Skip to main content

Monitored Meditation: Measuring the Intuitive System

A growing number of tech companies now sell devices that are marketed as mindfulness aids, ranging from heart rate monitors to breathing apps to sensors that measure brainwaves or electrical conductance of skin (GSR). These devices all operate under the assumption that they are measuring physiological indicators corresponding to the mental state of relaxation or awareness known as mindfulness. Mindfulness is a concept with many possible meanings; in a previous blog we gave our working definition of mindfulness as a combination of attention plus non-judgment. Considering the physiology of mindfulness adds another layer, which people sometimes describe as a state of deep relaxation, calm alertness, or being "centered" in one's own mind.

There is good evidence that meditation can produce a distinct physiological state characterized by slower breathing and heart rate, a smooth EEG wave function in the brain, and an improved immune response. This meditative state seems to be distinct from other types of relaxation. However, mindfulness is only one among many meditation practices, and modern Western takes on mindfulness may not convey the same physiological benefits as Eastern meditation practices rooted in a deep spiritual tradition. It is also unclear whether all of the smartphone apps and electronic devices that claim to promote mindfulness are actually collecting relevant data, analyzing them in a valid way, or providing useful feedback for people who want to improve their practice of mindfulness.

With a research assistant (my teenage daughter) I recently conducted an uncontrolled pilot study of several sensor devices to evaluate their potential utility as mindfulness tools. We looked at 3 questions: (a) do data from different devices provide similar information at the same points in time [reliability], (b) do data from various devices match the user's subjective experience of relaxation or centeredness over time [validity], and (c) do the devices register a disruption or disturbance of centeredness [sensitivity], which in our case was created by attempting to surprise the person meditating without prior warning. 

We tested 4 different devices by wearing them all simultaneously and attempting to meditate:
  • The Apple Watch 4, which includes a "breathe" app to promote mindfulness and monitors the user's heart rate continuously throughout the day even when not using the app
  • The Spire Stone (recently replaced by Spire Tag), which measures breathing rate throughout the day as long as its app is open on the user's phone, via pressure sensors in a belt-mounted device
  • The Muse headband, which measures brain activity with a 5-lead EEG and translates the results (alpha, beta, delta, and theta waves) into a single measure of active vs calm mental states using a proprietary algorithm
  • The HeartMath Inner Balance sensor, which tracks heart rate variability and converts the raw data to a "coherence" measure based on the pattern of heart rate measurements over time, also using a proprietary algorithm.

The image at the top of this post shows results from one trial, and this link gives results from two more (note that in Trial 1 the Spire sensor didn't provide any data, and in Trial 3 we looked at raw EEG data from the Muse device instead of the proprietary calculation of results). Each graph shows a time interval of about 5 minutes, reading from left to right, with an attempt made to visually line up the time scales on the different data stream. The blue line on each page shows timing of the pre-planned disturbance, demonstrating how it registered or didn’t on each of the sensor devices.

An examination of the data immediately suggests two key limitations: First, some measures provide more frequent data points than others, and second, the data are not available as a file download from most devices which makes it necessary to compare graphs visually rather than using more precise math. In view of these limitations a more sophisticated study could clearly be carried out to answer the research questions more conclusively. However, here are some preliminary thoughts about our findings:

  1. Basic reliability was supported. Reading from top to bottom of each page, there was agreement overall between measures, but variability in the moment-by-moment details. The two devices that used proprietary algorithms actually seemed to agree more with each other than the two devices that did not. This can be seen at the top of the page, where my mind wandered around minute 3, and both the Muse headband and the HeartMath sensor detected it (lower coherence on HeartMath = higher distraction on Muse, so the lines go in opposite directions, but the change happens at about the same time). The raw heart rate and breathing data collected via the other two sensors did not detect this loss of attention. This finding is more remarkable in view of the fact that the devices collect different biometrics, have different manufacturers, and presumably are not testing their results against each other during development.
  2. Validity of the biomarkers as measures of mindfulness also seems to have been supported. Reading from left to right, the devices mostly showed a progression from less relaxed/centered/coherent to more so, which matched both users’ subjective experiences. We did in fact feel that we were successful in meditating, with deeper states of mindfulness achieved as time went on.
  3. The measures’ sensitivity to change, on the other hand, varied across devices. The two sensors with proprietary algorithms did not respond in any dramatic way to an external disruption -- e.g., a loud noise or being lightly shaken while meditating. By contrast, the raw heart rate and breath rate data did. The third graph at the top of the page shows that my heart rate accelerated after being surprised, and in fact continued to climb until the end of the 5-minute monitoring period. This is a standard sympathetic nervous system (SNS) response, i.e. the fight-or-flight reflex. By contrast, the two sensors with proprietary algorithms agreed that I had a partial return to a mindful state, even though the SNS indicators remained elevated and I never returned to the deeper level of mindfulness seen on the graphs between minutes 2 and 3. Again, this fit with my experience.
What can we conclude from this very small-scale experiment? First, it appears that sensor data do in fact capture something meaningful about the mindfulness meditation experience. This is good news for anyone who uses these tools to meditate or relax. If the sensors didn't provide consistent or meaningful data they wouldn't be useful in efforts to train oneself on a mindfulness response. 

Second, there seems to be a level of value added by proprietary algorithms, which tracked more closely to our experience as users and agreed with each other better than with the raw heart rate or breathing rate data. In the final trial where we examined raw EEG data instead of the proprietary Muse output there was too much scatter to interpret the results, again suggesting that the type of smoothing and simplification offered by the proprietary algorithm is useful. One important caveat about the raw data is that both the Spire and Apple Watch provided much less frequent data points than the Muse or HeartMath sensors, which undoubtedly made them less sensitive to moment-by-moment changes in mindfulness. I would like to repeat this experiment with a Fitbit heart rate sensor, which collects data second-by-second and allows for easy user download (the Apple Watch data had to be manually copied off a smartphone screen). 

Finally, there is some suggestion that external disruptions aren't all that disruptive once one achieves a mindful state. In the graph at the top of the page, note that my mind actually started to wander before the surprising noise; the noise itself didn’t have much effect on the two algorithm-based measures. A surprise produces the expected SNS response of faster heart rate and breathing, but it may not actually interfere with mindfulness. If true, this would be consistent with something proposed by advocates of mindfulness, that a mindful response is about remaining centered or calm despite environmental stressors, and not about the avoidance of stress. In Jon Kabat-Zinn's words mindfulness involves "full catastrophe living," being OK with whatever might happen in life. If some sensors do in fact assess "coherence" aspects of mindfulness that are distinct from traditional measures of SNS activity, this might allow for interesting future studies that separate out the various conceptual aspects of mindfulness using physiological data.

Comments

Popular posts from this blog

Why Does Psychotherapy Work? Look to the Intuitive Mind for Answers

  Jerome Frank's 1961 book Persuasion and Healing  popularized the idea of "common factors" that explain the benefits of psychotherapy, building on ideas that were first articulated by Saul Rosenzweig in 1936 and again by Sol Garfield in 1957. Frank's book emphasized the importance of (a) the therapeutic relationship, (b) the therapist's ability to explain the client's problems, (c) the client's expectation of change, and (d) the use of healing rituals. Later theorists emphasized other factors like feedback and empathy that are sub-components of the therapeutic relationship, and that can be clearly differentiated from specific behavior-change techniques like cognitive restructuring or behavioral reinforcement . Additional aspects of therapy that are sometimes identified as common factors include the opportunity to confront difficult past experiences, the opportunity for a "corrective emotional experience" with the therapist, and the chance t

Ethical Improvement in the New Year

  Just after the first of the year is prime time for efforts to change our behavior, whether that's joining a gym, a "dry January" break from alcohol, or going on a diet. (See my previous post about New Year's resolutions for more health behavior examples). This year I'd like to consider ethical resolutions -- ways in which we try to change our behavior or upgrade our character to live more in line with our values.  Improving ethical behavior has been historically seen as the work of philosophers, or the church. But more recent psychological approaches have tried to explain morality using some of the same theories that are commonly used to understand health behaviors based on Narrative constructs like self-efficacy, intentions, and beliefs. Gerd Gigerenzer suggests that an economic model of " satisficing " might explain moral behavior based on limited information and the desire to achieve good-enough rather than optimal results. Others have used simula

Year in Review: 2023

Here’s my annual look back at the topics that captured my attention in 2023. Over the past year I taught several undergraduate mental health classes, which is not my usual gig, although it does fit with my clinical training. The Two Minds Blog took a turn away from health psychology as a result, and veered toward traditional mental health topics instead. I had posts on   mania   and   depression .  I wrote about   loneliness   as a risk for health problems, as well as   hopefulness   as a form of stress inoculation. I wrote about the “ common factors ” in psychotherapy, which help to improve people’s mental health by way of the intuitive mind (I was particularly happy with that one). I also shared findings from a recent study where my colleagues and I implemented a   burnout prevention   program for nursing students, and another new paper that looked at the incidence of mental and physical health problems among   back country search and rescue workers . Mental health has received more