This article is concerned with a study of the manner in which certain spatial relations are involved in the learning of an elevated simple T-maze. The relationships singled out for study are: (i) position of the food box with reference to other loci in the experimental room; (2) directional axis from the choice point to the food box; and (3) response (i.e., right turn or left turn) leading from the choice point toward the food box.These are aspects of the situation which may have an important influence on the development of spatial learning. Tolman, Ritchie, and Kalish (10), for example, have assumed that after several trials on a maze the rat develops a 'cognitive map' of the environment, and along with it the ability to go to the location in the room where food has been experienced. This kind of learning they have called 'place' learning, and the disposition to go to the food, a 'place disposition.'In the second article of the series, "Studies in Spatial Learning," these investigators (u) trained a group of rats on two alternate arrangements of a simple T-maze shown in Fig. I. The rats always found food, let us say, at location FI, but in half of the trials they started at Si and made a left turn at C, while in the remaining trials they started at Ss and made a right turn at C. Tolman, Ritchie and Kalish believe the learning shown by this group to be an example of pure 'place' learning.

No abstract

In 1944 the present writers reported a series of 10 experiments concerned with an analysis of choice point behavior in the white rat (1). In that study we attempted to learn the extent to which the rat's response at a choice point is determined by place and the extent to which it is determined by the sequence of movements leading up to the choice point.We found numerous conditions under which the response seemed to be directly determined by the locus of the choice point. For example, after rats had learned concurrently the two simple mazes shown in figure 1 they continued to turn right at locus Ci, and left at locus Gz on test trials in which the sequence of turns leading to the respective choice points was interchanged (see test indicated in figure 1). Likewise, when two mazes were learned concurrently, if opposite responses were to be made at a locus occupied successively by a choice point of the different mazes, marked inhibition developed at that choice point. This finding has been supported by Kalish (4), using the training method employed by us, and again, using the more efficient non-correction training method.However we found several contrary conditions. When the two critical responses had been established at two distinct loci, a gradual shift of maze positions bringing the loci to coincidence often failed to disrupt the subjects ability to perform correctly at C presumably on the basis of the preceding response sequence. Also, in our experiment 10, the response at a choice point locus was apparently governed by the sequence of movements preceding the point.Thus we obtained evidence for the operation of both place and response learning; but it seemed that in all cases we had failed to isolate the two aspects unequivocally. Indeed, it seemed likely that they were so intimately organized as to prevent any precise analysis.In 194(i Tolman, Ritchie and Kalish, began a series of studies of place learning closely related to the above investigation (5, C, 7, 8). In a clear statement these writers designated the aspects of learning which we had called place, and sequence, by the terms place disposition and response disposition. In one experiment (7) they purported to have separated the learning of place and response dispositions and to have shown that the learning of a place disposition is of the two, relatively quicker and probably more natural. However the present writers have subsequently challenged the effectiveness of the Tolman, Ritchie, and Kalish method and have contended that their findings fail to give a comparison of the rates of place and response disposition learning (2). We believe that the relatively slow learning shown by the Tolman, Ritchie, and Kalish response group is attributable to external inhibition deriving from the fact that spatially opposed responses occurred at the locus of the choice point.

The work reported here is an outgrowth of a study in which one of the writers (1) undertook an analysis of the extent and nature of integration of the maze response. The method employed in the earlier study involved the use of a type of maze which will be referred to as a double-pattern maze r 1 one having two start sections, distinct in place, or in pattern, a constant middle section, and two end sections. At that point in the maze where the common middle section joins the two end sections a critical choice is demanded; a left turn, let us say, if the run began at start section one, a right turn if it began at start section two. Now it may be noted (fig. 1, Maze 1) that the central part of the maze preceding the choice point, C, is the same regardless of the kind of response to be made at C. Hence, in our earlier work, it was assumed that a correct response at C could be made only if the maze response were integrated from point C backward through the common section to the different start sections. Also it was thought that the possible extent of such linkage could be studied by changing the length and complexity of the central section.Our results, in the main, were negative. Using both alley and elevated mazes of several degrees of length and complexity, we did not obtain evidence of an ability in white rats to make an appropriate response at the choice point. One exception to this statement must be made. In one instance, using a small maze having-a central section 32" in length, a 16" straight path followed by a ninety degree turn and another 16" straight path leading to C, eight out of a group of twenty-five animals learned the maze. Learners required approximately 30 practice cycles, that is 30 runs on SiKEi and 30 on S2KE2. Similar findings were obtained by Wilson (20), who used a maze like the above in general principle in a study of reaction time and its intervening behavior variables. Wilson reported that his maze, with a central alley 32" in length was extremely difficult and that this difficulty increased with lengthening of the alley.To the present writers these findings seemed to indicate (1) that there is only a slight degree of integration of the component acts in the maze response, or (2) that integration, while occurring in the spatial maze, is for some reason inhibited in the double-pattern maze. At once suggesting itself as a possible cause of inhibition is the fact that, in the double-pattern maze, spatially opposed responses are demanded at one and the same place. Another suggestion might be that the difficulty of the maze results from the absence of differential space 1 More properly it should be called the double-pattern fixed position maze. If this is understood there should be no objection to our using the shorter term.

No abstract