On the relative importance of attention and response selection processes for multi-component behavior - Evidence from...

Area of Science:

• Neuroscience
• Cognitive Science
• Artificial Intelligence

Background:

• Goal-directed actions often involve sequential steps, requiring complex cognitive processing.
• Understanding the neural basis of cognitive load in multi-component behavior is crucial.
• Predicting task demands from brain activity at an individual level remains challenging.

Purpose of the Study:

• To investigate the predictability of high vs. low cognitive demands in multi-component behavior using single-trial EEG.
• To identify specific neural processes and brain regions involved in predicting task demands.
• To determine the relative importance of sensory integration and attentional processes in decoding cognitive load.

Main Methods:

• Applied deep learning models to single-trial electroencephalography (EEG) data from 239 individuals.
• Utilized explainable artificial intelligence (XAI) for model interpretability.
• Employed temporal EEG signal decomposition and source localization techniques.

Main Results:

• Attentional selection and sensory integration in sensory association cortices predicted task demands with ~86% accuracy.
• Rule-based response selection and translation in parietal cortices predicted demands with ~70% accuracy, but only without sensory integration data.
• Sensory integration processes were critical for decoding demands related to response selection capacity.

Conclusions:

• Single-trial EEG combined with deep learning can predict cognitive demands in multi-component behavior at the individual level.
• Sensory integration processes play a key role in limiting multi-component behavior under high cognitive load.
• Attentional processes are vital for effective performance in complex, goal-directed tasks.