October 12, 2009

    • No lab meeting next week (10/19)
    • Check on flights to Banff 11/8-15 (Sunday to Sunday)
      • Skiing?
    • Look at chapter for Stacey/Lena about PCA/NMF
    • SfN meeting - posters on Saturday, motor meeting on Friday.

Lucas' Presentation

    • Does the nervous system reduce the dimension of sensory information?
    • Use cats to test this question
      • 2D stimulus
      • Sensory input dimension is assumed large due to many joints perturbed
      • We can separate sensory response, emg response and mechanical response due to the delays in the system
      • Joint angles are projected to sagittal and transverse plane
      • Doing PCA on the covariance was bad because...this emphasizes fitting data to noise
      • Doing PCA on the correlation was good because...only used variables that reconstruct non-noisy values
    • Results
      • Platform perturbation results in a higher dimensional perturbation
      • Dimension is higher in disturbance than in response