Figure 4.

Linear and nonlinear quantification of the neurons. Each color represents neurons from one group of vector field categories (D, C, DC, non-invariant; ID, IC, partially invariant; I, purely invariant), and each small asterisk represents one neuron. The ordinate represents the linear r ² value of a neuron (first class of kernels; see Materials and Methods), the abscissa represents the position invariance r ² value (second class of kernels), and the local tuning r ² value (third class of kernels) is approximately one minus the first two classes of kernels. The colored, open circles represent the mean linear r ² and invariance r ² values for each group of vector field categories. The orientation of the major and minor axis of each colored ellipse, for each group of vector field categories, corresponds to the eigenvectors of the covariance matrix of the data points. The length of each axis of the ellipse corresponds to 2 SD in Mahalanobis units, which is twice the square root of the eigenvalues.