TABLE IV. TIME REQUIRED TO ESCAPE FROM CAGE BY USING LATCH

Animals
Trials,
4 daily.		ABCEFGHIAv.Daily
Average
' "' "' "' "' "' "' "' "' "' "
(1):03:063:20:08:031:104:30:231:13
(2):051:10:591:502:00:102:461:001:00
(3)5:28:291:04:21:29:522:334:051:55
(4):311:453:52:14:06:232:402:301:30(1:25)
(5):15:102:25:03:10:07:31:50:34
(6):172:005:03:03:05:15:39:101:04
(7):08:20:36:31:03:11:30:46:23
(8):03:54:47:28:03:55:06:47:30(:38)
(9)1:12:34:39:17:03:04:21:20:26
(10):02:19:28:07:01:03:20:47:16
(11):02:51:09:14:03:021:46:47:29
(12):02:15:24:12:02:021:17:27:20(:23)
(13):03:19:22:07:06:04:48:11:15
(14):02:15:22:09:02:02:20:19:11
(15):08:06:12:05:02:02:03:06:06
(16):03:18:44:10:02:03:02:21:13(:11)
(17):02:17:41:03:07:07:11:06:12
(18):02:42:26:05:01:03:31:15:16
(19):04:13:48:04:01:02:15:14:13
(20):04:32:35:04:01:02:08:22:15(:14)
(21):03:13:10:13:09:011:10:16:17
(22):03:05:10:11:03:02:39:26:12
(23):02:07:17:05:03:15:20:26:12
(24):02:14:04:02:01:02:091:03:12(:12)
(25):01:03:05:04:01:02:31:24:09
(26):02:04:08:01:01:051:34:33:19
(27):02:03:03:03:10:05:15:39:10
(28):02:04:03:01:06:05:23:51:12(:12)
(29):02:04:10:03:02:06:23:21:09
(30):01:03:03:02:11:02:091:08:12
(31):21:03:11:03:01:03:28:11:10
(32):03:02:04:02:01:03:11:42:09(:10)
(33):02:02:03:10:01:03:18:17:07
(34):02:02:03:02:01:03:11:11:04
(35):01:02:03:02:01:02:18:09:05
(36):01:03:03:03:03:02:04:10:04(:05)
(37):01:02:07:02:04:01:11:09:05
(38):01:02:02:02:02:01:08:03:03
(39):02:04:11:01:02:02:14:08:06
(40):01:02:03:01:01:03:10:03:03(:04)

The periods here were similar to those given in the maze tests—the time was long at first, then it shortened very rapidly for a few trials, then more slowly but still constantly, until the act became thoroughly familiar. The process was one of "trial and error" throughout. As before, various lapses occurred, even although the animals were as persistent as usual in their efforts to escape. When the lever was moved to the side or back of the box, none of the animals could escape. In general, pigeons show less ingenuity in dealing with latches than do sparrows, according to Porter's[205] observations, although in some other tests they are equally apt.

B. Pecking Tests

The preceding series of trials proved the animals' ability to utilize certain touching or clawing movements, at first accidental, in making an escape, and showed that these could become highly specialized. Desiring to carry out similar tests in the case of pecking movements, which are quite as natural to pigeons, I arranged a contrivance by means of which the act of pecking at corn-grains fastened to a small piece of cardboard (placed just outside the cage, but within easy reach through an opening in the wire) would open the cage-door by making a delicate electrical contact. Four animals were tested.

On entering the cage they endeavored to escape as before; failing in this they began pecking about until they found the corn-grains and made the contact which opened the door and allowed them to escape to the food without. Three of the animals made their escape in this way several times; but the habit seemed to be one that could not be readily learned, as the successive periods showed little shortening.

Thinking that the pecking of things at a definite place perhaps complicated the matter, I removed the electrical apparatus and arranged to open the door myself by pulling a string whenever the pigeon pecked anywhere upon entering the box. Preliminary experiments with Bird J indicated some ability to profit by this kind of experience. As the act of pecking could be used to advantage in a series of imitation trials, this animal only was allowed to learn to escape by actually pecking; the others were reserved for the imitation tests next to be reported.

C. Imitation Tests

In the experiments already reported the animals were used individually and usually out of sight of the others, although in the same room and within hearing of them. When efforts were made in some of the experiments to test the animals in a separate room, signs of fear and discontent were often noticed, and it was necessary to return to the first room to continue the tests. Some instances were noticed in which a pigeon would do what it saw another doing. For example, one of my subjects would not eat one day, being ill apparently; but when I put two others into the compartment with it, and they began eating the food lying about, it also began pecking. Its act could not have been due to its only then happening to see the corn, for it had before looked toward the food when this was thrown to it.

Desiring to test, under definite conditions, the imitative ability of these animals, I arranged trials in which birds were allowed to see a useful, but simple act performed by another bird, and then were given an opportunity to execute the act themselves. Using the animal which I had trained for the purpose, I allowed its series of acts (entrance to box, pecking, release, and food-eating without) to be observed by another animal, confined in a small wire compartment (similar to the entrance compartment before used) attached to the side of the large cage. Care was taken to see that the confined bird was observing, or at least was looking toward the acting one; in case of doubt, the trial was repeated. Later the trained animal was replaced by the observing one, and the latter's reaction was noted. Five animals, in all, were tested, and each was given two opportunities to escape after having seen the trained animal perform the act ten successive times. None of them, however, showed any signs of trying to escape by repeating the movements so often performed by the bird familiar with the act, but each rushed against the sides of the cage and tried to push through at various places, just as the trained bird had done when first learning the habit.

As the act, or series of acts, was rather too complex to be easily observed and utilized by the other pigeons, I arranged two much simpler tests. In one case the leading bird was taught to open the cage-door by stepping upon a platform (the lowering of which made an electrical contact); in the other, to avoid a blind alley, enter a short passage, and ascend a wooden plane (inclined at an angle of thirty degrees) which led into another box containing food. In these tests it was more difficult for the series of acts to be viewed, but the animals, singly as before, were placed at a point of vantage and apparently saw the movements of the other animals.

Of the five birds tested in the platform experiment, four utterly failed to escape in the two trials given. The fifth, in its second test, went to the platform promptly and thus made its escape, but the success may have been accidental, or due to the animal's experience of seeing and approaching the platform in its first test. In the labyrinth experiment only one bird (second test) avoided the blind alley and went directly up the inclined plane to the food; this success was probably due to experience in the first test. There was certainly no evidence that the animals had grasped the nature of the problem; nothing to indicate that the performance of the trained animal had supplied data for the guidance of future conduct, that is, for the conditioning of the necessary movements, in this case those of pecking, stepping upon a platform, or avoiding a blind alley and ascending an inclined plane.[206]

These results are similar to those which other experimenters have secured in the case of chicks, cats, dogs, monkeys,[207] and also rats.[208] Although the method I employed is doubtless open to the criticism of being artificial,[209] some value at least should be attached to the results; if so it would seem probable that imitation in pigeons is not above the "instinctive" stage, and that learning depends entirely upon first hand experience, upon really doing the thing, and not upon merely seeing it done.

3. Position, Color and Form Tests. The apparatus used in these experiments consisted of small boxes, six inches in height, and open at the top. Sometimes they were exactly similar, and sometimes they differed in color or in form. They were moveably attached, six inches apart, to a board which was placed in a large wire-covered box, having an entrance compartment as in the case of the mazes and cages. Food was placed in one of the small boxes, and the pigeons were allowed to find it twice; later each bird was tested as to its ability to return to the food by depending upon the position of the box in the group, or upon its color or its form. Tests were given in series of six, and the box which was first approached was recorded as the animal's choice for that test. If it made a wrong selection, it was allowed to look about until it found the box containing the food, but in no case was it permitted to satisfy its hunger until the last test of the series. The animal apparently did not see the food until it approached the box, and subsequent tests demonstrated that it was not guided by odor.

A. Position Tests

In this series of trials I used at first six, later nine food-boxes, four inches square and covered with dark gray paper. The board to which the several receptacles were fastened was shifted at irregular intervals to various oblique angles; this was done to prevent the animal from being assisted by the position of the food-box in the larger box rather than in the group of similar boxes. After a bird had been tested sufficiently for one position it was then used, for a week or so, in some other experiment, and thus given an opportunity to forget, to some extent at least, the old experience before being taught to find the food in a box placed elsewhere in the group. For the positions 2, 3, and 4, in the group of six similar boxes, eight animals were each given thirty tests in series of six, as stated above. For the positions 5, 6, and 7, in the group of nine similar boxes, the experiments were shortened. Six animals were given twenty-four trials each, two animals for each of the three positions.

The animals quickly learned the position of the food-box and passed to it promptly when released from the entrance compartment. Changing the position of the board to which the food-boxes were attached did not affect the animals' ability to reach the food readily. They usually selected the proper box as before, although frequently they went around the end of the board and approached the food from the opposite side. The general distribution of choices in the case of positions 2, 3, and 4 is given in Table V; the rate of learning, in Table VI.

With a single exception (Bird B, box 3) the box containing the food was far more often chosen than any one of the empty boxes, and usually more often than all of them combined, the average right choices being: position 2, 62%; position 3, 57.7%; position 4, 53.3%. It will be seen that the animals were more successful in finding the food in the second position than in either the third or the fourth, that is, positions nearer the end were more easily located. But what is of greater interest to us is the rate of learning to go to the right box. This is indicated by the increasing number of right choices from series to series. (Table VI.) That the increase was small was due to the fact that the animals learned so quickly in the first series of six tests that little improvement could be made thereafter; what they could learn they acquired early in the experiment. There is some evidence of improvement after the first series in the case of box 4, a more difficult position, and the average for all three boxes shows a slight improvement from series 1 to series 5, although a falling-off is seen in the last series: 26, 27, 27, 31, 28. The same general features appear in the case of the incomplete tests (positions 5, 6, and 7). For each box there were, on the average, 26 right choices in the possible 48. There were more right choices in the case of box 7 than in the case of either box 6 or box 5, box 7 being nearer one end (box 9). There was little evidence of learning after the first series of trials.