OPERANT RESPONDING IN THE BOTTLENOSE DOLPHIN (TURSIOPS TRUNCATUS) 1

Abstract

An interesting facet of the analysis of behavior is the manner in which specific operants occur and stabilize in different species. This has been demonstrated in a wide variety of animals from fish to man (Cohen, 1969). While the dolphin has been the subject of a number of fairly exotic investigations, more basic be-havioral studies have been neglected. The present study sought to determine the ease with which a dolphin could be trained to press an underwater paddle for fish reinforcers delivered automatically , and if this response could be reliably maintained on intermittent schedules of reinforcement. The subject, a female Tursiops truncatus weighing 105 kg, had been in captivity for approximately 1.5 yr and was estimated to be 4 to 5 yr old. Throughout the period of testing the subject was maintained on its normal ration of 6.9 kg of fish per day. Approximately 2.3 kg of Columbia River smelt (Osmerus sp.) was earned during the daily testing, and an additional 2.3 kg of smelt plus 2.3 kg of Spanish Inackerel (Scomberomorus maculatus) was hand fed at the end of the day. The subject was 19 to 20 hr food deprived at the start of each test session. The subject was housed and tested in a 6 by 6 m floating pen enclosure (Steele, 1970), in Kaneohe Bay, Hawaii. A Davis Scientific Instruments Universal Feeder (Model #320) was mounted at one corner of the pen. During test sessions, a response paddle was mounted on one side of the pen 1.95 in from the feeder and sub-Illerged to a depth of 0.6 m. A push on the response paddle closed a microswitch that activated both the feeder and a reinforcer tone (700 Hz). The tone re-iliained on for the 4 sec required for the feeder to deliver one fish. Responses and reinforcements were recorded on a Rustrac Instruments four-track Event Recorder (Model #309) housed in an adjacent in-strument/observation van. The dolphin was trained to approach and press the paddle with the tip of its rostrum in a single training session by reinforcing successive approximations of the response. After shaping, the dolphin was exposed to three reinforcement schedules; continuous reinforcement (CRF) where every response was reinforced, a variable-ratio schedule (VR 6) where reinforcement occurred on the average of every sixth response, and a 'This work was sponsored by project order # P08-0383 of Naval Ordnance Systems Commlland to the Naval Undersea Research and Development Center (NUC), Hawaii Laboratory. fixed-ratio schedule (FR 38) where every thirty-eighth response was reinforced. Test sessions were initiated with the presentation of the response paddle. Sessions consisted of 60 reinforcements each and were conducted daily, five days per week. Seven sessions were run under CRF, with behavior apparently stabilized by the third session. Three extinction sessions were then administered. On the day after the CRF extinction series was completed, the subject was retrained to paddle-press before testing under the VR 6 schedule. Following 17 conditioning sessions under the VR 6 schedule, a number of extinction sessions were run. In all extinction sessions, a dual criterion of zero responses in 15 min or a single response in 20 min was employed. Following extinction under the VR 6 schedule, the animiial was again trained to paddle-press and the fixed ratio of responses to reinforcements was gradually increased (by two per session) until the animal was responding on an FR 38 schedule. Several attempts to extend the reinforcement ratio further resulted in the dolphin refusing to respond after a small number of reinforcements, therefore the animal's performance over 10 sessions on an FR 38 schedule were recorded. RESULTS AND DISCUSSION Acquisition under CRF reached a stable rate by the end of the third session, with the mean rate for the seven sessions at 6.73 responses per minute. This rate is somewhat higher than that shown in the only other reported study of free-operant feeding by the dolphin (Johnson, 1966). Johnson's lower rate (3.73) may be attributed in part to the fact that his subject was required to push a lever and retrieve fish at opposite ends of the tank. Conditioning under the VR 6 schedule resulted in a steady increase in response rate across sessions seen in Figure 1. After the first three VR 6 sessions, response rate under this schedule consistently exceeded the rate under CRF, and the mean rate for the 17 VR 6 sessions was 11.84 responses per minute. The maximum rate under VR 6 occurred in the final session when the subject responded on the average of every 2.69 sec throughout the session. The mean response rate obtained under the FR 38 schedule was substantially higher than that under VR 6 and similar to rates produced by conventional laboratory animals. Boren (1961) reported a mean response rate for the rat of approximately 30 per minute under FR 35, which is comparable to the mean rate of 33.84 obtained from the dolphin under FR 38. The 159 1972, 17, 159-160 NUMBER 2 (MARCH)