High-level cognition is supported by information-rich but compressible brain activity patterns

Area of Science:

• Neuroscience
• Cognitive Science
• Information Theory

Background:

• The brain must balance signal richness and robustness for efficient information processing.
• Typically, richer signals are less robust, posing a challenge for neural systems.
• Understanding how the brain manages this trade-off under varying cognitive demands is crucial.

Purpose of the Study:

• To investigate how neural signal informativeness and compressibility change with cognitive load.
• To determine if brain networks adapt to maintain both signal richness and robustness.
• To explore the relationship between cognitive engagement and neural data properties.

Main Methods:

• Applied dimensionality reduction and pattern classification to functional neuroimaging data.
• Analyzed neural data from participants listening to intact stories, scrambled stories, and during rest.
• Defined informativeness by peak decoding accuracy and compressibility by features needed for threshold accuracy.

Main Results:

• Peak decoding accuracy was highest during intact story listening.
• Fewer features were required to achieve high decoding accuracy for intact stories.
• Neural activity during high-engagement tasks is more informative and compressible.

Conclusions:

• Brain networks dynamically reconfigure based on cognitive demands.
• Higher-order cognition and engagement are associated with more informative and compressible neural patterns.
• This flexibility allows the brain to efficiently process complex information.