Classical (or Pavlovian) conditioning of motor responses is an experimental model, not only of associative learning, but also of learned response timing. For example, the eyeblink response can be conditioned to a neutral stimulus like a tone. This is achieved by repeatedly presenting the tone, followed by a puff of air to the cornea that elicits the reflex blink response (i.e. an unconditioned stimulus, or US, that elicits an unconditioned response, or UR). Initially the tone (the conditioned stimulus, or CS) elicits no eye blink response, but gradually as an effect of training, the tone CS will acquire the ability to elicit a learned blink response (i.e. a conditioned response, or CR).
Since the mid 1980’s it is known that this type of learning depends on the cerebellum, though the precise localization of the crucial cerebellar neural network and it’s plasticity mechanisms have remained elusive. However, already 15 years before this finding, the so called Marr-Albus theory of cerebellar learning (Marr, 1969; Albus, 1971) proposed that Purkinje cells might learn to respond with a learned pause in order to elicit a motor response. A pause is expected, since the Purkinje cells are inhibitory and in order to increase activity downstream this inhibition has to seize firing. According to the model, signals from the CS (tone) and US (air puff) are transmitted via several pathways in the brain, but two particularly important pathways converge on the dendritic tree of Purkinje cells in the cerebellar cortex. Mossy fibers and subsequently parallel fibers, constitute a CS pathway to the Purkinje cells, while the crucial US signal reaches the Purkinje cell via climbing fibers.
In a series of experiments, we have recorded Purkinje cell activity in vivo, during training with CS input from peripheral stimuli or direct activation of mossy fibers, paired with US input from peripheral stimuli or direct stimulation of climbing fibers. The existence of such a conditioned pause response in the Purkinje cell (a PcCR) was confirmed just prior to the launching of the CCL environment (see Jirenhed et al., Journal of Neuroscience, 2007)
- Experimental setup: In vivo recording of cerebellar Purkinje cell activity during classical conditioning. PC Purkinje Cell, GrC Granule Cells, AIN Anterior Interposed Nucleus, mf mossy fibers, cf climbing fibers. SC and IC Superior and Inferior Colliculus.
Since 2010, we have been investigating how Purkinje cells learn to produce adaptively timed PcCRs under different training conditions and how such conditions shape the learned PcCR. Most importantly, the PcCRs in a single cell display very complex temporal characteristics that can explain a wide range of behavioral timing phenomena described in the literature on motor learning.
Our data on conditioned Purkinje cell responses and the effects caused by varying the stimulus characteristics challenge current models of how the cerebellum controls timing of behavior. Most models of cerebellar timing assume that granule cells, which transmit the conditioned stimulus signal via parallel fibers to the Purkinje cells, have in-built firing patterns with varying delays. Another long-standing hypothesis is that the learned timing is realized by increasing or decreasing synaptic connections (LTD or LTP) between cells (see Hesslow et al., Neural Networks, 2013).
Our investigation of associative learning mechanisms have also resulted in important discoveries about other parts of the cerebellar network. Notably, we have shown that the number of spikes in the teaching signal (the US) in the climbing fibers, determines whether learning or extinction occurs. Given that it is commonly assumed that this signal is of an all-or-nothing nature, this finding will have important implications for future research. We have also begun to investigate how other types of neurons in the cerebellum contribute to learning. For example we have shown that Golgi cells are not directly involved in the generation of conditioned responses.
Future research directions
We are currently investigating the molecular mechanisms behind the Purkinje cell CR, using various pharmacological manipulations. Given that single cells can respond with a delay, we want to understand how this occurs in detail. Also, if we can obtain properly timed responses in the Purkinje cells with direct parallel fiber stimulation as the CS, this will help us to determine if the learning site is in the Purkinje cells or the interneurons.
In addition, we continue to investigate the properties of the entire cerebellar network, with particular focus on nucleo-olivary feedback which appears to have a key role in regulating cerebellar learning.
Peer reviewed articles since 2010
Wetmore D, Jirenhed DA, Rasmussen A, Johansson F, Schnitzer M, Hesslow G. (2013). Bidirectional plasticity of Purkinje cells matches temporal features of learning. Journal of Neuroscience. Accepted for publication Dec 19, 2013.
Rasmussen A, Zucca R, Jirenhed DA, Johansson F, Ortenblad C, Svensson P, Hesslow G. (2013). Golgi Cell Activity During Eyeblink Conditioning in Decerebrate Ferrets. Cerebellum. Aug 28.
Rasmussen A, Jirenhed DA, Zucca R, Johansson F, Svensson P, Hesslow G. (2013). Number of spikes in climbing fibers determines the direction of cerebellar learning. Journal of Neuroscience. 14;33(33):13436-40.
Hesslow G, Jirenhed D-A, Rasmussen A, Johansson F. (2013). Classical conditioning of motor responses: What is the learning mechanism? Neural Networks. 47:81–7.
Jirenhed D-A, Hesslow G (2011) Learning Stimulus Intervals – Adaptive Timing of Conditioned Purkinje Cell Responses. Cerebellum. 10: 523-535.
Jirenhed D-A, Hesslow G (2011) Time Course of Classically Conditioned Purkinje Cell Response is Determined by Initial Part of Conditioned Stimulus. Journal of Neuroscience. 31: 9070 –9074.
Poster presentations at international conferences
Johansson F, Hesslow G, Jirenhed DA & Rasmussen A. (2013). Distinguishing between conditioned Purkinje cell responses and synapse strength modulation In vivo. Soc For Neuroscience meeting, San Diego, CA.
Jirenhed DA, Johansson F, Hesslow G. (2013). Purkinje Cell Learning of Multiple Responses to a Single Stimulus. Gordon Research Conference on the Cerebellum. New London, NH.
Rasmussen, A., Zucca, R., Jirenhed, D., Johansson, F., & Hesslow, G. (2013). Direction of learning in cerebellar Purkinje cells dependent on number of climbing fiber impulses. Translational Sensorimotor Neuroscience Meeting. Stockholm, Sweden.
Zucca, R., Rasmussen, A., Johansson, F., Verschure, P. F., & Hesslow, G. (2012). Suppression of conditioned eye-blink responses through high frequency climbing fibres stimulation. Federation of European Neuroscience Societies meeting. Barcelona, Spain.
Rasmussen, A., Zucca, R., Jirenhed, D., Johansson, F., Örtenblad, C., Svensson, P., & Hesslow, G. (2012). Effects of the conditional stimulus on Golgi cell activity during classical conditioning of the eye-blink response in the decerebrate ferret. Society For Neuroscience meeting. New Orleans, LA.
Johansson, F., Jirenhed, D., Rasmussen, A., Zucca, R., & Hesslow, G. (2011). In vivo parallel fibre stimulation as conditional stimulus in Pavlovian conditioning of Purkinje neuron responses. Society For Neuroscience meeting. Washington, DC.
Rasmussen, A., Johansson, F., Zucca, R., & Hesslow, G. (2011). The neural foundation of overexpectation. Society For Neuroscience meeting. Washington, DC.
Zucca, R., Rasmussen, A., Herreros-Alonso, I., Verschure, P. F., & Hesslow, G. (2010). A closed-loop framework to investigate cerebellar learning in a decerebrated ferret preparation. Society For Neuroscience meeting. San Diego, CA.
Rasmussen, A., Jirenhed, D., & Hesslow, G. (2008). Olivary activity during Classical Conditioning. Motor Networks meeting. Lund, Sweden.
Hesslow, G. (2013). How does the cerebellum learn adaptively timed behavior? Gordon Research Conference on the Cerebellum. New London, NH.
Jirenhed, DA. (2013). Associatively learned responses in the Purkinje cell. Translational Sensorimotor Neuroscience Meeting. Stockholm, Sweden.
Collaborations outside of CCL
We have an active collaboration with Professor C.E. Yeo at University College London. He is studying classical conditioning on intact rabbits, which is a perfect complement to our setup and we plan to perform pharmacological experiments in intact animals in London.
We have also had a fruitful collaboration with Professor M.J. Schnitzer’s group at Stanford University.