seaturtle.org : MTN : ARCHIVES : Sign In

Marine Turtle Newsletter 75:4-6, © 1996

Food-Rewarded Operant Learning and Memory in the East Pacific Green Turtle

W. F. Angermeier¹ and Antonio Resendiz S. Hidalgo^2
1Marine Behaviour Research Station, Furnish Lettermullen, Co. Galway, Ireland
²Instituto Nacional de Pesca, Sea Turtle Project Research Station, Bahia de los Angeles, Baja California, Mexico

The development and extinction of food preferences (Grassman and Owens, 1982), food choice behavior (Steele et al., 1989), and some related topics have been studied in sea turtles, but there is a conspicuous lack of experimental studies of operant learning and memory in sea turtles. Operant learning can be defined as a "more or less permanent change of behavior, associated with the consequences of behavior" (Angermeier, 1994). It can be clearly distinguished from other forms of learning, such as classical conditioning, place learning, and observational learning, although consequences of behavior are also important in these types of learning paradigms (Angermeier and Peters, 1973; Angermeier, 1976).

Typically, in an operant learning paradigm, an organism learns that showing a certain type of behavior, such as manipulating its environment, can have positive consequences (e.g., food for a hungry animal). Thus, the response "produces" the reward, in contrast to classical conditioning, in which a stimulus leads to the response. In this context it is important to note that the manipulandum (e.g., lever, wheel, rod) itself cannot deliver the reward. If it does, we speak of place learning. In true operant learning, the reward is delivered independent of the manipulandum; that is, the reward is delivered by some mechanical device or, as in the study to follow, by hand.

During the summer of 1994, the authors tested 12 East Pacific green turtles (Chelonia mydas) for their ability to learn and remember food-rewarded operant responses. The study was conducted at the Sea Turtle Project Research Station in Bahia de los Angeles, Baja California, México. The animals were deprived of food for 1-2 days, removed from their holding tank, and placed singly into a 5 x 5 m testing tank. The depth of the water in the testing tank was 32-35 cm. A manipulandum made from a white plastic tube (approximately 5 cm in diameter) was fastened to the top of the tank wall, and inserted in such a way that the lower end was about 10 cm above the bottom of the testing tank.

Whenever the animal touched the manipulandum with any part of its body, it was rewarded with a small piece of fish (approximately 10 g). The reward was dropped in front of the animal's head, to ensure that s/he would see it. A total of 15 consecutive responses were rewarded in this fashion, of which 13 could be included in the analysis of the results (one animal refused fish from the beginning, and another refused the reward after the 13th trial). Table 1 shows the size (curved carapace length) and weight of the animals used.

Table 1. Curved carapace length (CCL) and weight of East Pacific green turtles (n=11) participating in a study of food-rewarded operant learning and memory.

Six weeks after the learning procedure was applied to all animals, they were tested in a single trial for long-term memory of the learned response. As soon as an animal had made one response and been rewarded for it, it was removed from the testing tank and transferred to its living tank. The major measurement taken during all testing sessions was the latency to respond (Angermeier, 1994). The rationale for this procedure was as follows. By comparing latencies between all the trials, one can easily determine when there is no more significant reduction of these latencies. That is the point at which the organism cannot improve his performance; e.g., the point at which he has learned the required response. By adding a single trial for one reward during the memory test, the latency of this trial can also be compared with all other latencies and thus yields information about the permanence of the learned behavior. A summary of the results is shown in Table 2.

Table 2. Mean latencies and standard deviations to a series of 13 food-rewarded operant responses of East Pacific green turtles (n=11).

A statistical analysis of the results (Repeated Measures ANOVA) showed that the difference between the time it took for the first reward to be obtained and all other rewards (2-13 and memory trial) was highly significant (F=5.03; df=13,130; p< 0.0001). On the other hand, there were no significant differences between any of the other latencies (2-13 and memory trial). We concluded that the animals learned during the first rewarded response and that they showed long-term retention of the learned behavior. One other observation may be interesting. Each animal performed the response to the manipulandum in its own characteristic way: some with the right flipper, some with the left flipper, and some with the head.

Acknowledgements: Thanks are due to Dr. Gerhard Piltz for assisting during the entire experiment and to Friederike Angermeier for photo documentation.

Angermeier, W. F. 1976. Kontrolle des Verhaltens. Springer: Heidelberg.

Angermeier, W. F. 1994. Operant learning, p.351-366. In: Encyclopedia of Human Behavior, Volume 3. Academic Press: San Diego.

Angermeier, W. F. and M. Peters. 1973. Bedingte Reaktionen. Springer: Heidelberg.

Grassman, M. A. and D. W. Owens. 1982. Development and extinction of food preferences in the loggerhead sea turtle, Caretta caretta. Copeia 1982:965-969.

Steele, C. W., M. A. Grassman, D. W. Owens and J. H. Matis. 1989. Application of decision theory in understanding food choice behavior of hatchling loggerhead sea turtles and chemosensory imprinting in juvenile loggerhead sea turtles. Experientia 45:202-205.