Inside the Thompson laboratory during the "cerebellar years" and the continuing cerebellar story.

This paper is based on the talk by one of the authors (DL) given at the symposium for the retirement of RF Thompson (RF Thompson: A bridge between 20th and 21st century neuroscience). We first make some informal observations of the historical times and research conditions in the Thompson laboratory when the cerebellum was found to play a critical role in eye lid classical conditioning, the "cerebellar years". These conditions influenced our collaborative international program on the phenomenon known as "transfer of training" or "savings". Our research shows that the appearance of "savings" is an artifact of the order of testing, and depends upon the functioning of the contralateral interpositus nucleus (IPN) in a way that is complementary to the role of the IPN in normal eyelid classical conditioning.

Dissociaton of conditioned eye and limb responses in the cerebellar interpositus.

Thompson and colleagues have demonstrated that the lateral interpositus nucleus of the cerebellum is the essential locus for the classical conditioning of the somatic eyeblink response. Preliminary studies reported that lesioning the cerebellar interpositus nucleus ipsilateral to the side of training also appears to abolish conditioned limb flexion responses. Previous studies have suggested that the interpositus nucleus is somatotopically organized with the eye being represented laterally and the hindlimb medially. Presently, we employed a double dissociation paradigm to examine the effects of muscimol (a GABA(A) agonist) injections on eyeblink versus limb flexion conditioned responses in the ipsilateral cerebellar interpositus nucleus of New Zealand white rabbits. For eyeblink conditioning, the conditioned stimulus (CS) was a 14-V lamp bulb and the unconditioned stimulus (US) was a 3-psi corneal airpuff to the left eye. For limb flexion conditioning, the CS was a 1-kHz, 85-95 dB SPL tone and the US was a 3- to 5-mA shock to the upper left hindlimb. Upon training on both responses to a 60-100% criterion, the rabbits were then tested on eyeblink and limb flexion responses after injections of muscimol (0.1-0.3 mul of a 0.01- to 1.0-M solution) into either the lateral (eyeblink) or medial (limb flexion) interpositus nucleus. We have been able to successfully decrease or abolish the percent conditioned responses (CRs) of both the eyeblink and limb flexion conditioning selectively without affecting the other. These results thus lend further support for the notion of the existence of a somatotopic map in the interpositus nucleus for learning.

Cooling the area postrema induces conditioned taste aversions in male rats and blocks acquisition of LiCl-induced aversions.

The experiments presented in this article were designed to examine whether area postrema (AP) lesions attenuate LiCl-induced conditioned taste aversions (CTAs) by disruption of information about the illness-producing properties of LiCl or by a lesion-induced malaise. Reversible lesioning of the AP caused by cooling induced a CTA in male rats. The cooling-induced CTA could be blocked if males were exposed to cooling for several days before acquisition day. Acquisition of a LiCl-induced CTA was blocked in males if they were exposed to cooling before acquisition day and during LiCl administration on acquisition day was attenuated but not blocked in males if they were exposed to cooling only before acquisition day, and was unchanged in males if they were exposed to cooling only during LiCl administration. Taken together these results indicate that the AP is important for acquisition of LiCl-induced CTAs but that inactivation of this area is so aversive it will induce CTAs that can obscure the attenuation of LiCl-induced aversions.

The importance of cerebellar cortex and facial nucleus in acquisition and retention of eyeblink/NM conditioning: evidence for critical unilateral regulation of the conditioned response.

Experiment 1 examined acquisition of the classically conditioned eyeblink response in rabbits (Oryctolagus cuniculus) during reversible cooling lesions of the cerebellar cortex (CX), facial nucleus (FN), or lateral parvocellular reticular formation (RF). Retention was then evaluated during periods of training during reversible cooling lesions and without cooling in rabbits that had acquired the conditioned response. We found that cooling the CX did not prevent acquisition, but did retard the acquisition rate. Cooling the FN during acquisition prevented the expression of the unconditioned and conditioned response, but did not prevent the acquisition when assessed during subsequent training without cooling. Cooling the RF had no effect on the acquisition or expression of the conditioned response. During subsequent retention testing, in well-trained animals, cooling the CX did not abolish the learned response. Cooling the FN abolished both the conditioned and the unconditioned response. The results from Experiment 1 indicate that the CX is more important for acquisition than retention of the conditioned response. The FN is not important for the acquisition of the conditioned response, but is essential for the expression of the conditioned and unconditioned response. Experiment 2 examined bilateral recordings from the cerebellum in well-trained rabbits, before and during interpositus or FN cooling. We found that cooling the interpositus abolished all learning related activity in the ipsilateral or contralateral cerebellum, but did not affect the stimulus evoked responses. Cooling the FN did not abolish stimulus evoked activity or learning related activity in the cerebellum. The results emphasize the critical importance of the ipsilateral cerebellum in classical eyeblink conditioning and suggest that the memory trace for this type of learning is mediated by unilateral circuitry.

The learning-related activity that develops in the pontine nuclei during classical eye-blink conditioning is dependent on the interpositus nucleus.

A growing body of research now implicates the cerebellum in the formation and storage of the critical neural plasticity that subserves the classically conditioned eye-blink response. Previous anatomical, physiological, and behavioral research suggests that auditory-conditioned stimulus information is routed to the cerebellum by the pontine nuclei. However, it has also been observed from multiple unit recordings that some populations of pontine cells, in addition to showing auditory-evoked responses, also show changes in activity that is learning-related. It is unknown whether this learning-related activity is generated by the pontine cells or whether it is generated by some other structure and projected to the pontine nuclei. Because the cerebellum has been implicated in the formation of the essential plasticity that subserves this learned behavior, we examined how multiple unit recordings of learning-related activity within the pontine nuclei are affected by reversible inactivation of the interpositus nucleus of the cerebellum. The results indicated clearly that when the interpositus nucleus was inactivated, the learning-related activity in the pontine nuclei was abolished completely and the auditory stimulus-evoked activity was unaffected. In contract, when the facial nucleus was inactivated, both the auditory stimulus and the learning-related activity were still present. These results indicate that the learning-related activity exhibited by some populations of pontine nuclei cells is dependent on the interpositus nucleus and may represent feedback from the cerebellum.

The effects of reversible inactivation of the red nucleus on learning-related and auditory-evoked unit activity in the pontine nuclei of classically conditioned rabbits.

The pontine nuclei carry auditory conditioned stimulus information to the cerebellum during classical conditioning of the nictitating membrane response in rabbits. In well-trained animals learning-related as well as stimulus-evoked unit activity can be recorded throughout the pontine nuclei but particularly in the lateral and dorsolateral pons. Recent work in our laboratory has provided evidence that the learning-related unit activity in the pons is dependent on the interpositus nucleus and that the pons is not a site of essential plasticity for the learned response. In the present study we considered the question of whether learning-related unit activity might be projected from the interpositus nucleus to the pons through the red nucleus, a primary output target of the interpositus and a structure known to be essential for expression of the learned response. Multiple unit recordings were taken from lateral and dorsolateral pontine locations in well-trained rabbits before and during cooling of the red nucleus. Analysis of pooled data for all recording locations within the lateral and dorsolateral pons indicated that reversible inactivation of red nucleus abolished both stimulus-evoked and learning-related unit activity. However, we also found discrete recording locations where stimulus-evoked and learning-related unit activity were attenuated but not abolished by red nucleus cooling.

The effects of cooling the area postrema of male rats on conditioned taste aversions induced by LiC1 and apomorphine.

Although permanent lesion studies have demonstrated that the area postrema (AP), a chemoreceptor trigger zone, is part of the neural mechanism for conditioned taste aversions (CTAs), its exact role remains questionable. It has been suggested that the attenuated acquisition of a CTA after permanent lesions of the AP is the result of an inability to recognize the conditioned taste as novel. The present series of experiments was designed to test the hypothesis that lesions of the AP interfered with LiCl processing and not recognition of taste novelty. This was accomplished by using the reversible lesioning procedure, cooling, only during administration of the illness-inducing agent. In Expt. 1, measurement of thermal lines around the tip of the cold probe in the AP indicated that our cooling procedures allowed the majority of the AP to be cooled to temperatures that suppress neuronal activity and transsynaptic transmission, but not axonal transmission. In Expts. 2 and 3, rats were injected with either LiCl or apomorphine after consumption of a 10% sucrose solution. Cooling of the AP was initiated 5 min before administration of one of the illness-inducing agents and was continued for 55 min after injection. The rats were tested later for acquisition while the neural function of the AP was preserved. Our experimental results demonstrated that cooling the AP could attenuate the CTA induced by LiCl, but had no effect on the CTA induced by apomorphine. Since the AP was functional when the rats encountered the novel sucrose solution both before and after conditioning, but not functional when LiCl was given, these results do not support the recognition of taste novelty hypothesis.

Neural unit activity in the trigeminal complex with interpositus or red nucleus inactivation during classical eyeblink conditioning.

During classical conditioning, many neurons in the trigeminal complex of rabbits exhibit activity that is related to the conditioned stimulus (tone), the unconditioned stimulus (airpuff), or to the conditioned response (eyeblink). For these reasons the trigeminal complex has been hypothesized to be a brainstem locus for the neuronal plasticity associated with conditioning. In this experiment, the learning-related activity (unit activity associated with the conditioned response) in the trigeminal is abolished when either the red nucleus or interpositus nucleus of the cerebellum is temporarily inactivated by cooling, but the stimulus-evoked activity is unaffected by cooling. This study and previous results support the suggestion that the learning-related activity seen in the trigeminal is driven by the interpositus by way of the red nucleus.

Unilateral inferior olive NMDA lesion leads to unilateral deficit in acquisition and retention of eyelid classical conditioning.

New Zealand White rabbits (Oryctolagus cuniculus) were trained for acquisition (N = 21) or retention (N = 10) of classical eyelid conditioning with unilateral or bilateral N-methyl-DL-aspartate chemical lesions of the rostromedial dorsal accessory inferior olive (rmDAO; multiple injections totaling 76 to 342 nmol). In all instances, subjects were unable to learn or retain conditioning on the side contralateral to the lesion. Learning rates were comparable for lesions outside of the rmDAO and sham operates. These findings demonstrate a specific unilateral deficit whereas in previous research the answer to this question was ambiguous since electrolytic lesions effectively cause bilateral olivary lesions. This research agrees with the concept that the inferior olive projects essential information about the unconditioned stimulus to a cerebellar locus of learning and memory for classical conditioning.

Transfer of learning but not memory after unilateral cerebellar lesion in rabbits.

Unilateral lesion of the cerebellum in rabbits completely and permanently abolishes previous learning and prevents new learning ipsilateral to the lesion. However, when training continues on the contralateral side, there is substantial savings in that it takes few trials to learn. This observation may imply that the memory survives the lesion. Rabbits were classically conditioned for an eyelid response and then the ipsilateral interpositus nucleus of the cerebellum was lesioned. Then the rabbits were trained on the contralateral side. There is no savings on the contralateral side without first trying to train on the lesioned side. The authors conclude that the survival of a memory after the lesion probably does not account for the rapid transfer, but rather that the act of trying to train on the lesioned side in previous studies first induces a new memory on the contralateral side.

Reacquisition of eyeblink classical conditioning following large cerebellar cortical lesions in Dutch belted rabbits.

The effects of cerebellar lesions on a classically conditioned eyeblink response was examined using Dutch Belted rabbits as subjects. Results confirmed our previous findings, using New Zealand White rabbits, that cerebellar cortical lesions do not abolish conditioned responding if the interpositus is intact. However, unlike our previous studies using New Zealand White rabbits, Dutch Belted rabbits failed to reach a pre-lesion level of performance. These results suggest a possible important strain difference in this paradigm.

Acquisition of classically conditioned eyeblink response following bilateral lesions of flocculus and paraflocculus.

Stimulation of regions of the cerebellar flocculus can elicit eyeblinks, but the relationship of this floccular eyeblink zone to eyeblink classical conditioning is unknown. In this experiment, New Zealand white rabbits received bilateral lesions of the flocculus and paraflocculus and were subsequently classically conditioned with tone and corneal airpuff on the left and then the right eye. All animals reached training criterion on both eyes, with the exception of one animal whose lesion included the superior cerebellar peduncle and who was unable to learn on the ipsilateral eye. The lesioned group was not significantly different from unlesioned controls in rate of acquisition or conditioned or unconditioned response amplitude. These results indicate that the flocculus and paraflocculus are not by themselves the essential site of plasticity for classical conditioning of the rabbit eyeblink response.

Cerebellar cortical lesions and reacquisition in classical conditioning of the nictitating membrane response in rabbits.

Portions of cerebellar cortex, particularly Larsell's hemisphere VI, have been identified as involved but not essential for the acquisition and retention of classical conditioning of the nictitating membrane (NM) response in rabbit. The present experiment was undertaken to examine the effect of lesions of cerebellar cortical regions projecting to anterior dorsolateral interpositus nucleus. Lesions of relatively equal size, including hemisphere VI and additionally more medial, anterior, or lateral cerebellar cortex were made after rabbits were classically conditioned. The effect of these cerebellar cortical lesions on retention to tone versus light conditioned stimulus (CS) and the timing of the conditioned response was evaluated. In spite of relatively large cerebellar cortical lesions, reacquisition of the conditioned NM response occurred quite rapidly. Whether the lesion was more medial and anterior or more lateral did not affect retention. Retention was significantly poorer with light CS than with tone CS. Timing of CRs was not affected by these lesions of cerebellar cortex, but the lesions spared most of the anterior lobe. The parasagittal zone C3 covering the longitudinal band of tissue from anterior lobe to the paramedian lobule and projecting to dorsolateral anterior interpositus requires additional exploration for its role in classical NM conditioning.

Reversible lesions of the red nucleus during acquisition and retention of a classically conditioned behavior in rabbits.

Previous lesion, recording, and stimulation studies implicated the cerebellum and its associated brain-stem circuitry as essentially involved in classical conditioning of discrete, somatic muscle responses. This is a companion to our study of interpositus cooling, which showed that the formation of a memory was prevented. In the present study, we assess the red nucleus for its role in the plasticity associated with learning and memory by using local cooling as a reversible lesion technique. A cooling probe was implanted lateral to the red nucleus. Recording electrodes were implanted in the right red nucleus and the left interpositus nucleus. Animals were trained for 5 days with the cooling probe activated. No behavioral conditioned responses (CR) developed, and multiple unit recordings related to learning did not develop in the red nucleus. However, a learning related model did develop in the interpositus. After 5 days of training while cooling, animals were given 5 days of normal training (cooling probe inactive) to assess retention. Substantial savings were evident when normal training was given. CRs appeared quickly on the first day of normal training, and multiple unit models were present in both red nucleus and interpositus nucleus. These results support the idea that the red nucleus is a necessary efferent for the memory trace formed in the cerebellum.

Mammalian brain substrates of aversive classical conditioning.

In this review, we have examined recent studies that have successfully identified neural circuits necessary for nonspecific and specific conditioned responses. This success is due in large part to the advantages of the classical conditioning paradigm for controlling stimuli and responses. Clearly, this research does not attempt to account for all forms of memory. The power of this approach is demonstrated by the distinction between essential and nonessential memory traces or engrams. Essential memory traces represent the circuitry responsible for forming the association in classical conditioning. Nonessential memory traces do not represent the essential association, but they are important for facilitating, adapting, and modifying the final performance of the learned behavior. The search for the engram for any learned behavior has been viewed with skepticism by some investigators who quote Karl Lashley: "This series of experiments has yielded a good bit of information about what and where the memory is not. It has discovered nothing directly of the real nature of the engram" (1950, pp. 477-78). However, these authors neglect to quote Lashley fully, for even he was less pessimistic about that search than in normally recognized. He continued, "I sometimes feel, in reviewing the evidence on the localization of the memory trace, that the necessary conclusion is that learning just is not possible. It is difficult to conceive of a mechanism which can satisfy the conditions set for it. Nevertheless, in spite of such evidence against it, learning does sometimes occur" (1950, pp. 477-78, emphasis added). Learning does indeed occur, and its neurobiological substrates can be localized.

Reversible lesions of the cerebellar interpositus nucleus during acquisition and retention of a classically conditioned behavior.

Previous lesion, recording, and stimulation studies have implicated the cerebellum and its associated circuitry as essentially involved in classical conditioning of discrete, somatic muscle responses. In 2 experiments, the interpositus nucleus of the cerebellum was assessed for the plasticity associated with learning and memory of the nictitating membrane (NM) response by using local cooling as a reversible lesion technique. In well-trained animals (Experiment 1), NM conditioned responses (CRs) were abolished during cooling of the interpositus but reappeared when the interpositus returned to body temperature. This cooling-warming protocol could be repeated many times. Cooling could be prolonged (one session, approximately 1 hr) with recovery of NM CRs as tested on the next day. Multiple-unit recordings related to learning were also absent in the interpositus and red nucleus during cooling. In naive animals (Experiment 2), both behavioral and unit CRs did not develop while training with cooling. There was no evidence of savings when training continued without cooling: Behavioral and unit CRs developed as if the animals were still naive. These results support the idea that the interpositus nucleus of the cerebellum is the critical locus for learning and memory of this classical CR.

Disruption of classical eyelid conditioning after cerebellar lesions: damage to a memory trace system or a simple performance deficit?

Over the past 10 years, a number of laboratories have reported that classically conditioned skeletal muscle responses, such as conditioned nictitating membrane/eyelid responses, are critically dependent on activity in the cerebellum. For example, unilateral lesions of the cerebellar interpositus nucleus have been shown to prevent acquisition and abolish retention of the conditioned eyelid response on the side ipsilateral to the lesions without affecting conditioned responding (CR) on the contralateral side. Also, recording studies involving the interpositus nucleus have consistently revealed patterns of neuronal discharge that predict execution of the CR. The lesion and recording studies have generally been cited as evidence that plasticity in the cerebellum is critically involved in the learning and memory of classically conditioned responses. This interpretation was recently challenged by Welsh and Harvey (1989a), who claimed that cerebellar lesions simply produced a performance deficit and speculated that the role of the cerebellum was not in learning and memory processes associated with the CR but only in performance of the eye blink response. Presented here are three experiments that provide additional strong evidence for a critical role of the cerebellum in the learning and memory of the Pavlovian CR. These experiments include (1) demonstrations of complete and permanent CR abolition after appropriate interpositus lesions, (2) a failure to find systematic or persisting decrements in the unconditioned response amplitude (i.e., the eye blink reflex) after appropriate interpositus lesion, and (3) observations of differential effects on the CR and unconditioned response after lesions were placed in populations of motoneurons responsible for executing the eye blink response. These data are discussed in the context of performance versus learning issues; evidence presented here rules out the possibility that interpositus lesion abolition of the eye blink CR is simply due to lesion effects on performance.

Are eyeblink responses to tone in the decerebrate, decerebellate rabbit conditioned responses?

Bloedel and associates recently claimed to have established conditioned eyeblink responses in the acute decerebrate, decerebellate rabbit. Their training procedure was extreme massed practice (mean intertrial interval of 9 s) and they used an idiosyncratic definition of the conditioned response (10% or more of the unconditioned response amplitude). They did not measure or control the excitability of their preparations and did not run any separate control groups for alpha responses, alpha conditioning or pseudoconditioning. Using normal animals we compared their training procedure with procedures standard in the field and analyzed the consequences of their scoring procedure. Our group trained at a 30-s intertrial interval (ITI) showed clear learning in the training session. In marked contrast, 3 groups trained at a 9-s ITI developed no conditioned responses. We also found that the method of scoring used by Bloedel and associates counts many spontaneous responses as conditioned responses (CRs) if unconditioned response (UR) amplitudes are low, excludes genuine CRs if UR amplitudes are high and does not control for the occurrence of spontaneous responses. It must therefore be concluded that the eyeblink responses to tone reported by Bloedel and associates to occur in the decerebrate, decerebellate rabbit are not associative CRs as they develop in the normal animal.

Lesions of the cerebellar interpositus nucleus abolish both nictitating membrane and eyelid EMG conditioned responses.

Both nictitating membrane extension and eyelid EMG activity were measured during classical conditioning of rabbits to tone-airpuff pairings. Both measures were highly correlated. Over trials, learning criterion was met earlier with eyelid EMG activity than with nictitating membrane extension. Within a trial, eyelid EMG preceded and was more robust than nictitating membrane extension. The rabbits were lesioned in the cerebellar interpositus nucleus and then trained for up to 26 days. Detailed analyses of tone alone trials demonstrate that the lesion abolished conditioned responses for both measures. These data confirm that conditioned responses are abolished by lesion of the cerebellar interpositus nucleus.

Acquisition of classical conditioning without cerebellar cortex.

The left cerebellar cortex was surgically aspirated in rabbits who were then subsequently trained for classical conditioning of the nictitating membrane. All rabbits were trained sequentially on both eyes. Rabbits with the lesion confined to the cerebellar cortex were able to learn with the eye ipsilateral to the lesion although it took many times longer than reported for either naive rabbits or for rabbits first trained on the unlesioned, contralateral side. Rabbits with lesions that included the cerebellar cortex and the cerebellar interpositus nucleus did not learn with the eye ipsilateral to the lesion. Learning with the eye contralateral to either type of lesion was always very rapid. It is now clear on the basis of this and previous studies that cerebellar cortex, unlike the cerebellar interpositus nucleus, is not essential for acquisition or relearning/retention of classical conditioning. However, cerebellar cortex normally plays an important role since acquisition of classical eyeblink conditioning is prolonged and of poor quality in its absence.