Learning is an enduring change in behavior arising from experience-induced structural changes in the brain. For example, during Pavlovian (classical) aversive learning, an organism learns that an auditory or visual stimulus (the conditioned stimulus, CS) that repeatedly precedes an electric shock (the unconditioned stimulus, UCS) predicts the occurrence of the shock. This learned association is reflected in a variety of adaptive cardiovascular responses to the CS and is a specific consequence of the associative relationship between the CS and UCS. The search for the structural changes of Pavlovian learning is governed by the assumption that an association between the CS and the UCS hinges on a convergence of information about both at a common brain structure or structures. It is this convergence that results in structural changes.
Researchers have used learned cardiovascular responses, particularly changes in heart rate (HR) in response to the CS during Pavlovian learning, as models for assessing learning and for identifying the brain regions in which learning-related structural changes occur. HR responses are especially useful because researchers have identified the location of the motor neurons that produce these responses and can thus identify the central brain structures that activate these neurons. This, then, sets the stage for determining the specific location(s) and nature of the structural changes.
The conditioned cardiovascular changes that occur during Pavlovian conditioning represent only one of a whole complex of nonspecific visceral responses that occur during classical conditioning. Acquisition of these learned autonomic adjustments occurs rapidly, often within just two to three CS/UCS presentations (Powell, McLaughlin, and Chachick, 2000). These changes may also accompany later-occurring specific learned somatomotor responses, such as the conditioned eye blink or the leg-flexion response. Learned autonomic changes are nonspecific responses because they occur regardless of the nature of the conditioning contingencies. A CS that signals an electric shock UCS delivered to either the orbital region or the footpad of animals, for example, results in a host of nonspecific CRs to the signals that are quite similar regardless of the application of the UCS. However, the learned somatomotor response is specific to the UCS, consisting of an eyeblink CR in the former case and leg-flexion CR in the latter. Hence such responses are usually referred to as specific conditioned responses.
There are many parametric differences between the acquisition of specific and nonspecific responses during classical conditioning. A wealth of research indicates that the the brain mechanisms responsible for the early-occuring heart rate and other nonspecific changes are quite different from those that mediate the later acquisition of somatomotor changes. We will concentrate on the former mechanisms, with specific reference to cardiovascular changes, although there is substantial evidence suggesting that there are interactions between the structures that mediate specific and nonspecific conditioning (Powell, McLaughlin, and Chachick, 2000).
Pathways Transmitting CS Information
Investigations of HR conditioning generally have used auditory CSs that precede an electric-shock UCS. Research by Joe LeDoux, using an auditory CS in the rat during Pavlovian aversive learning, has demonstrated that destruction of the inferior colliculus (a structure relaying auditory information from the most peripheral structures of the auditory system) blocks the expression of several nonspecific learned responses (LeDoux, Sakaguchi, and Reis, 1984). The inferior colliculus projects to the medial geniculate nucleus, and lesions of its magnocellular component (MGm) produce deficits in the acquisition of the learned bradycardiac responses (i.e., HR decreases) in the rabbit (McCabe, McEchron, Green, and Schneiderman, 1993). Destruction of the auditory cortex, which receives projections from the MGm, does not interfere with the initial acqusition of this learned response (Teich et al., 1988), suggesting that the auditory cortex is not necessary for learning the relationship between the CS and UCS. Thus, structures other than auditory cortex that receive MGm projections may be essential components of the circuit.
Central Structures Involved in Learned Bradycardia
There is substantial evidence that three separate but partially overlapping higher-level CNS structures mediate Pavlovian heart-rate conditioning; these include several subnuclei of the amygdala, the medial prefrontal cortex, and the cerebellar vermus. Interactions between these three structures are illustrated in Figure 1.
The work of David Cohen and associates (1984) on the pigeon, Bruce Kapp and colleagues on the rabbit (Kapp et al., 1991), and LeDoux and colleagues on the rat (LeDoux, 2000) have demonstrated a thalamic-amygdaloid circuit that is necessary for the elaboration of conditioned cardiovascular changes. These studies have shown that one recipient of projections from the MGm is the lateral nucleus of the amygdala (LeDoux, Farb, and Ruggiero, 1990), which projects to the amygdala central nucleus (Ace) and the baso-lateral (BL) nucleus. Cells in the somatic thalamus also project to the lateral nuclei. Since the UCS is electric shock, such inputs to the amygdala may well reflect UCS information; there is some evidence that the critical association between the CS and UCS takes place in this nucleus and is relayed to the ACe and BL nucleus.
The contribution of the ACe to learned bradycardia in the rabbit has been extensively investigated (Kapp et al., 1991). The ACe projects directly to the region of cardiodecelerative motor neurons within the medulla, and electrical stimulation of the ACe elicits vagal bradycardia and excitation of these neurons. Hence, the ACe is located between the CS pathway and the motor neurons that produce the response. Interference with its normal functioning markedly attenuates the learned response. Further, the activity of its neurons in response to the CS changes over repeated CS-UCS pairing; and, for some neurons, the greater the neuron's excitatory response to the CS, the greater the conditioned bradycardiac response. These combined observations suggest the ACe excites motor neurons leading to response expression.
While the ACe may be a critical structure within the circuit, as noted above, other structures, including the cerebellar vermis and medial prefrontal cortex, also appear to be important (Supple and Kapp, 1989; Powell, McLaughlin, and Chachick, 2000). Lesions of either produce a marked attenuation of the response, and the activity of neurons to the CS in each region changes as a function of repeated CS-UCS pairings. As with the ACe, in each region increased neuronal responses to the CS predicted greater CR response magnitude. Both regions are anatomically associated with the amygdala: the medial prefrontal cortex via a direct pathway to the BL and ACe, and the vermis via an indirect one. It has also been shown that efferents from the prefrontal cortex project directly to the cardiac nuclei that control HR changes in the medulla (Buchanan et al., 1994). There is also considerable evidence that the hippocampus may be intimately involved in the stimulus processing associated with classical conditioning of HR responses. There is evidence that the ventral hippocampus, possibly via its efferents to the medial prefrontal cortex, plays a role in conditioning of nonspecific responses (Powell, McLaughlin, and Chachick, 2000). Information processed in the hippocampus may thus be responsible for the participation of the medial prefrontal cortex in HR conditioning.
Do these three different areas mediate nonspecific responses under different parametric and behavioral circumstances? LeDoux has pointed out that automatic subcortical processing of fearful stimuli by the amygdala offers the advantage of allowing the organism to a make a quick response to dangerous environmental stimuli (LeDoux, 2000). This subcortical circuit is thus important in the immediate response to environmental signs of danger. On the other hand, a host of studies in human subjects have revealed that an emotional component associated with prefrontal processing guides and affects complicated decision-making. The work of Damasio (1999), for example, indicates that the prefrontal cortex is important as an emotional component of memory selection, which guides response selection based upon its consequences. The role of the prefrontal cortex in conditioned HR responding may, therefore, be related to its integration with more complicated somatomotor response selection. Prefrontal projections bypass hypothalamic and brain-stem mechanisms, thus projecting directly to autonomic regulatory mechanisms in the medulla and spinal cord. Such processing by the prefrontal cortex may be related to integration of autonomic activity that is part of the emotional processing involved in normal adaptive behavior.
The role of the vermis in learned HR changes is unclear, although cerebellar control of HR changes in a number of situations is well established (Ghelarducci and Sebastiani, 1996). For example, pronounced bradycardia occurs during the diving reflex. Moreover, such learned HR changes may be related to the integration of autonomic activity with somatomotor learned responses, which are controlled by the cerebellum. These conclusions are, however, highly speculative at the present time.
So where is the occurrence, within the components of this circuit, of the convergence of CS and UCS information that is necessary for the structural changes responsible for HR learning? The research suggests that multiple components may represent sites of structural change that depend on continual cognitive and somatomotor behaviors. On the sensory side, neurons within the MGm are responsive to both the CS and UCS, and, like more central structures, neuronal activity changes as a function of repeated CS-UCS pairings (Supple and Kapp, 1989). Likewise, neurons in the amydala, cerebellar vermis, and medial prefrontal cortex are responsive to the UCS, although the pathways by which the UCS is transmitted to these structures are in need of further analysis. The evidence therefore suggests that structural changes may occur at multiple sites along this critical circuit.
Buchanan, S. L., Thompson, R. H., Maxwell, B. L., and Powell, D. A. (1994). Efferent connection of the medial prefrontal cortex in the rabbit. Experimental Brain Research 100, 469-483.
Cohen, D. H. (1984). Identification of vertebrate neurons modified during learning: An analysis of sensory pathways. In D. L. Alkon and J. Farley, eds., Primary neural substrates of learning and behavioral change. Cambridge, UK: Cambridge University Press.
Damasio, A. (1999). The feeling of what happens: Body and emotion in the making of consciousness. New York: Harvest/Harcourt.
Ghelarducci, B., and Sebastiani, L. (1996). Contribution of the cerebellar vermis to cardiovascular control. Journal of the Autonomic Nervous System 56, 149-156.
Kapp, B. S., Markgraf, C. G., Wilson, A., Pascoe, J. P., and Supple, W. F. (1991). Contributions of the amygdala and anatomically-related structures to the acquisition and expression of aversively conditioned responses. In L. Dachowski and C. F. Flaherty, eds., Current topics in animal learning: Brain, emotion and cognition. Hillsdale, NJ: Erlbaum.
LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience 23, 155-184.
LeDoux, J. E., Farb, C., and Ruggiero, D. A. (1990). Topographic organization of neurons in the acoustic thalamus that project to the amygdala. Journal of Neuroscience 10, 43-54.
LeDoux, J. E., Sakaguchi, A., and Reis, D. J. (1984). Subcortical efferent projections of the medial geniculate nucleus mediate emotional responses conditioned to acoustic stimuli. Journal of Neuroscience 4, 683-698.
McCabe, P. M., McEchron, M. D., Green, E. J., and Schneiderman, N. (1993). Electrolytic and ibotenic acid lesions of the medial subnucleus of the medial geniculate prevent the acquisition of classically conditioned heart rate to a single acoustic stimulus in rabbits. Brain Research 619, 291-298.
Powell, D. A., McLaughlin, J., and Chachich, M. (2000). Classical conditioning of autonomic and somatomotor responses and their central nervous system substrates. In J. E. Steinmetz and D. S. Woodruff-Pak, eds., Eyeblink classical conditioning, Vol. 2: Animal models. Boston: Kluwer.
Supple, W. F., Jr., and Kapp, B. S. (1989). Response characteristics of neurons in the medial component of the medial geniculate nucleus during Pavlovian differential fear conditioning in rabbits. Behavioral Neuroscience 103, 1,276-1,286.
Teich, A. H., McCabe, P. M., Gentile, C. G., Jarrell, T. W., Winters, R. W., Liskowsky, D. R., and Schneiderman, N. (1988). Role of auditory cortex in the acquisition of differential heart rate conditioning. Physiology and Behavior 44, 405-412.
Revised byDonald A.Powell