Cerebellar contribution to spatial event processing: do spatial procedures contribute to formation of spatial declarative knowledge?

Subjects

Useful data were collected from 17 HCbed animals, when cerebellar symptomatology stabilized, and from 18 control animals. Experiment 1 included two different protocols that differed in the objects changed in the spatial change or in the object novelty phases. Nine HCbed animals and eight controls were tested according to protocol 1a, and eight HCbed animals and ten controls were tested according to protocol 1b.

Behavioural testing

In both protocols, the first five sessions were identical. To familiarize the animals with the apparatus, during session 1 (S1) each rat was allowed to move within the empty open field, and the baseline level of activity was measured. During S2–S5, the five objects were placed in a square arrangement with a central object (Fig. 2A). Four objects were placed in the middle annulus of the arena and one was placed in the central area. Habituation in object exploration was recorded. Before S6 (spatial change), the spatial configuration was changed by moving two objects.

Results (Experiment 1)

In this section, because S1 and the subsequent four habituation sessions were identical in both protocols, the related results will be pooled and described together.

In S1, control and HCbed animals exhibited a comparable level of anxiety. In fact, defaecation boluses did not show significant differences between groups (F_1,33 = 0.05). The other open-field parameter indicative of a high-stress state is the ‘freezing’ or motionless periods. Also with regard to this parameter no significant difference between groups was found (F_1,33 = 0.14). Of course, high statistical significance was displayed in motor parameters. The HCbed group exhibited a significantly lower number of rears (F_1,33 = 64.70, P < 0.0001) (Fig. 3A). With regard to the distance travelled within the arena, HCbed animals exhibited a significantly lower value. A two-way anova revealed significant effects of group (F_1,33 = 13.19, P < 0.0001) and session (F_8,264 = 24.32, P < 0.0001). Interaction was not significant (F_8,264 = 1.2). It is interesting to note that HCbed animals spent a significantly higher percentage of their path in the peripheral annulus (Fig. 3A). A two-way anova revealed significant effects of group (F_1,33 = 31.66, P < 0.0001) and session (F_8,264 = 15.01, P < 0.0001). Interaction was also significant (F_8,264 = 1.92; P < 0.05).

Figure 4 shows the mean time of contacts with objects in S2–S5 in the two experimental groups. Both groups of animals showed habituation (one-way anovas – controls: F_3,51 = 8.73, P < 0.0001; HCbed animals: F_3,48 = 5.58, P < 0.01). A two-way anova (group × session) revealed significant effects of group (F_1,33 = 28.11, P < 0.0001) and session (F_3,99 = 11.03, P < 0.0001). Interaction was also significant (F_3,99 = 4.81, P < 0.01).

Protocol 1b

In S6, control rats contacted displaced objects much more than nondisplaced objects (Fig. 4). This difference in time spent in contacting the two categories of objects was reduced, but maintained, in S7. Again, HCbed animals behaved as if no object displacement had occurred. In S6 and S7, they showed no significant difference between object categories. A three-way anova (group × session × object) revealed significant effects of group (F_1,16 = 7.87, P < 0.05), session (F_1,16 = 6.69, P < 0.05) and category of object (F_1,16 = 13.98, P < 0.01). A post-hoc analysis revealed a significant difference between displaced vs. nondisplaced objects only in control animals.

In S8, control animals reacted markedly to object substitution, contacting the novel object significantly more (Fig. 4). It is interesting to note that HCbed animals did not reach the significance level in contacting the new object in S8, but explored the novel object significantly more than the familiar ones in S9. A three-way anova (group × session × object) revealed significant effects of group (F_1,16 = 7.51, P < 0.05) and category of object (F_1,16 = 30.98, P < 0.0001). Session effect was not significant (F_1,16 = 2.55). Interaction among factors was significant (F_1,16 = 8.80, P < 0.01). A post-hoc analysis revealed a significant difference between new vs. familiar objects in both control and HCbed rats.

Pooling data obtained with both protocols, we analysed the procedural parameters of the two experimental groups (Fig. 5). The control animals mainly put into action an explorative strategy that traversed more than once the centre of the arena, to contact all objects (CE 5 Ob), regardless of their spatial localization. Conversely, HCbed animals mainly performed strategies linked to the exploration of the peripheral portions of the arena (C, PE 2 Ob, PE 5 Ob), and the central object was often completely neglected.

Discussion (Experiment 1)

In the presence of a cerebellar lesion, many postural and motor deficits are inescapably present. In spite of this, the HCbed animals succeeded in moving inside the arena and in contacting the objects. They habituated to the new environment as with the control animals. This observation is in agreement with data reported by Dahhaoui et al. (1992) in cerebellectomized rats, and by Caston et al. (1998) in intact or cerebellectomized Lurcher mutant mice, suggesting that the cerebellum is not involved in the habituation of exploration behaviour. The explorative pattern displayed by HCbed animals was markedly different from that of the controls. The HCbed rats persisted in exploring the outer annulus, even in the last session of the habituation phase, very rarely visiting the inner areas (Fig. 3A). They made few contacts, especially with the object in the centre of the arena. In sessions 6 and 8, cerebellar animals exhibited significantly decreased ability to react to a new arrangement (Fig. 4), whereas the control rats reacted to changes and spent more time contacting the displaced or new objects. This reduced amount of activity could be an indication of increased level of anxiety of HCbed animals. To test this hypothesis, defaecation and freezing were used. Defaecation, the parameter for which the open field was designed, is one, if not the most widely used, measure together with the prime index of emotionality, the validity of its role having been confirmed by factor analytic studies (Whimbey & Denenberg, 1967). The other parameter usually taken as indicative of a high-stress state is the absence of movement or freezing, a behavioural response occurring in the face of perceived danger. Elements such as spatial dislocation or abrupt change are reported to be particularly potent eliciting factors (Walsh & Cummins, 1976). However, both indexes failed to reveal any difference between HCbed and control animals. It is interesting to note also that nervous mutant mice that display a loss of Purkinje cells do not demonstrate freezing responses (Lalonde & Botez, 1985). Thus, the emotional reactivity seems to be not markedly altered by the cerebellar lesion, at least under some experimental conditions, in which, for example, before testing cerebellar as well as control animals were repeatedly handled to reduce fear reactions and timidity.

When an animal explored only some objects and neglected the others, it is reasonable to expect a reaction to changes in the former objects and not in the latter ones. To overcome the eventuality of bias in object exploration, we planned protocol 1b, in which only the most contacted objects were changed. In this protocol, the cerebellar animals never recognized spatial changes in S6 and S7 and succeeded in detecting object novelty only in S9. Behaviours characterized by prevalently peripheral exploration and by reduced (if any) reaction to environmental changes can be provoked by different factors. Theoretically, this poorly reactive spatial behaviour can result from a high level of anxiety, a low motivation to explore, or from an impairment in building the spatial place competence of the environment, perhaps linked to a procedural impairment in putting efficient explorative strategies into action. As mentioned above, in the open field task HCbed animals did not appear very anxious, considering either specific parameters, such as defaecation and freezing, or their general behaviour in the arena. The second factor inevitably affecting exploration behaviour is a motivational factor. Recently, it has been reported that Lurcher mutant mice, whose cerebellar cortex is lacking almost all Purkinje and granule cells, exhibited a low level of exploration in a hole-board matrix (Caston et al., 1998). Furthermore, nervous mutants exhibit deficits in spontaneous alternation rates (Lalonde & Botez, 1985). These deficits were interpreted as due both to a decreased motivation to explore a new environment and to spatial deficits. In fact, in addition to its role in cognitive functions (Schmahmann, 1997), such as executive function, memory, visuospatial analysis and language, it has been suggested that the cerebellum is also implicated in motivational states determining the motivational disorders associated with neurodevelopmental pathologies, such as autism (Pierce & Courchesne, 2001). Although it cannot be ruled out that some motivational factors contributed to affect the exploration of HCbed animals in the open field task, some consideration of the features of protocol 1b strengthen the hypothesis of a spatial deficit in the presence of a cerebellar lesion. In fact, protocol 1b, in which just the most contacted objects were changed, emphasizes even low levels of motivation, allowing reducing the motivational load in the explorative function. If a subject contacts only two objects, it is certainly possible that its reduced attitude to explore is provoked by low levels of motivation. Regardless, however low it may be, a motivational drive high enough to explore those particular objects is present. The same motivation should operate when just these objects are changed.

The third factor theoretically involved in the difficulty displayed by HCbed animals in reacting to environmental changes is the impairment in forming the declarative knowledge of the environment, suggesting that cerebellar networks could be involved in processes of forming and updating spatial relationships among environmental stimuli. However, before attributing this role to cerebellar circuits, the role played by procedural competencies in acquiring the spatial declarative competence must be analysed, because of the already demonstrated cerebellar function in gaining procedural competence. As repeatedly observed (Petrosini et al., 1996, 1998; Leggio et al., 1999), HCbed animals are severely impaired in acquiring the appropriate spatial explorative procedures (Fig. 5). Indeed, it is possible that the subject cannot acquire a declarative competence in the presence of a cerebellar lesion, because the cerebellar networks are not only involved in procedural learning but also in the representational phase of a new environment (neither declarative nor procedural learning). Conversely, it is possible that a subject that is not able to explore a new environment appropriately, in the presence of a cerebellar lesion, will not be able to represent it either (no declarative learning as a result of no procedural learning). To discriminate between hypotheses, the environment was modified according to the impaired spatial procedures of cerebellar animals.

Methods (Experiment 2)

Results (Experiment 2)

As in experiment 1, in S1, control and HCbed animals exhibited a comparable level of anxiety, as indicated by no significant difference in defaecation boluses (F_1,19 = 1.22) or in motionless periods (F_1,19 = 0.14). Of course, high statistical significance was obtained for motor parameters. Groups of animals significantly differed in rearing behaviours (F_1,19 = 38.5, P < 0.0001), because HCbed animals exhibited fewer rears than controls. HCbed animals displayed a lower value for distance travelled in the arena (Fig. 3B). A two-way anova revealed a significant effect of group (F_1,19 = 5.91, P < 0.05) and session (F_8,152 = 17.5, P < 0.0001). Interaction was also significant (F_8,152 = 2.05, P < 0.05). It is interesting to note that with the new spatial arrangement both experimental groups spent the highest percentage of their path in the peripheral annulus. A two-way anova revealed a significant effect only for session (F_8,152 = 2.57, P < 0.05). The effects of group (F_1,19 = 2.00) and interaction (F_8,152 = 1.45) were not significant, demonstrating that the exploration of the cerebellar animals in the arena was similar to that of controls. As shown in Fig. 6, which depicts the mean time of contacts in S2–S5 with all five objects, both groups of animals showed habituation (one-way anovas – control group: F_3,24 = 19.02, P < 0.0001; HCbed animals: F_3,33 = 27.66, P < 0.0001). A two-way anova revealed no significant group effect (F_1,19 = 3.25), whereas session effect (F_3,57 = 43.60, P < 0.0001) and interaction (F_3,57 = 3.82, P < 0.05) were significant. In S6, both control and HCbed rats contacted displaced objects significantly more than nondisplaced ones. This difference in contacting the two object categories was maintained in S7, although it was reduced. A three-way anova (group × session × object) revealed no significant effect of group (F_1,19 = 4.37) and session (F_1,19 = 0.02). Category of object displayed a significant effect (F_1,19 = 22.48, P < 0.0001). Interactions between and among factors were not significant. Post-hoc analysis revealed significant differences between displaced vs. nondisplaced objects in control and HCbed animals

A similar pattern was also observed in object novelty detection (Fig. 6). In S8, both control and HCbed animals reacted to object substitution, contacting the novel object significantly more often. This pattern was maintained in S9. A three-way anova (group × session × object) revealed no significant effect of group (F_1,19 = 0.48) and session (F_1,19 = 3.7). Category of object displayed a highly significant effect (F_1,19 = 31.7, P < 0.0001). Interactions between and among factors were not significant. Post-hoc analysis revealed significant differences between new vs. familiar objects in control and HCbed rats.

When procedural parameters were analysed, a marked analogy in the explorative patterns of the two experimental groups was noted (Fig. 5). Both control and HCbed animals mainly chose the explorative strategy that contacted all objects through peripheral trajectories, and avoiding ‘expensive’ and unmotivated passing through the centre of the arena.

Discussion (Experiment 2)

The findings of Experiment 2 indicate that HCbed animals were able to detect environmental changes once their altered procedures were favoured by object arrangement. Cerebellar animals reacted to both spatial and object changes, just as did controls. This finding strongly suggests that motivational factors, even if present, are not the main cause of the impaired exploration displayed in the Experiment 1. It is interesting to note that in this new spatial arrangement even their explorative strategies were identical. This indicates that a subject is not able to represent a new environment when it is not able to explore it appropriately, as in the case of a cerebellar lesion. This suggests that no declarative learning is possible without the appropriate procedural learning.

These data extend earlier findings on the involvement of cerebellar circuits in procedural spatial learning (Petrosini et al., 1996, 1998; Leggio et al., 1999; Mandolesi et al., 2001), showing that the spatial procedural impairment resulting from a cerebellar lesion conditions the processes of forming and updating spatial relationships among environmental stimuli. Altered explorative strategies make the acquisition of the declarative knowledge of the environment extremely demanding. This impairment is so marked that no detection of environmental changes was evidenced in Experiment 1, in which no attempt was made to foster the procedural explorative behaviour of HCbed rats. Conversely, when the environment was shaped according to the explorative abilities of the cerebellar animals, and thus their procedural deficits were bypassed, as in Experiment 2, cerebellar animals were able to react to environmental changes. Once they were allowed to explore a new environment keeping to their own strategies, no significant difference between experimental groups was found regarding reaction to change or procedural components put into action.

This finding indicates that the acquisition of the declarative competence of an environment does not require cerebellar involvement per se. Cerebellar mediation is fundamental in acquiring the procedural competence that appears to be a prerequisite of any localizatory knowledge. In fact, to understand an environment fully, in addition to learning to locate specific cues (localizatory learning), it is necessary to learn how to move in that environment (procedural learning). Procedural components include behaviours such as suppression of fear in an open space, leaving peripheral areas, development of efficient explorative strategies, and acquisition and utilization of snapshots of the target view and of the representation-forming procedures (Fenton & Bures, 1993). The localizatory and procedural aspects of spatial learning are widely inter-related, even if multiple and differentiated brain areas contribute to them. The most influential theory on spatial learning suggests that hippocampal areas have a selective role in high-order components, such as place learning (O'Keefe & Nadel, 1978; Jarrard, 1993; O'Keefe, 1993; Moser et al., 1995; Stupien et al., 2003) or processes underlying the establishment of relational representations (Sutherland & Rudy, 1989; Eichenbaum et al., 1994). By contrast, hippocampal-independent mechanisms, suggested to be involved in those forms of learning related to procedural components, bring brain structures into play that are primarily involved in sensorimotor functions, such as the cerebellum (Thompson & Kim, 1996; Guillou et al., 1999). These competencies represent the environment in a different manner. Through localizatory knowledge, the environment is described independently from action (for example, cue X is near cue Y) (Schenk & Morris, 1985). Conversely, through procedural knowledge, spatial competence is built into movements performed within the environment. In this case, the building of the spatial map may be defined within a distinct reference frame, defined by self-motion (for example, cue X is three steps forward from here). Thus, the contribution of spatial procedures to the mental representation of an environment is not location-based, but route-based. According to the present results, procedural aspects are necessary appropriately to organize spatial information and to represent an environment. It is usually held that mammals organize complex spatial information into coherent reference frames that are used to guide behaviour efficiently.

The present data indicate that efficient behaviour is used to guide the building of the internal representation of the environment.