Active visual search in non-stationary scenes: coping with temporal variability and uncertainty

Area of Science:

• Cognitive Neuroscience
• Human-Computer Interaction
• Signal Processing

Background:

• Studying neural processes in visual cognition often uses simplified, static environments.
• Natural visual environments are dynamic and non-stationary, posing challenges for current research methods.
• Understanding brain activity during active visual search in complex scenes is crucial for realistic applications.

Purpose of the Study:

• To investigate electroencephalography (EEG) correlates of visual recognition during active visual search in non-stationary scenes.
• To test the hypothesis that dynamic visual effects increase temporal uncertainty in cognition-related EEG activity.
• To develop and evaluate novel techniques for single-trial EEG detection and intention decoding.

Main Methods:

• Examined fixation-related EEG activity during an active visual search task.
• Utilized various stimulus appearance styles, including popping-up, fading-in, enlarging, and motion effects.
• Explored knowledge transfer methods to enhance EEG-based intention decoding across different experimental settings.

Main Results:

• Confirmed that dynamic visual content increases temporal uncertainty in EEG activity relative to fixation onset.
• Demonstrated that this temporal uncertainty challenges consistent decoding performance.
• Showcased promising EEG decoding performance using a knowledge transfer approach across settings.

Conclusions:

• The non-stationarity of visual scenes significantly impacts cognitive processes and ocular behavior during active search.
• The developed method for improving single-trial EEG detection enhances brain-computer interfacing capabilities.
• This research advances the application of brain-computer interfaces in human-computer interaction.