In the sea-slug Aplysia in which this has been studied, habituation is due to reduced transmitter release from the axon terminals of the sensory neurone (S) with repeated stimulation, perhaps as a result of partial inactivation of N-type Ca²⁺ channels in the presynaptic terminals. Dishabituation (sensitization) by a noxious stimulus, occurs when serotonergic endings (terminal with circle inside) located presynaptically on the sensory terminals, and acting via a cAMP-activated protein kinase (cAMP-PK), cause phosphorylation and hence inhibition of the K⁺ channels (S-type K⁺ channel). The action potential is therefore prolonged so that more Ca²⁺ can enter the terminal. At the same time, the 5-HT-receptor complex activates phospholipase C (PLC), and ultimately, protein kinase C (PKC). PKC acts with cAMP to promote mobilisation of vesicles into the pool ready for immediate release.

With repeated sensitizing stimuli, a persistent potentiation (Long Term Potentiation, LTP) can result, lasting for days. This occurs when the cAMP-activated protein kinase enters the nucleus of the sensory cell, and activates CREB (cAMP Response Element Binding protein) to trigger protein synthesis. This leads to formation of a variant of the same kinase (cAMP-PK), which is constitutively active, without the need for cAMP. The K⁺ channel will then be constantly inhibited. At the same time, genes for the synthesis of active-zone protein are promoted, causing increase of the effective synaptic area. This then is a form of long term memory, and illustrates how a short term effect can be converted eventually into a long-term change, in which protein synthesis is required for the transition.

Non-associative LTP: A sensitization-like form of Long Term Potentiation can be triggered in the CA3 and CA1 region of the hippocampus (left). In the CA1 region of the hippocampus, brief, high-frequency stimulation of many afferent axons (Shaeffer collaterals) at once, will cause enhancement of the PSPs in the pyramidal cells for weeks. Here LTP of the CA1 EPSP is elicited by two 10 sec bursts of stimuli at 100 Hz, 20 sec apart (Tetanus). The EPSP height increases because the strong, repeated stimulus opens NMDA channels, and allows massive Ca²⁺ inflow, activation of Ca²⁺-calmodulin, and the formation of various phosphorylated proteins involved in vesicular release.

Habituation and sensitization are limited forms of learning since they do not involve contingency. In classical conditioning, if presentation of an unconditioned stimulus (UCS) e.g. meat juice (producing the unconditioned response - salivation), is preceded repeatedly by a conditioned stimulus (CS) e.g. ringing of a bell (which causes no salivation), then soon, the ringing of the bell on its own will elicit salivation, which is then termed the conditioned reflex (CR). The pairing in the stated sequence is necessary for conditioning. Repeated presentation of the CS without reinforcement by pairing with the UCS (meat juice) will lead to attenuation and then extinction of the CR. To explain this type of associative conditioning, Hebb proposed that ..."If a neuron A weakly excites and effector B, but repeatedly assists in causing the discharge of B, then the strength of the excitatory effect of A on B will increase". A mechanism whereby Hebb's Postulate can be expressed is now known. Synapses which behave in this way (Hebbian synapses) have been described in the hippocampus. The hippocampus shows both non-associative and associative LTP.

Associative LTP: Tetanic stimulation of weak inputs on to the CA1 pyramidal cells results in no LTP (A). When separate weak inputs (releasing glutamate & acting via non-NMDA receptors) and strong synaptic inputs onto the same dendritic region of the pyramidal cells are simultaneously activated, the weak input will become strengthened. This (like the non-associative LTP) occurs because depolarization of the dendrite by the strong inputs, unblocks the NMDA receptors in the vicinity of the weak synapse, thereby allowing the glutamate to strongly activate these channels. The story, however, goes further.

The strong influx of Ca²⁺ into the post-synaptic terminal, causes activation of a Ca²⁺-calmodulin kinase and protein kinase C. It is believed that this triggers production of a retrograde messenger which promotes vesicular release and synaptic growth at the presynaptic terminal. The activated protein was thought to be arginase, which catalyses the production of NO, which diffused through the extracellular space to the presynaptic endings. NO would then potentiate any synapses which had recently been active (i.e. which still have elevated Ca²⁺ concentration). This is consistent with the fact that LTP here, is a presynaptic phenomenon, not involving prolonged increase in postsynaptic responsiveness. In fact, any nearby terminals that were recently active at the time of tetanic stimulation, not just those on the stimulated cell, may be potentiated. It is now thought that NO is unlikely to be the mediating agent, but that prostaglandins might play the necessary role.

Another well-studied form of associative learning is the eye-blink reflex in rabbits. A puff of air to the cornea causes nictitating membrane retraction. A tone causes no such response. Repeatedly preceding the puff by the tone, however, soon allows the tone alone to elicit the response. This rerouting of excitation triggered by the tone, involves the cerebellum, and can be abolished by lesioning areas along the portrayed pathway - including the interposed cerebellar nuclei. Note that activation of the Purkinje cells by strong climbing fibre inputs from the inferior olive, plays a key role in this form of reflexive learning.

Learning can be classified as declarative - types of learning that can be "brought to mind as an image or proposition" or procedural (reflexive) - types which are automatic, unconscious, reinforced by repetition. Some forms of declarative learning involve one-trial consolidation. The distinction between the two is important since declarative learning is impaired in amnesiacs (persons with memory loss) but procedural learning tends to remain intact. Thus, an amnesiac painstakingly taught to carry out a complex procedure, may deny being able to do it (i.e. have no declarative memory of it), but will be able to carry it out after coaxing.

Learning is not static. Memories are constantly being reorganised and regrouped. Two main stages of memory are short term memory and long term memory. Short term memory (working memory) differs from long term in being of limited capacity (but clustering of items may increase effective capacity), labile, and susceptible to disruption by factors which interrupt the electrical/synaptic activities of the brain. Hence electroconvulsive shock (used as the favoured treatment in some forms of depression) results in the loss of memory of events just preceding the treatment (a few hours, but even up to years before in some cases). Material can be maintained in the short term store for extended periods only by repetition (rehearsal). The hippocampus seems to be important in holding short term memories, and ensuring the proper indexing to permit accessibility, as they are consolidated into long term memories. Bilateral damage to the hippocampus results in anterograde amnesia in which new information cannot be acquired, but in which old memories remain accessible.

The hippocampus seems to be involved in organizing and indexing material for conversion to long term storage by a process involving protein synthesis. It seems to continuously reorganize and reinforce material in the long term store, with the effectiveness of the long term store building up over a period of several years. The long term store consolidates memories in a reorganised, rationalized form. Treatment with protein synthesis blockers up to about 1 hr after teaching animals a new task prevents retention in long term memory. After this period, detectable retention occurs. Long term memories are stored as changes in synaptic responses to stimuli and situations, in distributed overlapping, groups of neurones and pathways. There is no fixed locus for storing long term memories. Damage to certain structures however can severely impair the indexing and accessing of already stored memories (retrograde amnesia). So too can stress, distraction or simply fatigue.

Damage to the dorsomedial thalamus (frontal lobe projection) and mammillary bodies, causes anterograde amnesia, as in Korsakoff's syndrome in alcoholics with dietary problems (B1 deficiency). There is some loss of established memories as well. Destruction of the cholinergic cells of the basal forebrain (e.g. in Alzheimer's disease) affects ability to name objects and to link recognition of objects with recall of function or other associations, and ultimately inability to carry out simple procedures e.g. combing hair etc. (retrograde amnesia), and loss of cognitive function. Other cell types e.g. somatostatinergic cells in the frontal cortex, and hippocampal cells, also degenerate. Large doses of choline only help transiently. Cholinergic blockers such as scopolamine impair learning. Some agents such as caffeine and vasopressin have a potentiating effect on learning.

Damage to the prefrontal cortex leads to deficits in learning tasks that have a spatial component, and require a delay between responses. Clearly, many of the structures interconnected in the Circle of Papez (limbic system) play an important role in learning and memory. Perhaps this is because learning frequently involves some emotionally loaded component (reward, punishment). The strong involvement of olfactory stimuli in triggering old memories is a well known phenomenon, which is undoubtedly linked to the fact that olfactory inputs project directly into limbic structures (amygdala, septal nuclei etc.). Interestingly, degenerative changes are particularly marked in the olfactory system in Alzheimer's Disease.

Read Somjen, Neurophysiology - the essentials, Chapter 20.

Ganong, Review of Medical Physiology, Chapter 16.