To navigate in the world, animals use various cues [1–3]. Objects or landmarks constitute external cues in the layout ('landscape') of the environment, in reference to which an animal can locate itself [4, 5]. In general, motor behavior has a strong affinity with specific locations in the environment [6], and animals perceive the location of objects and use this information in exploration and navigation in the environment. For example, salient landmarks associated with specific locations function to help animals in returning to particular locations [1], and navigation in rats appears to be primarily based on the geometric arrangement of landmarks [7]. In the present study we manipulated an array of landmarks in order to examine how motor behavior in rats is affected by the number of objects, their density, their spacing and their location in the test environment. Specifically, we asked the question: What in the environment makes rats repeatedly travel to a specific object? Do the rats travel to a specific object or to a specific location? If traveling to a specific location, how is that location determined: in relation to the global room frame or in relation to some local cues?
The above questions were posed not only in the context of exploration and navigation of normal rats, but also in rats treated with the psychoactive drug quinpirole – a dopamine D2/D3 receptor agonist [8–10]. We chose to study navigation and exploration in rats under quinpirole for two reasons: 1) their repetitive locomotion; and 2) their relevance for the study of obsessive-compulsive disorder (OCD). Although quinpirole-treated rats may freely move in a given area, they repeatedly travel between only a few locations [11, 12]. We presumed that such repetitive behavior would unambiguously illuminate how environmental cues shape exploration and navigation – a target of the present study. Specifically, rats treated chronically with quinpirole ceaselessly move between the same few objects/locations, and their locomotion is therefore strongly coupled with the environment. After 10 repeated quinpirole injections (0.5 mg/kg per injection, with injections spaced at 3–4 day intervals) rats were sensitized to the drug, reaching a level of activity that could be as much as 16-fold higher than in controls. This elevated level of activity, however, was confined to a restricted portion of the arena, within which the rats locomoted hurriedly between a few objects/locations, seemingly exploring the environment with unbounded curiosity without habituation or fatigue [11]. Repetitive traveling between the same few objects raises the question of whether the rats travel from one specific object (landmark) to the next, or between specific locations in the environment. The former possibility was tested by moving objects apart, and by adding or removing objects. The latter possibility was tested by relocating the arena to different locations within the test room. Together, these tests were expected to reveal how environmental cues shape this behavior, and what is the contribution of the various facets of an object that make it a target in traveling: its location within the arena or in the room, its distance from other objects, or its distinctness against the background. These facets can be easily discerned in quinpirole-treated rats and compared with saline-treated rats in order to extend the results to exploration and navigation in general.
A second reason to study quinpirole-treated rats was that their behavior is considered a model of OCD [13], in meeting a set of ethologically-derived criteria of compulsive behavior [12]. These criteria were recently applied in studying rituals in OCD human patients [6], revealing that, as in quinpirole-treated rats, compulsive rituals in OCD patients are composed of relatively few motor acts that are organized in a flexible yet recurrent manner. Moreover, rituals in OCD patients feature a strong affinity to a few specific locations or objects [6, 14], as also shown in quinpirole-treated rats [12, 15]. This parallel between OCD patients and quinpirole-injected rats raised the question of what are the physical properties that account for the coupling of motor rituals with specific objects/locations. Identifying such properties was the target of the present study, and for this we tested how rats respond to environmental manipulations. Specifically, rats were expected to change their behavior following a change in object layout through modifying the number of objects, their spacing, density, or location. On a broader level, we asked how the spatiotemporal structure of locomotor behavior is modulated by objects in the immediate vicinity of the animal, and whether compulsive motor rituals are resistant to changes in the layout of local environmental landmarks.
Experiment 1: Rituals and attraction to objects: the effect of the number of objects and inter-object distance under 'elastic' and 'symmetrical' environment enlargement
Experimental design and rationale
Quinpirole-treated rats ceaselessly move between the same few objects/locations, and their locomotion is therefore strongly coupled with the environment. One possibility is that they move from object to object regardless of distance (spacing) between the objects. If so, spacing the objects further apart will not modify the tendency to travel between the objects – that is, piloting between local cues laid in a fixed geometric array. Another possibility, however, is that rats pilot between locations, regardless of the presence of an object in these locations. This implies that the rats are traveling fixed distances, and if an object is spaced further apart, the rats will not reach the new object location but keep traveling to the former location, relying on global cues. Finally, quinpirole-treated rats tend to select only 1–2 objects and move between them while ignoring the other objects [12]. Consequently, we asked why certain objects were selected as the goal in traveling.
The starting point of this set of experiments was at injections 9–10, when the rats were still being injected in the small (1 × 1 m) arena with four equispaced objects. At subsequent injections 11–16 the rats were tested in one of the following environmental settings:
(i) Elastic enlargement:- Large arena (2 × 2 m) with four equispaced objects. This setting reflected an "elastic" enlargement of the small arena in which rats had been tested in injections 1–10.
(ii) Symmetrical enlargement:- Large arena (2 × 2 m) with 16 objects. This setting reflected a constant object density with both area and object number being four-fold greater, as if the small arena had simply been multiplied.
(iii) Fixed area:- Small arena (1 × 1 m) with four equispaced objects as in injections 1–10. This provided a reference for the above enlargements.
Rats were divided into the above three test groups, each comprising: i) a subgroup of 5–7 rats treated with 0.5 mg/kg quinpirole; and ii) another subgroup of 5 rats administered with equivolume of saline. Test group #1 ('elastic' group) was tested first under elastic enlargement (injections 11–12), then under symmetrical enlargement (injections 13–14), and finally retested in the small arena (injections 15–16) where it had been tested in injections 1–10 (Figure 1, top row). Test group #2 ('symmetrical' group) was tested in a reciprocal order (Figure 1, center row), and test group #3 ('fixed' group) was tested in the same small environment throughout (Figure 1, bottom row). Since previous studies had revealed higher variability in quinpirole-treated rats [11], these groups had a greater number of rats than the control.
The rationale for this design was as follows: Test # 1 ('elastic' group) was designed to reveal whether rats travel between objects regardless of inter-object distance or the distance of the objects from arena walls. If the rats are traveling between objects, then the number of stops at objects and route shape will not be affected by this enlargement. However, if the rats travel from object to object without stopping, then interstop distance is expected to increase. Alternatively, they may stop on the way to the now more distant objects, in which case the total number of stops will increase. Test #2 ('symmetrical' group) was designed to reveal whether the rats are equally attracted to all objects. If so, then when the arena is symmetrically enlarged, rats will travel between more objects compared to their travel on arena with fewer objects. A comparison of groups # 1 and # 2 was designed to highlight the effect of testing order. Retesting in the small arena was designed to uncover whether behavioral changes were reversible, and test group # 3 ('fixed' group) provided a control for the above comparisons. Saline groups were tested to distinguish effects that were specific to quinpirole.
Results of Experiment 1
Traveled Distance
Quinpirole-treated rats traveled greater distances compared with saline-treated rats in the three test groups: the 'elastic' group (Two way ANOVA, between-group factor, F1,9 = 38.9; p < 0.001; Figure 2a, left), the 'symmetrical' group (Two way ANOVA, between-group factor, F1,10 = 30.3; p < 0.001; Figure 2a, center), the 'fixed' group (F1,8 = 35.2; p < 0.001; Figure 2a, right). Thus, quinpirole-treated rats, regardless of the setting and environmental changes, traveled greater distances compared with their saline-treated controls.
There were also differences between rats tested first under elastic and then symmetrical enlargement ('elastic' group; within-group factor, F3,27 = 6.3; p = 0.002; Figure 2a, left), and in rats tested first under symmetrical and then elastic enlargement ('symmetrical' group; F3,30 = 16.3; p < 0.001; Figure 2a, center). Indeed, traveled distance increased when rats were first introduced to an enlarged arena, either with four or 16 objects (Figure 2a). When rats in both these groups were retested in the setting of a small arena with four objects, traveled distance reverted to the initial level that these groups had displayed before the changes in environmental setting. This final level was identical with the final level in an environmental setting that was not modified ('fixed group'; Figure 2a, right – see the horizontal dotted reference line). Rats in the 'fixed group' preserved traveled distance at a constant level, similar to that observed when the other groups were tested in the same setting of a small arena with four objects. Thus, there was a typical level of activity in the small arena, which was not affected by previous exposures to a large arena.
The interaction between treatment (quinpirole vs. saline) and changes in environmental setting was not significant in all three test groups: i) in rats that were exposed first to elastic and then symmetrical enlargement ('elastic' group; F3,27 = 0.5; ns); ii) in rats that were exposed first to symmetrical and then elastic enlargement ('symmetrical' group; F3,30 = 3.6 ; ns); and iii) in rats that were repeatedly exposed to the same setting ('fixed' group; F3,24 = 0.6; ns). This constancy confirmed the similar effect of environmental changes on both quinpirole and saline groups (Figure 2a). However, the effect was faint in the less active, saline-treated rats compared with the hyperactive quinpirole-treated rats. In all, the increase in traveled distance was area-dependent, reversible, and independent of the number of objects.
Inter-stop Distance
Rats of the 'elastic' group significantly changed interstop distance over the various test settings (within-group factor in two-way ANOVA; F3,27 = 18.5; P < 0.001). Specifically, they increased interstop distance when introduced into a large arena with four objects spaced at a greater distance than in a small arena; and subsequently, when tested in a large arena with symmetrical object setting and shorter distance between objects, they reduced interstop distance. Thus, interstop distance seemed to be adjusted to inter-object distance (Figure 2b, left), as hypothesized above in the design of this test.
Rats in the 'symmetrical' group also changed interstop distance over the various test settings (within-group factor in two-way ANOVA; F3,30 = 26.2; P < 0.001). Specifically, they increased interstop distance when introduced into the large arena in which there were more objects spaced at a similar distance as that in the small arena. Subsequently, they preserved the same interstop distance when tested in the large arena with four-object setting and greater distance between objects. Thus, these rats seem to adjust interstop distance to arena size and not to inter-object distance (Figure 2b, center). There was no change in controls tested throughout in the small arena (Figure 2b, right).
The interaction between treatment (quinpirole vs. saline) and changes in environmental setting was not significant in all three test groups, confirming a similar effect of setting changes on both quinpirole and saline groups (Figure 2b). However, the effect was faint in the less active saline-treated rats compared with the hyperactive quinpirole-treated rats. In all, interstop distance seemed to be adjusted to inter-object distance in rats tested first under 'elastic' enlargement, and to arena size in the rats tested first under 'symmetrical' enlargement.
The difference between the behavior under 'symmetrical' and 'elastic' enlargement is shown in Figure 3. When quinpirole-treated rats first underwent elastic enlargment, their routes converged upon the objects and/or the left-bottom "start" corner, where the rats had been initially introduced into the arena (Figure 3a). The longer distance between objects resulted in a higher interstop distance (Figure 2b). When then introduced into the symmetrically enlarged arena, their routes still converged upon some of the objects at left-bottom area. The shorter distance among objects and to the left bottom corner in this setting compared with the previous elastic setting could now account for the shorter interstop distance (Figure 2b). When quinpirole-treated rats first experienced the symmetrical enlargement (Figure 3b), they displayed relatively long routes that did not converge upon objects, but passed in between them and typically converged only upon the starting corner. The same structure of routes was preserved when these rats were then introduced into the elastic setting. Since the routes in these rats typically did not converge upon specific objects, interstop distance was not affected by these changes, but adjusted solely to arena size (Figure 2b). Figure 3c depicts the routes of saline-treated rats. Despite their lesser activity, it is notable that routes converged upon objects in saline-treated rats tested first under elastic enlargment, whereas those that were first tested under symmetrical enlargment traveled more between objects (Figure 3c).
Figure 4 depicts the difference between rats that were first exposed to elastic enlargment and apparently traveled from object to object, and those that were first exposed to symmetrical enlargement and traveled in the spaces between the objects. As shown, exposure to a large arena with only four objects resulted in spending extended periods at one or two of the objects, with the time spent between objects not being significantly different from that spent at the most visited object. In contrast, in quinpirole- and saline-injected rats that were exposed to a large arena with 16 objects, time spent in the spaces between objects was significantly longer than time spent at the most visited object.
Experiment 2: Location in reference to arena walls or room setting: 'four-wall' shift or 'two-wall' shift, followed (respectively) by 'translocation' or 'removal' of objects
Experimental design and rationale
The present set of experiments was designed to evaluate how the location of objects in reference to room frame and their distance from the arena walls might affect the spatiotemporal structure of locomotor behavior. For this we preserved the distance between objects, keeping the same array of four objects and spacing as in the first 10 injections, but manipulated the location of the array within the arena, or the location of the arena within the room. The question posed here was whether rituals are performed at specific locations in the global environment, or whether they are coupled with specific objects regardless of their location in the near environment.
Saline-injected rats had become entirely habituated to the environment at this stage of testing (16 successive exposures to the open field). They hardly locomoted, and virtually did not respond to changes in environmental settings. Including the saline-injected groups in statistical comparisons could therefore result in significant differences compared with quinpirole-injected rats, but these differences would be meaningless considering the rats' dull behavior under saline. Therefore, although saline-treated rats were tested and analyzed, the following analyses refer only to quinpirole-treated rats. Behavior of these rats was analyzed to reveal whether they traveled in relation to object location (proximal cues), in relation to arena walls, or to room setting (distal cues). It is noteworthy that quinpirole-injected rats can be challenged over a long period of chronic injections: after the first 10 injections, they preserve their behavioral patterns as long as the chronic injection regime continues, up to 40 injections and even more [11]. To confirm here that they indeed preserve their behavioral patterns, rats were retested in the original 1 × 1 m arena at the beginning and end of this set of experiments. These experiments started with injections 15–16, when rats were injected in the small arena (1 × 1 m) with four equispaced objects. At injections 17–20, rats were tested in one of the following environmental settings:
(i) Two-wall shift:- In switching from a small (1 × 1 m) to a large arena (2 × 2 m) the four equispaced objects remained in the same location near the starting corner, as if two arena walls had been moved further away from the object array, and the objects remained in the same location in the room and in relation to two of the arena walls (Figure 5).
(ii) Four-wall shift:- In switching from a small (1 × 1 m) to a large arena (2 × 2 m) the four equispaced objects remained in the same location in relation to the room frame, as if the four walls had been moved further away from the object array while the array remained in the same location in the room, but now located further away from the arena walls (Figure 5).
(iii) Translocation:- the four-object array was moved to the opposite arena corner.
(iv) Removal:- the four objects were removed from the arena.
(v) Fixed area:- small arena (1 × 1 m) with equispaced four objects as at injections 15–16.
Rats that were injected with quinpirole at injections 1–16, were re-divided into three test groups, balanced for the previous groups of the previous experiment, and arranged in experimental groups as shown in Figure 5.
The rationale of this design was to distinguish whether quinpirole-injected rats were moving in reference to room setting (= a specific region of the room), in relation to arena walls (= a specific region of the arena), or in relation to the object array regardless of its location in the room or in the arena. In Test #4, the object array remained in the same sector in relation to room setting and start corner, but we moved two walls further away. If the rats were just traveling from the start corner to the object array, then the enlargement should not change their behavior, indicating that the increased activity in Experiment 1 was due to object relocation and not due to arena enlargement. In the subsequent phase, rats were challenged with translocation of the object array to the opposite corner. If behavior is solely dictated by the object array regardless of its location in the room and the arena, rats should transfer their activity to the new location of the array. However, if they also organize their behavior in relation to the room and/or arena, they should stretch their routes to the new location of the object array when this is relocated at the opposite corner. In Test group # 5, the same subject matter was addressed. In the four-wall shift, we tested whether the rats were moving in a certain sector of the room (the object array), or also in relation to the arena walls that were now further apart. Finally, the objects were removed to highlight their effect. At the end of this set of experiments, all rats were re-tested in the small arena (injections 21–22) to determine whether behavioral changes were reversible. Test group # 6 ('fixed' group) provided control for the above comparisons.
Results of Experiment 2
Four-wall shift and two-wall shift
Traveled distance significantly increased when the object array remained in the center while the four arena walls shifted away (within-group ANOVA; F3,30= 8.7; P < 0.001; Figure 6a, left), but not when the object array remained in the same location and distance to the start (left bottom) corner with only two walls shifted away (within-group ANOVA; F3,27 = 6.3; P = 0.002; Figure 6a, center). There was no significant change in the group injected in a fixed arena structure (within-group ANOVA; F3,24 = 1.1; P = ns; Figure 6, right). As shown in the trajectories of locomotion of these rats (Figure 7a), they extended their routes from the bottom/left walls to the center, where the object array remained under four-wall shift, but when the object array remained close to the bottom/left walls under two-wall shift, the rats remained in the same area, virtually ignoring the new space added (Figure 7b). Together, responses to these changes demonstrate that the rats were moving from the start corner to the object array, regardless of its location in the arena or in the room. Consequently, in two-wall shift when the object array remained near the start corner, the rats virtually ignore the added space of the enlarged arena (Figure 7).
Removal and Translocation
When the object array was translocated compared with its location under the two-wall shift, the rats profoundly increased traveled distance (Figure 6a, center). This is illustrated in the trajectories of locomotion: these rats remained in a confined sector of the arena when the objects were located near the start corner, extended their routes to the center of the arena when the objects were placed there, and further extended their routes to the further translocated objects (Figure 7). This result reconfirms the above finding that the rats were traveling from the start corner to the array regardless of its location.
When the array was removed, the rats preserved the greater traveled distance that they had displayed under four-wall shift (Figure 6a, left). Their routes in the removal test (empty arena) demonstrate meandering in the center without converging upon any specific location except for the start corner. To some extent, locomotion routes in these rats followed the arena walls, which in the lack of objects were the salient physical structure (Figure 7). Similar effect was obtained when the objects were removed from the small arena in injection 21 (Figure 6a,b; data not shown).
Interstop distance in this set of environmental changes (Figure 6b) corresponded to the changes in total traveled distance (Figure 6a), reconfirming the above finding that the rats were traveling from the start corner to the object array (within-group ANOVA; F3,30 = 17.9; P = 0.0003 for 'four-wall shift'; F3,27 = 14.1; P < 0.001 for 'two-wall shift'; F3,24 = 3.2; P = ns for 'fixed arena'). In other words, the further the objects were located in relation to the start corner, the greater were the overall traveled distance and inter-stop distance, with an insignificant effect of arena size. An illustration of the effect of array location on locomotor activity is shown in Figure 8.