Untrained Octopus vulgaris (observers) were allowed to watch conditioned Octopus (demonstrators) perform the task of selecting one of two objects that were presented simultaneously and differed only in color. After being placed in isolation, the observers, in a similar test, consistently selected the same object as did the demonstrators. This learning by observation occurred irrespective of the object chosen by the demonstrators as the positive choice and was more rapid than the learning that occurred during the conditioning of animals. The task was performed correctly without significant errors and further conditioning for 5 days. These results show that observational learning can occur in invertebrates.
Octopuses, like many other invertebrate species, are capable of a learned change in behavior as a result of experience[1-4], demonstrating that Octopus can integrate information to produce adaptive behavioral patterns. Among vertebrates living in social groups, learning can be facilitated by observation of another member of the same species (conspecific) performing a behavioral act[6,7].
Octopus vulgaris does not have social habits, which implies that it has little experience observing the behavior of conspecifics. However, some cephalopods are social, and specialized behavioral patterns are exhibited when octopuses come in contact with each other. Therefore, we tested whether octopuses can learn to perform a task by observation of other trained octopuses.
Individuals of Octopus vulgaris were conditioned to discriminate between two stimuli that were identical in shape and size but differed in color(9). Training was followed by a session of trials while an untrained Octopus visually observed the choices made by the conditioned demonstrator. Experiments to investigate the ability of Octopus vulgaris(10) to learn by observation were conducted in three phases: (i) training of demonstrators, (ii) observation of the task by untrained octopuses, and (iii) testing of observers.
Training of demonstrators(11) was realized by a series of trials where two balls were presented to the animal. We selected one ball to be the correct choice. When the animal attacked the correct ball it was rewarded, and each attack of the incorrect object was punished. The training of a demonstrator was complete when the animal made no mistakes in five trials.
Two groups (red balls and white balls) of demonstrators were trained. For the red group, the red ball was considered to be the correct choice. Its selection was rewarded with a small piece of fish attached to the side of the ball that was not visible to the animal. Selection of the white ball was punished by an electric shock(9). For the second group of octopuses the conditions were reversed. For the octopuses (n = 30) conditioned to choose the red ball, full training was reached at 16.83 [+ or -] 1.35 trials (mean [+ or -] standard error of the mean). The white group (n = 14) was trained after 21.50 [+ or -] 1.46 trials. Full training was reached at a significantly different number of trails for the two groups (Student t test = 2.11, df = 42, P < 0.05)(12).
During the observational phase, an...