A Meta-analysis on the Neural Correlates of Unconscious Visual Selective Attention

Sammanfattning

The relationship between attention and consciousness is a central question in cognitive neuroscience, with growing evidence that these processes can be dissociated. The present study conducted a systematic review and meta-analysis of electroencephalography (EEG) studies on unconscious visual selective attention, defined as attentional modulation occurring without conscious awareness of either the attentional cue or the target. A search across Scopus, Web of Science, and APA PsycInfo identified 25 EEG studies, of which 21 contributed quantifiable effect sizes. A three-level multilevel random-effects model was used to synthesize 55 effect sizes (Hedges’ g) from 20 studies in the main analysis and 1 additional study (Baier & Ansorge, 2020) in the sensitivity analysis only. The overall pooled effect was g = 0.60, 95% CI [0.50, 0.70], indicating a medium-sized neural signature of unconscious attentional selection. Subgroup analyses revealed broadly consistent effects across endogenous (g = 0.67) and exogenous (g = 0.55) attention, as well as across unconscious cue (g = 0.60) and unconscious target (g = 0.62) paradigms, with no significant moderation by attention type or unconscious element. Sensitivity analyses assigning g = 0 to non-significant results with unreported statistics confirmed these findings (g = 0.51). These results provide converging neural evidence that both top-down and bottom-up attentional mechanisms can operate without conscious awareness, supporting theoretical accounts that propose a partial dissociation between attention and consciousness, particularly the Recurrent Processing Theory. A funnel plot and formal tests suggested some potential publication bias, though sensitivity analyses supported the robustness of the pooled estimate.