Direct Two-Dimensional Access to the Spatial Location of Covert Attention in Macaque Prefrontal Cortex

Area of Science:

• Neuroscience
• Cognitive Science
• Machine Learning

Background:

• Direct neural decoding has enabled control of neuroprosthetics for motor functions.
• Accessing higher cognitive functions like covert attention directly from brain activity remains a challenge.
• Covert attention, internal cognitive focus, is typically inferred indirectly from behavior.

Purpose of the Study:

• To develop a method for direct, real-time decoding of covert attention from prefrontal cortical activity.
• To investigate the relationship between decoded attention and overt behavioral performance.
• To identify neural signatures associated with attentional failures.

Main Methods:

• Applied machine learning decoding methods to analyze ongoing prefrontal cortical activity in monkeys.
• Utilized a cued target-detection task to assess the predictive power of decoded attention.
• Examined neural population states, specifically noise correlations, during attentional lapses.

Main Results:

• Achieved direct two-dimensional, real-time decoding of covert attention location.
• Decoded attention accurately predicted behavioral responses, with detection probability increasing as decoded attention neared the stimulus.
• Identified high noise correlations in prefrontal cortical populations preceding attentional failures, suggesting a link to lapses.
• Observed misencoding of attention and cue locations during error trials.

Conclusions:

• Real-time decoding of covert attention from prefrontal cortex activity is feasible.
• High noise correlations in neural populations may represent a neural basis for attentional lapses.
• This work opens new avenues for studying the neural mechanisms of attention and developing neurofeedback interventions for attention enhancement or remediation.