Neurocomputational Dynamics of Pain Value-Based Decision Making

763 

Abstract Submission 

Léane Beaulieu-Laliberté¹, Mathieu Roy², Michel-Pierre Coll¹

¹Laval University, Québec, Quebec, ²McGill University, Montreal, Quebec

Léane Beaulieu-Laliberté  
Laval University
Québec, Quebec

Mathieu Roy, Dr.  
McGill University
Montreal, Quebec

Michel-Pierre Coll  
Laval University
Québec, Quebec

While pain is generally to be avoided, there are times when it is necessary to endure pain to achieve a goal, a reward or a more valuable outcome. Assigning too high a value to pain can lead to avoiding beneficial experiences (Crombez et al., 2012), while completely ignoring pain lead to in reckless behaviours. In a recent study, we used functional magnetic resonance imaging to understand how the brain independently represents the value of potential future pain in an economic context (Coll et al., 2022). However, the techniques used in this study don't allow for determining how the representation of pain evolves over time and is dynamically compared to potential rewards. Identifying the dynamic aspects of brain activity during decision-making in the presence of pain is crucial for a better understanding of the mechanisms involved in this process. Here, we aimed to investigate how the brain dynamically represents the prospect of pain in economic contexts and how this representation is integrated with external rewards to guide behaviour.

Fifty healthy adults (27 females, mean age = 24.72 +/- 4.14 ) took part in an EEG experiment (64 channels, Brain Vision Acticap). In the first phase, participants passively observed cues indicating that they would receive one of five levels of pain (from threshold to tolerance) or one of five monetary rewards (1-5 $CAD) before receiving a painful electric shock or feedback indicating monetary gain. This passive phase was used to measure brain activity during the anticipation of pain and monetary reward in isolation (Figure 1A). In the second phase (Figure 1B), participants had to decide at each trial to accept or reject offers combining different levels of pain and money. If they accepted the offer, they immediately received a painful stimulation and the amount of money was added to their potential earnings for the task. If they refused the offer, they did not receive any pain or opportunity to receive the money.

We preprocessed EEG data using a standard approach, including bandpass filtering, removal of bad channels, artifactual independent components and trials with high amplitude jumps. We first performed event-related potential analyses to confirm differential responses to cues indicating different levels of upcoming pain and money in the passive phase. We then followed with decoding analyses in sensor space that were used to create classifiers capable of discriminating between the anticipation of pain and money during the passive phase for each participant. We then applied these classifiers to EEG activity during decisions to attempt to predict whether participants would accept or reject offers.

Behavioural results replicate previous studies using a similar approach (Coll et al., 2022, Slimani et al., 2022) and show that participants' decisions and deliberation times were significantly impacted by both pain and money levels offered (Figure 1C). Event-related potential analyses confirmed that cues indicating different levels of pain and money led to a differential response emerging around 400 ms after cue onset (Figure 2A and 2B). EEG decoding analyses in sensor space showed that classifiers trained to discriminate cues indicating upcoming pain or money could accurately classify pain and money cues in the passive phase (accuracy: 0.69 +/- 0.1, p < 0.001). These classifiers were also able to predict participants' decisions to approach or avoid pain in exchange for a reward (balanced accuracy: 0.54 +/- 0.09, p = 0.03), suggesting that the representation of pain and rewards are involved in the decision to approach and avoid pain.

Our results show that the representation of pain and reward anticipation can be decoded from EEG data and used to predict choices during approach/avoidance decisions. This study will contribute to our understanding of the brain mechanisms involved in decision-making about potential future pain and to our comprehension of disorders marked by inadequate avoidance of pain.

Reward and Punishment ¹

Decision Making ²

Computational Neuroscience

Electroencephaolography (EEG)

Pain

Other - decision making