MEG Based Precision-Gain Index (PGI) Reveals Dual-task Costs in Vision and Audition

Le résumé fourni par la source

Perception allows humans to adaptively navigate their environment by integrating sensory signals into coherent representations and it depends on how precisely the brain weights prediction errors from sensory inputs. However, when multiple senses compete, it remains unclear how this precision is distributed and whether it predicts behavioral efficiency under attentional load. To further explore this question, we recruited thirty adults who performed auditory–visual oddball tasks during magnetoencephalography (MEG) under single-attention (‘pure load’) and dual-stream (‘dual load’) conditions. Behavioral efficiency was quantified using the inverse-efficiency score (IES), with modality-specific loss under dual load defined as the increase in IES from single- to dual-stream blocks. To capture neural precision, we derived a modality-specific precision-gain index (PGI) from the difference wave, computed as the normalized contrast between attended and ignored deviant–standard responses. At the group level, dual load impaired behavioral efficiency, with a larger cost for vision than for audition. In parallel, PGI increased under dual load for both modalities (vision Δ=+0.044, p=.0075; audition Δ=+0.033, p=.014). Finally, we examined whether PGI in the 100–200 ms window could predict these modality-specific efficiency losses. In vision, higher PGI predicted larger efficiency losses under dual load (r=.45, p=.0144). In contrast, in audition, a greater ΔPGI (dual minus pure) predicted smaller efficiency losses (r=−.44, p=.0149). Thus, within ~200 ms, PGI captures modality-specific, load-dependent precision and forecasts how well participants sustain efficiency under dual-task demands.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.