The present study grew out of a consideration of the fact that appropriately defined and objectively scored avoidable error indeces appear capable of assessing the impairment of intellectual functioning. In this research it was hypothesized that Progressive Matrices (1938) is not merely comprised of a process of series induction, but that embedded in this is a process of similarity perception. If this is in fact so, the similarity perception features involved in the test may act to facilitate (or conversely, to render more difficult) the task of completing the arrays for some sUbjects. That is, the contaminating features involved in the test, may attenuate the differences in abstract reasoning in various pathological groups that might otherwise obtain. Instead of presenting the standard test, and then using a modified scoring technique, as outlined above, it may be more efficient to present a modified version of the test itself i.e. eliminating the similarity perception features involved in the standard form. In this manner, it may be possible to obtain a clearer picture of the differences in abstract reasoning related to different pathological groups than is the case when "avoidable error" measures are used. Thus, the present study is an exploratory attempt to determine whether or not the standard form of Raven's Progressive Matrices (1938) is contaminated by similarity perception features.