Data for this project was collected during 2021 as part of Ido Ben Artzi's Master's thesis. **Abstract:** To maintain accurate action-outcome associations, observers need to refrain from assigning value to outcome-irrelevant features of the environment. However, reinforcement learning studies have mostly ignored the role of attentional control processes on credit assignment. Here, we examined to what extent working memory, a system allowing filtering and blocking of irrelevant information in mind, predicts credit-assignment to outcome-irrelevant task features. 174 individuals completed working memory capacity assessment and a reinforcement learning task under varying working memory load conditions. The reinforcement learning task required individuals to choose cards to gain rewards, while only the cards’ visual features predicted reward, but not the response-keys used to report selection. We found a consistent tendency to assign value to the task’s response keys reflecting outcome-irrelevant learning at the group level. However, we also found substantial individual differences such that 55% of participants demonstrated this effect while the rest did not. Importantly, working memory capacity significantly mediated individual differences in outcome-irrelevant learning, showing that individuals with higher capacity demonstrated lower credit assignment to the outcome-irrelevant response keys. We discuss the influence of working memory on outcome-irrelevant learning in light of depleted cognitive resources needed for appropriate task representations.