Self-prioritization in response selection: Computational evidence for biased evidence accumulation.

The self-prioritization effect (SPE) is well-documented for sensory representations in working memory (WM), yet its extension to response selection remains unclear. This study investigated how self-referential processing modulates response selection within WM using a novel rule-based paradigm. Participants first learned associations between social labels (self/friend) and visual features (shapes/colors), then maintained stimulus-response mappings that linked social labels to specific keypress responses and applied them to the following target stimulus. We introduced a voluntary-selection context where participants can freely choose between self- and friend-referential responses, alongside a traditional forced-selection context. Across three experiments controlling for perceptual and strategic confounds, results consistently revealed a robust SPE: participants exhibited significantly higher selection proportion and faster reaction times for self-referential responses. Hierarchical drift-diffusion model revealed the computational architecture of this self-bias. In the voluntary-selection context, the SPE was driven by an increased drift rate toward the self-referential response, rather than a group-level starting-point bias. In the forced-selection context, self-prioritization was characterized by a concurrent increase in both drift rate and decision threshold, suggesting that enhanced processing efficiency for self-related responses was accompanied by heightened response caution. These findings demonstrate that the SPE is not confined to perceptual content but fundamentally reconfigures the decision-making stage within WM. We propose that self-relevance acts as a salient contextual anchor that optimizes the activation and implementation of action-oriented representations. PUBLIC SIGNIFICANCE STATEMENT: This study reveals that individuals are naturally drawn to actions associated with their personal identity. When given the freedom to choose, people consistently prefer self-related actions and execute them more efficiently. Computational modeling shows that this advantage stems from faster evidence accumulation in the brain. Interestingly, this efficiency is balanced by increased caution: individuals set a higher evidence threshold for self-related decisions, likely to ensure accuracy for personally significant information. These findings explain why self-centered choices feel intuitive yet are handled with specific care, providing new insights into how self-reference shapes daily behavior.