A machine learning approach towards the differentiation between interoceptive and exteroceptive attention

Area of Science:

• Neuroscience
• Psychology
• Machine Learning

Background:

• Interoception, the awareness of internal bodily states, is crucial for emotion and wellbeing.
• Current measurement of interoceptive sensibility relies solely on self-report, lacking objective validation.
• Objective markers are needed to advance mental health research concerning interoceptive abilities.

Purpose of the Study:

• To develop and validate machine learning classifiers for distinguishing interoceptive from exteroceptive attention using functional magnetic resonance imaging (fMRI).
• To assess the potential of these classifiers in identifying periods of sustained interoceptive focus.
• To explore the impact of interoceptive training on the ability to sustain interoceptive attention.

Main Methods:

• Utilized fMRI data from a randomized controlled trial of interoceptive training (N=44 scans).
• Employed a neuroimaging paradigm manipulating attention (breath vs. visual) and reporting demands (active vs. passive).
• Applied machine learning algorithms for classification of attention states and analysis of sustained attention tasks.

Main Results:

• Machine learning accurately classified interoceptive vs. exteroceptive attention (within-session ~80%, out-of-sample ~70%).
• Classifiers identified periods of active engagement during sustained interoceptive attention tasks.
• Interoceptive training improved participants' ability to sustain interoceptive attention.

Conclusions:

• Neural patterns reliably differentiate interoceptive and exteroceptive attention.
• Developed classifiers show promise for objectively demarcating interoceptive focus.
• Findings support the development of objective markers for interoceptive sensibility in mental health.