Active tactile discrimination is coupled with and modulated by the cardiac cycle

Abstract

Perception and cognition are modulated by the phase of the cardiac signal in which the stimuli are presented. This has been shown by locking the presentation of stimuli to distinct cardiac phases. However, in everyday life sensory information is not presented in this passive and phase-locked manner, instead we actively move and control our sensors to perceive the world. Whether active sensing is coupled and modulated with the cardiac cycle remains largely unknown. Here, we recorded the electrocardiograms of human participants while they actively performed a tactile grating orientation task. We show that the duration of subjects' touch varied as a function of the cardiac phase in which they initiated it. Touches initiated in the systole phase were held for longer periods of time than touches initiated in the diastole phase. This effect was most pronounced when elongating the duration of the touches to sense the most difficult gratings. Conversely, while touches in the control condition were coupled to the cardiac cycle, their length did not vary as a function of the phase in which these were initiated. Our results reveal that we actively spend more time sensing during systole periods, the cardiac phase associated with lower perceptual sensitivity (vs. diastole). In line with interoceptive inference accounts, these results indicate that we actively adjust the acquisition of sense data to our internal bodily cycles.

Keywords: active inference; active sensing; cardiac cycle; human; interoception; neuroscience; predictive coding; tactile discrimination; touch.

Figure 1.. Task and schematic illustration of one trial with cardiac and tactile events.

(A) The participants’ task was to touch gratings (one per trial) with the index finger to determine their orientation (vertical or horizontal, probability of each orientation = 0.5). There were seven levels of difficulty according to the gratings’ widths. Also, there was a movement control condition with a flat stimulus and no orientation judgement required. Participants were free to start, hold, and end the touch when they felt like, and their electrocardiograms (ECGs) were co-registered. (B) Upper panel: for each touch, we computed the start, stationary hold, and end of touch in degrees relative to the entailing heartbeat: for example, for an R–R interval of time t_R, where the touch started at time t_E, we calculated t_E/t_R × 360. Then, each type of event was subjected to circular averaging. Lower panel: we also computed the proportion and duration of touches starting and ending in the systole and diastole phases of the cardiac cycle. We used the systole length of each cardiac cycle to define a time window of equal length (to equate events’ probability in diastole; see e.g., Al et al., 2020), which we located at the end of the cardiac cycle.

Figure 2.. Duration, correct responses, and touches as a function of the whole cardiac cycle.

(A) The mean proportion of correct responses changed as a function of task difficulty. Performance in the first four levels of difficulty was similar and significantly greater than in the remaining levels of difficulty (see main text). (B) The duration of participants’ stationary holds for the first four levels of difficulty was similar and significantly shorter than in the remaining levels of difficulty. The duration of stationary holds in the control movement condition (flat stimulus) was significantly shorter than those of the gratings condition (see main text); error bars show standard deviations (SDs). (C) Circular means showing the distribution of the start, holding, and end of touches for the grating stimuli across the cardiac cycle (interval between two R-peaks at 0/360°). Points depict subjects’ mean degrees, the central arrows in v-shape point to the overall mean degree and its length indicates the coherence of individual means. The grey outer lines depict the circular density of individual means. Overall, the start, mean point of stationary hold, and end of touches occurred at 151°, 202°, and 278°; n = 46 (D) Touch data for the control movement condition as analysed and depicted in panel C. Here, the start, mean point of stationary hold, and end of touches occurred at 101°, 168°, and 243°; n = 45.

Figure 3.. Proportion of touches, correct responses, and duration as a function of the cardiac phase.

(A) In the grating and control movement conditions, the proportion of touches starting in systole was greater than that of touches ending in systole. Also, the proportion of touches ending in systole was significantly smaller than the chance level (0.5, depicted with a dashed line); see main text. Hence, more touches ended in diastole. (B) In the gratings condition, the proportion of correct responses was similar regardless the touches were started or ended in systole or diastole. (C) In the grating condition, the duration of touches was greater when participants started to touch in systole vs. diastole (M_diastole = 1143, M_systole = 1093). (D) Conversely, for the control condition, the duration of touches was similar when participants started to touch during systole and diastole (M_diastole = 708, M_systole = 717). Right panels show the difference between the duration of touches (stationary holds) initiated in diastole and systole, that is, for each participant, the holding time of touches started in diastole minus touches started in systole. Error bars show SDs; n = 46 gratings condition, n = 45 control movement condition; p-values adjusted using the Holm–Bonferroni method.

Figure 4.. Duration and variability of touches initiated in each cardiac phase.

(A) Across all levels of difficulty (1 easiest, 7 most difficult), the duration of touches when initiated in each cardiac phase. (B) Heatmap showing all participants ranked as a function of touch variability in the whole experiment. Each row shows one participant and each column one level of difficulty (e.g., participants at the top varied the least the duration of their touches). The duration of touches changed with task difficulty. (C) Differences between the duration of touches started in diastole and systole (Dia. − Syst.) across all levels of difficulty. Each dot represents one participant, and the colouration denotes their variability in the whole experiment (e.g., participants who varied the least are shown as dark blue dots). (D) Data from panel C are depicted as a heatmap with participants ranked as a function of touch variability in the whole experiment. Participants who varied the duration of their touches often spent more time touching in systole (vs. diastole) in the higher levels of difficulty (Task difficulty × Cardiac phase × Touch variability: p = 0.002). Error bars depict standard errors; n = 46.

Figure 5.. Interbeat interval (IBI) before, during, and after the touch of the stimuli.

(A) For the gratings condition, the IBI of the heartbeat where the touch was initiated (0 on the x-axis) was significantly longer than the preceding and succeeding heartbeats (p < 0.0001). Right panels: with similar statistical effects, this difference was driven by the elongation of the diastole phase of the cardiac cycle; n = 46. (B) For control movement condition, the IBI of the heartbeat where the touch was initiated was significantly longer than in the two preceding heartbeats (p < 0.015). Right panels: with similar statistical effects, these differences were driven by the elongation of the diastole phase of the cardiac cycle; n = 45. All p-values adjusted using the Holm–Bonferroni method. The box plots depict the interquartile range (IQR). Given the number of Post Hoc comparissons, see main text and Supplementary file 1.