Predicting Attentional Focus: Heartbeat-Evoked Responses and Brain Dynamics During Interoceptive and Exteroceptive Information Processing

Abstract

A coherent representation of the sense of self hinges on defining the limits between oneself and the outside. Therefore, the ability to recognize external and internal signals could yield essential information on the level of awareness of individuals, which has major implications in the context of uncommunicative patients’ diagnoses. Attention plays a significant role in shaping our consciousness content and perception by increasing the probability of becoming aware and, or, better encode a selection of the incoming inner or outer sensory world. In this study, we designed a task to engage interoceptive and exteroceptive attention by orienting healthy participants to their heartbeats or to salient auditory stimuli and measured EEG, ECG, and respiration, while the effects of attention on passive encoding were probed using concealed noise repetitions. We investigated whether brain dynamics and evoked brain responses accurately distinguished interoceptive from exteroceptive covert attention at the subject level using AdaBoost classifiers with decision trees as base estimators. An overall gain in auditory processing during exteroceptive attention was observed, as indexed by an increased cortical response to target sounds and a better encoding of noise repetitions. Interoceptive attention positively modulated the heart-evoked potential (HEP), and the HEP features successfully classified the attentional state of 17 out of 20 participants. Exteroceptive attention was characterized by an overall flattening of the power spectrum across the 1-30 Hz frequency range accompanied by an increase in the bandwidth of the beta power peak, whereas during interoceptive attention there was a decrease in complexity and an increase in frontal connectivity and oscillations in the theta range. Subject-level classifiers based on Kolmogorov complexity, permutation entropy, and weighted symbolic mutual information features showed comparable accuracy and exhibited a synergic behavior together with the HEP features. Power features demonstrated exceptional performance, effectively classifying the attentional state of all participants. Our findings demonstrate that directing attention to bodily rhythms and to the external world elicits distinct brain dynamics that can be employed to track covert attention at the individual level. Importantly, the brain markers studied in this work could provide multiple layers to explore information processing and awareness in uncommunicative patients.