Aug 022009

Someone wrote to me about inkblots. Apparently, the topic has become hot, in response to the decision by Wikipedia editors to make the Rorschach blots available online. Attempts by some to suppress this information using, among other things, questionable copyright claims, are of a distinctively Scientologist flavor (made all the more curious by Scientology’s rejection of conventional psychoanalysis.) They do have a point, though… the validity of the test could be undermined if test subjects were familiar with the inkblots and evaluation methods. On the other hand, one cannot help but wonder why such an outdated test is still being used in daily practice. It certainly gives credence to those who consider psychoanalysis a pseudoscience.

I am also wondering… suppose I build a sophisticated software system with optical pattern recognition, associative memory, and a learning algorithm. Suppose the software is buggy, and I wish to test it. Would I be testing it by running the recognition program on meaningless symmetric patterns? The behavior of the system would be random, but perhaps not completely so; it may be a case of ordered chaos with a well defined attractor. Would running the recognition program on a few select images reveal anything about that attractor? Would it reveal enough information to determine reliably if the attractor differs from whatever would be considered “normal”?

More importantly, do practitioners of the Rorschach test know about chaos dynamics and do they have the correct (mathematical, computer) tools to analyze their findings?

I am also wondering how such a test could be conceivably normalized to account for differences in life experience (or, to use my software system example, for differences in the training of the learning algorithm) but I better shut up now before my thoughts turn into opinionated rantings about a subject that I know precious little about.

 Posted by at 2:36 pm