DEFINICIÓN

Consolidation is the process by which already formed associations move from a state of lability to one of stability. Müller and Pilzecker (Müller, 1900) were the first to formalise a description of this process. They proposed the existence of continuing neural events that are required for the consolidation of a memory trace of recently acquired material. They hypothesised that such events could be prevented from occurring by external interference.

Subsequently, experiments using electro-convulsive shock (ECS) revealed that processes occurring after the presentation of stimuli were indeed essential for the formation of a long-lasting memory. The first of these used human subjects

undergoing electroshock therapy (Zubin, 1941). Lists of words were presented to subjects on the day before ECS application and on the morning of the actual day. Subjects were then tested on recall after the ECS and it was found that those words on the list presented in the morning before could not be recalled. Lists presented the previous evening to subjects receiving ECS and lists presented in the morning to control subjects who did not receive the ECS were easily recalled. This dissociation of effect strongly suggests that the formation of memories for the morning list was

interfered with by the ECS immediately after learning. It also demonstrates that such memories are not so fragile if they have been given time to consolidate.

This phenomenon has been re-observed on numerous occasions since in animal subjects learning much simpler tasks. For instance, rats were found to be unable to leam an active avoidance task, in which a light CS informed them of impending footshock, if ECS was delivered directly after the task (Duncan, 1949). Many other learning tasks have been used to successfully replicate these findings (Leukel, 1957) (Thompson, 1957).

It has been argued that these experiments may not prevent consolidation but rather that they demonstrate the aversive qualities of the ECS (Miller, 1955). If presented after successful active avoidance, the ECS could act as an even greater negative stimulus than the footshock, and effectively punish the successful formation of an association. To test this possibility subjects were trained to associate a mouthshock with approach to food. The ECS was delivered at varying intervals after this. If ECS prevents consolidation then the subjects should have been unable to learn not to approach the food. However, it was found that the ECS compounded the negative reinforcement of the mouthshock and the subjects would not approach the food at all (Miller, 1955). It is therefore a possibility that the ECS experiments do not provide evidence for consolidation. These contrary findings are tempered somewhat by the fact that human subjects do not report any aversive qualities to the ECS when it is delivered properly (Stainbrook, 1948). However, we could attribute this declaration to the fact that humans expect the delivery of ECS, whereas animal subjects are not aware of the impending shock.

Nonetheless, the concept of consolidation has persevered and propagated, and

memory has subsequently been found to be sensitive to other post-training insults than just ECS. These include anoxia (Hayes, 1953), anaesthesia (Leukel, 1957),

temperature changes (Ransmeier, 1954), direct brain stimulation (Mahut, 1958) (Glickman, 1958) (Thompson, 1958) and perhaps most significantly protein synthesis inhibition (Agranoff, 1964) (Flexner et al., 1963a). It could be argued that all of these serve as potentially aversive stimuli in their own right. However, there are some non- experimental pieces of evidence to suggest that consolidation is a real and common process in memory.

On the phenomenon of retroactive amnesia

Damage to the temporal lobes in humans causes retrograde amnesia (Squire, 1981). This amnesia is graded such that more recent episodic memories are completely lost while remote memories of events long since past are more likely to be retained. This observation is considered to be evidence for processes of consolidation. The temporal lobes house a phylogenetically old component of the forebrain cortex that was

mentioned at the beginning of this introduction; the hippocampus. This has long been suspected to act as a kind of episodic memory module (O'Keefe and Nadel, 1978) (Tulving, 1983). Other surrounding areas such as the entorhinal cortex are additional candidate sites for memory formation and storage.

The graded effect of damage to these areas in retrograde amnesia suggests that remote episodic memories are at least partially stored within other regions. The reduced vulnerability of remote memories to temporal lobe damage suggests that one form of memory consolidation requires the movement of stored information from temporal lobe structures to other regions of the brain over time (McClelland et al., 1995). This theory raises a contentious issue in memory research. How can associations be shifted from one population of cells to another without losing their meaning and efficacy relative to the outside world? The process is a mystery and it has received significant discussion over the course of the previous decade (Nadel and Moscovitch, 1997) (Alvarez and Squire, 1994). Nevertheless, the reduced vulnerability of remote memories provides a strong indication that memories change over time, and ties in well with Müller and Pilzecker’s notion of consolidation.

On retroactive inhibition and interference

Müller and Pilzecker’s experiments at the end of the nineteenth century focused on human subjects and their ability to memorise lists of nonsense syllables. They formulated the theory of memory consolidation on the basis of the observation that memorising one list of syllables could be impaired by presentation o f a second list of syllables almost immediately afterwards. Similarly, presentation of pictures after a list of syllables also reduced the number of syllables that could be recalled in a later test. This impingement on memory processes by subsequent competing presentations is described as retroactive inhibition.

Retroactive inhibition was initially taken as evidence that activity within the memory system persisted after initial learning and that this was required for permanent

memory formation. Müller and Pilzecker described such activity with the German word ‘perseveration’. This label perhaps corresponds to what Pavlov described as a ‘trace’ (Pavlov, 1927) or what Hebb described as ‘reverberatory activity’ (Hebb, 1949). In the retroactive inhibition of an information set, the processing of a novel but related information set might infiltrate the perseverative trace because it relies upon some of the same circuitry as the first set. This interpretation is corroborated by the finding that the closer qualitatively the second set of information is to the first then the greater the impingement upon the memory of the first.

An alternative but related explanation of these findings is interference theory. Interference theory posits that the second information set competes for storage sites, perhaps at the level of individual synapses, with the first. This issue is now confused. Modern descriptions of consolidation are unable to distinguish fully between

perseverative inhibition and storage interference. We cannot dissect localised consolidation from system level consolidation even though the term ‘consolidation’ almost certainly refers to two separate processes. One o f these processes takes only a few hours, is sensitive to various pharmacological manipulations, and relates to changes in the quality of memory, and another alters the quantity or distribution of memories over the course of months and years. It is interference with this latter

consolidation process that accounts for the gradation seen in retrograde amnesia. Retroactive inhibition is likely to occur due to interference with the former process.

Motor learning requires consolidation

Most pertinently to the studies detailed in this thesis, motor learning has been shown to be subject to retroactive interference. In a key study human subjects were trained to use a double-hinged manipulandum to direct a cursor across a monitor to a designated target (Brashers-Krug et al., 1996). This task was found to observe a similar

interference profile to that seen in the early studies of Müller and Pilzecker. In the control case, subjects had difficulty in overcoming the perturbation imposed by the manipulandum, but they eventually accomplished the task reliably and were able to carry it out effectively twenty-four hours later without much further learning.

Experimental subjects also learned this first task but then received further training on a similar but slightly different task.

This second task involved all the same components as the first but required an opposing trajectory for the cursor. Learning the second task soon after the first prevented subsequent recall of memory of the first. It appears that learning a second competing motor task retroactively interferes with consolidation of the first. An unusual and counter-intuitive finding that accompanied this was that learning of the second task seemed to be interfered with by the first task. This phenomenon is converse to retroactive inhibition and is known as negative transfer.

Consolidation is aided by intervals in training

It is also a common anecdote of those that work on animal learning in a variety of species that memory will suddenly expose itself after an interval in training.

J.Z. Young observed this phenomenon in his octopus subjects as they learned visual discriminations (Young, 1965). His suggestion at the time was that this effect is due to an initial suppression of an already learned response during the training that is then expressed in a subsequent training session. Most people now attribute the apparent benefit of a fallow period in training to a consolidation effect. Some theories have tied this consolidation effect to sleep (Buzsaki, 1998). Indeed, learning of visual

discrimination tasks in humans has been found to be impaired by preventing REM sleep (Kami et a l, 1994) (Stickgold et a l, 2000) and there is some indication that motor learning may be aided by sleep (Hobson et a l, 1998).

Conditioned eyeblinks require consolidation

It has been shown with eyeblink conditioning that spacing of training over days greatly reduces the number of trials required to acquire CRs (Kehoe and Gormezano, 1974). Delay pairings of a tone CS with a periocular shock US were delivered to four different groups of rabbits in daily sessions consisting of one, five, ten or fifty trials. The group receiving one trial per day were found to perform a first CR within twenty trials on average and reached maximal performance, known as asymptote, within forty trials. In contrast, those subjects receiving fifty trials per day had not shown CRs after this many trials. Of course, it is worth noting that the one trial-a-day group took many more days to acquire CRs than the fifty trial-a-day group, but the experiment nonetheless demonstrates the efficacy of spaced training compared with massed training.

It is not clear whether the animals receiving spaced training are benefiting from sleep in between each session and the next or if the trials within each session for those receiving massed training are causing retroactive inhibition. These alternatives could potentially be tested if a time-window for conditioned eyeblink consolidation could be identified. Trials could then be spaced across an appropriate period of time during a single day to allow consolidation to occur. Comparison of these animals with those receiving a single trial-a-day would reveal whether sleep is playing an important role.

Consolidation of conditioned eyeblinks has been investigated using systemic drug infusions after training (Scavio et a l, 1992). Infusions of a variety of dmgs before and after NMR conditioning in rabbits have been found to affect the rate of CR

acquisition. Scopolamine, amphetamine and chlorpromazine retard the rate of conditioning with respect to control subjects. These substances variously interfere with neurotransmitter systems in the brain that have been suggested to act in a modulatory fashion during memory formation. Scopolamine interferes with the cholinergic system by antagonising muscarinic acetylcholine receptors.

Amphetamines prevent the re-uptake pre-synaptically of the catecholamines, dopamine and noradrenaline. Chlorpromazine blocks the action of these

catecholamines post-synaptically, as well as other monoamines such as serotonin (5- HT) (Rang et ah, 1999). The result of amphetamine action is an increase in the length of time that these catecholamines remain in the synaptic cleft and should, in theory, facilitate their action in the short-term while chlorpromazine should reduce their action by blocking post-synaptically. In theory we might expect chlorpromazine to have a contrasting effect to amphetamine.

Amphetamine has been shown to facilitate eyeblink conditioning (Gormezano and Harvey, 1980) and other forms of animal learning (Evangelista and Izquierdo, 1971). However, amphetamine, along with chlorpromazine and scopolamine is found to retard NMR conditioning when infused after training. It seems that amphetamine only has a facilitatory effect when delivered before training. Amphetamine may alter the properties of the conditioning stimuli such that the encoding process rather than the memory formation itself is aided. In contrast, post-training infusion of the NMDA receptor blocker ketamine, was found to significantly accelerate both acquisition and extinction learning in NMR conditioning (Scavio et al., 1992). This latter finding suggests a genuine role for NMDA receptors in the consolidation o f NMR conditioning. This receptor type has recently been implicated in ‘perseverative’ consolidation processes in the hippocampal-prefrontal cortex system (Steele and Morris, 1999) (Shimizu et al., 2000) (Day et al., 2001). However, adult Purkinje cells do not express NMDA receptors, so the enhancing effect of post-training ketamine on NMR conditioning provides no support for the notion that information storage in eyeblink conditioning occurs at parallel fibre-Purkinje cells synapses.

On the experiments described within this thesis

My experiments, as detailed in this thesis, target the consolidation processes involved in eyeblink conditioning. They accomplish this through localised infusions of a pharmacological reversible inactivation agent, the GABAa receptor agonist

muscimol, into previously delineated components of eyeblink circuitry within HVI of the cerebellar cortex and the AIP ipsilateral to the trained eye. These regions are considered to be components of a functional loop within the eyeblink circuitry. To

overcome the problems associated with analysis of ‘on-line’ inactivations of this loop during NMR conditioning I have used post-training infusions in my experiments. This approach restricts any effect on the system to processes involved in consolidation. If consolidation of NMR conditioning relies solely on ‘perseverative’ processes then post-training reversible inactivations will have the same effect on the system as pre­ training infusions in that secondary and perhaps tertiary effects elsewhere in the essential circuitry will accompany primary effects at the site of infusion. The loop­ like nature of the circuitry would render analysis of the location of memory storage impossible. However, if consolidation has some component that relates to changes in the quality of a memory at the site of storage, then muscimol may affect these changes and cause storage interference as opposed to perseverative interference. Post-training muscimol infusions could thereby serve to dissect out a site of memory storage within the loop.

In the first instance this approach reveals that there is a dissociation in the effect that muscimol has on consolidation in the cortex and in the nuclei, suggesting either storage interference or only very localised perseverative interference. This finding greatly enhances our knowledge of the location of memory formation in the

cerebellum and goes some way towards identifying an engram for associative learning in the mammalian CNS. Secondly, I have been able to identify a time window for the consolidation of NMR conditioning using a similar approach. My final experiments accompany a more detailed consideration of the mechanisms that may be involved in this consolidation process and can be found in the final chapter of this thesis.