Behavioural Brain Research, 38 (1990) 7-18 Elsevier BBR 01036
Classical conditioning of the eyeblink reflex in the decerebrate-decerebellate rabbit T h o m a s M. Kelly, C h e n g - C i Z u o * a n d J a m e s R. B l o e d e l Division of Neurobiology, Barrow Neurological Institute, Phoenix, A Z 85013 (U.S.A.) and Department of Physiology, University of Arizona College of Medicine, Tucson, A Z (U.S.A.) (Received 12 June 1989) (Revised version received 9 November 1989) (Accepted 10 November 1989)
Key words: Cerebellum; Nictitating membrane reflex; Classical conditioning; Learning; Brainstem; Eyeblink; Rabbit
The purpose of these experiments was to test the hypothesis that a conditioned nictitating membrane reflex can be acquired in decerebrate rabbits in the absence of the cerebellum. Experiments examining the effects of large cerebellar lesions on the acquisition and performance of the conditioned reflex were performed in acutely prepared decerebrate rabbits. Most lesions encompassed all of the cerebellar nuclear regions ipsilateral to the eye receiving the unconditionedstimulus. In all rabbits included in this study the continuity between the cerebellar nuclei and the bralnstem was interrupted, even in those preparations in which small regions of the nuclei were present in the lateral hemisphere. The findings demonstrate that these animals could acquire the conditioned reflex independent of whether conditioning had occurred prior to the cerebellectomy. Strong associativity was found between the latency of the conditioned response and the interstimulus interval between the conditioned and unconditioned stimuli. The behavior of the conditioned reflex observed in the decerebrate-decerebellate animals differed from that reported for awake intact rabbits in two ways. Once the conditioned behavior had been acquired, the percent of trials showing conditioned responses was somewhat less in decerebrate-decerebellate rabbits and was also more variable in these animals. The data demonstrate that the nictitating membrane reflex can be classically conditioned in the absence of the cerebellum, indicating that this structure is neither necessary nor sufficient for the acquisition of this type of conditioned behavior. In addition, an hypothesis is presented which addresses the difference between the data reported here and those previously reported by other laboratories based on observations in awake intact animals.
INTRODUCTION
During the past decade principally two lines of investigation have been used to address the general issue of the neuroanatomical substrate required for motor learning: the adaptation of the vestibuloocular reflex and the classical conditioning of the nictitating membrane/eyeblink reflex in the rabbit. Although the majority of the most recent studies examining the substrate for
vestibulo-ocular reflex (VOR) adaptation (referenced below) indicate that the plastic changes underlying this phenomenon likely occur in the brainstem, the data from the experiments examining the substrate required for conditioning the nictitating membrane reflex have been interpreted as indicating that the necessary memory trace is established within the cerebellum. The experiments of McCormick and Thompson 2s,29 were among the first to indicate that lesions restricted
* Present address: Department of Biology, Nanjing University, People's Republic of China. Correspondence: J.R. Bloedel, Division of Neurobiology, Barrow Neurological Institute, 350 West Thomas Road, Phoenix, AZ 85013, U.S.A. 0166-4328/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
to specific regions of the cerebellum result in the elimination of this conditioned reflex while leaving the characteristics of the unconditioned reflex unchanged. In these studies lesions in the lateral interposed/medial dentate region of the cerebellar nuclei were the most effective in permanently eliminating either the acquisition or the execution of the conditioned reflex. Yeo and colleagues 41-43 reported similar observations. However, these investigators argued that the critical area was localized to lobule HVI of the cerebellar cortex. Subsequent to these observations several lines of corroborative findings have been reported. For example, the acquisition and execution of the nictitating membrane reflex was disrupted by lesions of the red nucleus 34,35, the structure receiving inputs from the interposed and dentate nuclear regions ablated in the studies of McCormick and Thompson 28'29. These findings now have been supported by related findings in several laboratories 7,8,19-21,36.
Our studies were motivated by several observations from our laboratory as well as by a re-interpretation of the existing data in the literature. First, the findings implicating the cerebellum as the site of memory storage required for nictitating membrane reflex conditioning are negative in nature, i.e. they are based on the failure to elicit conditioned responses following cerebellar lesions. These reported effects could be due to modified interactions at extracerebellar sites receiving projections from the cerebellar nuclei rather than the consequence of eliminating the location of the critical memory trace. Decerebellation has long been known to produce changes in the tonic excitability of neurons in several brainstem nuclei that feasibly could play a role in the acquisition of conditioned behavior 9. Second, the evolving research examining the substrate for vestibulo-ocular reflex adaptation progressively points more towards the plastic changes occurring at sites within the brainstem rather than the cerebellum 22'23'3°'31 despite the arguments to the contrary early in the study of this question. Third, deficits in motor memory are not commonly reported following lesions of the cerebellum in patients 13. Fourth, studies from our laboratory provide direct evidence that the cerebellum is not
required for the acquisition for at least some types of conditioned behavior. These experiments were based on studies showing that decerebrate ferrets ambulating on a treadmill could be conditioned to step over a bar placed repeatedly in the trajectory of the limb's swing phase 24. Ferrets that underwent near-total cerebellectomy at least two months before still could acquire this conditioned forelimb movement6. Last, our investigations into the action of the climbing fiber system 4"5'~°'~1 revealed that the mechanism most frequently touted as the basis for establishing the engrams within the cerebellum, namely a long-term suppression of Purkinje cell excit ability p roduc ed b y the s e afferent s ~'~2. ~~- ~~. 26, does not occur under physiological conditions. Rather these afferents were shown to produce a short-lasting increase in the responsiveness of Purkinje cells to the mossy fiber-evoked parallel fiber input. These data were acquired using paradigms based on either spontaneously occurring or naturally evoked climbing fiber inputs rather than climbing fiber inputs evoked electrically at unusually high rates and train durations. An operational rather than a memory function for the climbing fiber system is also supported by recent studies investigating the relationship of climbing fiber activity to characteristics of volitional arm movements. In two different paradigms complex spike responses of Purkinje cells were found to be related to features of the movement other than the time course of its learning or acquisition 25'37. In studies done in parallel with those reported here, Welsh and Harvey 38 investigated the additional possibility that the effects of cerebellectomy on the conditioned nictitating reflex response were due to a performance deficit rather than a deficit in the animal's capacity to acquire the conditioned behavior. This premise was supported in part by showing parallel changes in the conditioned as well as the unconditioned reflexes. The study reported in this manuscript addresses the critical role of the cerebellum in the plastic changes required for the classical conditioning of the rabbit's nictitating membrane reflex by asking a very fundamental question using a decerebrate rabbit preparation: namely whether
the cerebellum is in fact both necessary and sufficient for the acquisition and/or execution of this conditioned reflex. An acute decerebrate preparation was chosen as the experimental model for several reasons. First, it provided an opportunity to evolve a preparation that could be used for prolonged, stable intracellular recording during the acquisition, performance and extinction of the conditioned reflex. Second, using an acute decerebrate preparation made it possible to examine the effects of cerebellectomy immediately after the removal of the cerebellum before the occurrence of any modifications in reflex characteristics resulting from adaptive changes in the brainstem. Last, the decerebrate preparation eliminated interactions with structures above the colliculi. Consequently if the reported effects of cerebellectomy on conditioned reflexes were related to alterations in the excitability of the conditioned reflex pathways outside the cerebellum, decerebration may fortuitously modify those tonic changes in the opposite direction, thus increasing the likelihood of observing a conditioned reflex. The data will show that in acutely decerebrate rabbits the nictitating membrane reflex can be classically conditioned even in the absence of the cerebellum. Differences between the behavior of the conditioned reflex in these preparations and that described in the intact rabbit are also presented and discussed. The data provide strong evidence against the fundamental notion that the plastic changes required for the conditioning of the nictitating membrane reflex of the rabbit occur within specific locations of the cerebellar cortex and/or nuclei. Although the acute ablation of the cerebellum unequivocally modifies properties of these reflexes, it does not eliminate either the capacity to acquire or execute this conditioned behavior. MATERIALS AND M E T H O D S
The experiments were conducted in white, adult New Zealand rabbits. Prior to administering halothane anesthesia, the animals were given 0.1 ml of atropine subcutaneously in order to
suppress tracheal secretions following intubation. Once the animal was adequately anesthetized with Halothane, a tracheostomy was performed, and the animal was artificially respired with continuous administration of the anesthetic. Next, a glue cap attachment was constructed onto the anterior part of the cranium so that a Kopf semichronic headholding device could be employed to stabilize the animal's head during the experiment. Following a craniectomy the posterior halves of both cerebri were exposed. The occipital and part of the parietal lobes overhanging the colliculi were then removed carefully by suction. Next a stereotaxic coagulating electrode was positioned at the anterior edge of the superior colliculus and angled toward the mamillary bodies. Decerebration was performed by placing coagulation lesions in a 1.5 x 1.5 mm grid across the entire cross-section of the brainstem. Also while under anesthesia an extensive craniectomy was performed over the cerebellum, exposing enough of this structure to permit the removal of cerebellar regions by suction. The lesion included most of the cerebellar tissue ipsilateral to the eye receiving the air puff stimulus. The lesion was extended lateral enough to disconnect any remaining cerebellar structures associated with the lateral hemisphere from continuity with the superior peduncle. Conditioned stimuli consisted of a tone, usually at a frequency of 500 Hz, that was applied from the onset of the trial until the unconditioned stimulus was terminated (delay conditioning paradigm). Two types of unconditioned stimuli were used with no difference in the characteristics of conditioning. In some experiments, airpuffs 100 ms in duration were applied to the cornea of the eye. In the other studies the eyeblink reflex was evoked using a bipolar electrical stimulus applied to the posterior part of the upper and lower eyelids. Stimuli usually consisted of a single pulse 0.5 ms in duration at an intensity just above threshold for producing an eyeblink on every trial. Intensities that evoked movements in other facial or neck regions were not used. Typically an interval of 350 ms between the onset of the tone and the onset of the unconditioned stimulus was employed. However, this interval varied depending upon the specific objectives of the experiment (see
10 Results section). Intertrial intervals of 8-10 s were used throughout these experiments. Two methods were employed to assess the reflexes recorded in these experiments. In some experiments (4 of the animals included in this study) the eyeblink reflex was measured using either a very thin wire directly inserted into the orbicularis oculi muscle or a flexible wire electrode that was hooked over the upper eyelid under just enough tension to produce a mild retraction. In the remaining experiments (4 animals) the retraction of the eyeball was monitored directly using an infrared emitter-receiver system that was placed in close proximity to the eye receiving the airpuff stimulus. Retraction was detected by recording changes in the reflection of the infrared light signal from the corneal surface. In both techniques the criterion for designating a response as a CR was its occurrence within the C R - U C R interval with an amplitude at least 10~/o of the average UCR. Because the objective of this study was confined to determining whether or not conditioning of the eyeblink reflex could occur in the decerebellatedecerebrate preparation, a detailed analysis of the amplitude changes of the CR and U C R was not included in this report. The animal was permitted to recover from anesthesia for a period of approximately one-half hour before any trials were initiated. A control period followed in which several blocks of unpaired tone and airpuff or electrical stimuli were applied. These data were carefully analyzed to ensure that the changes in the presumed conditioned response reported below were not due to pseudoconditioning or sensitization. When the control trials were completed, paired trials were initiated in order to determine the capacity of the decerebrate animal to acquire the conditioned behavior before the effects of cerebellectomy were assessed. Conditioning was measured by determining the number of conditioned responses present in blocks of 10 trials. If conditioning were not present in the decerebrate preparation with the cerebellum intact, no cerebellectomy was performed, and the experiment was discontinued. However, if the animal was capable of acquiring the conditioned behavior, halothane anesthesia
was again administered, and the cerebellum was removed. Immediately after the surgical procedure the conditioned behavior and/or its acquisition was evaluated. Consequently in most animals paired stimuli were reinitiated directly after the cerebellectomy procedure. In two animals the above protocol was modified in order to determine the capacity of the naive decerebrate-decerebellate preparation to acquire the conditioned behavior. In these experiments no paired trials were applied until after the decerebellation, which was performed immediately following the decerebration before the anesthesia was discontinued. After each experiment the animals were injected with an overdose of barbiturate and perfused intracardially with 10~o formalin solution. Transverse sections 4 0 # m thick were taken through the site of the cerebellar lesions and stained with Cresyl violet. The borders of the lesions were carefully drawn in order to determine the cerebellar regions removed at the time of the experiment. RESULTS
In general it was difficult to maintain decerebrate rabbits in a stable condition in which consistent acquisition and performance of the nictitating membrane conditioned reflex could be observed. Although not systematically studied, this difficulty appeared to be primarily related to the effects of decerebration on the animal's respiratory system. Because respiratory failure led to the early termination of many preparations, assessing the incidence of conditioning in cerebellectomized animals as a percent of the total number of animals used in the study would be misleading. It is more meaningful to characterize this feature of the experiments by calculating the percent of decerebrate-decerebellate animals that acquired the conditioned behavior relative to the total number of animals that also displayed the behavior prior to cerebellectomy. Consequently, as stated in Methods, the experiment was discontinued in any decerebrate animal incapable of acquiring the conditioned eyeblink reflex. Using this as a criterion, 5 of 5 decerebrate animals that acquired
11 the conditioned reflex behavior also manifested the acquisition and performance of the conditioned reflex following cerebellectomy. In the other 3 animals included in this study, conditioning was not attempted until after cerebellectomy. As will be reported below, these animals also acquired the conditioned reflex behavior. The data in Fig. 1 illustrate the conditioning that occurred in these decerebrate-decerebellate preparations as well as the type of experimental observations that were measured in order to obtain the plots in the subsequent figures. The records in A and B were obtained from trials after pairing of the conditioned and unconditioned stimuli was initiated but before maximal conditioning had been achieved. In this experiment the conditioned stimulus was a 500-Hz tone and the unconditioned stimulus was an air puff. The reflex was monitored in this experiment using an infrared emitter/sensor that detected retraction of the eyeball. In A only an unconditioned response was evoked in this paired trial during the initial acquisition phase. A trial in which a conditioned
A
response was evoked is shown in B. Characteristically there was usually a modest reduction in the amplitude of the unconditioned response when the conditioned response was evoked at a relatively long latency following the tone. Once the conditioned behavior was acquired, conditioned responses could be evoked in a tone-only trial during the initial stages of extinction (C). As the tone-only trials continued, extinction became complete, and the conditioned stimulus no longer evoked the reflex. A plot characterizing the time course of the acquisition and extinction of the eyeblink reflex following cerebellectomy is shown in Fig. 2 for an animal that had been conditioned prior to this surgical procedure. Data were quantified by plotting the percent of trials having conditioned responses over each block of 10 trials. In this experiment conditioning occurred over a total of 80 trials (8 blocks). Typical of these preparations the percent of trials in which there were conditioned responses varied from block to block much more dramatically than in intact preparations. This feature will be addressed in greater detail below. Nevertheless there was clearly a dif-
C Tone Only
CR Failure
80"
Tone
Control Paired
÷r
+r
ts
B CR-UCR
D
60-
Tone Only-Extinction
nO
~o 9 r
~ s
o
m.,,c
t'
6oo
T
Fig. 1. Conditioned and unconditioned responses in the decerebrate, decerebellate rabbit obtained from a series of trials in one conditioning-extinction sequence. A: in this paired stimulus trial, only an unconditioned response was evoked prior to conditioning. B: after conditioning, a conditioned response occurred after the tone but before the air puff and was followed by a smaller unconditioned response evoked by the air puff. C: a conditioned response unaccompanied by an unconditioned response was evoked by the tone in the early stages of extinction. D: no response to the conditioned stimulus was observed after extinction had taken place. T: the conditioned stimulus, a 500 Hz tone. S: the unconditioned stimulus, an air puff directed at the cornea. Responses were recorded with a coupled infrared emitterdetector system aimed at the cornea.
2O
0
20
40 Block
60
8'0
(10 trials/block)
Fig. 2. Acquisition and extinction of the conditioned eyeblink response in a decerebrate, decerebellate rabbit. The percentage of conditioned responses is shown for each block of 10 trials. Control blocks were comprised of alternating conditioned (tone) and unconditioned (air puff) stimuli. During paired trials the air puff was applied 200 ms after onset of the tone. The conditioned behavior was extinguished by presenting the animal with a series of tone-only trials. Measurements of the reflex behavior were made using the E M G technique described in the Methods.
12 ference in the percent of trials with conditioned responses between the control period and the period in which the conditioned behavior was maximal. This difference was statistically significant in all animals considered to display conditioned behavior. In this experiment the difference was statistically significant at a level of P < 0.001. When the paired trials were discontinued, the conditioned behavior clearly extinguished with the occurrence of conditioned responses returning to control values. The characteristics of the conditioning and extinction were very repeatable in each animal. Consequently once the behavior was extinguished it could be reacquired when the conditioned and unconditioned stimuli were again paired. An example of this is shown in Fig. 3A. As was the case in all but one of the animals studied, the conditioned responses were not evoked in the first paired trials immediately following cerebellectomy. Customarily this surgical procedure took approximately 20 min and was always performed while 0.5~o halothane anesthesia was administered. These factors may have contributed to the lack of savings in most preparations. In the experiment shown in Fig. 3A, conditioned responses were present only in a small percentage of trials initially. The percentage increased over 80 trials to a mean percentage of 94 ~o once the conditioned behavior was acquired. When only the conditioned stimulus was applied, the behavior extinguished. Once extinguished the behavior could be reconditioned by resuming the pairing of the conditioned and unconditioned stimuli. In this preparation this was repeated 3 times before the condition of the animal deteriorated. Similar to the data in Fig. 2, once a maximal level of conditioning was achieved there was a considerable variability in the percent of trials having conditioned responses compared with the same phenomenon in the intact rabbit. In order to provide an overview of the time course of conditioning and its extinction in the decerebellate-decerebrate preparation, a composite plot was calculated from the measurements in 4 animals. The graph was assembled by first dividing the learning curve into 4 phases: (1) control, (2) acquisition, (3) plateau, (4) extinction.
Paired
Tone Only 100 80 ¸ o
60 40
Q.
20 0
5
A °
10
15 Block
2'0
2'5
3'0
1 O0
120
(50 trials/block) Paired
Tone Only .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
100" 80" 60" c
~,
40-
n
20-
B
2'0
40
60 Block
8'0
(10 trials/block)
Fig. 3. Time course of repeated cycles of conditioning and extinction in a decerebrate, decerebellate rabbit and the variability of the conditioned behavior. A: the sequence of acquisition and extinction of the conditioned reflex is plotted with a resolution of 50 trials/block. B: plot of the first two conditioning-extinction cycles shown in A at an increased resolution of 10 trials/block. This animal had been conditioned prior to cerebellectomy. Note the rapid reacquisition of the conditioned behavior following cerebellar ablation. Responses were measured using the infrared emitter system.
Both the control and acquisition phases were aligned on the onset of acquisition. The 10 blocks prior to the onset of acquisition comprised the control portion of the graph, and the length of the other phases equalled the mean number of blocks taken by the 4 animals to complete each phase of the behavior. For each animal, individual phases of the learning curve were normalized to the average phase length. Then the various phases were averaged across the 4 animals to form the mean learning curve. This composite plot is shown in Fig. 4. Notice that the time course for conditioning and ex-
13 Paired 90" 80" 70-
¢
60-
(3 50"
~" 30" 20" 10" 0
10
20 Block (10 trials/block)
30
4'0
mal, the behavior in the decerebrate-decerebellate rabbit differed from the behavior in the intact rabbit ~4 in tw6 important ways. First, there was appreciably more variability from block to block in the reduced preparations. Second, the percentage of trials displaying conditioned responses did not approximate 100~o in most animals. Although not a primary objective of this study, one consistent finding across all animals (illustrated in Fig. 3) provides some insight into the basis for both of these differences. Fig. 3B was plotted from the same data as that used in'Fig. 3A; however, the data were grouped
Fig. 4. Average time course of conditioning. The normalized conditioning curves from 4 animals were averaged according to the procedure discussed in the text. Each point on the graph is the mean number of conditioned responses in a ten trial block. Error bars indicate standard deviation.
500
Decerebrate Rabbit x=418.1
400
tinction are approximately symmetrical, occurring over approximately 8 blocks of trials. The plateau calculated across these 4 animals also illustrates another characteristic finding: namely that the level of conditioning never reached 100 ~/o. Based on the data presented in Figs. 2-4, it is apparent that once the conditioning of the nictitating membrane/eyeblink reflex became maxi-
E
300 ¸
//
t=-9,244 p
Classical conditioning of the eyeblink reflex in the decerebrate-decerebellate rabbit T h o m a s M. Kelly, C h e n g - C i Z u o * a n d J a m e s R. B l o e d e l Division of Neurobiology, Barrow Neurological Institute, Phoenix, A Z 85013 (U.S.A.) and Department of Physiology, University of Arizona College of Medicine, Tucson, A Z (U.S.A.) (Received 12 June 1989) (Revised version received 9 November 1989) (Accepted 10 November 1989)
Key words: Cerebellum; Nictitating membrane reflex; Classical conditioning; Learning; Brainstem; Eyeblink; Rabbit
The purpose of these experiments was to test the hypothesis that a conditioned nictitating membrane reflex can be acquired in decerebrate rabbits in the absence of the cerebellum. Experiments examining the effects of large cerebellar lesions on the acquisition and performance of the conditioned reflex were performed in acutely prepared decerebrate rabbits. Most lesions encompassed all of the cerebellar nuclear regions ipsilateral to the eye receiving the unconditionedstimulus. In all rabbits included in this study the continuity between the cerebellar nuclei and the bralnstem was interrupted, even in those preparations in which small regions of the nuclei were present in the lateral hemisphere. The findings demonstrate that these animals could acquire the conditioned reflex independent of whether conditioning had occurred prior to the cerebellectomy. Strong associativity was found between the latency of the conditioned response and the interstimulus interval between the conditioned and unconditioned stimuli. The behavior of the conditioned reflex observed in the decerebrate-decerebellate animals differed from that reported for awake intact rabbits in two ways. Once the conditioned behavior had been acquired, the percent of trials showing conditioned responses was somewhat less in decerebrate-decerebellate rabbits and was also more variable in these animals. The data demonstrate that the nictitating membrane reflex can be classically conditioned in the absence of the cerebellum, indicating that this structure is neither necessary nor sufficient for the acquisition of this type of conditioned behavior. In addition, an hypothesis is presented which addresses the difference between the data reported here and those previously reported by other laboratories based on observations in awake intact animals.
INTRODUCTION
During the past decade principally two lines of investigation have been used to address the general issue of the neuroanatomical substrate required for motor learning: the adaptation of the vestibuloocular reflex and the classical conditioning of the nictitating membrane/eyeblink reflex in the rabbit. Although the majority of the most recent studies examining the substrate for
vestibulo-ocular reflex (VOR) adaptation (referenced below) indicate that the plastic changes underlying this phenomenon likely occur in the brainstem, the data from the experiments examining the substrate required for conditioning the nictitating membrane reflex have been interpreted as indicating that the necessary memory trace is established within the cerebellum. The experiments of McCormick and Thompson 2s,29 were among the first to indicate that lesions restricted
* Present address: Department of Biology, Nanjing University, People's Republic of China. Correspondence: J.R. Bloedel, Division of Neurobiology, Barrow Neurological Institute, 350 West Thomas Road, Phoenix, AZ 85013, U.S.A. 0166-4328/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
to specific regions of the cerebellum result in the elimination of this conditioned reflex while leaving the characteristics of the unconditioned reflex unchanged. In these studies lesions in the lateral interposed/medial dentate region of the cerebellar nuclei were the most effective in permanently eliminating either the acquisition or the execution of the conditioned reflex. Yeo and colleagues 41-43 reported similar observations. However, these investigators argued that the critical area was localized to lobule HVI of the cerebellar cortex. Subsequent to these observations several lines of corroborative findings have been reported. For example, the acquisition and execution of the nictitating membrane reflex was disrupted by lesions of the red nucleus 34,35, the structure receiving inputs from the interposed and dentate nuclear regions ablated in the studies of McCormick and Thompson 28'29. These findings now have been supported by related findings in several laboratories 7,8,19-21,36.
Our studies were motivated by several observations from our laboratory as well as by a re-interpretation of the existing data in the literature. First, the findings implicating the cerebellum as the site of memory storage required for nictitating membrane reflex conditioning are negative in nature, i.e. they are based on the failure to elicit conditioned responses following cerebellar lesions. These reported effects could be due to modified interactions at extracerebellar sites receiving projections from the cerebellar nuclei rather than the consequence of eliminating the location of the critical memory trace. Decerebellation has long been known to produce changes in the tonic excitability of neurons in several brainstem nuclei that feasibly could play a role in the acquisition of conditioned behavior 9. Second, the evolving research examining the substrate for vestibulo-ocular reflex adaptation progressively points more towards the plastic changes occurring at sites within the brainstem rather than the cerebellum 22'23'3°'31 despite the arguments to the contrary early in the study of this question. Third, deficits in motor memory are not commonly reported following lesions of the cerebellum in patients 13. Fourth, studies from our laboratory provide direct evidence that the cerebellum is not
required for the acquisition for at least some types of conditioned behavior. These experiments were based on studies showing that decerebrate ferrets ambulating on a treadmill could be conditioned to step over a bar placed repeatedly in the trajectory of the limb's swing phase 24. Ferrets that underwent near-total cerebellectomy at least two months before still could acquire this conditioned forelimb movement6. Last, our investigations into the action of the climbing fiber system 4"5'~°'~1 revealed that the mechanism most frequently touted as the basis for establishing the engrams within the cerebellum, namely a long-term suppression of Purkinje cell excit ability p roduc ed b y the s e afferent s ~'~2. ~~- ~~. 26, does not occur under physiological conditions. Rather these afferents were shown to produce a short-lasting increase in the responsiveness of Purkinje cells to the mossy fiber-evoked parallel fiber input. These data were acquired using paradigms based on either spontaneously occurring or naturally evoked climbing fiber inputs rather than climbing fiber inputs evoked electrically at unusually high rates and train durations. An operational rather than a memory function for the climbing fiber system is also supported by recent studies investigating the relationship of climbing fiber activity to characteristics of volitional arm movements. In two different paradigms complex spike responses of Purkinje cells were found to be related to features of the movement other than the time course of its learning or acquisition 25'37. In studies done in parallel with those reported here, Welsh and Harvey 38 investigated the additional possibility that the effects of cerebellectomy on the conditioned nictitating reflex response were due to a performance deficit rather than a deficit in the animal's capacity to acquire the conditioned behavior. This premise was supported in part by showing parallel changes in the conditioned as well as the unconditioned reflexes. The study reported in this manuscript addresses the critical role of the cerebellum in the plastic changes required for the classical conditioning of the rabbit's nictitating membrane reflex by asking a very fundamental question using a decerebrate rabbit preparation: namely whether
the cerebellum is in fact both necessary and sufficient for the acquisition and/or execution of this conditioned reflex. An acute decerebrate preparation was chosen as the experimental model for several reasons. First, it provided an opportunity to evolve a preparation that could be used for prolonged, stable intracellular recording during the acquisition, performance and extinction of the conditioned reflex. Second, using an acute decerebrate preparation made it possible to examine the effects of cerebellectomy immediately after the removal of the cerebellum before the occurrence of any modifications in reflex characteristics resulting from adaptive changes in the brainstem. Last, the decerebrate preparation eliminated interactions with structures above the colliculi. Consequently if the reported effects of cerebellectomy on conditioned reflexes were related to alterations in the excitability of the conditioned reflex pathways outside the cerebellum, decerebration may fortuitously modify those tonic changes in the opposite direction, thus increasing the likelihood of observing a conditioned reflex. The data will show that in acutely decerebrate rabbits the nictitating membrane reflex can be classically conditioned even in the absence of the cerebellum. Differences between the behavior of the conditioned reflex in these preparations and that described in the intact rabbit are also presented and discussed. The data provide strong evidence against the fundamental notion that the plastic changes required for the conditioning of the nictitating membrane reflex of the rabbit occur within specific locations of the cerebellar cortex and/or nuclei. Although the acute ablation of the cerebellum unequivocally modifies properties of these reflexes, it does not eliminate either the capacity to acquire or execute this conditioned behavior. MATERIALS AND M E T H O D S
The experiments were conducted in white, adult New Zealand rabbits. Prior to administering halothane anesthesia, the animals were given 0.1 ml of atropine subcutaneously in order to
suppress tracheal secretions following intubation. Once the animal was adequately anesthetized with Halothane, a tracheostomy was performed, and the animal was artificially respired with continuous administration of the anesthetic. Next, a glue cap attachment was constructed onto the anterior part of the cranium so that a Kopf semichronic headholding device could be employed to stabilize the animal's head during the experiment. Following a craniectomy the posterior halves of both cerebri were exposed. The occipital and part of the parietal lobes overhanging the colliculi were then removed carefully by suction. Next a stereotaxic coagulating electrode was positioned at the anterior edge of the superior colliculus and angled toward the mamillary bodies. Decerebration was performed by placing coagulation lesions in a 1.5 x 1.5 mm grid across the entire cross-section of the brainstem. Also while under anesthesia an extensive craniectomy was performed over the cerebellum, exposing enough of this structure to permit the removal of cerebellar regions by suction. The lesion included most of the cerebellar tissue ipsilateral to the eye receiving the air puff stimulus. The lesion was extended lateral enough to disconnect any remaining cerebellar structures associated with the lateral hemisphere from continuity with the superior peduncle. Conditioned stimuli consisted of a tone, usually at a frequency of 500 Hz, that was applied from the onset of the trial until the unconditioned stimulus was terminated (delay conditioning paradigm). Two types of unconditioned stimuli were used with no difference in the characteristics of conditioning. In some experiments, airpuffs 100 ms in duration were applied to the cornea of the eye. In the other studies the eyeblink reflex was evoked using a bipolar electrical stimulus applied to the posterior part of the upper and lower eyelids. Stimuli usually consisted of a single pulse 0.5 ms in duration at an intensity just above threshold for producing an eyeblink on every trial. Intensities that evoked movements in other facial or neck regions were not used. Typically an interval of 350 ms between the onset of the tone and the onset of the unconditioned stimulus was employed. However, this interval varied depending upon the specific objectives of the experiment (see
10 Results section). Intertrial intervals of 8-10 s were used throughout these experiments. Two methods were employed to assess the reflexes recorded in these experiments. In some experiments (4 of the animals included in this study) the eyeblink reflex was measured using either a very thin wire directly inserted into the orbicularis oculi muscle or a flexible wire electrode that was hooked over the upper eyelid under just enough tension to produce a mild retraction. In the remaining experiments (4 animals) the retraction of the eyeball was monitored directly using an infrared emitter-receiver system that was placed in close proximity to the eye receiving the airpuff stimulus. Retraction was detected by recording changes in the reflection of the infrared light signal from the corneal surface. In both techniques the criterion for designating a response as a CR was its occurrence within the C R - U C R interval with an amplitude at least 10~/o of the average UCR. Because the objective of this study was confined to determining whether or not conditioning of the eyeblink reflex could occur in the decerebellatedecerebrate preparation, a detailed analysis of the amplitude changes of the CR and U C R was not included in this report. The animal was permitted to recover from anesthesia for a period of approximately one-half hour before any trials were initiated. A control period followed in which several blocks of unpaired tone and airpuff or electrical stimuli were applied. These data were carefully analyzed to ensure that the changes in the presumed conditioned response reported below were not due to pseudoconditioning or sensitization. When the control trials were completed, paired trials were initiated in order to determine the capacity of the decerebrate animal to acquire the conditioned behavior before the effects of cerebellectomy were assessed. Conditioning was measured by determining the number of conditioned responses present in blocks of 10 trials. If conditioning were not present in the decerebrate preparation with the cerebellum intact, no cerebellectomy was performed, and the experiment was discontinued. However, if the animal was capable of acquiring the conditioned behavior, halothane anesthesia
was again administered, and the cerebellum was removed. Immediately after the surgical procedure the conditioned behavior and/or its acquisition was evaluated. Consequently in most animals paired stimuli were reinitiated directly after the cerebellectomy procedure. In two animals the above protocol was modified in order to determine the capacity of the naive decerebrate-decerebellate preparation to acquire the conditioned behavior. In these experiments no paired trials were applied until after the decerebellation, which was performed immediately following the decerebration before the anesthesia was discontinued. After each experiment the animals were injected with an overdose of barbiturate and perfused intracardially with 10~o formalin solution. Transverse sections 4 0 # m thick were taken through the site of the cerebellar lesions and stained with Cresyl violet. The borders of the lesions were carefully drawn in order to determine the cerebellar regions removed at the time of the experiment. RESULTS
In general it was difficult to maintain decerebrate rabbits in a stable condition in which consistent acquisition and performance of the nictitating membrane conditioned reflex could be observed. Although not systematically studied, this difficulty appeared to be primarily related to the effects of decerebration on the animal's respiratory system. Because respiratory failure led to the early termination of many preparations, assessing the incidence of conditioning in cerebellectomized animals as a percent of the total number of animals used in the study would be misleading. It is more meaningful to characterize this feature of the experiments by calculating the percent of decerebrate-decerebellate animals that acquired the conditioned behavior relative to the total number of animals that also displayed the behavior prior to cerebellectomy. Consequently, as stated in Methods, the experiment was discontinued in any decerebrate animal incapable of acquiring the conditioned eyeblink reflex. Using this as a criterion, 5 of 5 decerebrate animals that acquired
11 the conditioned reflex behavior also manifested the acquisition and performance of the conditioned reflex following cerebellectomy. In the other 3 animals included in this study, conditioning was not attempted until after cerebellectomy. As will be reported below, these animals also acquired the conditioned reflex behavior. The data in Fig. 1 illustrate the conditioning that occurred in these decerebrate-decerebellate preparations as well as the type of experimental observations that were measured in order to obtain the plots in the subsequent figures. The records in A and B were obtained from trials after pairing of the conditioned and unconditioned stimuli was initiated but before maximal conditioning had been achieved. In this experiment the conditioned stimulus was a 500-Hz tone and the unconditioned stimulus was an air puff. The reflex was monitored in this experiment using an infrared emitter/sensor that detected retraction of the eyeball. In A only an unconditioned response was evoked in this paired trial during the initial acquisition phase. A trial in which a conditioned
A
response was evoked is shown in B. Characteristically there was usually a modest reduction in the amplitude of the unconditioned response when the conditioned response was evoked at a relatively long latency following the tone. Once the conditioned behavior was acquired, conditioned responses could be evoked in a tone-only trial during the initial stages of extinction (C). As the tone-only trials continued, extinction became complete, and the conditioned stimulus no longer evoked the reflex. A plot characterizing the time course of the acquisition and extinction of the eyeblink reflex following cerebellectomy is shown in Fig. 2 for an animal that had been conditioned prior to this surgical procedure. Data were quantified by plotting the percent of trials having conditioned responses over each block of 10 trials. In this experiment conditioning occurred over a total of 80 trials (8 blocks). Typical of these preparations the percent of trials in which there were conditioned responses varied from block to block much more dramatically than in intact preparations. This feature will be addressed in greater detail below. Nevertheless there was clearly a dif-
C Tone Only
CR Failure
80"
Tone
Control Paired
÷r
+r
ts
B CR-UCR
D
60-
Tone Only-Extinction
nO
~o 9 r
~ s
o
m.,,c
t'
6oo
T
Fig. 1. Conditioned and unconditioned responses in the decerebrate, decerebellate rabbit obtained from a series of trials in one conditioning-extinction sequence. A: in this paired stimulus trial, only an unconditioned response was evoked prior to conditioning. B: after conditioning, a conditioned response occurred after the tone but before the air puff and was followed by a smaller unconditioned response evoked by the air puff. C: a conditioned response unaccompanied by an unconditioned response was evoked by the tone in the early stages of extinction. D: no response to the conditioned stimulus was observed after extinction had taken place. T: the conditioned stimulus, a 500 Hz tone. S: the unconditioned stimulus, an air puff directed at the cornea. Responses were recorded with a coupled infrared emitterdetector system aimed at the cornea.
2O
0
20
40 Block
60
8'0
(10 trials/block)
Fig. 2. Acquisition and extinction of the conditioned eyeblink response in a decerebrate, decerebellate rabbit. The percentage of conditioned responses is shown for each block of 10 trials. Control blocks were comprised of alternating conditioned (tone) and unconditioned (air puff) stimuli. During paired trials the air puff was applied 200 ms after onset of the tone. The conditioned behavior was extinguished by presenting the animal with a series of tone-only trials. Measurements of the reflex behavior were made using the E M G technique described in the Methods.
12 ference in the percent of trials with conditioned responses between the control period and the period in which the conditioned behavior was maximal. This difference was statistically significant in all animals considered to display conditioned behavior. In this experiment the difference was statistically significant at a level of P < 0.001. When the paired trials were discontinued, the conditioned behavior clearly extinguished with the occurrence of conditioned responses returning to control values. The characteristics of the conditioning and extinction were very repeatable in each animal. Consequently once the behavior was extinguished it could be reacquired when the conditioned and unconditioned stimuli were again paired. An example of this is shown in Fig. 3A. As was the case in all but one of the animals studied, the conditioned responses were not evoked in the first paired trials immediately following cerebellectomy. Customarily this surgical procedure took approximately 20 min and was always performed while 0.5~o halothane anesthesia was administered. These factors may have contributed to the lack of savings in most preparations. In the experiment shown in Fig. 3A, conditioned responses were present only in a small percentage of trials initially. The percentage increased over 80 trials to a mean percentage of 94 ~o once the conditioned behavior was acquired. When only the conditioned stimulus was applied, the behavior extinguished. Once extinguished the behavior could be reconditioned by resuming the pairing of the conditioned and unconditioned stimuli. In this preparation this was repeated 3 times before the condition of the animal deteriorated. Similar to the data in Fig. 2, once a maximal level of conditioning was achieved there was a considerable variability in the percent of trials having conditioned responses compared with the same phenomenon in the intact rabbit. In order to provide an overview of the time course of conditioning and its extinction in the decerebellate-decerebrate preparation, a composite plot was calculated from the measurements in 4 animals. The graph was assembled by first dividing the learning curve into 4 phases: (1) control, (2) acquisition, (3) plateau, (4) extinction.
Paired
Tone Only 100 80 ¸ o
60 40
Q.
20 0
5
A °
10
15 Block
2'0
2'5
3'0
1 O0
120
(50 trials/block) Paired
Tone Only .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
100" 80" 60" c
~,
40-
n
20-
B
2'0
40
60 Block
8'0
(10 trials/block)
Fig. 3. Time course of repeated cycles of conditioning and extinction in a decerebrate, decerebellate rabbit and the variability of the conditioned behavior. A: the sequence of acquisition and extinction of the conditioned reflex is plotted with a resolution of 50 trials/block. B: plot of the first two conditioning-extinction cycles shown in A at an increased resolution of 10 trials/block. This animal had been conditioned prior to cerebellectomy. Note the rapid reacquisition of the conditioned behavior following cerebellar ablation. Responses were measured using the infrared emitter system.
Both the control and acquisition phases were aligned on the onset of acquisition. The 10 blocks prior to the onset of acquisition comprised the control portion of the graph, and the length of the other phases equalled the mean number of blocks taken by the 4 animals to complete each phase of the behavior. For each animal, individual phases of the learning curve were normalized to the average phase length. Then the various phases were averaged across the 4 animals to form the mean learning curve. This composite plot is shown in Fig. 4. Notice that the time course for conditioning and ex-
13 Paired 90" 80" 70-
¢
60-
(3 50"
~" 30" 20" 10" 0
10
20 Block (10 trials/block)
30
4'0
mal, the behavior in the decerebrate-decerebellate rabbit differed from the behavior in the intact rabbit ~4 in tw6 important ways. First, there was appreciably more variability from block to block in the reduced preparations. Second, the percentage of trials displaying conditioned responses did not approximate 100~o in most animals. Although not a primary objective of this study, one consistent finding across all animals (illustrated in Fig. 3) provides some insight into the basis for both of these differences. Fig. 3B was plotted from the same data as that used in'Fig. 3A; however, the data were grouped
Fig. 4. Average time course of conditioning. The normalized conditioning curves from 4 animals were averaged according to the procedure discussed in the text. Each point on the graph is the mean number of conditioned responses in a ten trial block. Error bars indicate standard deviation.
500
Decerebrate Rabbit x=418.1
400
tinction are approximately symmetrical, occurring over approximately 8 blocks of trials. The plateau calculated across these 4 animals also illustrates another characteristic finding: namely that the level of conditioning never reached 100 ~/o. Based on the data presented in Figs. 2-4, it is apparent that once the conditioning of the nictitating membrane/eyeblink reflex became maxi-
E
300 ¸
//
t=-9,244 p
