Introduction
The findings presented in Chapter 2 suggest that further investigation of consolidation in NMR conditioning should focus on the cerebellar cortex. In order to characterise the consolidation processes that may occur at this site it is first necessary to identify the period over which they occur. The mechanism by which memories are made stable will only become clear if observations are made both in the right place and at the right time. The experiments detailed in Chapter 2 demonstrate that at least some component of memory pertaining to NMR conditioning is stored in the cerebellar cortex. Rostro-medial HVI presents itself as a strong candidate site for this memory formation because previously conditioned responses are temporarily abolished by its inactivation (Attwell et al., 1999), as are both acquistion (Attwell et al., 2001) and consolidation of these responses (Chapter 2). There is as yet, however, no direct dissection of the progression of this memory formation over time.
The study presented in this chapter took a similar approach to that presented in the previous chapter. Muscimol was infused into lobule HVI after NMR conditioning and the effects on acquisition rate were taken as an indication o f consolidation
impairment. The study differed in that delays were imposed on muscimol infusions so as to create time windows in which the consolidation might be allowed to occur. It was hoped that this approach would reveal a crucial period during which cerebellar cortical-dependent information storage occurs.
There is a significant body of evidence to suggest that consolidation processes are time-dependent. The time-course of consolidation was first considered to be an important characteristic of memory formation when the phenomenon o f retroactive interference was recognised by Müller and Pilzecker ((Müller, 1900) and see
(Glickman, 1961) or (Lechner et al., 1999) for review). Their original experiments on consolidation made use of a variation on Ebbinghaus’ (Ebbinghaus, 1885) lists of nonsense syllables. Human subjects were asked to memorise lists of syllable pairs over a fixed number of presentations. Subjects were later asked to recall one half of a
syllable pair when cued by the other. The number of errors made in recall was used to quantify memory formation. Müller and Pilzecker found that consolidation was impaired by presentation of a competing syllable list almost immediately after presentation of the first. They described this impairment as retroactive interference. Delaying the presentation of the second competing list by six minutes spared memory of the first. The concept of a perseverating memory trace arose from this observation.
More recently, the time-course of consolidation in motor learning has been
investigated in a similar psychophysical fashion. A time window for motor memory consolidation has accordingly been identified. This time-window is, however, very different in length to that found by Müller and Pilzecker for their nonsense syllables. Human subjects were trained on a task in which they used a multi-joint
manipulandum to move a cursor across a screen. Two variations on this task involved opposing relationships between the joystick movement and the consequent cursor movement (Brashers-Krug et al., 1996). Presentation of one task soon after the other led to a failure in memory consolidation. Various experimental groups received training on the second task at different time-points after the first. Retroactive interference was imposed on the first task to a greater or lesser extent depending on when the training for the second occurred. If the training on the second task
immediately followed the first then subjects did not retain a motor memory of the first task at all and had to relearn it completely twenty-four hours later. Leaving a one-hour delay between the two tasks reduced retroactive interference to some extent and increased memory retention on the twenty-four hour test. Imposition of a four-hour delay reduced the retroactive effect even more.
These findings are fairly consistent with the literature on consolidation in animals. The stability of various forms of animal memory has been tested using a variety of different insults. Electroconvulsive shock (ECS) has been found to have a more or less deleterious effect on inhibitory avoidance retention when delivered at different points after training. ECS seems to cause retrograde amnesia for inhibitory avoidance if delivered soon after training but not if delivered an hour later (Alpem and
The application of protein synthesis inhibitors (PSIs) corroborates this finding in a variety of learning paradigms and species. Application of PSIs just prior to training, or just after training, prevents memory recall any time from an hour after training onwards (see (Davis and Squire, 1984) for review). Careful PSI application has no ostensible effect on learning or memory recall within this first hour. A phase of consolidation must therefore occur within the first hour after training that uses newly synthesised proteins to perpetuate the changes that occurred during learning.
Application of PSIs after an hour to an hour and a half has been shown to have no effect on long-term memory formation in most systems. A second time window of memory sensitivity to PSIs, however, has been identified in chicks learning a passive avoidance problem. A long-lasting association between a coloured bead and the bitter- tasting substance coating it fails to form if a PSI is injected either between half an hour and an hour and a half, or between four and five hours after training. Intervening injections have no such effect (Freeman et al., 1995). Although this experiment does not focus on a mammalian system, it is one of the best demonstrations of the time- dependency o f memory consolidation. As with the human literature we can see that there is a great variation in the duration and spacing of consolidation time-windows from species to species and task to task. More complex forms of memory, such as visual discrimination learning, seem to be susceptible to pharmacological treatments after intervals of up to three hours after training (Krivanek and McGaugh, 1969).
The temporal properties of consolidation in animal motor memory and, in particular, NMR conditioning have also been investigated in this regard. Systemic infusions after training of amphetamine, chlorpromazine and scopolamine affect consolidation of NMR conditioning in a time-dependent fashion (Scavio et al., 1992). All of these substances manipulate neurotransmitter systems that are thought to act in a broad modulatory fashion. These neurotransmitters, dopamine, noradrenaline, serotonin and acetylcholine have been accorded a variety of roles in learning. Noradrenaline, for example, has been posited to act as a reinforcing signal in a model of motor learning in the cerebellum (Gilbert, 1974). Amphetamine, chlorpromazine and scopolamine all retard NMR conditioning relative to controls when infused immediately after training These various substances do not exert an effect on learning if delivered two hours
after training. A time window during which the consolidation of eyeblink conditioning is sensitive to them should, therefore, close two hours after training.
During this same period ketamine, an NMDA receptor blocker, facilitates the consolidation of NMR conditioning. Ketamine also does not have this effect two hours after training (Scavio et al., 1992). Memory has also been shown to be open to facilitation after learning by a variety of substances. Strychnine is a drug that has traditionally been used to enhance learning. When injected at low doses prior to training (McGaugh and Petrinovich, 1965), or directly after training (Krivanek and Hunt, 1967) this glycine receptor antagonist has been shown to enhance various forms of learning. Infusion of the same drug half an hour or longer after learning has no such effect, demonstrating again a progression of memory in its sensitivity over time.
Drugs that act on the GABAergic system such as picrotoxin, bicucilline or flumazenil have been shown to enhance memory consolidation. Picrotoxin and bicucilline are both G A B Aareceptor antagonists, and they act in different ways on the G A B Aa
receptor but both enhance learning in a variety of paradigms (Castellano and McGaugh, 1989) (McGaugh et al., 1990) (Brioni et al., 1989). Flumazenil is an antagonist of benzodiazepines. These substances normally enhance GABA binding to the G A B Aareceptor by binding at their own specialised site (see (Izquierdo and
Medina, 1991)). Systemic flumazenil is found to enhance both non-associative and associative forms of learning when infused prior to training (Lai et al., 1988) (Pereira et al., 1989) (Izquierdo et al., 1990).
These findings, coupled with the fact that benzodiazepines themselves (Thiebot, 1985), and GABAa agonists such as muscimol (Brioni et al., 1990) (Wilensky et al., 1999) (Zanatta et al., 1997) impair learning, strongly suggests a down-regulatory role for the GABAergic system in memory formation. There is currently no indication that the GABAergic system plays a time-dependent role in memory storage. The results presented in the previous chapter are consistent with the view that the GABAergic system has a down-regulatory role in cerebellar cortical processes, although
alternative conclusions can be drawn. The results presented in this chapter manipulate GABAergic transmission in the cerebellar cortex in order to identify time-dependent consolidation processes.
M ethods
For methods see previous chapter (Chapter 2 - location). All surgery, behaviour, histology and autoradiography were conducted in the same fashion. Reversible inactivations were similar but delivered at a different time after each training session. Statistical analysis was also similar for phase comparisons. One comparison was made for a single session on transformed data. These square root values were normally distributed and allowed for the application of a parametric one-way
ANOVA. This consisted of an all pair-wise comparison in which every possible pair of groups was compared. A Tukey post-hoc test was applied to determine which of these comparisons revealed significant differences.
Experimental design
The design of this experiment is very similar to that presented in study 1 of Chapter 2. Two groups of subjects were cortically implanted with injection cannulae (see
methods). NMR conditioning protocols were also similar to those in Chapter 2. Two new experimental groups were conditioned and were compared with the acquisition rate of the original control group (CON). The data collected from the CON group, and also the CTX group, were re-used in the comparisons presented in this chapter for the sake o f minimising expense. Matched controls may be regarded as ideal in any form of experiment, but achieving a replication of the original result was not considered important enough to warrant the cost of using up to 15 more animals.
Each phase comprised 40-trial sessions of paired conditioning stimuli, interspersed with 10 unpaired CS probes presented 1 every 10 trials, each day for four days of a week. After every block of conditioning in phase 1 subjects received a localised muscimol infusion (Tnmoles) of the same volume (2p,l) as those delivered in the Chapter 2, However, the phase 1 treatments in this experiment differed in that muscimol infusions were delayed after training. One group, the 45CTX group, received muscimol 45 minutes after each training session while another, the 90CTX group, received muscimol 90 minutes after each session (fig. 3.01). Phases 2, 3 and 5