In Japan, rapid population ageing and declining birth rates are intensifying the dementia-care load on younger generations. Early detection of mild cognitive impairment (MCI), the prodromal phase of dementia, is challenging because functional loss is subtle and conventional imaging or clinical EEG are costly and burdensome for screening. This study aimed to extract MCI-related features from high-frequency EEG recorded with a low-cost, two-sensor head-band device. EEG was collected from healthy younger and older adults while they performed dementia-related tasks, and spectral power across 1–130 Hz was quantified using continuous wavelet transforms. Older adults exhibited significantly greater power than younger adults in every band for all tasks. Notably, high-γ activity (70–130 Hz) was markedly elevated and its post-task suppression was prolonged in the elderly, providing clear age-group separation. These findings suggest that head-band EEG combined with high-frequency wavelet analysis offers a rapid, non-invasive biomarker for early cognitive decline.