Fig. 2.
Example of a one-factorial design with 3 stimuli shows the influence of noise on sample correlations. (A) The five measures for each of the 3 conditions consist of the corresponding true pattern component U and noise E. (B) Dependence of sample correlations (gray line) and of the estimates from the component model (dashed line) on the noise variability (σε2). Correlations between stimulus 1 and 2, stimulus 1 and 3, and stimulus 2 and 3 are shown. The true value is indicated by a line. (C) Multi-dimensional scaling of similarity structure, based on 1-r as a distance metric. The true similarity structure can be represented as one dimension.