Generalization of learning hand-weight changes - Rehabilitation Institute of Chicago

Skip to Content

Wed July 20

Speaker: Kunlin Wei, Ph.D., Associate Professor, Dept of Psychology, Lab of Motor Control and Virtual Reality, Peking University, Beijing, China
Title: Generalization of learning hand-weight changes AND Serotonin as a transmitter of force gain in the spinal cord

Abstract: Generalization of learning hand-weight changes

Most experimental studies of reaching adaptation perturb movements using either a visuomotor transformation or a robot that pushes the hand off target. Here we introduce a new way of perturbing the hand. We attached a water-filled cylinder underneath the hand and unexpectedly change the weight of the hand between trials. This allows us to conduct adaptation experiments with 3D reaches that involve neither a virtual reality setup nor a robot and thus, arguably, are closer to the way by which the nervous system normally operates. We examined how learning of a perturbation (hand weight change) in one direction would affect movements into other directions. In Experiment 1, we use a trial-by-trial design where we randomize the distribution of perturbations over time. By evaluating the hand deviations in the vertical direction across trials, we can assess adaptation and generalization. In Experiment 2, we use a block-design where subjects are trained for a sequence of trials before a test trial.

We found that for trial-by-trial perturbations generalization is spatially tuned but with a much wider generalization curve than those found in previous studies. Generalization is observable even in the directions that are opposite to the training direction. For block learning (Experiment 2), we found that the generalization of a loading hand is significant but, interestingly, uniform across the space, i.e., with a flat generalization curve.

The transition from uni-modal to uniform generalization suggests that the widely observed uni-modal generalization is associated with the uncertainty about the applied perturbations, which can be reduced or learned over time. Overall the present study found a much wider generalization, compared to previous studies on visual transformations or external forces. This may suggest that the hand's weight is a more global variable in movement control than other control variables such as visuomotor gains. It is worthy to note that limited generalization found previously are mostly based on artificial perturbations in lab settings with the aid of virtual reality or simulated force fields, while the wider generalization reported here is based on unconstrained real-life movements which are still underrepresented in the area of sensorimotor control.

Abstract:  Serotonin as a transmitter of force gain in the spinal cord.

Sensory stimuli vary over multiple orders of magnitude, and a central question in neuroscience is how neurons with limited bandwidth can encode such widely varying stimuli. Gain control, which is present in all sensory systems, helps solve this problem. Our motor system faces an analogous problem, but from an inverse perspective. The forces we produce vary over multiple orders of magnitude, but motor commands from the brain to the spinal cord are transmitted by noisy neurons with limited bandwidth. The resulting fluctuations need to be minimized across the full force range to allow precise control. However, it is not currently known if the spinal cord also uses mechanisms of gain control. Here, we show that neuromodulators such as serotonin define such a gain control signal in humans. The spatiotemporal patterns of force interactions across limbs show clear reflections of the predicted gain control mechanisms. Importantly, drugs that change the effect of serotonin systematically modulate these interactions. Simulations, psychophysics and pharmacology suggest that gain control is not only important in the sensory systems but also for force production in the spinal cord where it is affected by serotonin.


Speaker: Sunday Francis, Ph.D., Post-Doctoral Candidate

Title: Two Systems:  From Genetics to Behavior

Abstract: Outcomes, especially in neuroscience, are often the result of networks or systems. These systems can be studied at different levels from molecular to behavioral.  We studied motor behavior by examining the effects of learning on a defined network consisting of primary motor (MI) and dorsal premotor (PMd) cortices.  The two forms of motor learning studied were sensorimotor learning (SML) and motor skill acquisition (MSA).  Utilizing a visual gain change paradigm we studied the effects of SML, which was defined as the ability to form new associations between learned movements and sensory cues.  MSA was defined as creating a new temporal pattern among simple, learned movements; this was accomplished by repeating a sequence of targets in a specific pattern.  Given the differences in these forms of motor learning we predicted task dependent activity in both areas, but a higher proportion of units displaying learning effects in PMd for SML, and a higher proportion of learning cells in MI for MSA.

Using primates implanted with multielectrode arrays (MI and PMd), single unit activity was simultaneously collected, during the controls and experimental paradigms.  This data yielded four neural categories: memory, learning, multiactivity and non-categorized.  We confirmed the prediction that more learning effects would be observed in the experimental tasks as compared to controls.  However, both learning paradigms equally influenced PMd and MI.  After further analysis we noted that PMd displayed task dependent activity in the firing rates of learning cells, while MI did not.

Next, I moved on to research the behaviors and phenotypes associated with autism spectrum disorders (ASD).  Impaired social interactions, communications and the presence of restricted and repetitive behaviors characterize ASD.  Due to their roles in modulating social behaviors across many species of mammals, oxytocin, vasopressin and their pathways have been hypothesized to play a role in ASD.  There have been nearly a dozen reports of genetic linkage and association of OXT, AVP and their receptor genes in independent samples of ASD.  Our work examines copy number variations (CNVs) within the neuropeptide hormone genes of OXT, AVP and their receptors.

CNVs are deletions, insertions or translocations of DNA on the kilo- to megabase scale and can be observed by comparing reference genomic DNA to a sample.  We will apply microarray technology and other genetic techniques to the study of autism, possibly revealing genomic changes in the DNA of ASD individuals.  Additionally, given our access to complete families, we have the ability to identify de novo versus inherited CNVs.  These CNVs may cause varying gene function that could contribute to the heterogeneity observed in oxytocin plasma levels and autism phenotypes.

Host: Dr. Jayaraman