Skipping a beat: heartbeat-evoked potentials reflect predictions during interoceptive-exteroceptive integration

Abstract

Several theories propose that emotions and self-awareness arise from the integration of internal and external signals and their respective precision-weighted expectations.Supporting these mechanisms, research indicates that the brain uses temporal cues from cardiac signals to predict auditory stimuli, and that these predictions and their prediction errors can be observed in the scalp heartbeat-evoked potential (HEP). We investigated the effect of precision modulations on these cross-modal predictive mechanisms, via attention and interoceptive ability. We presented auditory sequences at short (perceived synchronous) or long (perceived asynchronous) cardio-audio delays, with half of the trials including an omission. Participants attended to the cardio-audio synchronicity of the tones (internal attention) or the auditory stimuli alone (external attention). Comparing HEPs during omissions allowed for the observation of pure predictive signals, without contaminating auditory input.We observed an early effect of cardio-audio delay, reflecting a difference in heartbeat-driven expectations. We also observed a larger positivity to omissions of sounds perceived as synchronous than to omissions of sounds perceived as asynchronous when attending internally only, consistent with the role of attentional precision for enhancing predictions.These results provide support for attentionally-modulated cross-modal predictive coding, and suggest a potential tool for investigating its role in emotion and self-awareness.

Heartbeat-evoked potentials during interoceptive-exteroceptive integration are not consistent with precision-weighting

“…Attention is thought to increase the precision of the prediction errors of the attended sensory channel via synaptic gain control, enhancing model updating (Friston, 2009; Hohwy, 2012). Consistent with the role of attentional precision in modulating predictive mechanisms, previous research demonstrated attentional modulation of HEPs, supporting its interpretation as a precision-weighted prediction error signal (Banellis & Cruse, 2020; Mai et al, 2018; Montoya et al, 1993; Petzschner et al, 2019; Villena-González et al, 2017; Yuan et al, 2007).…”

“…A final sample of 34 participants were included for both parts of subsequent EEG analyses (Median age = 20 years, Range = 18-35 years). This sample size was chosen in advance, as it provides 95% power to detect the same effect size (Cohen’s d’ = 0.58) as the within-subjects interaction between attention and cardio-audio delay observed in our previous experiment (preregistered analysis: M=0.0349, SD=0.0598; alpha=.05; note that the effect size in the final published version of that study [Banellis & Cruse., 2020] was slightly larger [0.61] due to pre-processing changes suggested by peer reviewers; GPower, (Faul et al, 2007)).…”

“…As recommended by SPM12, we low-pass filtered the resulting HEP below 20Hz (again, with EEGLAB’s pop_neweegfilt). In deviation from our pre-registration, but following discussions with peer reviewers and investigation of similar decisions in previous studies of HEPs (Azzalini et al, 2019; Babo-Rebelo et al, 2016, 2019; Banellis & Cruse, 2020; Park et al, 2014; Petzschner et al, 2019), we chose not to apply any baseline correction to our data as cardiac activity is cyclical by nature and may therefore insert artefactual effects in post-R data. This decision also allows direct comparison with published results of Banellis and Cruse (2020).…”

“…first percentile of the R->sound intervals for the participant with the earliest perceived synchronous delay, thus before >99% of anticipated tones) and a second set of analyses relative to the onset of the omitted sound (part 1: 0-250ms relative to the most rated and least rated synchronous delay time; part 2: 95-138ms post-omission (i.e. significant attention and delay interaction from Banellis & Cruse., 2020)). This allows the investigation of cardio-audio expectation and unfulfilled expectation mechanisms separately.…”

“…For example, tones presented at short delays from the heartbeat (~250ms) are typically perceived as synchronous, while those presented at longer delays (~550ms) are typically perceived as asynchronous with the heart. Furthermore, when participants listen to sequences of such synchronous or asynchronous sounds, we previously observed an HEP effect in the period between the heartbeat and the expected sound, potentially reflecting cardio-audio integrated expectations (Banellis & Cruse, 2020). In support of attentional modulation of integrated predictive mechanisms, we also observed a larger positivity to unexpectedly omitted sounds in a sequence of cardio-audio synchronous sounds only when participants were attending to their heartbeat.…”