Computing Reviews

Fractals and Ravens
McGreggor K., Kunda M., Goel A. Artificial Intelligence215(1):1-23,2014.Type:Article
Date Reviewed: 12/28/16

Startling! This paper really offers something to think about. The authors also seem surprised by the success of their counter-intuitive juxtaposition of a fractal-based “visual analogical reasoning” algorithm to solve all major variants of the Raven’s Progressive Matrices (RPM) (nonverbal IQ) tests. One would expect some modest overlap, which would then give the authors room to explain a degree of success and to suggest some directions for improvement. However, after a comprehensive introduction to RPM and highly articulated disclosure of their algorithm (including a detailed example), this carefully presented reporting of total success seems beyond anything that any of us are prepared for. Now, the tasks are many: to understand what the RPM suite is really testing, how the RPM construction reflects something in brain “architecture” or neurophysiology, what the reported algorithm might actually be transforming, how that transformation is related to intelligence, whether the brain actually performs that transformation, and so forth. Of course, these results need to be independently reproduced. Then one wonders if they will extend to levels of RPM tests that humans have difficulty solving, or even guessing. Would some fractal human-front-end filter improve the human’s intelligence, like add-on lenses improve vision? Are there 3D or multisensory equivalents to the Raven’s matrices? This is a carefully constructed must-read research report!

Reviewer:  Chaim Scheff Review #: CR144978 (1703-0189)

Reproduction in whole or in part without permission is prohibited.   Copyright 2024 ComputingReviews.com™
Terms of Use
| Privacy Policy