PARALLELS BETWEEN CEREBELLUM- AND AMYGDALA-DEPENDENT CONDITIONING

Transcription

1 PARALLELS BETWEEN CEREBELLUM- AND AMYGDALA-DEPENDENT CONDITIONING Javier F. Medina, J. Christopher Repa, Michael D. Mauk and Joseph E. LeDoux Recent evidence from cerebellum-dependent motor learning and amygdala-dependent fear conditioning indicates that, despite being mediated by different brain systems, these forms of learning might use a similar sequence of events to form new memories. In each case, learning seems to induce changes in two different groups of neurons. Changes in the first class of cells are induced very rapidly during the initial stages of learning, whereas changes in the second class of cells develop more slowly and are resistant to extinction. So, anatomically distinct cell populations might contribute differentially to the initial encoding and the long-term storage of memory in these two systems. EXTINCTION The reduction and cessation of a predictive relationship and behaviour after the omission of a reinforcer (negative prediction error). Howard Hughes Medical Institute, Department of Physiology, W.M. Keck Foundation Center for Integrative Neurobiology, University of California, 513 Parnassus Avenue, Room HSE-808, San Francisco, California , A. Correspondence to J.F.M. jmedina@phy.ucsf.edu DOI: /nrn728 Minds made out of the soft wax are good at learning, but apt to forget; and those employing the hard wax are the reverse. Plato, Theaetetus Where in the brain are memories stored? This question represents one of the greatest challenges facing the field of neuroscience, and it remains a chief source of debate and controversy. Research in this area has benefited tremendously from the development of simple behavioural procedures to assess learning, as well as from advances in anatomical, physiological, pharmacological, molecular and imaging techniques that have been used to elucidate the structural and functional organization of neural circuits and their contribution to memory formation. Some of the clearest evidence on the localization of memory within the mammalian brain has come from studies of associative learning, especially through the use of classical conditioning. The classical-conditioning model (also known as Pavlovian conditioning) first emerged from Pavlov s studies of alimentary responses in dogs 1, and it has become a general paradigm for studying associative learning, and its neural basis, in a wide variety of response systems. In a typical classicalconditioning procedure, animals learn to express a conditioned response (CR) to a predictive or conditioned stimulus () that is paired with a significant unconditioned stimulus (). Here, we focus on two of the most widely used forms of Pavlovian learning in neurobiological studies in mammals: eyelid and fear conditioning. In both cases, the standard procedure is delay conditioning, in which the onset of the precedes the onset of the, but the two stimuli overlap in time. For eyelid conditioning, the is typically a tone, which is followed after 100 1,500 ms by a, such as a puff of air applied to the eye 2,3. Initially, the tone does not elicit an eyelid response, but after a few hundred tone-plus-air-puff training trials, the animals come to express a defensive CR, which ultimately involves a precisely timed closure of the eyelid, milliseconds before the predicted arrival of the air-puff 4 6. These conditioned eyelid responses persist for long periods in the absence of exposure to the 7, but can be extinguished quickly by presentation of the without the 4. However, as noted by Pavlov, extinguished CRs can recur spontaneously, can be revived by exposure to certain stimuli, and can be retrained in fewer trials than originally needed, indicating that memory is not erased by EXTINCTION 8,9. Fear conditioning shares some similarities with eyelid conditioning, but there are several important 122 FEBRUARY 2002 VOLUME 3

2 Parallel fibres Golgi cell Pontine nuclei Vestibulocochlear nucleus GOLGI CELLS Cerebellar interneurons located in the granule cell layer. Their axonal terminals form part of the cerebellar glomeruli. PURKINJE CELLS Inhibitory interneurons in the cerebellum that use GABA as their neurotransmitter. Their cell bodies are situated beneath the molecular layer, and their dendrites branch extensively in this layer. Their axons project into the underlying white matter, and they provide the only output from the cerebellar cortex. Mossy fibre Tone Basket cell Granule cell Stellate cell Interpositus nuclei Red nucleus Facial nucleus Purkinje cell Climbing fibre CR Inferior olivary nucleus cell Trigeminal nucleus Puff of air delivered to eyelid Cerebellar cortex Figure 1 Neural circuits engaged during eyelid conditioning. The conditioned stimulus () pathway is shown in green. In the case of a tone stimulus, it involves a mossy fibre projection that arises in the pontine nuclei and reaches both the interpositus neurons of the deep cerebellar nuclei and, through activation of granule cells, the Purkinje cells of the cerebellar cortex. The unconditioned stimulus () pathway is shown in blue. For an air-puff stimulus, it is transmitted to the same interpositus and Purkinje cells through activation of the climbing fibres that originate in the inferior olivary nucleus. The conditioned response (CR) leads to eyelid closure after exposure to the, and is generated by activation of the interpositus nucleus and its downstream targets, which include cells in the red nucleus and ultimately in the facial nucleus. Plus and minus signs indicate whether the particular connection is excitatory or inhibitory. differences. A typical fear-conditioning experiment involves the paired presentation of a tone (which, as in eyelid conditioning, serves as the ) with an aversive footshock (the ). Although these conditioning stimuli are often similar to those used in eyelid conditioning, the CRs differ in the two paradigms, as does the amount of training required to condition the responses 10. In fear conditioning, a wide range of bodily responses comes under the control of the within a single pairing, or at most after a few pairings, with the These responses include gross reactions of the skeletal musculature, such as immobility or freezing, and changes in various autonomic (cardiovascular, respiratory, alimentary and so on) and endocrine (for example, pituitary adrenal hormone) systems. So, eyelid conditioning is essentially a form of fear conditioning (or to use Pavlov s term, defence conditioning) that is extended over trials to the point at which the diffuse responses that are initially expressed are gradually shaped into a precise adaptive response. As in eyelid conditioning, extinguished conditioned fear responses are not erased, because they can be revived in various ways 14. Considerable progress has been made in elucidating the brain systems that are responsible for these two simple forms of associative learning. Studies of eyelid conditioning indicate that the deep cerebellar nuclei and overlying cerebellar cortex mediate the acquisition and storage of discrete motor memories 15,16. By contrast, evidence from fear conditioning points to the lateral nucleus of the amygdala as a key component of the brain system that is responsible for the formation of aversive emotional memories (see REFS 11,17 19). We begin this review by summarizing the basic findings on the localization of learning and memory processes in these systems. We will then consider recent data that point to a trigger-and-storage model for memory formation, in which the initial encoding and subsequent long-term storage of memory are mediated by separate groups of neurons. These findings indicate that the location of memory storage might differ at different times in the learning process. It has long been believed that the neural basis of classical conditioning involves the convergence of synaptic inputs onto individual cells from pathways that process the and. Because of the relative simplicity of the stimuli used and the stereotyped nature of the responses measured in most classical-conditioning studies, the search for sites of convergence and for the regions that control the CRs has been fairly straightforward. Before we consider the roles of the cerebellum and amygdala in learning and memory, we will discuss how the and, and the CR, engage the input and output pathways of these regions in eyelid and fear conditioning, respectively (FIGS 1 and 2). Areas engaged during eyelid conditioning pathway for eyelid conditioning. The mossy fibre input to the cerebellum, which arises from many sources in the brainstem and spinal cord, transmits a variety of types of information. In particular, mossy fibres that originate in the pontine nuclei are activated by stimuli, such as tones, that are frequently used as s during eyelid conditioning 20. After the anatomical and physiological work of Ito, Eccles, Szentagothai, Llinas and others, we now have a detailed description of how this mossy fibre pathway transmits information to the cerebellum (FIG. 1). In short, mossy fibres make direct excitatory synapses onto the output cells of the deep cerebellar nuclei, and they also project to the cerebellar cortex, where they branch profusely and make excitatory synapses onto a large number of granule cells and GOLGI CELLS. The granule cells, the activity of which is also influenced by inhibition from Golgi cells, provide excitatory input to the PURKINJE CELLS, which in turn inhibit cells in the deep cerebellar nuclei. The divergence NATURE REVIEWS NEUROSCIENCE VOLUME 3 FEBRUARY

3 Auditory thalamus Auditory cortex Tone Amygdala LA CE Footshock Somatosensory thalamus Somatosensory cortex Purkinje cells. This cortical connection is striking, because each Purkinje cell receives input from a single climbing fibre, which, when activated, produces a transient cell-wide increase in intracellular calcium. This is believed to have a role in the induction of plasticity at granule to Purkinje synapses 29,30. Inactivation and stimulation studies also support the idea that air-puff activation of climbing fibres is necessary for eyelid conditioning. When the climbing fibre input to the cerebellum is inactivated in a welltrained animal, and tone-plus-air-puff trials are carried out, the conditioned eyelid response extinguishes gradually, as might be expected if the tone had been presented in the absence of the air-puff 31 (but see REFS 32,33). In addition, direct stimulation of the dorsal anterior region of the inferior olive (the area that contains climbing fibres that are activated by air-puff stimuli) can serve as a normal and lead to the acquisition of conditioned eyelid responses 34. INFERIOR OLIVARY NUCLE A nucleus situated in a bulge on the ventral medullary surface of the brainstem. Its neurons form very strong excitatory synapses with those of the cerebellum. CG LH Freezing Blood Hormones pressure Figure 2 Neural circuits engaged during fear conditioning. During fear conditioning, the conditioned stimulus () and unconditioned stimulus () are relayed to the lateral nucleus of the amygdala (LA) from thalamic and cortical regions of the auditory and somatosensory systems, respectively. As shown in FIG. 3, the inputs enter the dorsal subregion of the LA, where interactions with the induce plasticity in two functional cell types (so-called trigger and storage cells). information is then transmitted through further stations in the LA to the central nucleus of the amygdala (CE). Interactions between the lateral and central amygdala are more complex than illustrated, and involve local-circuit connections (see main text). The LA also communicates with the CE by way of connections with other amygdala regions (not shown), but the direct pathway seems to be sufficient to mediate fear conditioning. CG, central grey; LH, lateral hypothalamus; PVN, paraventricular hypothalamus. PVN of the mossy fibre granule cell Purkinje cell pathway is so great that each Purkinje cell receives ~100,000 granule cell synapses. In addition to the evidence that tone stimuli activate mossy fibres, several studies indicate that tone activation of mossy fibres is necessary for eyelid conditioning. For example, it has been shown that lesion or temporary inactivation of the mossy fibre system prevents learning and completely abolishes previously conditioned eyelid responses, causing the animals to behave as if the tone were not being presented 24,25. In addition, electrical stimulation of the mossy fibre input to the cerebellum can substitute for the tone, and can lead to the normal acquisition of the eyelid response when the stimulation is paired with an air-puff 26,27. pathway for eyelid conditioning. Stimuli such as airpuffs, which are typically used as the during eyelid conditioning, are known to activate climbing fibres that originate in the INFERIOR OLIVARY NUCLE 28. In addition to sending an excitatory collateral to the deep cerebellar nuclei, climbing fibres represent an important source of input to the cerebellar cortex, where they contact convergence in the cerebellum. The evidence presented in the previous two sections clearly indicates that during eyelid conditioning, information about the and converges in specific regions of the deep cerebellar nuclei, as well as in the overlying cerebellar cortex. A number of recording studies has shown that this convergence occurs at the level of individual neurons in both interpositus and Purkinje cells In addition to the cerebellum, however, there are other brain regions that receive input about the and, and these represent potential sites of plasticity during eyelid conditioning. Perhaps the most compelling evidence for the hypothesis that eyelid conditioning is caused by the convergence of and stimuli in the cerebellum (and not in other, pre-cerebellar areas of convergence in the brain) has come from studies in which direct stimulation of mossy and climbing fibres is substituted for tone and air-puff presentations, respectively, leading to the normal learning of the conditioned eyelid response in the complete absence of external stimuli 40. More recent experiments, which will be discussed below, have begun to provide insights into the relative contributions made by the two sites of convergence within the cerebellum. CR pathway for eyelid conditioning. Can the cerebellum generate the specific movement (eyelid closure) associated with the conditioned eyelid response? It is known that the deep cerebellar nuclei transmit the entire output of the cerebellum to brainstem centres, where various spinal descending pathways originate and are responsible for motor control, both directly, through connections with the spinal cord, and indirectly, through connections to motor regions of cerebral cortex via the thalamus 41,42. Indeed, a map of the entire body exists at the level of the deep cerebellar nuclei, and it has been shown that stimulation of the anterior portion of the interpositus nucleus can elicit eyelid movement 36,43. So, increased activation of cerebellar interpositus output by the is ideally suited to drive the expression of the conditioned eyelid response. As we shall see, the idea that deep cerebellar 124 FEBRUARY 2002 VOLUME 3

4 PARABRACHIAL AREA A nucleus situated in the pons that transmits information from the viscera to the hypothalamus and amygdala. STRIA TERMINALIS One of the main efferent projections of the amygdala. It innervates regions that include the nucleus accumbens and the hypothalamus. nuclei cells are activated by the as a result of learning is supported by the observation that during eyelid conditioning, neurons in the cerebellar interpositus nuclei develop activity that closely models the CR 36,43. Areas engaged during fear conditioning pathways for fear conditioning. The pathways through which inputs reach the amygdala have been studied extensively in recent years. Much of the work has involved the auditory modality, which is focused on here. For studies of visual pathways, see Shi and Davis 44, and Linke et al. 45, and for studies of contextual pathways, see Maren 19, Fanselow 46, and Anagnostaras and colleagues 47. Auditory and other sensory inputs to the amygdala terminate mainly in the lateral amygdala (LA) (FIG. 2) (see REFS 48 52), and damage to this region interferes with fear conditioning to an acoustic 53,54. Auditory inputs to the LA come from both the auditory thalamus and auditory cortex (see REFS 48 50,52,55), and fear conditioning to a simple auditory can be mediated by either of these pathways 56. It seems that the projection to the LA from the auditory cortex is concerned with more complex auditory stimulus patterns 57,but the exact conditions that require this projection are poorly understood 58. Although some lesion studies have questioned the ability of the thalamic pathway to mediate conditioning 59,60, single-neuron recordings show that the cortical pathway conditions more slowly over trials than the thalamic pathway 61,62, indicating that plasticity in the amygdala occurs initially through the thalamic pathway. Recent functional magnetic resonance imaging (fmri) studies in humans have shown that the amygdala shows activity changes during conditioning that correlate with activity in the thalamus but not the cortex 63, further emphasizing the importance of the direct thalamo-amygdala pathway. pathway for fear conditioning. Given that the LA is the site of termination of pathways that carry acoustic inputs, it is important to ask whether inputs might also reach this area and induce plasticity in this region. Thalamic areas that receive afferents from the spino-thalamic tract 64 project to the LA 50 (FIGS 2 and 3). Furthermore, cells in the LA are responsive to nociceptive and auditory stimulation 65. So, the substrate for conditioning (convergence of and information) exists in the LA, and as we shall see below, conditioning induces plasticity in -elicited responses in this area. Cortical areas that process somatosensory stimuli, including nociceptive stimuli, project to the LA and other nuclei in the amygdala (see REFS 48,49,51,52,66). Recent behavioural studies indicate that inputs to the amygdala from either thalamic or cortical areas might be sufficient to mediate conditioning 60, a finding that would parallel the conclusions concerning inputs that are mentioned above. The central amygdala (CE) receives nociceptive inputs from the PARABRACHIAL AREA 67, and also directly from the spinal cord 68. Although the CE does not receive major inputs from sensory areas that process acoustic s, it is a direct recipient of inputs from the LA and from other regions of the amygdala, such as the basal and accessory basal nuclei, both of which receive inputs from the LA (see below). inputs to the CE could be involved in higher-order integration. For example, representations created by convergence in the LA, when transferred to the CE directly or by way of connections from the basal or accessory basal amygdala, might converge with and be further modified by nociceptive inputs to the CE. convergence in the amygdala. As described above, and pathways both project to the LA, and cells in the LA respond to both auditory and nociceptive stimulation 65. So, the substrate for conditioning (the convergence of and information) exists in the LA, and as we shall see, conditioning induces plasticity in -elicited responses in this area. However, convergence does not occur in the entire LA. The LA consists of three subregions (dorsal, ventrolateral and medial) 69, and convergence occurs only in the dorsal subregion 65. Interestingly, this subregion is also the site of initial plasticity and long-term storage (see the trigger-and-storage model below). CR pathway for fear conditioning. The CE projects to brainstem areas that control the expression of fear responses. Whereas damage to the CE interferes with the expression of conditioned fear responses, damage to areas that the CE projects to selectively interrupts the expression of individual responses For example, damage to the lateral hypothalamus affects blood pressure but not freezing responses, and damage to the periaqueductal grey interferes with freezing but not with blood pressure responses 71. Similarly, damage to the bed nucleus of the STRIA TERMINALIS has no effect on either blood pressure or freezing responses 71, but disrupts the conditioned release of pituitary adrenal stress hormones 77. Intra-amygdala pathways. From the findings described above, it would seem that information about a simple (such as a tone paired with shock) is directed towards the CE (where response execution is initiated) by way of pathways that originate in the LA. The LA projects to the CE, both directly and through the basal and accessory basal amygdala 69,78. Although the LA and CE communicate in a complex manner that involves local circuits 79,80, the direct projection seems to be sufficient, because lesions of the basal and accessory basal amygdala have no effect on simple fear conditioning to a tone An alternative view was proposed by Killcross et al. 84, who argued that a direct projection to the CE that bypasses the LA can mediate conditioning. However, as we have already seen, fibres from auditory areas terminate mainly in the LA. Moreover, auditory response latencies in the LA are shorter than in the CE, both before and after conditioning (see next section), indicating that the CE depends on the LA for its inputs. It is important to point out, however, that the task used to rule out the LA as a way station to the CE differed in several ways from the tasks used to implicate the LA, including the use of hundreds instead of tens of training NATURE REVIEWS NEUROSCIENCE VOLUME 3 FEBRUARY

5 trials 85. It is possible that the further training trials in the Killcross study allowed the brain to learn in a way that is not normally used when fewer trials are given. At most, a direct pathway to the CE would be an alternative rather than the main route of transmission through the amygdala. Recent studies have revealed some details of the local circuit connections that are involved in the flow of information from the LA to the CE 80,but much remains unknown about these processing pathways. Areas necessary for learning Much of the initial evidence that implicated specific brain regions in eyelid and fear conditioning was based on the results of studies that examined whether learning was possible after surgical removal or destruction of various areas. More recently, techniques involving pharmacological inactivation have allowed more control over the temporal extent of the lesion, and have provided a way to distinguish learning deficits caused by the inability to learn from those that simply reflect the inability to perform or express the CR. Eyelid conditioning. The demonstration that decorticate and decerebrate animals were able to learn classically conditioned eyelid responses led to intense investigation into the role of neural circuits in the brainstem and cerebellum in eyelid conditioning Subsequently, the cerebellum was clearly implicated in a lesion study in rabbits, which showed that removal of a 1-mm 3 portion of the anterior interpositus cerebellar nucleus completely prevents learning of the conditioned eyelid response 43,89. Although these studies indicate an essential role of the cerebellum in eyelid conditioning, the results were initially challenged on the grounds that cerebellar lesions produced a performance deficit rather than an impairment of learning Indeed, because cerebellar output is required to drive the conditioned eyelid response, it was suggested that learning occurred somewhere upstream of the cerebellum, and that the cerebellar lesions simply blocked its expression. This issue has recently been addressed in several reversible inactivation studies. Temporarily inactivating either the cerebellar nucleus and/or overlying cerebellar cortex with a local anaesthetic such as lidocaine, or with the GABA (γ-aminobutyric acid) agonist muscimol, prevents learning, as evidenced by the lack of CRs after the inactivation By contrast, temporary inactivation of the efferent fibres from the cerebellum prevents the expression of conditioned eyelid responses during inactivation, but does not prevent learning, as shown by the presence of conditioned eyelid responses after the inactivation 96,97. Although lesion and inactivation studies have implicated the cerebellum in eyelid conditioning, it has been much more challenging to tease apart the relative contributions made by the cerebellar cortex and deep cerebellar nuclei to the learning process. The hypothesis that memory is localized in the deep cerebellar nuclei was initially supported by studies showing that in well-trained animals, lesions restricted to the cere- bellar cortex had no effect on previously conditioned eyelid responses 36,43. By contrast, others found that lesions of the cerebellar cortex permanently abolished the conditioned eyelid response, and they proposed that the memory must be stored there 98,99. Adding more fuel to the controversy, recent studies have indicated that when the lesions include the anterior lobe of the cerebellar cortex, the timing of the CR is affected 100,101. That is, instead of closing the eyelid just before the time when the air-puff is expected, animals with anteriorlobe lesions close the eyelid immediately after the onset of the tone stimulus. These results are consistent with the hypothesis that eyelid conditioning is mediated by two sites of plasticity within the cerebellum. Plasticity in the cerebellar cortex might be required for learning to time the CR appropriately, whereas plasticity in the cerebellar nuclei might be responsible for the shortlatency responses that are observed after lesions of the anterior lobe. The intriguing possibility that the cerebellar cortex and deep cerebellar nuclei could contribute to different aspects of the learning process is discussed further below. Fear conditioning. Lesion studies in the amygdala are limited by the small size of this region in the brains of the animals most often used in fear-conditioning studies, namely rats. Nevertheless, considerable progress has been made with these techniques. Early studies showed that large lesions of the amygdala prevented fear conditioning 102, but later work showed that damage limited to or including the CE was sufficient to prevent the acquisition or expression of learning 70,73,76,103.However, subsequent anatomical and physiological studies indicated that the LA was the primary region within the amygdala to receive inputs from sensory systems that process the auditory 50, , indicating that this region might be a key link in the pathway before the CE. Indeed, following these anatomical and behavioural clues about the role of the LA as the sensory interface of the amygdala, several studies showed that electrolytic lesions confined to the LA, or neurotoxic lesions that included the LA and adjacent areas, prevented auditory conditioning 53,54,59,81,82. As mentioned above, permanent lesions raise questions about the role of the damaged area in the performance of the response as opposed to the acquisition of the association. To address this issue, pharmacological agents have been used to temporarily inactivate the amygdala during learning. These studies show that inactivation of the LA disrupts learning in a dose-dependent way 107,108. To rule out the possibility that drugs infused before training linger and have their effects on posttraining consolidation processes 109,110, studies were also carried out with infusions immediately after training, a procedure that had no effect 111,112. Because the subjects were all tested drug-free one or more days after training, the effects of the inactivation are limited to the training time, ruling out a performance deficit at the time of the test as an explanation. These studies indicate that functional activity in the amygdala, presumably in the LA, is required for conditioning to occur. 126 FEBRUARY 2002 VOLUME 3

6 MITOGEN-ACTIVATED PROTEIN KINASE Any member of a family of protein kinases that are important for relaying signals from the cell membrane to the nucleus. PROTEIN KINASE A Also known as cyclic-ampdependent protein kinase. One of a class of enzymes that use ATP as a phosphoryl-group donor to phosphorylate hydroxyl or phenolic groups on their target proteins. The issue of whether the amygdala is required for the performance of the CR has been investigated in other ways. Recent studies, for example, have shown that neurotoxic lesions of the LA and the adjacent basal nucleus disrupt the fear response elicited by a but not by a, namely the odour emitted by a predator 113. As the conditioned and unconditioned response was the same (freezing), the failure to express the response to the cannot be due to a performance deficit. In summary, lesions of specific regions of the cerebellum and amygdala block the learning that occurs in eyelid and fear conditioning, respectively. In both cases, the effects of the lesions are consistent with a disruption of the associative processes that underlie learning, rather than with an inability to express the CR. Neuronal plasticity during learning Progress in memory research has benefited greatly from the development of recording and imaging techniques that allow investigators to listen to the brain as the animal learns or remembers. Because learning is associated with a change in behaviour, neurons that change their activity during conditioning are good candidate cells for memory storage. It is important to remember, however, that not all neurons that show activity changes with learning represent sites of plasticity. For example, the activity of cells downstream of a site of plasticity will still change during the learning process, even though the neurons themselves might not contribute to memory formation. Plasticity during eyelid conditioning. Two sets of cerebellar neurons change their activity during eyelid conditioning, and so represent potential sites of plasticity. Several recording studies in rabbits have shown that after learning, Purkinje cells in the cerebellar cortex, as well as neurons in the anterior interpositus nucleus, show activity that precedes and closely resembles the conditioned eyelid response 35,36,38,43,114. In general, functional imaging studies in humans have provided evidence that is consistent with the results of the animal studies. For example, metabolic changes in the ipsilateral cerebellar nuclei and bilateral cerebellar cortex have been shown to correlate with the degree of learning These results, together with the evidence discussed in the previous sections, strongly indicate that changes in activity in both the cerebellar cortex and cerebellar nuclei might have a role in eyelid conditioning. Although the cellular pathways that underlie these learning-dependent changes are not fully understood, tremendous progress has been made in unravelling the molecular mechanisms that contribute to plasticity in the cerebellum 118. Much of this research has focused on the long-term depression (LTD) of the granule to Purkinje cell synapse 29,30, although it is now clear that many other cerebellar cells and synapses, including the deep cerebellar nuclei 119,120, are plastic and can participate in the learning process 118. An important challenge for the future will be to determine how the growing list of plastic synapses in the cerebellum contribute to the different aspects of learning and memory storage during eyelid conditioning. Plasticity during fear conditioning. Neuronal responsivity in the LA is altered by fear conditioning. Studies in rats and cats, using various electrophysiological techniques, including single-cell and field-potential recordings in awake, behaving animals 61, , as well as in vitro studies of synaptic currents 125, all indicate that LA neurons become more responsive to the after pairings with a. Further in vitro studies have made considerable progress in identifying the cellular and molecular mechanisms involved in plasticity in the LA, indicating that NMDA (N-methyl-D-aspartate) receptors and L-type voltage-gated calcium channels are the source of intracellular signals that initiate the activation of protein kinases (especially MITOGEN-ACTIVATED PROTEIN KINASE (MAPK) and PROTEIN KINASE A (PKA)), which in turn leads to protein-synthesis-dependent changes in plasticity (see REFS 18,126,127). Behavioural studies have implicated the same signalling pathways in fear conditioning 127. Imaging studies in humans also show that amygdala activity increases during fear conditioning So, the anatomical, inactivation and electrophysiological evidence all point to the amygdala, and specifically the LA, as a crucial site for fear conditioning. In general, the physiological changes develop quickly and are expressed at short latencies, consistent with the hypothesis that such changes underlie the learning of behavioural responses. Changes in neural activity also occur in the CE 70, but the latencies of the CRs seem to be longer than in the LA, indicating that these changes might be secondary to LA plasticity. Alternatively, as the CE receives inputs that are separate from those arriving in the LA, it is possible that plasticity in the CE occurs independently of the LA. The trigger-and-storage model of memory Recent results from eyelid and fear conditioning raise the intriguing possibility that for both behaviours, anatomically distinct populations of neurons might mediate the initiation of learning and long-term storage of memory. These studies provide a first step towards understanding how the different anatomical components of a particular brain system might interact during the learning process. Trigger and storage neurons in eyelid conditioning. Recent evidence indicates that for eyelid conditioning, the cerebellar cortex might have a role in the initiation of learning, whereas the deep cerebellar nuclei are more involved in long-term memory storage. This hypothesis is based on behavioural observations that indicate that plasticity in the cerebellar cortex is induced even before the first conditioned eyelid response is produced 132 (which typically requires tone-plus-air-puff trials). By contrast, plasticity in the cerebellar nucleus, as shown by the presence of short-latency conditioned eyelid responses after inactivation of the cerebellar cortex, develops more slowly over the training period 8. Indeed, the time course for induction of the plasticity responsible for the short-latency response is indistinguishable from the time course for acquisition of properly timed conditioned eyelid responses. Interestingly, plasticity in the NATURE REVIEWS NEUROSCIENCE VOLUME 3 FEBRUARY

7 a b c Fear conditioning D V LAm LAvl LAd Purk Trigger stage Cerebral cortex Corpus callosum Hippocampus Third ventricle Thalamus Internal capsule Striatum Hypothalamus Amygdala e f Eyelid conditioning DCN nucleus persists even after conditioned eyelid responses have been fully extinguished. That is, if the cerebellar cortex is inactivated after CRs have been extinguished (which typically requires tonealone presentations), short-latency responses, which provide evidence of plasticity in the cerebellar nuclei, are unmasked during the inactivation. For most of the rabbits in the study, cerebellar nuclear plasticity was diminished but still detectable even after 3,000 tonealone trials. Furthermore, the rate at which CRs were reacquired during subsequent tone-plus-air-puff trials was strongly correlated with the amount of residual plasticity that was left in the cerebellar nuclei after extinction. These studies indicate that plasticity in the cerebellar cortex and cerebellar nuclei might contribute to different aspects of the learning process. FIGURE 3 illustrates the way in which these two sites of plasticity might contribute to memory formation. The induction of plasticity in the cortex occurs quickly, during the initial training trials, and so might serve as a trigger signal for the induction of plasticity in the nucleus 132. By contrast, cerebellar nuclear plasticity is induced more slowly, persists long after conditioned eyelid responses have been extinguished, and seems to have a role in the savings (fast learning) seen during reacquisition 8. It is not yet known whether these changes can be induced independently of one another, with the induction of plasticity in the cerebellar nuclei proceeding at a slower rate than in the cerebellar cortex, or whether the induction of plasticity in the cerebellar cortex is necessary to drive changes in the cerebellar nuclei. d LAm CE Storage stage Figure 3 The trigger-and-storage model. a Hemisection of the rat brain at the level of the amygdala (included within the dashed lines). b The area of the amygdala is enlarged to show the three main subdivisions of the lateral nucleus: dorsal (LAd), medial (LAm) and ventrolateral (LAvl). The conditioned and unconditioned stimuli ( and ) converge on single cells in the LAd. c During the trigger stage of fear conditioning, convergence of the and initiates rapid but transient plasticity in cells of the dorsal LAd. d The induction of plasticity in the dorsal (D) LAd leads to more slowly acquired, longer-lasting changes in the storage cells of the ventral LAd. Ventral (V) LAd cells would then communicate with LAm cells that project to the central nucleus of the amygdala (CE) and are responsible for driving the CR. e The and used in eyelid conditioning converge on the Purkinje cells of the cerebellar cortex (Purk), as well as on the interpositus cells of the deep cerebellar nuclei (DCN). f According to the trigger-and-storage model of eyelid conditioning, plasticity is first induced in the Purkinje cells, even before any conditioned responses can be detected. g Subsequently, plasticity is induced in the interpositus nucleus, which is responsible for the long-term storage of memory and the expression of the conditioned eyelid response. g CR Trigger and storage neurons in fear conditioning. Of the three LA subregions 69, only the dorsal region (LAd) is a site of convergence 65. Recent studies indicate that two distinct sets of neurons in the LAd could contribute differentially to the initiation and storage of fear-conditioning memories 121. Cells in the dorsal tip of the LAd, which tend to show more robust and shorter-latency auditory responses before training than cells in more ventral LAd regions, develop conditioning-induced enhancement of their auditory responses very early in training, even before behavioural responses are evident. This plasticity is evident only at the outset of training, as auditory responsivity in these cells resets towards baseline levels with repeated training. By contrast, cells in more ventral regions of the LAd take longer to reach maximal response levels, but the enhanced response persists throughout training and even after behavioural extinction of the CRs. In addition, these cells in the ventral LAd are much more likely than dorsal cells to show certain conditioning-induced biochemical changes, such as the activation of MAPKs, that are crucial for the consolidation of long-term memory of fear conditioning 133. So, as in eyelid conditioning, plasticity might be triggered and stored in different populations of cells during fear conditioning. In addition to storage within these ventral LAd cells, storage of certain aspects of fear memories might be distributed across 128 FEBRUARY 2002 VOLUME 3

8 several brain systems, including the cortex and thalamus, both of which show evidence of lasting neural changes that result from fear conditioning 134. Interestingly, current evidence indicates that most of these changes depend on the amygdala 62, The exact manner in which the so-called trigger and storage cells participate in plasticity is still a matter for speculation. One hypothesis, depicted in FIG. 3, shows the trigger and storage cells in a simple linear arrangement, with plasticity in the trigger cells leading to longer-lasting changes in the storage cells. This model can be extended to account for further data. Specifically, as mentioned earlier, auditory inputs reach the LA from both the thalamus and cortex, with the former providing a faster but cruder representation of the than the latter. As the trigger and storage regions both receive and input, it is possible that plasticity of the fast and crude thalamic input primes and enables plasticity in the slower but more precise representation provided by the cortical input. The role of these two functional cell types in the LAd remains to be clarified. Although some have questioned the contribution of the amygdala to memory storage 110, the discovery of cells in the LA that store the effects of learning, even beyond behavioural extinction, provides support for the idea that long-term memory of fear conditioning is stored, at least in part, by enduring changes in synaptic transmission within the LA. Recent studies of the cellular and molecular mechanisms that underlie these synaptic events have led to the development of a specific hypothesis about how the synaptic changes are manifested 18,127. As mentioned above, one of the key steps, activation of MAPK, seems to occur preferentially within the ventral storage cells of the LAd 133.It will be interesting to further elucidate the manner in which the trigger and storage cells within the LAd can be related to specific cellular and molecular mechanisms that contribute to plasticity during fear conditioning. Future research directions Recent studies in eyelid and fear conditioning indicate that during memory formation in these two simple forms of learning, soft plasticity might have a role in the initial encoding, whereas hard plasticity contributes to the long-term storage of memory. However, many of the details underlying this process remain unknown. For example, it is not clear whether there is a complete transfer of memory from the soft trigger cells to the hard storage cells. This might be the case for fear conditioning, because the activity of dorsal LAd neurons is modulated only during the initiation of learning, and subsequently resets to baseline levels with repeated training 121.For eyelid conditioning, however, cells of the cerebellar cortex might have a role not only at the beginning of training, but also later on, because lesions of the cerebellar cortex are known to disrupt the timing of the conditioned eyelid response in well-trained animals 100,101,138. Another issue that deserves further attention pertains to the role of the trigger cells during the learning process. It is not known, for example, whether plasticity in the trigger cells is required for the induction of plasticity in the storage cells, or whether the induction of soft and hard plasticity can proceed independently, but at different rates. For eyelid conditioning, this issue remains controversial, because one set of results has shown that mutant mice lacking Purkinje cells, the presumed trigger neurons of the cerebellar cortex, are impaired but still able to acquire conditioned eyelid responses 139,140. By contrast, more recent studies indicate that mutant mice with deficient cerebellar LTD do not learn the CR at all 141, which is consistent with earlier reports showing that lesions of the cerebellar cortex prevent both acquisition and extinction 101,142. Given the potential for compensatory mechanisms, both in animals with permanent lesions and in mutant mice, addressing the controversial role of the cerebellar cortex in the induction of plasticity in the cerebellar nucleus will probably require the use of reversible lesions or conditional mutations 95. Within the amygdala, LAd cells might be important in the initiation of plasticity in other amygdala regions, such as the basal or central nucleus, which are downstream of the LA. Whether the dorsal LAd trigger cells have this role for ventral LAd storage cells only, or for other amygdala regions as well, is not known. There are also several important questions that remain unanswered about the cellular/molecular mechanisms that are responsible for memory formation during eyelid and fear conditioning. Although bidirectional plasticity at granule-to-purkinje synapses in the cerebellar cortex has been widely studied and seems well suited to drive learning 29,30,118, the role of plasticity in the cerebellar nucleus, as well as the conditions required for its induction and the molecular pathways involved, is less well understood 119,120,143. Similarly, considerable progress has been made in implicating certain molecules in plasticity and fear conditioning in the amygdala (see REFS 18,127), but much remains to be done in this area. Finally, it is important to consider the broader implications of this work. The simple versions of eyelid and fear conditioning described here have been extended in several ways. For example, procedural variations, such as insertion of additional s that have to be differentiated from one another (discriminated conditioning paradigm) or that interfere with the primary (blocking or latent inhibition paradigms), alterations in the timing of the and (trace conditioning), the use of qualitatively different kinds of (for example, contextual conditioning), or requiring an instrumental response as opposed to a Pavlovian one (avoidance conditioning), leads to predictable changes in the brain systems that are involved in fear or eyelid conditioning or both. Much less is known about the circuitry and cellular mechanisms that are involved in these paradigms; nevertheless, they offer the opportunity for researchers to build on the foundation laid by the exacting studies of the simple versions of these tasks in the quest to achieve an ever-growing understanding of how the brain learns and stores information in a variety of neural networks. NATURE REVIEWS NEUROSCIENCE VOLUME 3 FEBRUARY

9 1. Pavlov, I. P. Conditioned Reflexes. An Investigation of the Physiological Activity of the Cerebral Cortex (Oxford Univ. Press, London, 1927). 2. Gormezano, I., Prokasy, W. F. & Thompson, R. F. Classical Conditioning (Lawrence Erlbaum Associates, Hillsdale, New Jersey, 1987). 3. Gormezano, I., Kehoe, E. J. & Marshall, B. S. Twenty years of classical conditioning research with the rabbit. Prog. Psychobiol. Physiol. Psychol. 10, (1983). 4. Schneiderman, N., Fuentes, I. & Gormezano, I. Acquisition and extinction of the classically conditioned eyelid response in the albino rabbit. Science 136, (1962). 5. Mauk, M. D. & Ruiz, B. P. Learning-dependent timing of Pavlovian eyelid responses: differential conditioning using multiple interstimulus intervals. Behav. Neurosci. 106, (1992). 6. Millenson, J. R., Kehoe, E. J. & Gormezano, I. Classical conditioning of the rabbit s nictitating membrane response under fixed and mixed intervals. Learn. Motiv. 8, (1977). 7. Schreurs, B. G. Long-term memory and extinction of the classically conditioned rabbit nictitating membrane response. Learn. Motiv. 24, (1993). 8. Medina, J. F., Garcia, K. S. & Mauk, M. D. A mechanism for savings in the cerebellum. J. Neurosci. 21, (2001). Showed that plasticity located outside the cerebellar cortex persists after behavioural extinction. The amount of plasticity that remained after extinction was a reliable predictor of the rate of reacquisition. 9. Napier, R. M., Macrae, M. & Kehoe, E. J. Rapid reacquisition in conditioning of the rabbit s nictitating membrane response. J. Exp. Psychol. Anim. Behav. Process. 18, (1992). 10. Lennartz, R. C. & Weinberger, N. M. Analysis of response systems in Pavlovian conditioning reveals rapidly versus slowly acquired conditioned responses: support for two factors, implications for behavior and neurobiology. Psychobiology 20, (1992). 11. LeDoux, J. E. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, (2000). 12. Brown, J. S., Kalish, H. I. & Farber, I. E. Conditioned fear as revealed by magnitude of startle response to an auditory stimulus. J. Exp. Psychol. 41, (1951). 13. Bolles, R. C. & Fanselow, M. S. A perceptual defensive recuperative model of fear and pain. Behav. Brain Sci. 3, (1980). 14. Bouton, M. E. Conditioning, remembering, and forgetting. J. Exp. Psychol. Anim. Behav. Process. 20, (1994). 15. Kim, J. J. & Thompson, R. F. Cerebellar circuits and synaptic mechanisms involved in classical eyeblink conditioning. Trends Neurosci. 20, (1997). 16. Raymond, J. L., Lisberger, S. G. & Mauk, M. D. The cerebellum: a neuronal learning machine? Science 272, (1996). 17. Fanselow, M. S. & LeDoux, J. E. Why we think plasticity underlying Pavlovian fear conditioning occurs in the basolateral amygdala. Neuron 23, (1999). 18. Blair, H. T., Schafe, G. E., Bauer, E. P., Rodrigues, S. M. & LeDoux, J. E. Synaptic plasticity in the lateral amygdala: a cellular hypothesis of fear conditioning. Learn. Mem. 8, (2001). 19. Maren, S. Neurobiology of Pavlovian fear conditioning. Annu. Rev. Neurosci. 24, (2001). 20. Aitkin, L. M. & Boyd, J. Acoustic input to the lateral pontine nuclei. Hear. Res. 1, (1978). 21. Eccles, J. C., Ito, M. & Szentágothai, J. The Cerebellum as a Neuronal Machine (Springer, New York, 1967). 22. Llinas, R. in Handbook of Physiology II. The Nervous System (ed. Brokks, V. B.) (Am. Physiol. Soc., Bethesda, Maryland, 1981). 23. Voogd, J. & Glickstein, M. The anatomy of the cerebellum. Trends Neurosci. 21, (1998). 24. Solomon, P. R., Lewis, J. L., LoTurco, J. J., Steinmetz, J. E. & Thompson, R. F. The role of the middle cerebellar peduncle in acquisition and retention of the rabbit s classically conditioned nictitating membrane response. Bull. Psychon. Soc. 24, (1986). 25. Lewis, J. L., LoTurco, J. J. & Solomon, P. R. Lesions of the middle cerebellar peduncle disrupt acquisition and retention of the rabbit s classically conditioned nictitating membrane response. Behav. Neurosci. 101, (1987). 26. Steinmetz, J. E. et al. Classical conditioning of the rabbit eyelid response with a mossy-fiber stimulation. I. Pontine nuclei and middle cerebellar peduncle stimulation. Behav. Neurosci. 100, (1986). 27. Hesslow, G., Svensson, P. & Ivarsson, M. Learned movements elicited by direct stimulation of cerebellar mossy fiber afferents. Neuron 24, (1999). 28. Sears, L. L. & Steinmetz, J. E. Dorsal accessory inferior olive activity diminishes during acquisition of the rabbit classically conditioned eyelid response. Brain Res. 545, (1991). 29. Linden, D. J. & Connor, J. A. Cellular mechanisms of longterm depression in the cerebellum. Curr. Opin. Neurobiol. 3, (1993). 30. Ito, M. Cerebellar long-term depression: characterization, signal transduction, and functional roles. Physiol. Rev. 81, (2001). 31. McCormick, D. A., Steinmetz, J. E. & Thompson, R. F. Lesions of the inferior olivary complex cause extinction of the classically conditioned eyeblink response. Brain Res. 359, (1985). 32. Yeo, C. H., Hardiman, M. J. & Glickstein, M. Classical conditioning of the nictitating membrane response of the rabbit. IV. Lesions of the inferior olive. Exp. Brain Res. 63, (1986). 33. Welsh, J. P. & Harvey, J. A. Acute inactivation of the inferior olive blocks associative learning. Eur. J. Neurosci. 10, (1998). 34. Mauk, M. D., Steinmetz, J. E. & Thompson, R. F. Classical conditioning using stimulation of the inferior olive as the unconditioned stimulus. Proc. Natl Acad. Sci. A 83, (1986). 35. Hesslow, G. & Ivarsson, M. Suppression of cerebellar Purkinje cells during conditioned responses in ferrets. Neuroreport 5, (1994). 36. McCormick, D. A. & Thompson, R. F. Neuronal responses of the rabbit cerebellum during acquisition and performance of a classically conditioned nictitating membrane-eyelid response. J. Neurosci. 4, (1984). The first clear evidence that the cerebellum is involved in eyelid conditioning. Lesions of the cerebellum prevented learning; recording data showed cells within the interpositus nucleus that developed activity resembling the conditioning response; stimulation of those interpositus sites caused eyelid closure. 37. Berthier, N. E. & Moore, J. W. Activity of deep cerebellar nuclear cells during classical conditioning of nictitating membrane extension in rabbits. Exp. Brain Res. 83, (1990). 38. Berthier, N. E. & Moore, J. W. Cerebellar Purkinje cell activity related to the classically conditioned nictitating membrane response. Exp. Brain Res. 63, (1986). 39. Gould, T. J., Sears, L. L. & Steinmetz, J. E. Possible and pathways for rabbit classical eyelid conditioning: electrophysiological evidence for projections from the pontine nuclei and inferior olive to cerebellar cortex and nuclei. Behav. Neural Biol. 60, (1993). 40. Steinmetz, J. E., Lavond, D. G. & Thompson, R. F. Classical conditioning in rabbits using pontine nucleus stimulation as a conditioned stimulus and inferior olive stimulation as an unconditioned stimulus. Synapse 3, (1989). 41. Middleton, F. A. & Strick, P. L. Cerebellar output channels. Int. Rev. Neurobiol. 41, (1997). 42. Ito, M. The Cerebellum and Neural Control (Raven, New York, 1984). 43. McCormick, D. A. & Thompson, R. F. Cerebellum: essential involvement in the classically conditioned eyelid response. Science 223, (1984). 44. Shi, C. & Davis, M. Visual pathways involved in fear conditioning measured with fear-potentiated startle: behavioral and anatomic studies. J. Neurosci. 21, (2001). 45. Linke, R., De Lima, A. D., Schwegler, H. & Pape, H. C. Direct synaptic connections of axons from superior colliculus with identified thalamo-amygdaloid projection neurons in the rat: possible substrates of a subcortical visual pathway to the amygdala. J. Comp. Neurol. 403, (1999). 46. Fanselow, M. S. Contextual fear, Gestalt memories, and the hippocampus. Behav. Brain Res. 110, (2000). 47. Anagnostaras, S. G., Gale, G. D. & Fanselow, M. S. Hippocampus and contextual fear conditioning: recent controversies and advances. Hippocampus 11, 8 17 (2001). 48. Mascagni, F., McDonald, A. J. & Coleman, J. R. Corticoamygdaloid and corticocortical projections of the rat temporal cortex: a Phaseolus vulgaris leucoagglutinin study. Neuroscience 57, (1993). 49. Romanski, L. M. & LeDoux, J. E. Information cascade from primary auditory cortex to the amygdala: corticocortical and corticoamygdaloid projections of temporal cortex in the rat. Cereb. Cortex 3, (1993). 50. LeDoux, J. E., Farb, C. & Ruggiero, D. A. Topographic organization of neurons in the acoustic thalamus that project to the amygdala. J. Neurosci. 10, (1990). 51. Amaral, D. G., Price, J. L., Pitkänen, A. & Carmichael, S. T. in The Amygdala: Neurobiological Aspects of Emotion, Memory, and Mental Dysfunction (ed. Aggleton, J. P.) 1 66 (Wiley Liss, New York, 1992). 52. McDonald, A. J. Cortical pathways to the mammalian amygdala. Prog. Neurobiol. 55, (1998). 53. LeDoux, J. E., Cicchetti, P., Xagoraris, A. & Romanski, L. M. The lateral amygdaloid nucleus: sensory interface of the amygdala in fear conditioning. J. Neurosci. 10, (1990). The first demonstration of the role of the LA in fear conditioning. It shifted the emphasis from the central to the lateral nucleus in the search for sites of plasticity. 54. Goosens, K. A. & Maren, S. Contextual and auditory fear conditioning are mediated by the lateral, basal, and central amygdaloid nuclei in rats. Learn. Mem. 8, (2001). 55. Doron, N. N. & LeDoux, J. E. Cells in the posterior thalamus project to both amygdala and temporal cortex: a quantitative retrograde double-labeling study in the rat. J. Comp. Neurol. 425, (2000). 56. Romanski, L. M. & LeDoux, J. E. Equipotentiality of thalamoamygdala and thalamo-cortico-amygdala projections as auditory conditioned stimulus pathways. J. Neurosci. 12, (1992). 57. Jarrell, T. W., Gentile, C. G., Romanski, L. M., McCabe, P. M. & Schneiderman, N. Involvement of cortical and thalamic auditory regions in retention of differential bradycardia conditioning to acoustic conditioned stimuli in rabbits. Brain Res. 412, (1987). 58. Armony, J. L., Servan-Schreiber, D., Romanski, L. M., Cohen, J. D. & LeDoux, J. E. Stimulus generalization of fear responses: effects of auditory cortex lesions in a computational model and in rats. Cereb. Cortex 7, (1997). 59. Campeau, S. & Davis, M. Involvement of the central nucleus and basolateral complex of the amygdala in fear conditioning measured with fear-potentiated startle in rats trained concurrently with auditory and visual conditioned stimuli. J. Neurosci. 15, (1995). 60. Shi, C. & Davis, M. Pain pathways involved in fear conditioning measured with fear-potentiated startle: lesion studies. J. Neurosci. 19, (1999). The first attempt to map the pathways to the amygdala. Indicates that as in the case of the auditory pathways, the reaches the amygdala by way of thalamic and cortical somatosensory routes. 61. Quirk, G. J., Repa, J. C. & LeDoux, J. E. Fear conditioning enhances short-latency auditory responses of lateral amygdala neurons: parallel recordings in the freely behaving rat. Neuron 15, (1995). The first demonstration of cellular plasticity in the LA. Latencies indicated a role for direct thalamic inputs to the LA from the auditory thalamus. 62. Quirk, G. J., Armony, J. L. & LeDoux, J. E. Fear conditioning enhances different temporal components of toned-evoked spike trains in auditory cortex and lateral amygdala. Neuron 19, (1997). 63. Morris, J. S., Ohman, A. & Dolan, R. J. A subcortical pathway to the right amygdala mediating unseen fear. Proc. Natl Acad. Sci. A 96, (1999). Presented evidence for the existence of a subcortical transmission route to the amygdala in the human brain, mirroring the findings in rats. 64. LeDoux, J. E., Ruggiero, D. A., Forest, R., Stornetta, R. & Reis, D. J. Topographic organization of convergent projections to the thalamus from the inferior colliculus and spinal cord in the rat. J. Comp. Neurol. 264, (1987). 65. Romanski, L. M., LeDoux, J. E., Clugnet, M. C. & Bordi, F. Somatosensory and auditory convergence in the lateral nucleus of the amygdala. Behav. Neurosci. 107, (1993). 66. Turner, B. H. & Zimmer, J. The architecture and some interconnections of the rat amygdala and lateral periallocortex. J. Comp. Neurol. 227, (1984). 67. Bernard, J. F. & Besson, J. M. The spino(trigemino)pontoamygdaloid pathway: electrophysiological evidence for an involvement in pain processes. J. Neurophysiol. 63, (1990). 68. Burstein, R. & Potrebic, S. Retrograde labeling of neurons in the spinal cord that project directly to the amygdala or the orbital cortex in the rat. J. Comp. Neurol. 335, (1993). 69. Pitkanen, A., Savander, V. & LeDoux, J. E. Organization of intra-amygdaloid circuitries in the rat: an emerging framework for understanding functions of the amygdala. Trends Neurosci. 20, (1997). 70. Kapp, B. S., Whalen, P. J., Supple, W. F. & Pascoe, J. P. in The Amygdala: Neurobiological Aspects of Emotion, Memory, and Mental Dysfunction (ed. Aggleton, J. P.) (Wiley Liss, New York, 1992). 71. LeDoux, J. E., Iwata, J., Cicchetti, P. & Reis, D. J. Different projections of the central amygdaloid nucleus mediate autonomic and behavioral correlates of conditioned fear. J. Neurosci. 8, (1988). 72. Gentile, C. G., Jarrell, T. W., Teich, A., McCabe, P. M. & Schneiderman, N. The role of amygdaloid central nucleus in the retention of differential Pavlovian conditioning of bradycardia in rabbits. Behav. Brain Res. 20, (1986). 73. Davis, M. The role of the amygdala in fear-potentiated startle: implications for animal models of anxiety. Trends Pharmacol. Sci. 13, (1992). 74. Hitchcock, J. & Davis, M. Lesions of the amygdala but not of the cerebellum or red nucleus block conditioned fear as measured with the potentiated startle paradigm. Behav. Neurosci. 100, (1986). 75. Iwata, J., LeDoux, J. E., Meeley, M. P., Arneric, S. & Reis, D. J. Intrinsic neurons in the amygdaloid field projected to by the medial geniculate body mediate emotional responses 130 FEBRUARY 2002 VOLUME 3