Bounded Kolmogorov Complexity Based on Cognitive Models

Computable versions of Kolmogorov complexity have beenused in the context of pattern discovery [1]. However, these complexitymeasures do not take the psychological dimension of pattern discoveryinto account. We propose a method for pattern discovery based on aversion of Kolmogorov complexity where computations are restricted toa cognitive model with limited computational resources. The potentialof this method is illustrated by implementing it in a system used tosolve number sequence problems. The system was tested on the numbersequence problems of the IST IQ test [2], and it scored 28 out of 38problems, above average human performance, whereas the mathematicalsoftware packages Maple, Mathematica, and WolframAlpha scored 9, 9,and 12, respectively. The results obtained and the generalizability of themethod suggest that this version of Kolmogorov complexity is a usefultool for pattern discovery in the context of AGI