A simple Google search of the term ‘mindfulness’ provides countless search results detailing various tips on how to practice mindfulness in your daily life and its endless health benefits. Meditation and mindfulness seem to have become catch-all phrases for mental and physical well-being, associated with emotional regulation and everyday self-awareness. 

Practicing mindfulness is shown to have improved health benefits associated with blood pressure, cardiovascular health, and obesity. However, its advantages aren’t just limited to physical benefits — the ability to practice self-awareness and attentiveness through meditative practices has been linked to lowered stress levels and reduced symptoms for mental disorders.

Meditation has clear psychological advantages in helping individuals deal with increased stress levels, improve attention, and build resiliency. However, a problem often arises with researching these claims: the lack of quantitative data to back up these statements.

While studies have explored the effects of meditation and have found mindfulness to be somewhat effective in aiding emotional regulation, these studies often lack appropriate measures to determine the significance of their results. 

Recently, a new U of T-affiliated study published in the Frontiers journal highlighted the use of machine learning and functional magnetic resonance imaging (fMRI) technology to obtain quantifiable data that measures the mental states of individuals practicing meditation. It provides a promising approach to objectively assessing the effects of mindfulness on overall well-being.

The two types of attention

The lack of quantifiable data within mindfulness research makes it difficult to research what the meditating mind goes through — namely, how one’s attention is affected. The U of T-affiliated study focused on trying to identify one type of attention called internal attention in the brain through fMRI technology and machine learning. 

In an interview with The Varsity, Dr. Norman Farb, a professor of psychology and a co-author of the study, noted that internal attention states are all about “what’s happening inside your own mind or inside your own body. It’s experiences that wouldn’t be accessible to someone else.” This is contrasted with external attention, where you pay attention to stimuli from the outside world like sights and sounds.

The term ‘internal attention’ was first coined in a 2011 study from Yale University and Massachusetts Institute of Technology psychologists. They described it as the predominant mode of attention used to access information already stored in the brain, such as retrieving long-term memories or making abstract decisions, like what to eat for lunch. 

At the interface between internal and external attention is working memory, used for making decisions that involve immediate sensory information. For example, a student reading an essay will try to piece together the structure of the argument as they read.

Attention and mindfulness

Mindfulness practices often centre on evaluating the usefulness of thoughts and emotions. Farb explained that this kind of cognitive processing is actually an internal attention state because we are trying to access information already in the mind: our thoughts. 

“If the thought is wrong or unhealthy, or misconstruing a situation, then everything that comes after it is going to be wrong,” he said. “So that’s the source that you want to attend to, like what’s happening internally and how I perceive a situation.”

He further elaborated that meditation is “really a process of introspection.” 

“In the classic sense, I’m really looking internally at what’s playing out inside me to understand the building blocks of much more obvious responses that are full blown emotions or a full blown move into hostility or avoidance,” he said. This process of introspection supports the emotional regulation that is commonly the aim of practicing mindfulness.

The study suggests that being able to measure internal attention states would allow researchers to identify moments when participants lose focus on mindfulness tasks, and would provide crucial insight into the effectiveness of the meditation process for mindfulness practitioners of all experience levels.

Using machine learning to measure attention states

Being able to monitor these mental states would more objectively indicate whether meditational practices impact behaviour or alleviate stress, as mindfulness activities largely focus on emotional regulation and healthy self-reflection. 

The study employs the use of machine learning analysis and fMRI technology in order to predict and track patterns of activity within a participant’s mind based on pre-labelled data. This process of machine learning involves feeding a set of data into an algorithm with predetermined conditions based on the specific factor being measured.

In this study, researchers asked 16 participants to go through a set of meditation activities designed to create internal attention states. Their brain activity was measured using an fMRI machine. Afterward, an algorithm was trained on the data until it could detect the distinct patterns of neural activity associated with meditation.

The study found that significant neural patterns associated with different attention states were present within all participants. It recorded specific patterns associated with meditative breathing for 87.5 per cent of participants through this methodology, including novice participants. 

Once trained, the algorithm was able to accurately detect internal attention from the data 42 per cent of the time — which was considered significant because a random guess would be successful 20 per cent of the time. This shows that the choices the algorithm made were not due to chance, but rather a genuine ability to recognize meditation from brain activity data. 

As Farb explained, “we want to validate that when we do know the right answer, that [the algorithm is] getting the right answer above chance, [that] it actually is making informed decisions. And then, we can try to apply it to situations where we’re not so sure what’s happening.” 

While Farb cautioned that the relatively small number of study participants makes this result more of a proof of concept than a landmark breakthrough, the results of this study still provide an encouraging future for the use of machine learning in neuroscience.

The tested methods may be used to further understand mindfulness training and provide quantitative evidence that supports the effectiveness of mindfulness. Meditation may work for some mental health concerns; for others, medication or therapy might be more effective. Either way, an improved understanding of meditation’s effects will help us to prescribe the right course of action.

As Farb put it, “there isn’t a one-size-fits-all [approach to] mental health care, but understanding how each of these things works might make us better at guessing.”