Abstract

In this thesis, we examined whether the recursive structure of a design task is mirrored by phase-locked changes in the fractal dynamics of cortical activity and evaluated which preprocessing pipeline best preserves those dynamics under realistic noise. Twenty participants completed an open-ended environmental-design exercise while a 64-channel EEG was recorded. Idea-generation and evaluation intervals were identified retrospectively. After preprocessing with PREP, HAPPE, or
artifact-subspace reconstruction (ASR)—pipelines previously benchmarked against graded EMG, eye-blink, and ECG artifacts—Higuchi fractal dimension (HFD) and detrended fluctuation analysis
(DFA) were extracted, and t-tests assessed whether these metrics differed significantly between the idea-generation and evaluation phases.
The observations confirm a phase-specific modulation of neural complexity. During evaluation, HFD increased reliably at frontocentral sites (FC3, FC1, C1, FC4, FC2, FCz; partial η2 ≈ 0.20–0.38), consistent with heightened executive control. Conversely, DFA rose in C1 during generation, indicating stronger long-range correlations that accompany spontaneous associative thought.
Pipeline choice, rather than noise amplitude, proved the principal determinant of metric fidelity: PREP preserved waveform morphology; ASR maximised signal-to-noise at the cost of fine detail under severe artifacts, and HAPPE offered a balanced compromise.
Taken together, the findings strengthen the claim that fractal logic operates across brain and behavior and provide practical guidance for researchers seeking to measure it. An open-source,
stand-alone GUI toolbox offering validated HFD and DFA routines accompanies the thesis.