Hearing VS. Listening: Attention Changes the Neural Representations of Auditory Percepts

Abstract

Making sense of acoustic environments is a challenging task. At any moment, the signals from distinct auditory sources arrive in the ear simultaneously, forming an acoustic mixture. The brain must represent distinct auditory objects in this complex scene and prioritize processing of relevant stimuli while maintaining the capability to react quickly to unexpected events. The present studies explore neural representations of temporal modulations and the effects of attention on these representations. Temporal modulation plays a significant role in speech perception and auditory scene analysis. To uncover how temporal modulations are processed and represented is potentially of great importance for our general understanding of the auditory system.

Neural representations of compound modulations were investigated by magnetoencephalography (MEG). Interaction components are generated by near rather than distant modulation rhythms, suggesting band-limited modulation filter banks operating in the central stage of the auditory system. Furthermore, the slowest detectable neural oscillation in the auditory cortex corresponds to the perceived oscillation of the auditory percept.

Interactions between stimulus-evoked and goal-related neural responses were investigated in simultaneous behavioral-neurophysiological studies, in which we manipulate subjects' attention to different components of an auditory scene. Our experimental results reveal that attention to the target correlates with a sustained increase in the neural target representation, beyond well-known transient effects. The enhancement of power and phase coherence presumably reflects increased local and global synchronizations in the brain. Furthermore, the target's perceptual detectability improves over time (several seconds), correlating strongly with the target representation's neural buildup. The change in cortical representations can be reversed in a short time-scale (several minutes) by various behavioral goals.

These aforementioned results demonstrate that the neural representation of the percept is encoded using the feature-driven mechanisms of sensory cortex, but shaped in a sustained manner via attention-driven projections from higher-level areas. This adaptive neural representations occur on multiple time scales (seconds vs. minutes) and multiple spatial scales (local vs. global synchronization). Such multiple resolutions of adaptation may underlie general mechanisms of scene organization in any sensory modality and may contribute to our highly adaptive behaviors.