AS Psych: Fagen (elephant learning)

Psychology Being Investigated
1. Operant conditioning is learning through the consequences of our actions
2. Negative reinforcement is the removal of a negative stimulus to reward a wanted behavior
3. Positive reinforcement is the addition of a positive stimulus to reward a wanted behavior
4. A primary positive reinforcer is a reward for a behavior that fulfills a biological need
5. Secondary positive reinforcer (SPR) training is training in which a secondary reinforcer e.g. a sound marker is used then followed by the administration of a primary positive reinforcer โ†’ animals then associate specific noises with food
6. Shaping occurs when reinforcers guide behavior toward closer and closer approximations of the desired behavior
7. Behavioral chaining is linking together individual behaviors into a more complex sequences in response to cues

Background

– Concerns for captive animal welfare and keeper safety has increased interest in reward-based training over punishment
– SPR has been used successfully with several animal species (pandas and primates), but not with elephants
– SPR allows shaping captive animal’s behavior to improve health and well-being
– Purpose of using SPR in this study is to detect tuberculosis (TB) in elephants
– TB is best detected through taking a sample from the elephant’s trunk through a “trunk wash”
– However, getting elephants to perform a trunk wash correctly can be challenging, resulting in many samples being insufficient for testing

Aims

1. To investigate whether secondary positive reinforcement can be used to train free-contact, traditionally trained elephants to participate in a trunk wash to test for tuberculosis.

Method

Research Method: Observational study
Sampling technique: opportunity

Sample

– 5 elephants
– 4 juveniles, 1 adult (in her 50s)
– Housed at an elephant stable in Nepal
– No previous experience with SPR
– Traditionally trained

Housing & Care
– Went into the jungle to graze under control of manhouts from 5-7am and 1030-4pm daily
– The rest of the day: leg chained (front legs chained together) in open stalls
– Diet: fresh grasses and dhana
– Had access to river for water during grazing hours only
– This was the standard practice โ†’ no alterations were made for the purpose of this study (to prevent confounding variable)

Procedure

– Training method: SPR
– Primary reinforcer: chopped banana
– Secondary reinforcer: short whistle blow
– No elephant went >2 days without a training session
– Training took place in the morning (7:30-10:00 am) and/or afternoon (4-7:00pm)
– Elephants could choose to participate by turning or walking away from the trainer
– Manhouts (elephant trainer/keeper) were present for the safety of the trainer, but they did not speak/signal to the elephants

Actual training
– Marker-reward relationship (whistle-banana) first established

Training methods
1) Capture: waiting for the animal to naturally perform a behavior and then โ€œcapturingโ€ it by marking it with a reward
2) Lure: for non-natural behaviors, animal is “lured” into a certain body position by placing a reward in a certain place
3) Shaping: after starting either capture or lure, rewards are given for the behaviors that are the โ€œbestโ€ โ†’ guides behavior closer and closer to the goal

Behavioral tasks
1) Trunk here: to allow saline/water to be instilled into trunk
– Place end of trunk in trainer’s hand
– LURE (chopped bananas placed in trainer’s hand)
2) Trunk up: to allow saline/water to run to base of trunk
– Elephant lifts trunk upwards
– LURE (trainer lifts chopped banana above head) and SHAPING (only rewarded for the “best” behaviors)
3) Bucket: to allow placement of trunk in the bucket ready to blow
– Elephant places trunk in bucket
– LURE (chopped bananas placed at bottom of the bucket)
4) Blow: to exhale the sample for collection
– Elephant exhales through trunk into bucket
– CAPTURE (wait for elephant to naturally exhale) and SHAPING (only reward for more forceful exhales)
5) Steady: to allow the elephant to hold any position for a required amount of time
– Elephant holds the position they were previously asked to do
– SHAPING (extending the period of time the behavior was held for with reward)

– 3 other behavioral tasks (trunk down, trunk out, targeting) were discarded as they were deemed unnecessary for the trunk wash
– Each task was paired with a verbal cue (one-syllable, distinctive, no meaning in English/Nepali)
– Individual behavioral tasks were put together into small sequences using chaining
– Varying sequences were put together to form the trunk wash
– Trunk wash: trunk here โ†’ steady โ†’ trunk up โ†’ steady โ†’ bucket โ†’ blow
– The syringe and fluid were introduced incrementally using desensitization and counterconditioning methods (during the trunk-here position)
– Rewarded for tolerating the syringe touching the outside of their trunk โ†’ tolerate it inside the trunk โ†’ insert syringe
– Built tolerance of liquid from 1mL โ†’ 15mL โ†’ 60mL (amount required for trunk wash)
– Elephants were offered water to drink before the session so they wouldn’t drink the saline solution
– No time limit for training

Data Collection
– An assistant recorded
1) Length of each session (point where elephant was offered her first cue to after she performed the last cue)
2) Total number of times the elephant was given a cue/offer for behavior
3) Success rate for each behavior and sequence after session 10
– After session 10, the elephants were tested every five sessions on previously taught behaviors
– Passing score was 80% (means a pass for a behavioral sequence, and means training is complete if achieved for the whole trunk wash)

Results

Juveniles
– All 4 elephants successfully completed the trunk wash
– Took 25-30 sessions to learn the trunk wash
– Mean session duration was 12 minutes (range 10-13 mins)
– Mean total training time was 367 minutes (sessions ranged from 257-451 mins)
– Elephants 2 and 4 did not fully passed the “steady” test, but managed to pass the full trunk wash

Adult
– Did not successfully complete the trunk wash within the learning time available
– Not tested
– Including her training time in the total training time increases mean time to 378 mins

Overall
– The amount of practice required for specific behavioral tasks differed significantly (some tasks were more harder than other) e.g. “trunk here” required 295 offers vs “blow” which only required 61 offers
– Mean success rate for individual behaviors and sequences increased from 39% after 10 training sessions to 89.3% after 35 sessions

Conclusion
1. Juvenile, free-contact, traditionally trained elephants can be trained to voluntarily and reliably participate in trunk washing using SPR training techniques only.

Ethical Issues
โœ” Housing – elephants were housed in open stalls and only their front legs were chained so they could still move about
โœ” Given food and water daily, as per normal
โœ” Reward, deprivation, adversive stimuli – chopped bananas were used to reward a desired behavior rather than punishment
โœ” Pain, suffering, distress – elephants could choose whether or not to participate by turning or walking away from the trainer, so they were not psychologically distressed by being made to participate when they did not want to

Application to Daily Life

– Zoos can use SPR techniques to train elephants to participate in a trunk wash to test for TB
– SPR can also be used to train animals to perform other desired behaviors/tasks (e.g. blood tests), instead of using punishment, thus promoting captive animal welfare

Scroll to Top