Objective: The objective of this study was to investigate the role of the high-frequency cochlear dysfunction in the cognitive-ear link.Methods: Seventy-four presbycusis patients (PC group) and seventy-one age-, sex-, and education-level matched normal hearing controls (NH group) were recruited in this study. Participants underwent a battery of cognitive tests estimated by Montreal Cognitive Assessment (MoCA), Stroop Color-Word Interference Test (Stroop), Symbol Digit Modalities Test (SDMT), Auditory Verbal Learning Test (AVLT), and Trail-Making Test (TMT-A and B), as well as auditory tests including distortion product otoacoustic emission (DPOAE), pure tone (PT) thresholds, and speech reception thresholds (SRT). Data were analyzed using the factor analysis, partial correlation analysis, multiple linear regression models, and mediation models.Results: Distortion product otoacoustic emission detection amplitudes and PT thresholds performed worse gradually from low to high frequencies in both the NH and PC groups. High-frequency DPOAE (H-DPOAE) was significantly correlated with cognitive domains in the PC group (AVLT: r = 0.30, p = 0.04; SDMT: r = 0.36, p = 0.01; Stroop: r = –0.32, p = 0.03; TMT-A: r = –0.40, p = 0.005; TMT-B: r = –0.34, p = 0.02). Multiple linear regression models showed that H-DPOAE predicted cognitive impairment effectively for aspects of memory (R2 = 0.27, 95% CI, 0.03 to 1.55), attention (R2 = 0.32, 95% CI, –6.18 to –0.40), processing speed (R2 = 0.37, 95% CI, 0.20 to 1.64), and executive function (TMT-A: R2 = 0.34, 95% CI, –5.52 to 1.03; TMT-B: R2 = 0.29, 95% CI, –11.30 to –1.12). H-DPOAE directly affected cognition and fully mediated the relationship between pure tone average (PTA)/SRT and cognitive test scores, excluding MoCA.Conclusion: This study has demonstrated that the high-frequency cochlear amplifier dysfunction has a direct predictive effect on the cognitive decline and makes a large contribution to the cognitive-ear link.