Modeling dynamic brain responses of sustained pleasure and pain based on flexible basis functions

2518 

Abstract Submission 

Soo Ahn Lee^1,2,3, Tor Wager⁴, Choong-Wan Woo^1,2,3

¹Department of Biomedical Engineering, Sungkyunkwan University, Suwon-si, Korea, Republic of, ²Center for Neuroscience Imaging Research, Institute for Basic Science, Suwon-si, Korea, Republic of, ³Department of Intelligent Precision Healthcare Convergence, Sungkyunkwan University, Suwon-si, Korea, Republic of, ⁴Department of Psychological and Brain Sciences, Dartmouth College, Hanover, NH

Soo Ahn Lee  
Department of Biomedical Engineering, Sungkyunkwan University|Center for Neuroscience Imaging Research, Institute for Basic Science|Department of Intelligent Precision Healthcare Convergence, Sungkyunkwan University
Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of

Tor Wager, PhD  
Department of Psychological and Brain Sciences, Dartmouth College
Hanover, NH

Choong-Wan Woo  
Department of Biomedical Engineering, Sungkyunkwan University|Center for Neuroscience Imaging Research, Institute for Basic Science|Department of Intelligent Precision Healthcare Convergence, Sungkyunkwan University
Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of

Both sustained pleasure and pain induce dynamic affective brain states for longer time scales. Because these brain states slowly emerge and fluctuate over time, it is hard to specify the timings of the emergence and recovery of these state-related brain responses (Robinson et al. 2010). Here, we conducted a functional Magnetic Resonance Imaging (fMRI) experiment in which participants received the oral delivery of chocolate and capsaicin fluids. While experiencing multiple combinations of sustained pleasure and pain conditions, participants continuously rated their subjective pleasantness and unpleasantness. To model these brain responses more effectively, we used spline functions as flexible basis sets in the GLM analysis to identify encoding models of dynamic affective states.

We collected 14-minute fMRI data from 61 participants (28 females, mean of age = 22.74, SD of age = 2.66). We delivered capsaicin or chocolate fluids to participants' oral cavity through the MR-compatible fluid delivery system (Octaflow II, ALA Scientific Instruments Inc., Westbury, NY) and a custom-built MR-compatible mouthpiece. In the 'Capsaicin-Capsaicin (CC)', 'Capsaicin-Sweet (CS)', and 'Sweet-Sweet (SS)' conditions, we delivered capsaicin (Jinmifood, Inc.) or sweet chocolate (Hershey's, Inc.) fluids sequentially during each scan. The first and second deliveries occurred 90 seconds and 7 minutes after the start of the scan, respectively. In addition, there was 'Capsacin-water (C0)' condition, where we delivered capsaicin fluids only once 90 seconds after the start of the scan. During scans, we asked participants to continuously rate their moment-by-moment changes of subjective pleasantness or unpleasantness, based on the general Labeled Magnitude Scale (Bartoshuk et al. 2004). We conducted the spline-based univariate analysis that flexibly captures the fMRI signals related to pleasant and unpleasant states, obtaining the areas under the curve (AUCs) that indicate the overall activation magnitudes for each brain voxel.

We used the 265-region whole-brain parcellation combining Schaefer cortical atlas (Schaefer et al. 2018), subcortical and cerebellar regions from the Brainnetome atlas (Fan et al. 2016), and brainstem regions (Beissner et al. 2014; Roy et al. 2014). We examined brain response patterns based on the spline-based event regressors, toward initial sustained pain (i.e, the first capsaicin deliveries across CC, C0, and CS conditions) and initial sustained pleasure (i.e., the first sweet chocolate deliveries in SS conditions). There were significant AUCs in the dorsal posterior insula, cingulate cortex, somatosensory cortex, and ventromedial cortex for the initial capsaicin deliveries. There were significant AUCs in the amygdala, ventral anterior insula, and frontal parts of the medial prefrontal cortex for the initial chocolate deliveries. The results were consistent with the GLM results based on the canonical hemodynamic function.

We categorized the brain regions based on the a priori criteria about their response patterns, similar to our previous study (Lee et al. 2023). (1) We found 58 pain-responsive brain regions that showed significant AUCs to initial capsaicin and (2) 49 pleasure-responsive brain regions that showed significant AUCs to initial sweet chocolate. 22 brain regions overlapped that satisfied both (1) and (2), and we found (3) 13 brain regions of affective intensity (weak vs. strong) that showed the same signs of AUCs and (4) 9 brain regions of affective valence (pleasant vs. unpleasant) that showed opposite signs of AUCs. The 22 brain regions showed large spatial overlaps with the predictive regions of the multivariate encoding models of the affective intensity and valence (Lee et al. 2023), tracking the overall trajectories of our hypothetical response patterns.

This study characterized brain regions based on their response patterns toward sustained pleasure and pain, showing dynamic changes of brain responses over time.

Emotional Perception

Activation (eg. BOLD task-fMRI)

Univariate Modeling ²

Perception: Pain and Visceral ¹

Emotions

FUNCTIONAL MRI

Modeling

Pain

Perception

Univariate

Other - Pleasure