Decoding load or selection in visuospatial working memory?

Area of Science:

• Cognitive Neuroscience
• Psychology

Background:

• Flexible updating of information in Visual Working Memory (VWM) is essential due to its limited capacity.
• Previous studies indicate that removing irrelevant information from VWM is a time-consuming process.

Purpose of the Study:

• To investigate the temporal dynamics of information removal from VWM.
• To track the reduction in cognitive load using behavioral and neurophysiological measures.
• To examine the effectiveness of Multivariate Pattern Analysis (MVPA) in decoding VWM load.

Main Methods:

• Participants performed a visuospatial retro-cue task with manipulated Cue-Target Intervals (CTI).
• Behavioral performance was assessed to determine the time course of retro-cue utilization.
• Electroencephalography (EEG) data were collected and analyzed using MVPA to decode VWM load.

Main Results:

• Behavioral data indicated that approximately 500 milliseconds are needed to fully benefit from retro-cues.
• MVPA successfully decoded VWM load shortly after encoding, replicating prior findings.
• Post-retro-cue, MVPA classifiers appeared more sensitive to attentional selection than to VWM load itself.
• Potential confounds like visual cue variations and eye movements were identified as influencing decoding accuracy.

Conclusions:

• The process of updating information in VWM, specifically the removal of irrelevant items, unfolds over approximately half a second.
• Caution is advised when employing MVPA for decoding VWM load, as classifiers may inadvertently capture selection processes or other confounding neural/behavioral signals.
• Future research should carefully control for factors such as visual stimuli characteristics and oculomotor activity when using MVPA to infer VWM load.