Two rough-toothed porpoises (Steno bredanensis) were individually trained to emit novel responses, which were not developed by shaping and which were not previously known to occur in the species, by reinforcing a different response to the same set of stimuli in each of a series of training sessions. A technique was developed for transcribing a complex series of behaviors on to a single cumulative record so that the training sessions of the second animal could be fully recorded. Cumulative records are presented for a session in which the criterion that only novel behaviors would be reinforced was abruptly met with four new types of responses, and for typical preceding and subsequent sessions. Some analogous techniques in the training of pigeons, horses, and humans are discussed.The shaping of novel behavior, that is, behavior that does not occur or perhaps cannot occur, in an animal's normal activity, has been a preoccupation of animal trainers for centuries. The fox-terrier turning back somersaults, the elephant balancing on one front foot, or ping-pong playing pigeons (Skinner, 1962) are produced by techniques of successive approximation, or shaping. However, novel or original behavior that is not apparently produced by shaping or differential reinforcement is occasionally seen in animals.

No abstract

Two specimens of Stenella attenuata, trained to chase a winchtowed lure, reached a top speed of 11.03 meters per second (21.4 knots) in 2.0 seconds. The maximum power output, occurring 1.5 seconds after the start, was calculated from measured values of acceleration and drag coefficient. The maximum power output per unit body weight was 50 percent greater than for human athletes. The measured drag coefficient, obtained from periods of coasting, was approximately the same as that of an equivalent rigid body with a near-turbulent boundary layer.

No abstract