Distinct Neural Representations of Content and Ordinal Structure in Auditory Sequence Memory

Area of Science:

• Cognitive Neuroscience
• Auditory Perception
• Working Memory Research

Background:

• Working memory (WM) is crucial for retaining sequential information, involving both content (e.g., tone frequency) and structure (e.g., ordinal position).
• The neural mechanisms underlying the maintenance of content and structure in auditory WM are not well understood.
• Existing research lacks clarity on how these two information types are represented and interact during WM processes.

Purpose of the Study:

• To investigate the distinct neural representations and coding characteristics of content and structure in auditory working memory.
• To determine if content and structure are dissociatively stored and maintained throughout the WM process.
• To explore the relationship between neural reactivation of stored information and subsequent working memory behavior.

Main Methods:

• Employed electroencephalography (EEG) in two studies involving human participants.
• Utilized a transient perturbation approach to probe the "activity-silent" WM state.
• Applied decoding techniques and time-resolved representational dissimilarity analysis to analyze reactivated WM information.

Main Results:

• Demonstrated that auditory WM stores content (frequency) and structure (ordinal position) in a dissociative manner with distinct neural codes.
• Showed that structural retrocues reactivated structure but not content, while white noise triggered content but not structure.
• Revealed stable structure representation versus dynamic content transformation during memory retention, with content reactivation correlating with behavioral performance.

Conclusions:

• Auditory working memory maintains content and structure through distinct, dissociable neural representations.
• This dissociation facilitates efficient memory formation by allowing stable structure to generalize across new auditory contents.
• The findings provide insights into accessing silently stored WM information and support a dual-code model for auditory sequence memory.