (minimum of 15 trials session-1), Castaway was performing the response with approximately 100% success. However, Castaway was still under rehabilitative care, so the criteria for targeting were broad. Due to the large size of Castaway’s lagoon, a physical approach by staff to acquire and guide her to station was necessary. Though Castaway’s feeding station and times were consistent over six months, she never anticipated a feed while at MMC.
Once fist targeting was established, the transition to a target pole was virtually effortless. The presentation of the target pole was paired with a fist and contact with both elicited primary reinforcement. After the first pairing, no other approximations were necessary, because Castaway was targeting on the pole upon presentation with 100% success.
Recall
The laborious exercise of entering Castaway’s enclosure at MMC to retrieve her to station for feeding resulted in the implementation of a strobe light for recall purposes. Castaway’s inability to hear precluded the use of traditional auditory recall methods, so the strobe light was substituted as a visual cue. The strobe was initially presented within visual range and contact with the light source was paired with primary reinforcement. Shaping involved the presentation of the light at various distances (approximate range = 1.5 – 9 m) and orientation towards the strobe eventually developed into a directed approach. Though the strobe proved to be an effective recall, it was limited in range (i.e. required proximate presentation), and its effectiveness was heavily dependent upon the physical conditions of her natural lagoon (e.g. turbidity). Thus, the necessity of entering the water to station and recall Castaway remained unavoidable.
When Castaway was transferred to Dolphins Plus, an attempt was made to solve the problems associated with a “long-distance” recall by employing the use of a toy water cannon. This was implemented, because Castaway was noted to respond to proximate water displacement or disruption while at MMC. In order for the water cannon to be effective, the splash needed to fall within her visual range, or contact her directly. Castaway’s orientation towards the splash was consistently rewarded with primary reinforcement, and this application remains in place as the most effective recall method at Dolphins Plus to date.
Pattern Recognition
While at MMC, Castaway’s ability to recognize patterns, and thereby exhibit an anticipatory response to a repeated sequence, was tested via rudimentary A to B feeding trials. Two feeders were separated by approximately 0.5 m at her feeding station. Each feeder paired the presentation of the strobe light in close range of her eye with primary reinforcement, alternating sides of the bilateral axis. The hypothesis was that Castaway would eventually anticipate the presentation of the strobe and reward by swinging her head side to side. This protocol was replicated over the course of two weeks, up to seven times day-1, and no anticipation was noted (multiple observers).
Castaway’s capacity to recognize patterns was re-tested at Dolphins Plus in October 2007. Trials (n = 9) were designed to replicate the MMC experiment, using two trainers separated by 1.2m at Castaway’s feeding station. In place of a strobe, each trainer alternated slapping the surface of the water, followed by the presentation of primary reinforcement. Each trial involved 40 presentations, and the response time was limited to five seconds. After the last presentation, both trainers remained motionless, and Castaway’s response was recorded. Castaway exhibited no response in three trials and an opposite response in one. However, she successfully anticipated the pattern in five of the nine trials. Thus, this experiment was a verification of her improved cognitive condition.
Vertical Positioning
Once stationing was achieved, Castaway was trained to accept food in a vertical position, with her head and eyes above the water. This proved to be a very arduous task. Castaway initially exhibited a reluctance or inability (reasoning is unclear) to lift her head above the water. While at MMC, and during her first month at Dolphins Plus, Castaway almost never acknowledged above-water activity, and she was never observed craning, spy-hopping, or bowing when this training commenced. Due to the fact that an appropriate bridge had not yet been established, primary reinforcement was used to convey success. The target pole was utilized as a tool, and slight approximations were “jack-potted” to impart this seemingly difficult concept. Her progress during this initial stage of training was painstakingly slow, and consistent, focused training for two and a half weeks was necessary to achieve rostral contact with the pole presented just above the surface of the water. Once contact with the target pole, while in a vertical position, was established as a successful response, regular increases in the distance of the target pole from the surface of the water resulted in 100% success of the targeted behavior in four days.
The Bridge
Once at Dolphins Plus, the development of a non-auditory bridge was crucial to Castaway’s development. The coastal Atlantic bottlenose dolphin population at Dolphins Plus is conditioned to respond to athletic whistles as an auditory bridge. These whistles are very effective in terms of dolphin response and are also utilized to convey the termination of behavior to guest swimmers. Consideration of an appropriate non-auditory bridge for Castaway included the use of pneumatics, a board or paddle, and a pointing gesture. The point bridge has been utilized with great success in other marine mammals (e.g. polar bears, Belting et. al 1998; sea lions, MaGee 2007, personal communication) but proved to be a very difficult application with Castaway. The point bridge involved the pairing of an exaggerated, direct, finger point towards Castaway’s dominant, right eye with primary reinforcement. Castaway’s learning curve was extremely slow, and four months of consistent replication was required to obtain an 85% success rate. Castaway’s biggest impediment to improvement was what appeared to be signs of short-term memory loss. For example, when she appeared to understand the concept of the point bridge while at station, she could not apply that concept to other behaviors. Thus, the bridge training had to be re-instated with every new behavior. Plausible theories for Castaway’s deficient/impaired ability to learn include her inability to hear, neurological impairment, and her classification as an offshore bottlenose dolphin.
There are inherent limitations of a point bridge, predominantly the necessity of proximate application. Castaway must be within close range of the presentation of the point bridge and must be looking at her trainer in order for it to be effective. Thus, capturing behaviors at appropriate or apex moments is very difficult. A vast majority of auditory bridges function at relatively long distances between the trainer and the animal and are also omni-directional. Thus, the necessity of utilizing visual cues undeniably puts Castaway at a learning disadvantage.
Secondary Reinforcement
While at MMC, Castaway presented as an overtly tactile animal and was frequently observed rubbing fences, platforms, limestone rocks, and ropes. Thus, a scrub brush was applied as secondary reinforcement. Vigorous scrubbing from her trainers was utilized as a reward for remaining at station and completing various tasks. The limitations of using only primary reinforcement with Castaway involved her desire to minimize the time between performing a response and the delivery of food. Thus, the use of the scrub brush as a secondary reinforcer was pivotal in Castaway’s training, because she learned to hold requested responses and positions (e.g. husbandry behaviors) for longer periods of time. This aspect of her training resulted in a dramatic improvement in her ability to learn behaviors, including moving rubs and dorsal and pectoral presentations.
The effectiveness and “value” of the scrub brush as a secondary reinforcer was stimulated by pairing scrubbing with trainer swim interactions. These interactions eventually became more reinforcing to Castaway than primary reinforcement. Her trainers entered the water after successful training sessions, wearing mask, fins, and snorkel, and essentially “scrubbed” her. This interaction became a tool to entice a very sedentary animal to exercise, because she was eventually required to actively swim to receive a brush-down. To date, this appears to be Castaway’s preferred activity and serves as a great relationship-builder between Castaway and her trainers.
Enrichment
While at MMC, environmental enrichment devices (EEDs), such as hula-hoops, balls, and bath toys, initially elicited little to no response from Castaway, with the exception of the rub rope. The activities that consistently prompted a response from Castaway were trainer swim interactions. Thus, while at MMC, staff swims with Castaway were encouraged, accounting for three to eight swims day-1 for one month. This activity provided a source of enrichment and prepared Castaway for the Dolphins Plus public natural swim program. During a natural swim, the dolphins are not at control and are not offered any reinforcement beyond that received from human-dolphin swim interactions. Though Castaway’s response to individuals other than her trainers was slow at both MMC and Dolphins Plus, interactions between staff were encouraged with a fluke-tow. During the tow, trainers pulled Castaway backwards by her flukes for a few seconds, which elicited an active swimming response upon release. Eventually, the fluke-tow was phased out, and Castaway actively sought out swimmers, frequently engaging in side-by-side swim interactions.
Behavioral Inventory
With the application of the point bridge, and the use of the scrub brush as an effective secondary reinforcer, Castaway accomplished the following behaviors with an 85% success rate over the course of three months at Dolphins Plus: open/close mouth, tongue presentation, eye cup, high and low targets, blowing bubbles, toy retrieval (e.g. ball, ring, propeller), bilateral pectoral presentation, short distance A to Bs, dorsal presentation, and moving dorsal rubs. These behaviors were trained using shaping and capturing. This learning “explosion” is attributed in part to the possibility of improved neurological function and hearing and increased social stimulation.
Acknowledgements
We would like to acknowledge the founder and president of the Marine Mammal Conservancy, Robert Lingenfelser, for his dedication to Castaway’s rehabilitation and well-being. Staff and volunteer veterinarian Robert Stevens was crucial in maintaining Castaway’s health and remains a consistent source of support and guidance. We would like to extend our deepest gratitude to all of the MMC volunteers and Dolphins Plus staff for their hard work and participation with care and training, most notably Art Cooper, MMC Board Chairman and Dolphins Plus Vice President and Curator. Funding for this work and Castaway’s care was provided by Dolphins Plus, Inc.
References
Belting, T., Joseph J., Messermith M. & Nordone L. (1998) Teaching an old polar bear program new tricks. Soundings 23(2): 12-14