Is There a Relationship Between Learning and Neurochemical Changes in Rodents?


“Memory refers to an organized collection of representations of events and of relationships between events. The formation of sensory or mnemonic representations is the result of a process of internalization of the properties of the world…” (Doyère et al. 1993). But just how are these representations encoded in the brain on a chemical level? This is the question that I asked when I began researching. But there is a problem with this research question – how does one study a memory? Memories are very complex and multifarious things; it would be impossible to isolate one type of memory to study. It would also be impossible to control how an event is represented as a memory and even whether or not a memory has been formed. Therefore, I modified my research question slightly to a form that is more realistically studied: “What is the relationship between learning and changes in neurochemistry in rodents?”

This topic is of particular interest to me and to the field of psychology because the knowledge of how learning takes place in rodents would help give researchers a starting point in the quest to discover how higher order thinking is represented on a neurochemical level in humans and our closer relatives. Learning and the formation of memories are very basic – and essential – functions of the brain. Imagine how useless the brain would be if it could never retain anything that one has experienced. Incoming stimuli would be meaningless because they couldn’t be related to previous stimuli. But what kind of changes lead to learning and the formation of memories? This article will address part of this question: whether or not these changes are neurochemical changes – either partially or entirely.

In the studies that I found, learning refers to classical conditioning or active avoidance learning. Classical conditioning is the learned association between two paired stimuli – for instance a tone and a footshock – in which the unconditioned response (UCR) to the unconditioned stimulus (UCS) will be elicited by a previously neutral stimulus (NS) after the conditioning (Tocco et al. 1992). A slight variation to classical conditioning that is used in one study is called differential classical conditioning. In this case, the subject is randomly presented two stimuli – a reinforced stimulus (CS+) and a nonreinforced stimulus (CS-). In the case of the tone and footshock, two tones of different frequencies would be randomly presented, but the footshock would only follow one of the two. The subject learns to discriminate between the two stimuli as only one of them predicts the footshock (Edeline et al. 1990). Active avoidance learning refers to the learning to avoid a punishment that is consistently presented a few seconds after a stimulus. This type of learning involves more participation on the part of the subject than classical conditioning because now, instead of the response being unconditioned, the subject must discover how to avoid the punishment and actively perform this behavior when the stimulus is presented (Pöǧün et al. 1992).

Pöǧün et al. 1992

This study was conducted to investigate how the level of D2 receptor binding changes in a rat’s brain following active avoidance learning. Seven areas of the brain were studied for changes in binding: the cerebellum, hippocampus, corpus striatum, frontal lobe, occipital lobe, parietal lobe, and temporal lobe. Active avoidance learning was used on the experimental group with a tone preceding a footshock and a vertical pole as an area safe from the shock. The control group received identical learning trials except that the pole was removed. This corrected for any changes that were due to classical conditioning and not active avoidance learning. Chemical tests were used at the termination of the experiment to determine the level of D2 receptor binding in different parts of the brain in the two groups. The results of this experiment showed an increase in D2 receptor binding in the experimental group over the control group only in the hippocampus and the corpus striatum.

In this study, only the D2 receptor binding was studied. So although there appears to be a causal relationship between learning and D2 receptor binding, this knowledge may be useless. One reason for this is that the level of D2 binding and the availability of dopamine are inversely related (Pöǧün et al. 1992). Therefore it may be that the availability of dopamine is affected by learning and the level of dopamine affects D2 receptor binding. In this case, variations in D2 receptor binding would be just a “byproduct” of the changes that take place during learning. A flaw that I found with this study deals with the experimental procedure. Although the rats were divided up randomly into two groups, the groups were not treated equally. The rats that were to undergo active avoidance learning were kept in an enriched environment for 100 days prior to the learning trials whereas the control group was kept in a standard environment for the 100 days prior to the trials (Pöǧün et al. 1992). This created a preexisting difference between the groups – a second variable that isn’t dealt with. Because of this confound, the results of this study cannot show a causal relationship between learning and D2 receptor binding.

Tocco et al. 1992

This research studies the regional changes in binding of the NMDA and AMPA receptors following nictitating membrane (NM) classical conditioning. The rabbits were divided up into 3 groups: rabbits in group I were conditioned classically with the CS being a tone, the NS being an air puff and the UCR being NM and eyelid response; those in group II received pseudoconditioning where their conditioning trials were identical to the rabbits in group I except that the NS and UCS were unpaired; and the rabbits in group III received no handling. Tritiated AMPA, CNQX and TCP were used to study binding through autoradiography (locating a radiolabel within a solid specimen by exposure to a layer of detector material) with a tritium-sensitive film. The results of this study showed no difference between groups II and III, illustrating that any experiment-related stress did not affect the results. In group I, there was an increase in AMPA binding in the hippocampus and a decrease in CNQX binding in the hippocampus after classical conditioning. There was no difference between any of the three groups in TCP binding nor any significant difference in binding of the three chemicals in any other areas of the brain.

In analyzing this study, it is important to understand the chemicals involved and their roles. NMDA and AMPA are two subtypes of glutamate receptors. TCP is a ligand that binds specifically to the NMDA receptors. AMPA is an agonist for the AMPA receptors whereas CNQX is an antagonist for the AMPA receptors. With this knowledge, this study clearly shows that classical conditioning leads to an increase in the binding of the AMPA receptors in the hippocampus and no modification of the binding of the NMDA receptors. The fact that the agonist binding is increased while the antagonist binding is decreased suggests a modification of the configuration of the AMPA receptors during classical conditioning – not a change in the number of binding sites. This is because a change in the number of binding sites would push both AMPA and CNQX binding in the same direction while maintaining the same ratio between them – not in opposite directions like the results of this study showed (Tocco et al., 1992). This knowledge that learning involves a change in configuration and not just a change in number of receptors suggests possible future experiments: What is the spatial change in configuration (as opposed to the quantitative one)? Are these changes consistent over time, and if not, how do they change with time? Will extinction of the classical conditioning reverse these changes?

Edeline et al. 1990

The purpose of these investigations was to test the retention of learning induced changes formed during classical conditioning. Differential classical conditioning sessions were held for two groups with the CS+ being a 1 kHz tone and the CS- being a 7kHz tone for one group. The second group received the same two tones, but with the opposite frequency corresponding to the CS+ and the CS-. The UCS was a footshock for both groups and the UCR was any neurochemical responses. Extinction sessions were held forty-five days after discriminatory classical conditioning. Responses in the hippocampus, medial geniculate and auditory cortex were recorded throughout the habituation, conditioning and extinction sessions. On the first day of conditioning, the hippocampus showed no difference in response to the CS+ versus the CS-. The response to the CS+ progressively increased above that for the CS- throughout the twelve days of conditioning. On the first day of extinction, the CS+ response in the hippocampus was considerably larger than the almost imperceptible CS- response, and by the second day of extinction, there was absolutely no response to the CS-. These same responses were also shown in the auditory cortex. In the magnocellular nucleus of the medial geniculate body (MGm), the CS+ was much greater than the CS- response from the first day, and neither one changed throughout the conditioning trials. On the first and second days of the extinction, the CS+ response was less than during conditioning, but present, whereas the CS- response was nonexistent.

In these experiments, the MGm showed markedly faster effects than the other two areas studied, but it showed no increase in response after more conditioning trials like the other two areas, and it also showed a decrease in response from the last day of conditioning until the first day of extinction (Edeline et al., 1990). This shows the MGm’s ability to quickly distinguish between two stimuli, but that its discriminatory response is also the first one to erode with time. These properties may lead to evidence that the MGm is responsible for certain types of learning that are initiated very quickly such as the development of phobias. This study suggests that the changes caused by learning are not associated with one particular structure that is designed for long-term storage of information – but rather with many of the structures of the brain where changes can be detected during learning. This is a valuable piece of knowledge in understanding the workings of the brain – memory is not centralized like in a computer, but rather it is localized in multiple areas of the brain.

Hennevin et al. 1993

The purpose of this study was to determine whether or not neurons in the medial geniculate nucleus (MG) respond to a conditioned tone during paradoxical sleep. Rats were divided up into two groups – a conditioned group and a pseudoconditioned group. The conditioned group received classical conditioning in which the UCS was a footshock, the NS was a tone and the UCR was the response of the MG neurons. The rats were then presented the same tones during paradoxical sleep. The results of this study show that during conditioning the level of evoked discharges in the MG increased above the baseline, whereas there was no change in the pseudoconditioned group.

This study showed the responsiveness of the MG during classical conditioning by studying the evoked discharges that take place during classical conditioning. This confirms some of the findings that Edeline et al. (1990) made regarding the changes that take place in the MG during classical conditioning.

Weinberger et al. 1995

This research studied whether facilitation of the magnocellular nucleus of the medial geniculate body (MGm) affects the receptive field plasticity in the auditory cortex. The experiment paired electrical stimulation of the MGm – or Sham stimulation in the case of the control group – with an auditory stimulus and recorded changes in the contralateral auditory cortex. In this case the UCS was the stimulation, the NS was the tone and the UCR was the plasticity of the primary auditory cortex. Classical conditioning was used in this experiment because it has previously been shown to produce a highly specific receptive field plasticity in the primary auditory cortex. The research showed that all subjects who received MGm stimulation had long-term facilitation of click-evoked potentials (EPs). Facilitation occurred as early as 1.3 minutes into the two hour session – but more often at 2.6 minutes – and continued to increase throughout the entire session. The control group developed no facilitation of click EPs. In fact the control group developed a reduction in click EPs throughout the two hours.

In looking at these results, one immediately notices the reduction in click EPs in the control group. The researchers used this to improve the correlation that they found. If the control group is considered the baseline from which all measurements are made, then the MGm stimulation group appears to have a decreasing baseline. When this is corrected for, the correlation is even better (Weinberger et al., 1995). It is important to raise the point made in this article that the MGm facilitation may not have directly facilitated the click-evoked potentials. It is very possible that the MGm affected other nuclei in the thalamus that actually caused the facilitation. Another explanation is that the MGm played no part in facilitation whatsoever, that the current actually spread to the nearby ventral nucleus of the medial geniculate body (MGv) – which also projects directly to the primary auditory cortex – and it is the MGv that actually caused the click EP facilitation (Weinberger et al.). A possible course of further study to address some of these issues would be to study the effects of MGm stimulation on the MGv and other surrounding nuclei.


From these studies, one can learn a lot about the relationship between learning and neurochemical changes. For one, these studies illustrate that the areas in the brain that show the highest level of responses during learning are the medial geniculate body (Edeline et al. 1990m Hennevin et al. 1993), the hippocampus (Edeline et al. 1990, Pöǧün et al. 1992, Tocco et al. 1992) and the corpus striatum (Edeline et al. 1990, Pöǧün et al. 1992). These studies also show that the D2 and AMPA receptors in the hippocampus and the D2 receptors in the corpus striatum play a major role in these changes (Pöǧün et al. 1992, Tocco et al. 1992). Although these studies only show a quantitative change in binding and click-evoked potentials, they do show that, in the case of the AMPA receptors, this change is due to a modification of the configuration of the receptors and not a change in the overall number (Tocco et al. 1992). These studies do not show, however, just how these changes take place or what types of configuration changes they are. Another piece of information that comes out of these studies is the fact that learning takes place in multiple areas – that information can be stored in the same place that it is detected and/or needed (Edeline et al. 1990). This is very useful when thinking about how learning is represented in the brain and when designing future experiments to further explore learning.

Based on these studies, I can confidently answer my research question that yes, there is a relationship between learning and neurochemical changes in rodents. The only qualification of this answer that I need to make is that these studies cannot answer just what types of changes take place or how these changes occur. This is because of the complexity of the brain and the chemicals it contains. Changing one chemical in one part of the brain may affect other chemicals that have not been studied yet in parts of the brain that are spatially unrelated. The complexity of a system where any chemical change can lead to countless other ones makes it very difficult to draw any conclusions from this type of research. As in many of these studies, one can safely say that learning leads to a certain change in receptor binding or neurotransmitter availability, but that may simply be the result of some other change in chemical availability. In the research of Weinberger et al. (1995), the MGm is directly stimulated and the auditory cortex plasticity is directly measured, but the actual mechanism for this change in plasticity may be a result of countless other thalamic nuclei that cannot be disregarded. Because of the incredible complexity of this system, one can safely say that learning causes a neurochemical change, but as far as the actual mechanism is concerned, there is – as of yet – no way to tell.


Doyère V, Burette F, Negro CR, Laroche S. (1993). Long-Term Potentiation of Hippocampal Afferents and Efferents to Prefrontal Cortex: Implications for Associative Learning. Neuropsychologia, 31, 1031-1053.

Tocco G, Annala AJ, Baudry M, Thompson RF. (1992). Learning of a Hippocampal-Dependent Conditioning Task Changes the Binding Properties of AMPA Receptors in Rabbit Hippocampus. Behavioral and Neural Biology, 58, 222-231.

Edeline JM, Neuenschwander-El Massioui N, Dutrieux G. (1990). Discriminative Long-Term Retention of Rapidly Induced Multiunit Changes in the Hippocampus, Medial Geniculate and Auditory Cortex. Behavioral Brain Research, 39, 145-155.

Hennevin E, Maho C, Hars B, Dutrieux G. (1993). Learning-Induced Plasticity in the Medial Geniculate Nucleus Is Expressed During Paradoxical Sleep. Behavioral Neuroscience, 6, 1018-1030.

Pöǧün, S., Kanit, L., & Okur, B. E. (1992). Learning-Induced Changes in D2 Receptors of Rat Brain Are Sexually Dimorphic. Pharmacology Biochemistry and Behavior, 43, 71-75.

Weinberger NM, Javid R, Lepan B. (1995). Heterosynaptic Long-Term Facilitation of Sensory-Evoked Responses in the Auditory Cortex by Stimulation of the Magnocellular Medial Geniculate Body in Guinea Pigs. Behavioral Neuroscience, 1, 10-17.