Human eeg responses to exact and statistical fractal patterns.
Paper in proceedings, 2012
Attention Restoration Theory point to nature as a particularly good candidate for restoring the ability to focus and inhibit distractions. However, little effort has been directed to elaborating on what makes nature's visual pattern unique for restoration purposes. One suggested explanation is the fractal properties of natural patterns (Hagerhall, 2005). Both perception studies (Hagerhall et al., 2004) and studies using quantitative electroencephalography, qEEG (Hagerhall et al., 2008) have pointed to that visual patterns with mid fractal dimension, D, seem to be the most relaxing and preferred. However, it has so far not been investigated if these responses are being driven by fractal geometry in general or by the specific form of fractal geometry found in nature. In natural scenery it is so called statistical fractals that are common because of nature's integration of randomness with the underlying fractal scaling properties. A natural fractal hence looks different from an artificial exact fractal (where the patterns repeating are identical at all scales), even if the two patterns have the same D value. We have here conducted the first study in which we consider both types of fractals and morph one type into the other.qEEG was recorded from 35 subjects viewing 9 images, combining 3 fractal dimensions and 3 levels of randomness. Each image was shown for 60 seconds, interspaced by 30 seconds of a neutral grey image. The data was mainly treated by analysis of variance (repeated measures design), including both within-group (fractal dimension, level of randomness) and between-group (order) variances. Bonferroni correction was carried out and the significance level set to p<.005. The study confirmed our hypotheses in that it showed that alpha power (an indicator of a wakefully relaxed state and internalized attention) was largest for fractals found in nature, i.e. fractals incorporating high randomness and mid to low fractal dimension. Statistically a majority of the results were tendencies (with p_.05) but the reaction patterns showed consistencies which point to that human response to the patterns, and the parameters randomness and fractal dimension, have some fundamental base. The results must be considered as preliminary, but encouraging given the relative visual simplicity of the patterns.