choosing a turn to the left into alley D, leading to food compartment F, at one tone, and a turn to the right into alley D', to food box, F', at the other tone. The stimulus tone may be sounded while the animal is in the home-box. A, and the animal released after it has been damped; or the tone may be sounded after the animal's re- lease, say one-half second before he can reach the end of the intro- ductory alley. If punishment is to be administered the animal should receive it the moment he makes the wrong turn.
Yerkes' apparatus for testing hearing in frogs (air and water). — Fig. 13 shows the simple apparatus for testing the auditory responses of frogs under the following condi- tions: When tympanic membranes are (1) in air, (2) half submerged, and (3) wholly submerged. As will be shown on p. 388, no direct response to sound is exhibited by frogs. In order to show that auditory stimuli are effee- OLFACTOEY APPARATUS 89 tive it is necessary to utilize an indirect method. When tactually stimulated the frog will jerk up its leg. Re- peated observations show the magnitude of this reaction. When a bell is sounded at definite intervals (i.e., within one second) before the tactual stimulus is given the height of the leg reaction may be (1) reinforced or (2) inhibited, Fig. 13. Auditory Appaeatus for Testing Hearing in Air and IN Water A, aquarium; B, electric bell; T, tactual stimulus apparatus; K, hand key for giving stimuli; W, weight to hold leg. (After Yerkes, Jour. Comp. Neu. and Psych., 1905, XVI, p. 297.)
depending upon the temporal relations of the stimuli. This apparatus may be used in connection with a pendulum for automatically regulating the intervals between bell stroke and tactual stimulus.
Apparatus for obtaining olfactory and cutaneous stimuli. — Fig. 14 shows a rough apparatus for studying (1) olfactory responses. It can be used only where there is a steady air supply. It is designed to be used in con- nection with the air system sketched above. In addition to its use as a device for testing olfaction it may be used for testing sensitivity to (2) differences in temperature, (3) differences in the moisture content of two air columns, and (4) differences in the pressure of two air columns.
90 APPARATUS AND METHODS The manometer shown in Fig. 10 should be used to control the stimuli in all this work. As drawn the apparatus is ready for use in olfactory work.^ Fig. 14. Olfactory Apparatus The whole apparatus rests upon a glass plate and the inside walls of the stimulus compartment are lined with glass. The cover is also of plate glass for ease in observing the behavior of the animal. Two air columns sho^^^l by the two arrows on the extreme right of the lower drawing come directly from the air system, each being connected with a separate manometer. They pass over the sur- face of two fluids, in which appropriate olfactory substances have been dissolved, B^, B-. After passing over the olfactory substances the 8 As may readily be inferred, we may pass one air source over a dry chamber and another over water if our purpose is to test sensitivity to differences in moisture content. Or if it is to test differences in tem- perature, we may pass one through a small constant temperature air chamber. On the other hand, if we are testing differences in pressure, the two air sources are sent directly into the stimulus compartment.
TEMPERATURE APPARATUS 91 air is piped separately into the stimulus box througli the fine pointed nozzles, Oj and Og. The nozzles are placed at an angle such that the two air columns escaping from them converge at a point, X — the apex of the two glass partitions, GP. It is at X that the animal is stimulated simultaneously by the two stimuli. Electric grills may be placed in front of 0^ and O^. At EF is shown a funnel, to wdiich are attached two tubes. Ihese tubes are connected with an aspirator attached to any convenient water faucet. This serves to partially prevent diffusion of olfactory particles. As sketched above, two serious defects are inherent in the apparatus. There is no con- venient way of admitting the animal and there are no exit com- partments. Food should not be used in the chamber but in the exit boxes. The entrance should probably be placed underneath the box in such a position that the animal would have to enter near X. Since opening this door would introduce variations in the direction of the currents Oi and O2, the animal should be placed first in the ante-chamber, the door to which should immediately be closed. The air should then be turned on at once and the animal admitted. Immediately upon its correc.t response the proper exit door should be opened. Unfortunately no careful work has so far been published upon these forms of sense modalities, and investigators are working more or less blindly in trying to devise proper apparatus. Dr. Jean Weidensall carried out a series of experiments upon rats with this apparatus but the work was never completed. It was carried far enough to show that with the modifications suggested we may safely begin work upon olfactory and cutaneous sensitivity. There are many methods of conducting simple qualitative tests upon smell, taste, and cutaneous sensitivity. Such methods will be taken up in the chapter devoted to those senses.
Yoakum's temperature apparatus. — Fig. 15 shows the apparatus used by C. S. Yoakum for testing the tempera- ture sense of gray squirrels and white rats.
The construction of the apparatus is as follows: Two galvan- ized iron boxes, A and B, are made with outside dimensions of 9 X 9 X 24 inches. The inner opening or tunnel, running the entire length of the box, is 5x5 inches. The space between the outer and inner jackets is 2 inches deep. It entirely encloses the central cavity with the exception of the ends. This enclosed space is suppKed with three vents, an inlet, I, an outlet, 0, for the water circulation, and an air vent, V, to relieve the pressure when water is admitted to the system. The outlet is placed at the bottom of the encircling jacket and brought to a level with the upper part of the box. Thus arranged it can be used to syphon out the water after the day's experiment is over. In the bottom of each tunnel small cubical food pans are sunk flush with the floor. They are placed near the end farthest from that used by the animal as an entrance. The back of each box is closed with a piece of ground glass. Separate electric lights EL, illuminate the tunnels. The temperature of these boxes is regulated by forcing hot or cold water of the desired temperature through them. A triple faucet is used; two vents to give hot and 92 APPARATUS AND METHODS cold water, respectively; the third to give a mixture of hot and cold. The respective temperatures of the two boxes cannot be maintained with extreme accuracy. The variation is sometimes as great as 2° C above or below the standard. The two boxes can be inter- changed by means of a rope and pulley, P. The animal is admitted from behind, the arrow showing its probable course.
Fig. 15. Yoakum's Temperature Apparatus. [Jour. Comp. 'Neu.
A and B, galvanized iron boxes; I, inlet for water supply; 0, outlet; V, air vent to relieve pressure when water is admitted; EL, electric lights illuminating the tunnels (through ground glass); P, attachment for cords used in shifting position of boxes.
TV. Methods of Studying Motor Habits Introduction. — Motor habits do not differ fundamentally from sensory habits, yet they are distinct from them in one PROBLEM BOXES 93 respect. In sensory habits the investigator attempts in general to bring out particularly data bearing upon the stimuli to which the receptors of the animal are sensi- tive. In motor habits he attempts to bring out particu- larly data concerning the action system of the organism. In thus examining first one end of the reflex arc and then the other there is introduced no change in standpoint or in method of interpretation. Further discussion of the re- semblances and the differences between sensory and motor habits may very well be left to Chapter VI, where the con- crete experiments upon them are presented.
The problem box method. — The problem box method in animal behavior in one form or another is as old as civilization. It remained for Thorndike, in 1899, to per- fect the method and to make it serviceable for bringing out certain facts connected especially with what one roughly calls learning in animals. The forms of puzzle boxes in use in the various laboratories are very numerous. Their size has had to be varied with the size of the animal and their nature to suit the particular instinctive capacity of the animal. It is obviously impossible for the English spar- row to learn to manipulate all of the problem boxes that may be offered to the monkey. On account of the differ- ences in the action systems of animals no standardization of types has thus far been suggested. Each experimenter has devised problem boxes which he thinks will be suited to his animals' capabilities. On account of this difference in the boxes themselves and in the technique of experimenta- tion there is no uniformity in results even in the same laboratory where different animals are used; and no hope at present of work being duplicated in other laboratories. Much better results could be obtained if groups of boxes could be devised which would be suitable for a large range of animals. E.g., it is possible to devise boxes of the same dimensions and which are manipulated in an identical way, that may be commonly used for mammals of small size — monkeys, cats, dogs, raccoons, squirrels, porcupines, etc. Another set could be used for birds; still another for fish, etc. The method of using problem boxes is somewhat as 94 APPARATUS AND METHODS follows: some form of general stimulus is provided. This has usually been food particularly sought by the animal. It is placed inside the box. Some form of restraining cage is usually placed over the box (see Fig. 19). Usually one allows the animal to feed in the open box for a few days before attempting to give it the problem of opening the door. The general stimulus to work is thus, strong by the time the animal begins its attack upon the specific problem. The mode of the animal's attack upon the box while solving the problem depends upon its instinctive equipment and upon the type of problem selected. The animal is allowed to eat a small amount of food after opening the box and is then removed in preparation for another trial. The number of trials given per day alters both the form of the learning curve and its length. The subject of the distribution of effort in forming motor habits will be taken up on p. 228. The recording of errors in the formation of habits of manipulation is almost im- possible. The time curve furnishes the best criterion of the learning of the simple boxes usually offered.
Description of boxes. — Below we show some of the boxes which have been used with success in the animal work. All Fig. 16. The Sawdust Box that we maintain for them is that they are serviceable and have a wide range of usefulness. "We do not maintain that they are better adapted to the study of motor habits than PROBLEM BOXES 95 others which have been employed. Nor do we claim that placing the food inside the box is more advantageous than Thorndike's method of placing the animal inside and allow- ing it to escape.
Box I (Fig. 16) — the " sawdust box " — is one of the simplest forms. It has been widely used in work upon rodents. It may be looked upon as standing between the maze and more definite forms of problem boxes in regard to the type of activity demanded for its solution. The sides of the box are constructed usually of wire mesh. This extends to the floor of the containing cage on two sides and one end. On the remaining end, E, the wire netting extends only to the floor of box I, The animal enters under this floor and pushes its head into the interior of the box through the opening, F. It is usually banked upon all sides with sawdust to a height cor- responding to the level of its floor.
Box II (Fig. 17) — the "latch box" — requires on the part of the animal a narrowly circumscribed movement. It is made of the same material and in the same general way as box I, The fasten- ing, L, shows clearly in the drawing. Care must be taken to have Fig. 17. The Latch Box the hinges on the door work very easily, to have the socket which holds the latch very shallow, and the spring which opens the door work very gently. This box has been used largely on rodents and dogs.
Box III (Fig. 18) — the "inclined plane" box — is shown in two forms. In its simplest form (Fig. 18), as used by the author and by Richardson, it does not work well. The animal may chew the 96 APPARATUS AND METHODS string or it may throw the plane by pulling on the string without necessarily touching the plane. Basset has greatly improved upon the original plan. Fig. 19 shows the form employed in his work Fig. 18. The Inclined Plane Box and the details in construction and method of operation. The food box, A, is formed of wood, 11x12 inch base, 11 inches in height, and is covered on top and sides with % inch heavy wire mesh. It is fitted with a hard rubber door, d, 3-16 inch thick, 5 inches high, and 414 inches wide. To the inner side of the door is fastened a cord which passes over a pulley, p, and is weighted at the other end with a piece of lead, 1, of sufficient weight to insure the opening of the door upon releasing the latch. B, shows the device for latching and releasing the door. A short distance above the door is fastened a 3-inch electrical magnet, m; directly below that is a steel wire, s.w, surmounted by a steel disc, s.d, of the same diameter as the core of the magnet. The steel wire holds the door by dropping through holes in two brass plates, g, which serve as guides, to a point, behind another brass plate, which is set at the top of and behind the door, 11/2 mm. below the top. The setscrew, s.s, placed on the steel wire above the lower guide prevents any further drop. When the steel wire holds the door the disc is 2 mm. below the magnet; when the disc is dra^^^l up to the magnet % ^^™- clearance is allowed for the door to swing back. Back of the feeding box, A, is placed the inclined plane, LP. The inclined plane has a hard rubber base % inch thick, 6 inches long, and 2% inches wide. Upon pivot standards rising from the middle of the base rests the plane itself. The plane is of wood fibre and of the same dimensions as the base. It is weighted at the end nearest the feeding box in order to insure its return to position after use. At the end opposite the weight and farthest from the feeding box, platinum electrical contacts, e.e, are placed in both base and plane. The power is provided through wires connecting the regular electric lighting system, 115 volts, direct current, with the wired apparatus. A 32 candle-power lamp is placed in the series in order to avoid any danger of short-circuiting. To make the contact and allow the current to pass through the magnet, thus raising the steel wire and releasing the door^ it is necessary for the rat to step on the PROBLEM BOXES 97 point of operation, o, wliich lies well out towards the end of the plane. On account of a certain amount of latency in the operation of the magnet, the rat must not only make the contacts touch, but must also inhibit further action, remaining on the point of operation until click of the disc meeting the magnet is heard. Over the food box and plane is placed a cage, C, constructed of l^-inch heavy wire mesh, the base measurements of which are 24 X 24 inches and the height 14 inches. This allows the rat ample Fig. 19. The Modified Inclined Plane Box (After Basset, Behavior Monograph, Ser. No. 9, p. 25) room to explore all sides of and above the food box. When the rat is placed within, the entrance, e, to the cage is closed.
Box IV (Fig. 20) shows the box used by Porter in the work on birds. This likewise is made of wire netting. As used by him the box was tilted forward so that the door would swing open when released. The bar, L, drops into a shallow socket. A string is attached to the end of the bar. This string runs upward through a ring at the upper edge of the box, thence along a wire arm, through a ring on the end of this arm, down to a loop, B, which is placed within reach of the animal. Another method of opening the box is shown. The string attached to the latch is fastened to a stud, C. The bird, in climbing over the side of the box, pushes in the string, raises the latch, and the door swings open.
Box V (Fig. 21) shows a relatively simple combination of fasten- 98 APPARATUS AND METHODS ings. The animal, in order to open this box, must: bear down lever at left (1), push in bar at right (2), lift up latch in front (3), and pull out string behind (4).
Fig. 20. Peoblem Box Used by Porter on Birds (Am. Jour. Psych. 1906, XVII, p. 257) Box VI (Fig. 22) shows a similar but more complicated set of fastenings. It is an interlocking device. The door is fastened by a vertical hook (5) at the upper left-hand corner which cannot be un- FiG. 21. Showing Box with Simple Combination of Fastenings hooked until the button (4) near it is pushed back. But the button cannot be moved until the bolt on the right side of the door (3) is pushed back. This in turn requires the removal of the plug at the THE MAZE EXPERIMENT 99 left (2) which again cannot be moved until (1) is pulled out. The only order which brings success is 1, 2, 3, 4, 5. Once worked, how- ever, the fastenings cannot accidentally be replaced.
Fig. 22. Box with More Complicated Fastenings (Both boxes after Kinnaman, Ame7\ Jour. Psych., 1902, p. 123) Description of maze experiments. — The technique of the maze experiment is quite similar to that of the problem box. The animal is fed for a few days in the center of the maze without being allowed to explore any other part than the food compartment. After habituation to this environment, regular tests are begun: the animal is ad- mitted to the maze and allowed to find its way to the food. Both the errors (false turns, returning, and partial returning, etc.) and the time spent in going from the entrance to the food should be recorded. A record of errors made in the maze can be taken very accurately and con- veniently by two different methods. In the writer's study of the terns the maze floor was lightly covered with a thin layer of coral sand. The tracks of the bird could be seen plainly and the length of the excursions into the alleys, etc., could be measured. The tracks were obliterated by brushing the sand lightly with a whisk-broom. A better method for obtaining the same results is the camera lucida method, which will be described in connection with the circular maze.
APPARATUS AND METHODS Fig. 23. The Circular Maze With Camera Lucida Attachment SB, starting box; M, Mi, mirrors; L, lens; IM, image of maze.
The Circular Maze: The circular maze shown in Fig. 23 is made with wooden base and aluminum walls. The base is 150 cm. in diameter and 4 cm. in thickness, and is constructed as follows: Two wooden discs 150 cm. in diameter and 2 cm. in thickness are first sawed out. These two discs are finally glued together. Before gluing, however, the upper disc is marked off into a series of con- THE CIRCULAR MAZE 101 centric circles. The diameter of each of the circles is as follows, be- ginning with the outermost one: 140 cm., 120 cm., 100 cm,, 80 cm., 60 cm., 40 cm., and 20 cm. The circles are then sawed out upon a band saw. The width of the saw is made just equal to or slightly wider than the aluminum sheets used for the walls. After sawing, the disc as a whole is converted naturally into a series of concentric rings. These are fastened down to the lower disc with hot glue and screws. The lower surface of the base is thus solid, while the upper surface shows a series of grooves into which the aluminum Avails may be slid. Soft aluminum bought in rolls is used for the latter. The height of the aluminum is 18.5 cm., the thickness,.8 mm. The aluminum is unrolled and cut into the proper lengths. Each strip is cut just 10 cm. shorter than the length of the circular groove into which it is to be fitted. This gives an opening into the alley. By means of this arrangement it is possible to slide the aluminum around in its groove and thus to place the entrance in any desired position. Fig. 23 shows clearly the construction of the maze, the number of alleys, the placing of the entrances, and the radial stops. This maze offers several desirable things in work of this character: in the first place, it can be used on the unit plan, in that only the home-box and the surrounding segments need be used when a very simple maze is desired. The addition of other segments, then, merely increases the complexity in an, at present, unknown ratio. The coefficient of increasing complexity could be determined by allowing one group of animals to learn the maze in its simplest form, another in its next most complex, etc. Secondly, the ease with which complications can be introduced makes the maze very desirable. This is brought about by the flexibility in the arrangement of the entrances and radial stops. The camera lucida attachment is easily installed: it is simple and permanent. A large plate-glass mirror, jNIj, 91 cm. wide, 121 cm. in length, is placed at an angle of 45° directly over the center of the maze. This mirror is strapped by small clamps to the edge of the surrounding framework. At a certain distance from this mirror a second mirror, M, 60 cm. X '^ cm., is placed at an angle of 45° above the maze and at such a distance from Mi that the light reflected downward from M falls outside of the maze. Below M, and in the path of the light reflected from it, is placed a single achromat, L, 6 cm. in diameter and 50 cm. focus. The lens is placed in a barrel and the barrel is attached to a wooden disc 30 cm. in diameter. This board is attached to an iron collar which slides freely up and down the rod, CR. This gives a very easy means of adjusting the size of the image, focusing, etc. BeloAv this first disc will be found a second disc similar in character and controlled in the same way. A pad of circular paper is laid upon this disc.^ A reduced image, IM, of the maze appears upon this paper. Extraneous light is ex- cluded by means of a soft black flannel curtain attached to the disc which supports the lens, L. As may readily be seen from the figure, the maze must be illuminated quite highly in order to produce a clear image. The illumination is obtained by means of six lights 9 It is convenient to cut out several sheets upon a disc cutter and to stamp a hole 13 mm. in diameter in the center of each for the reception of a stud 13 mm. in diameter and 1 cm. high placed in the center of the board, IM.
APPARATUS AND METHODS placed symmetrically around the maze and by one light in the center of the maze. The six lights on the periphery are supplied with opaque half-shades, the light in the center of the maze with a similar opaque circular shade. These shades are of aluminum, black- ened on the upper surface. The floor of the maze is covered with imported white linoleum. This serves to reflect the light upward to Ml, thence to M. Passing through the lens the rays are brought to a focus at IM. The ratio between the maze and the image is 6.4 to 1. The image appearing at IM is extremely clear when proper precau- tions are used to sensitize the eye. Even the smallest mouse can Fig. 24. Chartometer be seen quite clearly. The movements of the animal are traced upon white paper with a soft pencil. In the early stages of learning several sheets of paper are used on each animal at any given trial in order to avoid a too complicated tracing. Each sheet is marked with the number of the animal, the number of the trial, and the serial number of the tracing. The length of the lines so traced is measured by means of a chartometer (Fig. 24). This instrument is surprisingly accurate even in measuring lines which are tortuous in their course. The error in measuring the length of the charted line is about 1 per cent.^° Other Forms of Maze: Fig. 25 shows the "modified Hampton Court maze," which is too well known to require description. It has 10 Yerkes and Kellogg have described a similar apparatus. See Journal of Animal THE HAMPTON COURT MAZE been used extensively in the animal work — upon rats, birds, squirrels, monkeys, and man. It is generally considered to be a very compli- cated maze. Yet actual experience shows that the circular maze just described is probably equally difficult, at least for the rat. The dotted line shows the true pathway.
Fig. 25. The Hampton Court Maze Complicated blind alleys, or culs de sac, are shown at A, B, D, E, F. It will be noted that C is not a blind alley but an alternate and longer way to the food. The animal is admitted at 0. Food is placed at H. After the maze has been learned with the alleys lying in the directions indicated tne maze as a whole may be turned 90° — 180°, etc., to test the effect upon the animal of changing the absolute direction of the turns. Fig. 26 shows a maze constructed for the special purpose of determining the function of the kinsesthetie sense of the animals.
As constructed for the rat the alleys are of wood 6 inches in width and 6 inches in height. The maze was constructed so that it could be sawed across the dotted lines a-b. It will be seen that by remov- ing or inserting the middle section, a-b, we can convert the maze respectively into its short or long form. The change merely increases the length of the four alleys without altering the relations of the turns leading to or from them. Particular attention is called to the fact that Q in the lengthened form comes to the point Q' in the APPARATUS AND METPIODS shortened form; and that the distance S-B in the extended maze equals the distance S-B' in the shortened maze.^^ Fig. 26. Special Form of Maze Which Can Be Lengthened oe Shortened Without Altering the Relations of Turns Apparatus for the study of the " delayed reaction." — The delayed reaction has recently been studied at the Uni- versity of Chicago by Hunter. On p. 224 we present the more important results which have been obtained. Fig. 27 shows the needed apparatus.
It consists of a chamber, G, of suitable size, made with glass sides. The animal is confined here while being stimulated by lights coming from the food compartments, L, L, L. The entrances to the three food boxes are equidistant from G. Any one of the food boxes may be illuminated at will by closing the switch at M. Grills for elec- trical stimulation are shown at a, b, c. The current from the second- ary of a Porter inductorium may be sent through any one of these at will by the switch system shown at N. The use of the apparatus in brief is as follows: The animal, previously habituated to the apparatus, is confined in G. A light, e.g., in box A, is turned on for a definite interval and then turned out. After an interval (during which no visual stimulation is offered) the animal is released at G. His problem is to enter only the box in which the light last appeared