When intelligence is just a matter of copying

摘要

The paper investigates a general approach that allows us to solve IQ tests based on Raven's progressive matrices automatically. The 3×3 Raven matrices exhibit 8 geometric pictures displayed in 8 cells of the matrix, which have to be logically completed with a 9th picture to be put in the remaining empty matrix cell. In these tests, a set of candidate pictures for the solution is also given. The suggested approach is based on a logical view of analogical proportions (i.e., statements of the form "A is to B as C is to D"). The reading of Raven matrices in terms of such proportions can be applied to a feature-based description of the pictures, but also, in a number of cases, to a very low level representation, i.e., the pixel level. It appears that the analogical proportion reading just amounts here to a recopy of patterns of feature values that already appear in the data (after checking that there is no conflicting patterns). Implementing this principle, the proposed algorithm computes the 9th picture, without the help of any set of candidate solutions, but only on the basis of the 8 known cells of the Raven matrices. A comparison with other approaches is provided, and we emphasize the generality of the approach which is able to provide a simple and uniform mechanism applicable in many situations.