Human subjects
The EEG dataset was collected from 257 human subjects (females: 106) aged 50 to 88 years as part of the Tata Longitudinal study of aging (TLSA), of which usable data was obtained from 244 subjects (227 healthy, 12 MCI and 5 AD; see Murty et al., 2021 for detailed selection criteria). Subjects were recruited from urban Bengaluru communities and were evaluated by trained psychiatrists, psychologists and neurologists affiliated with National Institute of Mental Health and Neuroscience (NIMHANS) and M.S. Ramaiah Hospital, Bengaluru. Cognitive performance was evaluated using ACE (Addenbrooke’s Cognitive Examination), CDR (Clinical Dementia Rating) and HMSE (Hindi Mental State Examination) tests (see Murty et al., 2021 for more details). Clinical scores and their relationship with age are shown in the Supplementary Fig. 1, following the format mentioned in the study by Farina and colleagues (2020). The AD group was very small, and most subjects did not have appreciable gamma, and were therefore not considered for further analysis. As this was a community-based study, in which subjects were recruited based on advertisements, we did not have any knowledge of their cognitive status prior to the experiment. Their clinical identity was determined and was revealed to us after the experiments were over. Therefore, one limitation of this study is that no power calculation could be done for arriving at the MCI subject count used for the study. We further discarded 10 subjects (9 healthy and 1 MCI) because data was collected using 32 channels. All results are based on the remaining 218 healthy and 11 MCI subjects. As with our previous study (Murty et al. , 2020), we divided the healthy subjects into two groups: middle-aged (50-64 years) and elderly (>64 years). All subjects took part against monetary compensation and provided signed informed consent. All the procedures were approved by The Institute Human Ethics Committees of Indian Institute of Science, NIMHANS, Bengaluru and M.S. Ramaiah Hospital, Bengaluru.