Power from the people: Commercial headsets such as Muse deliver insights on aging from large-scale EEG datasets.
For decades scientists have focused on collecting EEG data from small groups of participants in tightly controlled lab environments to demonstrate specific effects of interest. Typical sample sizes across most studies range from 15 to 50 subjects. But the status quo is now changing. EEG is no longer restricted to the lab, to a select few (see Taking Neurotechnology out of the Lab). At the click of a button you can purchase an EEG headset of your very own to use for your personal interest. Systems like this don’t just collect and store data at an individual level but collate large-scale datasets.
Population-scale datasets
Muse, for instance is a neurofeedback-based EEG system utilizing 4 electrodes (TP9, TP10, AF7, and AF8) and is marketed to enhance your meditative practices by receiving feedback from the headband directly to your mobile device .
In doing so Muse has compiled and studied over 6000 EEG recordings from individuals living in America, Europe and Asia, including both male and female users, spanning age 8 to 88.In collaboration with researchers from McMaster University in Ontario, Canada, they have looked at how aspects of the EEG differ across the population with dimensions of gender and age. The data analyzed spans two of their standard tasks – a “Category Exemplar Task” and a meditation task. In the exemplar task users are required to think up as many examples of items from a particular category with their eyes closed, whilst in the meditation task they have to focus on counting their breathing whilst relaxing with their eyes closed. See the original paper here.
Shifts in EEG activity with age.
The power spectrum showed systematic shifts with age.
Figure: Alpha and beta power plotted against age for males (gray symbols) and females (white symbols) in the frontal sites AF7 and AF8. Each point represents the mean for that age; symbol size represents how many individuals were used to compute the mean. Regression was used to compute the best-fitting curves separately for males and females, and the shaded regions represent 95% confidence intervals
The most prominent pattern was an increase in the power of the higher frequency alpha (7-14Hz) and beta band (14-30 Hz) frequencies in the frontal sites with increasing age in both males and females (see above). There was a concomitant decrease in the lower frequency theta (3–7 Hz) and delta (0–2 Hz) bands in the frontal sites between 20-40 years, followed by a leveling off or an increase. This suggests that when older individuals were meditating, or thinking up category exemplars, there was a shift in the distribution of alpha activity across the scalp compared to the younger members of the population group, manifesting itself as progressively less power between 7 and 30 Hz at temporoparietal regions, and progressively greater power in these bands at frontal regions.
Girl Power
EEG power in the alpha and beta bands (0-30 Hz) was also greater in women than men in the frontal regions (and less so across temporoparietal regions), something that has been proposed by previous studies, potentially due to functional and/or structural differences.
Variability in the alpha oscillation
They also looked at whether there was a noticeable alpha peak (8–13 Hz) in both tasks, reflecting alpha oscillatory activity. They found that ∼88% of people presented an alpha peak at temporoparietal sites, and 50% presented one at frontal sites (although this frontal peak did not show up on the averaged spectrum). The fact that a peak was not found in all individuals in these brain regions is in line with previous data from Sapien Labs who have shown significant variability in the presence of an alpha peak across individuals, particularly in frontal and temporo-parietal regions.
Slowing with age
Despite large individual variability, the peak frequency of the alpha oscillation, when detected, was systematically lower with age across both frontal and temporoparietal regions. Similar findings have been found in other studies. For example, Richard Clark from Flinders University in Australia and colleagues showed a decrease in the alpha peak with age (especially at frontal regions) and also a positive correlation between peak alpha frequency and performance on a digit span working memory test. One explanation for the decreasing peak frequency is that it could be due to physiological changes in the brain (e.g. neural degeneration), or alternatively reflect a general decline in the speed of cognitive processing. However, the life experiences of someone who is 20 will differ from someone who is 60 and it could also be the case that situational and lifestyle factors may also play a role. Future studies will be needed to pull out the various contributions of biology versus experience to provide the answers.
1000s not 10s.
This study is an illustration of how you can use commercially available EEG headsets to collect large-scale datasets which allow you to study 1000s of individuals (or 6029 in this case), rather than just 10s, and deliver meaningful results. While such headsets may have lower signal quality than research grade wet electrode systems, they can make up for this in the scale of the data that is made possible. Only by studying such large groups of people are you able to accurately pry apart how our brains are the different. And although the population demographics in studies like these are self selected, and may not always reflect a true cross section of the population, they are springboard for further forays into the diversity of the human brain.
