vis.frequency.movshon2005_fit

  MOVSHON2005_FIT - fit a Movshen et al. 2005 frequency response function

  [P,RFIT,MSE,RSQUARED] = MOVSHON2005_FUNC(F, R, USEC)

  Computes a least squares fit of responses (R) to frequencies (F) according to
  Movshon et al. 2005, J Neurosci 25:2712-2722.

  F is a vector of frequencies over which to make the calculation.
  R is a corresponding vector of responses for each value of F.
  If USEC is present and is 1, then P will also contain the constant.
  Outpus:
  P is a vector with the parameters where
    k - P(1) - scaling factor
    fc - P(2) - characteristic temporal frequency
    fh - P(3) - corner frequency of the low-frequency limb
    B - P(4) - slope of the low-frequency limb
  MSE - mean squared error (per point)
  RSQUARED - the RSQUARED value of the fit.

  The function has the form:

     R(f) = k * exp(-(f./fc).^2) ./ (1+(fh./f).^B)

  See MOVSHON2005_FUNC (vis.frequency.movshon2005_func)

  Example:
    [cell1,cell2] = vis.frequency.movshon2005_cells();
    [P1,rfit1,mse1,rsq1] = vis.frequency.movshon2005_fit(cell1(:,1),cell1(:,2));
    [P2,rfit2,mse2,rsq2] = vis.frequency.movshon2005_fit(cell2(:,1),cell2(:,2));
    freq_points = logspace(log10(0.02),log10(64),50);

    figure;
       % the data points
    plot(cell1(:,1),cell1(:,2),'ro'); 
    hold on;
       % the fit
    plot(freq_points,vis.frequency.movshon2005_func(freq_points,P1),'r-'); 
       % the second data points
    plot(cell2(:,1),cell2(:,2),'bo');
       % the second fit 
    plot(freq_points,vis.frequency.movshon2005_func(freq_points,P2),'b-'); 
    box off;
    ylabel('Response (ips)');
    xlabel('Temporal Frequency (Hz)');
    set(gca,'xscale','log');