Higher-dimensional neural representations predict better episodic memory.

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Cognitive Psychology

Background:

• Episodic memory allows vivid retrieval of experiences, but its neural underpinnings remain unclear.
• Understanding the brain's representational format is crucial for explaining memory capacity.

Purpose of the Study:

• To investigate the neural representations supporting episodic memory.
• To determine the role of representational dimensionality (RD) in face-name associative memory.

Main Methods:

• Utilized a large fMRI dataset (n=468) involving face-name associative memory tasks.
• Employed principal component analysis to quantify neural representational dimensionality (RD).
• Analyzed the influence of frontoparietal activity on neural representations.

Main Results:

• The human brain exhibits high-dimensional representations of faces, organized hierarchically.
• Increased RD positively correlated with subsequent memory performance, specifically for episodic memory.
• Frontoparietal activity was observed to suppress low-dimensional fluctuations, thereby increasing RD.
• RD's contribution to episodic memory was complementary to item-specific pattern similarity.

Conclusions:

• Neural representational dimensionality (RD) plays a key mechanistic role in supporting accurate episodic memory.
• High-dimensional neural representations are essential for effective encoding and retrieval of experienced events.
• Frontoparietal networks actively shape neural representations to enhance memory function.