Major depressive disorder (MDD) represents a prevalent mental health condition characterized by prominent attentional biases, particularly toward negative stimuli. While extensive research has established the significance of negative attentional bias in depression, critical gaps remain in understanding the temporal dynamics and valence-specificity of these biases. This study employed eye-tracking technology to systematically examine the attentional processing of emotional faces (happy, fearful, sad) in MDD patients (n = 61) versus healthy controls (HC, n = 47), assessing both the initial orientation (initial gaze preference) and sustained attention (first dwell time). Key findings revealed the following: (1) while both groups showed an initial vigilance toward threatening faces (fearful/sad), only MDD patients displayed an additional attentional capture by happy faces; (2) a significant emotion main effect (F (2, 216) = 10.19, p < 0.001) indicated a stronger initial orientation to fearful versus happy faces, with Bayesian analyses (BF < 0.3) confirming the absence of group differences; and (3) no group disparities emerged in sustained attentional maintenance (all ps > 0.05). These results challenge conventional negativity-focused models by demonstrating valence-specific early-stage abnormalities in MDD, suggesting that depressive attentional dysfunction may be most pronounced during initial automatic processing rather than later strategic stages. The findings advance the theoretical understanding of attentional bias in depression while highlighting the need for stage-specific intervention approaches.

The prevalence of sleep disorders has been rising over recent decades. Sleep disorders are frequently associated with disruptions in resting-state functional brain networks. Despite these associations, the direction and genetic basis of the relationships between brain functional networks and sleep disorders remain unclear. Utilizing bidirectional two-sample Mendelian randomization (MR), we examined bidirectional associations between 191 resting-state functional magnetic resonance imaging (rsfMRI) phenotypes (n = 34,691 individuals) and 8 sleep disorder phenotypes (N = 307,843–410,385 individuals). Forward MR identified 14 rsfMRI traits that showed significant associations with sleep disorder phenotypes. Notably, the inverse variance-weighted method showed that increased amplitude in the precuneus or occipital regions within the default mode or central executive network was associated with 13.2% higher odds of sleep disorders (combined; odds ratio = 1.132, 95% confidence interval: 1.064–1.204, P = 8.30 × 10 ⁻⁵ ). In addition, reverse MR identified 6 sleep disorder phenotypes with significant associations involving 18 rsfMRI traits. For example, higher genetic liability to sleep disorders (combined) was associated with a 14.2% lower strength of functional connectivity between precentral, frontal, or supplementary motor and frontal networks (odds ratio = 0.858, 95% confidence interval: 0.795–0.927, P = 9.62 × 10 ⁻⁵ ). This study identified 32 bidirectional MR associations between sleep disorders and brain functional networks, providing candidate neural circuits for future mechanistic and interventional studies of sleep disorders.