Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neurocognitive Signatures of Reinforcement Learning Proficiency: Behavioral and Electrophysiological Insights
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
This poster is part of the Sandbox Series.
Muhammad Adeel Khan1, Jueyao Lin1, Jiayi Lu1, Caicai Zhang1; 1POLYU, HK
Individual differences in reinforcement learning (RL) proficiency may have important implications for second language (L2) acquisition, particularly in adults, where successful learning often depends on efficiently processing corrective feedback and adapting linguistic representations (Ellis, 2006). However, the neurocognitive mechanisms that underlie individual differences in RL proficiency—and their potential impact on feedback-driven language learning—remain poorly understood. Therefore, this study presents preliminary findings from an ongoing project using an RL task to examine behavioural and neural signatures of RL proficiency, focusing on how performance aligns with established neural markers of feedback processing from prior studies (Tobias et al., 2014; Eisma et al., 2021; Giustiniani et al., 2020; Cavanagh & Frank, 2014). More importantly, the established paradigm will lay the groundwork for future analyses linking RL mechanisms to L2 acquisition within the same cohort, whose language data are currently being collected. Forty young adults participated in the experiment. In the RL task, participants stopped a moving dot (90°/s) at a target location on a circle, with each point linked to a certain probability of reward (+1) or penalty (-1). Reward probabilities were linearly distributed, with the highest probability at the reward centre (80%) and the lowest (20%) at the opposite end. Each participant completed 120 trials, with feedback displayed on the monitor after a 1-second delay. Participants were divided into two RL proficiency groups based on learning trajectories, with positive curves indicating high proficiency and negative curves indicating low proficiency. We then examined whether RL proficiency modulated neural responses, focusing on event-related potentials (ERPs) and oscillatory activity time-locked to the feedback onset. A significant interaction between groups and outcomes (reward vs. penalty) was observed in feedback-related negativity (FRN) amplitude (p = 0.0143), with the low RL proficiency group showing greater FRN amplitude differences between the two outcomes. Time-frequency analysis of frontal theta power also revealed a significant interaction between groups and outcomes (p = 0.045138), with the low RL proficiency group again exhibiting greater frontal theta power differences between the outcomes. Interestingly, both groups showed enhanced P300 amplitude in response to rewards compared to penalties (p < 0.00001) during feedback processing. FRN reflects early, transient feedback processing. The low proficiency RL group exhibited hyper-reactivity to penalties and overfocused on negative feedback. In contrast, the high proficiency RL group showed balanced feedback processing and faster feedback-based learning. Frontal theta power indicated increased cognitive control demand in the low RL proficiency group, while the high RL proficiency group showed more efficient feedback processing. The enhanced P300 amplitude to rewards suggests increased motivation and attentional focus on positive outcomes during feedback processing. These findings suggest that RL proficiency is reflected in distinct neurocognitive profiles. By establishing an RL paradigm sensitive to individual neural differences, this work lays a foundation for future investigations linking feedback-based learning mechanisms to second language acquisition. While analyses directly connecting RL proficiency to L2 outcomes are pending, the current findings highlight the importance of efficient feedback processing and adaptive learning strategies—mechanisms likely to be crucial in language learning contexts.
Topic Areas: Methods, Language Development/Acquisition