Animal Models and Learning Paradigms
By Jing Lu


__Learning and memory__ is a crucial ability for organisms which allows them to acquire and retain salient information
Experimental mice
for survival. Animal models are non-human organisms that have been widely used to elucidate the cellular and molecular mechanisms underlying learning and memory, and have played an important role in understanding human cognition. Often, the experiments involving human subjects are confounded by ethical concerns about the possibilities of harmful effect that experimental procedures might exert. Therefore, animal models would be able to offer scientists better opportunities to extensively study the physiology of learning and memory. So far, a variety of animals, both vertebrates and invertebrates, have been used to model different behaviours. These animals display unique characteristics which make them particularly suitable for examining diverse aspects of neural phenomena. Aside from the behavioural studies, pathological conditions such as neuropsychiatric disorders, neurodevelopmental disorders, stroke, and epilepsy can be modeled as well (1). Although it is very difficult to build perfect animal models for all human conditions, they are making irreplaceable contributions. In addition, studies using animal models greatly depend on carefully designed experimental paradigms. Employment of a variety of learning paradigms among different animals will not only enable the recognition of mechanisms underlying memory formation, but also allow comparisons between different types of learning behaviour(2).

  1. Model Circuits1 and behaviours in Animal models
1.1 Invertebrate models
1.1.1 Gastropod Mollusks
1.1.1a. //Aplysia// Gill-Siphon Defensive Withdrawal Reflex
1.1.1b. //Lymnaea// Breathing Bahaviour
1.1.2 Drosophila
1.2 Vertebrate models
1.2.1 Rodents
  1. Learning paradigms in common animal model
2.1 Associative learning in Mollusks
2.1.1 Classical conditioning
2.1.2 Operant conditioning
2.2 Hippocampal-dependent (Spatial)
2.3 Hippocampal-independent (Non-spatial)

Model circuits and behaviours in Animal models

The process of learning and memory is a fundamental feature shared among many organisms. It involves experience-dependent modifications of synapses to acquire, process and store the information (4). Generally, memory can be classified into two categories: explicit and implicit memory. __Explicit memory__ refers to the conscious retrieval of information about events such as who, where, when, and what. Implicit memory, on the other hand, is known as a form of unconscious recall of events mostly associated with motor skills (4, 5). These two modes of memory also differ in their localizations in neural circuits. In mammalian brains, explicit learning is associated with medial temporal lobe of the cerebral cortex. Implicit memory, on the other hand, would be dependent on cerebellum, striatum and amygdala (6). Although, explicit memory is still examined mostly in more complex mammalian systems, implicit memory is commonly studied in the simple reflex systems found in higher invertebrates (4, 5).

In vertebrate models

Invertebrate models have been playing powerful roles in the examination of mechanisms underlying learning and memory (5). Due to the simplicity of their neural circuits, researchers can trace the alterations down to synaptic or even molecular levels while analysing primitive forms of behavioural plasticity (4). Similar types of analyses are very hard to perform in mammalian brains because of their circuit complexity. It was also demonstrated that both associative (such as conditioning) and non-associative learning (such as habituation) can be assessed in invertebrate systems (4, 5).
In addition, many of the reflex pathways studied in invertebrate systems display some features analogous to the mammalian learning behaviours, suggesting mechanistic similarities between the two systems (4, 5). Hence, the cellular and molecular changes that occur in invertebrate learning can have great implications for understanding the memory formation in mammalian brains (4).

  1. Gastropod mollusks
Gastropod mollusks have been extensively studied for the neural mechanisms of learning and memory processes. This is largely because the synaptic and cellular alterations are easily identified in parallel with the changes in behaviour. There are several well-defined neural circuits where the individual neurons and their synaptic connections can be identified and visualized (11, 14). By investigating the simple associative and non-associative memory in these systems, researchers can identify molecular pathways involved in synaptic plasticity and changes in the cellular excitability that affect formation of memory (2).
Two of the commonly used molluscan model systems:

  • Aplysia gill siphon defensive withdrawal reflex
imagesCA0QE2OU.jpgAplysia californica is a marine mollusk that has been used to study both associative and non-associative learning behaviours (4). Its nervous system is consisted of approximately 20,000 neurons that can be individually identified (2, 4, 5). To elucidate the synaptic and molecular mechanisms underlying short-, intermediate- and long-term memory, several physiological response pathways have been frequently studied in detail (4). Among the pathways, the gill and siphon withdrawal reflex is most commonly used because it displays good correspondence between cellular and behavioural modifications (14). This reflex can be evoked by a tactile stimulus to the siphon, and the stimulation will cause a defensive withdrawal of both gill and siphon back into the mantle cavity. Since the level of neural response is graded depending on the activity, different learning behaviours, such as sensitization, habituation and conditioning, can be examined (5).
The circuitry of this reflex pathway is composed of distinct sets of sensory and motor neurons in the abdominal ganglia which can be easily identified. The LFS motor neurons controlling gill and siphon movement and some excitatory and inhibitory interneurons are connected by monosynaptic innervations from primary afferent neurons and LE mechanosensory neurons (4, 5). Two forms of non-associative memories, sensitization and habituation, found in this model circuit are well –studied.
Aplysia gill siphon withdrawal reflex neural circuit

Habituation is a behavioral plasticity which is accomplished by repeated delivery of light touch stimuli (2). The withdrawal responses following the stimuli will gradually diminish overtime, and they remain depressed for minutes to hours. This depression, known as homosynaptic depression (HSD), is usually accompanied by a decrease in the neurotransmitter release from pre-synaptic mechanosensory neurons. Consequently, the reduction of neurotransmitter will suppress the post-synaptic potential in motor neurons (2, 4, 5).
In contrast, sensitization is an opposite process caused by an increase in the release of neurotransmitter. After the habituation is accomplished, a single strong stimulus to another site of the body would be able to evoke immediate dishabituation (or sensitization) and restore the reflex (4, 5). This is primarily due to the facilitatory effect of interneurons that form synapses on the pre-synaptic terminals of mechaosensory neurons. The consequence of such heterosynaptic facilitation would be the increase of glutamate release at the sensory-motor junctions, and the prolonged EPSP. It was also shown that motor neurons become more excitable because of the rise in intracellular cAMP level (2, 4).
In addition, sensitization and habituation processes can be consolidated by continuous training over longer time (days) to be converted into long term memory that lasts for weeks (2, 4, 5). It was previously believed that the mechanisms of long term habituation and sensitization are similar to the ones found in short term memory which involves only pre-synaptic changes. In more recent studies, it was demonstrated that some post- synaptic mechanisms are also contributing to long term plasticity (2, 4, 14). For instance, long term sensitization is facilitated by PKC-dependent pre-synaptic changes and Calcium-/ CaMKII-dependent post synaptic processes, but the short term form is mediated by PKA pre-synapticly (4, 14).

Aerial respiration pathway in Lymnaea

  • Lymnaea breathing behaviour
Lymnaea stagnailis is an aquatic snail specie that can breathe in bi-modal fashion: cutaneously and aerially. When the water is depleted of oxygen, they will come to the surface and perform aerial respiration (7). This breathing behaviour is accomplished by repeated opening and closing of a structure called pneumostome. Respiration cycle is initiated by a set of mantle muscles surrounding the pneumostome, and their contraction would expose the lungs. After the breathing, another set of muscles will be responsible for pneumostome closure (5, 7). The rhythmic contractions of mantle muscles are governed by the 3-neuron central pattern generator (CPG). It is consisted of RPeD1, IP3 and VD4 interneurons which make their projections to the I, J and K motor neurons controlling the opener and closer muscles (2, 7). Hypoxic condition in the environment will activate the sensory receptors and trigger the depolarization of RPeD1. Subsequently, RPeD1 activates IP3 and inhibits VD4. The activated IP3 will then stimulate further depolarization of RPeD1 and activate other downstream motor neurons that innervate opener muscles. Similar to RPeD1, the IP3 interneuron also inhibits VD4, but the combination of two inhibitions activates VD4. The VD4 activity will close off pneumostome and provide negative feedback to RPeD1 and IP3 (2, 5, 7).

Drosophila melanogaster is commonly used in genetic and molecular analysis of learning and memory

Drosophila melanogaster is a commonly used model organism in many fields of research. It has a small and easily manipulated genome which carries genes homologous to the ones found in mammalian systems (5). Currently, there are many comprehensive methodologies developed for genetic and biochemical analysis in Drosophila. This offered scientists with new opportunities that cannot be obtained from other animal models (1, 5). By screening genetic mutants of every gene in drosophila, researchers can isolate factors that contribute to molecular pathways underlying different aspects of learning and memory (5).
There are several experimental methodologies that are frequently used to assess the associative learning in Drosophila. In operant conditioning, adult fly and larvae can learn an odor-shock association after pairing an aversive electric shock with a type of odor. As a result, flies will reduce the number of visits into the side of apparatus containing the paired odorant (5, 8). In other studies, a visual learning flight simulator was used to investigate associative conditioning. In the apparatus, a fly is tethered to a yaw torque that records the orientation of flight (1, 10). Flies can be trained to use visual cues on the circular arena (different patterns) to orient themselves accordingly. Both classical and operant conditioning can be studied with this apparatus (8, 10).

Vertebrate Models

Although vertebrate models are much more complex compared to invertebrate systems, some vertebrate species are still actively used in the laboratory because they can better resemble the physiology and behaviours in humans. For instance, primates are usually used to study complex memory processes such as the visual working memory (15); rodents are commonly used to study spatial memory and fear conditioning (12). By integrating the knowledge obtained from invertebrate and vertebrate studies, researchers can make effective comparisons, and propose better causal conclusions for the human behaviours and conditions (5).

Rodents are very important in modeling behavioural plasticity during learning and memory formation. They are the mammalian systems that can provide insights into not only cellular mechanisms, but also genetic contributions to the behaviours (6, 12). In other areas of neuroscience, rodents are also used to model conditions such as trauma, stroke, neurodevelopment and neuropsychiatric disorders (12). Although there are many experimental paradigms for assessing learning and memory, we generally divide them into two domains: hippocampal-dependent and nonhippocampal-dependent. Hippocampal-dependent paradigms are usually for studying spatial memories by employing variants of mazes. There are a wide variety of experimental designs for nonhippocampal-dependent memories because it can be memories related to emotion, motor tasks, languages and so on (6, 12).

Learning paradigms in common animal models

Associative learning in mollusks

Original experiment by Pavlov. Classical conditioning a tone with salivation reflex.
cal conditioning is a learning paradigm where a predictive external stimulus can elicit a particular reflexive behaviour. For example, in Pavlov’s original experiment, the dog being train overtime will respond a tone (which predicts food) with salivation reflex while anticipating the food (16).

In the gill withdrawal pathway of Aplysia, an associative form of learning can be achieved by administering light tactile stimulus to the siphon before the strong noxious stimulus (usually electrical shocks) used for sensitization. (See aversive classical conditioning). As a result, trained animals will produce a stronger monosynaptic post synaptic potential in the LFS motor neurons just by applying the tactile stimulation. This will subsequently increase the firing of motor neurons and elicit stronger siphon withdrawal response compared to the controls (2). The synaptic mechanism that underlies these changes is called activity dependent pre-synaptic facilitation. Many studies showed that PKA is the primary molecule mediating the plasticity. However, other researchers argued that Hebbian LTP is facilitating the increased response because blockage of NMDA receptors inhibited the conditioning. Later evidences also revealed that both pre and post synaptic mechanisms are involved in this form of plasticity (2, 4).

Molecular pathways involving Short-term memory (left), and long term memory ( right )

In operant conditioning, animals learn to modify their behaviours accordingly by associating the consequence of its behaviour with a positive or negative reinforcer (10, 17). Unlike the reflexive pathways used in classical conditioning, operant paradigms require spontaneous act by the animal itself. A positive reinforcement usually result in an increase in particular autonomous action, whereas a negative reinforcement eliminate such behaviour (3).
In molluscan models, the CPG circuit in Lymnaea aerial respiration is frequently used for operant conditioning studies. This is because the respiratory behaviour can be paired with a touch stimulus to the pneumostome as a form of negative reinforcement. Snails are usually trained in an oxygen-deficient condition which promotes the breathing (2, 4). When the snails are about to open the pneumostome, a touch stimulus is applied so that they are forced to stop the aerial breathing (4). Overtime, trained Lymnaea will decrease its aerial respiration compared to the yokes and controls. Depending on the conditioning protocols, this short term reduction can be converted into either intermediate memory (ITM, lasting a few hours) or long term memory (LTM, lasting more than six hours). It was demonstrated that these two forms of memory have different underlying cellular and molecular mechanisms since ITM and LTM responded differently to transcription and translation inhibitors. In general, ITM formation relies on new protein synthesis, whereas LTM require alteration in the gene expression in addition to the new proteins (2).
In the cellular level, the CPG interneuron RPeD1 after the conditioning showed much less activity and ability to generate rhythmic respiration compared to controls (7). At the same time, IP3 neuron also reduced its activity. In order to elucidate a causal relationship of memory formation and RPeD1, scientists removed the soma of RPeD1 (4). In mollusks, soma ablation would not affect the survival of primary neurites and the remaining compartments still maintain their local synthesis of new proteins. The removal of the cell body os RPeD1 neuron did not seem to affect the fomraiotn of the ITM, but it inhibited LTM formation, suggesting that RPeD1 neuron is necessary for formation and storage of LTM (2, 4, 7).

Spatial and non-spatial memory in rodents
In mammalian brains, spatial memory, or explicit memory in general often rely on the function of hippocampus. It is involved processes like encoding, retaining and retrieving information, and it is one of the first structures to be affected as human ages. On the other hand, implicit memory (non-spatial) relies on structures such as cerebellum and amygdala depending on the type of memory to be processed (12).
  • Spatial memory
Spatial memory is a brain function related to the answers for question like “where”. Usually, the spatial information is obtained by exploratory behaviours. Although this type of memory is conserved among many species, their external expressions are very different (6, 12). Humans, for example, can use language and other symbols to represent the spatial cognition. Animals, on the other hand, cannot represent such knowledge verbally or symbolically. Therefore, researchers need to find another way to recognize the acquisition of spatial knowledge in animal models (12).
The first experiment on spatial memory in rodents was conducted by Edward Tolman in 1948. He placed starved rats on a maze which is consisted of a paths and a blind alley with food rewards, and observed that the rats made much less mistakes after several trials (12). Decades later, many more behavioural paradigms became available for the assessment for hippocampal-dependent memory in rodents. In addition, these paradigms can be combined with genetic analysis to reveal the underlying genetic and molecular contributions of learning behaviours (6, 12).

A schematic of a T maze

T-Maze: T-maze is a simple paradigm for examining spatial memory in rodents. It is consisted of T-shaped paths where rodents can make a single choice with two alternatives. The underlying principles for T-maze are based on the normal exploratory behaviours displayed by the animals. Since rodents usually prefer to explore the less visited arm, we can measure the number and order of visits they make. This leanring paradigm was also shown to produce robust results in testing the action of drugs and chemicals that alters the spatial memory (12).

A 8-arm radial maze

Radial-arm maze: This learning paradigm is made of about 8 to 12 arms radiate out from a central platform. The basic principle is similar to that of the T-maze. Animals are expected to use their spatial memory to go to the arms they have not visited, and usually a food reward is placed at each end of the arm for them to collect. The number of times rodent visits empty arms are coded as errors (6, 12).

Schematic of morris water maze

Morris water maze: This behavioural paradigm is one of the most commonly used methods in assessing spatial memory in rodents (6). It includes a pool of water that is made opaque, and it is surrounded by a wall with different patterns at various directions. Animals are expected to fund the hidden platform under the water using these visual cues. After several trials, the platform is removed, and scientist will measure the time that animals spent in the quadrant (that previously had the platform) to evaluate their spatial memories (12).

  • Non-spatial memory
Contextual fear conditioning: The classical fear conditioning was often used as a behavioural paradigm in examining non-hippocampal dependent memories in mammalian brain. This model system provided us insights into the circuitry underlying fear assotiation and helped us understand some fear-related mental disorders in humans. There are cumulating evidences showing that amygdala is the major structure in the brain responsible for associative fear memory (13).
A freeze monitor used in contextual fear conditioning

One of the first fear conditioning experiment was performed on a human infant named Little Albert. The 11-month-old infant was conditioned to associate a fearful stimulus (the sound of hammering a steel bar) with the presence of the white rat so that he will cry every time a white rat was placed beside him (18). Due to the ethical concerns, the modern experiment on the fear conditioning is largely performed in animal models. The general protocol is to pair a footstock with a conditioned stimulus such as tone or a visual cue, and the number of times a rodent displays fearful response (e.g. Freezing behaviour) is recorded (13).


  1. Brembs, B. 2011. Spontaneous decisions and operant conditioning in fruit flies. Behavioural Processes. 87: 157-164
  2. Benjamin, P.R. and Kemenes, G. Behavioural and circuit analysis of learning and memory in Mollusks. Elsevier LTD. Pp. 587-604
  3. Nargeot, R. and Simmers, J. 2011. Neural mechanisms of operant conditioning and learning-induced behavioural plasticity in Aplysia. Cell. Mol. Life Sci. 68 803-816
  4. Hawkins, R.D., Kandel, E.R. and Bailey, C.H. 2006. Molecular mechanisms of Memory storage in Aplysia. Biol. Bull. 210: 174-191
  5. Carew, T.J. and Sahley, C.L. 1986. Invertebrate learning and memory: From behaviour to molecules. Ann. Rev. Neurosci. 9: 435-487
  6. Sharma, S., Rakoczy, S. and Brown-Borg, H. 2010. Assessment of spatial memory in mice. Life Sciences. 87:521-536
  7. Lukowiak, K., et al. 2003. A molluscan model system in the search for the engram. Journal of Physiology. 97: 69-76
  8. Brembs, B., and Heisenberg, M. 2001. Conditioning with compound stimuli in Drosophila melanogaster in the flight stimulator. The Journal of Experimental Biology. 204: 2849-2859
  9. Brembs, B. 2011. Spontaneous decisions and operant conditioning in fruit flies. Behavioural Processes. 87: 157-164
  10. Brembs, B. and Heisenberg, M. 2000. The operant and the classical in conditioned orientation of Drosophila melanogaster at the flight simulator. Learn. Mem. 7: 104-115
  11. Michel, M., Green, C.L., Eskin, A. and Lyons, L.C. 2011. PKG-mediated MAPK signaling is necessary for long-term operant memory in Aplysia. Learn. Mem. 18: 108-117
  12. Paul, C.M., Magda, G. and Abel, S. 2009. Spatial memory: Theoretical basis and comparative review on experimental methods in rodents. Behavioural Brain Research. 203: 151-164
  13. Sah, P., Westbrook, R.F. and Luthi, A. 2008. Fear conditioning and long-term potentiation in the amygdala. Ann. N.Y. Acad. Sci. 1129: 88-95
  14. Sharma, S.K. and Carew, T. J. 2004. The roles of MAPK cascades in synaptic plasticity and memory in Aplysia: Facilitatory effects and inhibitory constraints. Learn. Mem. 11: 373-378
  15. Miller, E.K., Erikson, C.A. and Desimone, R. 1996. Neural mechanisms of visual working memory in pre-frontal cortex of the macaque. Journal of Neuroscience. 16 (16): 5154-5167
  16. Pavlov, I.P. 1927. Conditioned Reflexes: An Investigation of the Physiological Activity of the Cerebral Cortex. Oxford University Press, Oxford, UK
  17. Skinner, B.F. 1938. The Behaviour of Organisms. Coplet Publishing Group, United States
  18. Phillips, R.G. and LeDoux, J.E. 1992. Differential contribution of amygdala and hippocampus to cued and contextual fear conditioning. Behav. Neurosci. 106(2): 274-285