A new paper of mine with my colleagues, and Algorithmic Nature Lab members, Nicolas Gauvrit and Fernando Soler-Toscano just came out.
Using previously generated and new experimental data together with new methods to calculate the algorithmic complexity of 2-dimensional objects, we were able to find that when humans assess the complexity of an image (a small 4×4 pattern), their rating is correlated to the algorithmic complexity of the image mediated by the probability that such pattern appears in real world scenes. In other words, humans are biased both towards patterns in the world and algorithmic complexity, but also patterns in the world are correlated to algorithmic complexity. This strengthens my claim for an algorithmic world, where patterns can be accounted for by an algorithmic production process.
The journal (Visual Cognition) allows 50 free electronic copies of the paper to be downloaded. Should you be interested in this paper and can’t access it otherwise, you can have a free copy, using the following e-print link.