Flexible top-down control in the interaction between working memory and perception.

Area of Science:

• Cognitive Neuroscience
• Experimental Psychology

Background:

• Successful goal-directed behavior requires integrating sensory input with working memory (WM) maintenance.
• The control mechanisms governing WM-perception interactions remain poorly understood.

Purpose of the Study:

• To investigate if WM-perception interactions employ proactive and reactive control modes, analogous to conflict-adaptation tasks.
• To determine if these control modes are explained by reinforcement learning principles.

Main Methods:

• Participants completed a delayed orientation recall WM task concurrently with an orientation discrimination task.
• Stimulus congruity between WM and perception tasks was manipulated.
• A reinforcement learning model was used to quantitatively explain observed control effects.

Main Results:

• Proactive control was evidenced by sensitivity to previous trial congruity (Gratton effect).
• Reactive control manifested as repulsive serial-dependence with incongruent stimuli.
• Reinforcement learning model parameters (tonic predicted conflict, phasic control prediction error) quantitatively captured these effects.

Conclusions:

• WM-perception interactions are regulated by distinct proactive and reactive control mechanisms.
• These control mechanisms align with those governing conflict processing in other cognitive domains.
• Reinforcement learning provides a quantitative framework for understanding trial-to-trial control adjustments in WM-perception.