Learning to Overexert Cognitive Control in a Stroop Task

Area of Science:

• Cognitive psychology
• Neuroscience
• Decision-making

Background:

• Cognitive control allocation is crucial for task performance.
• The Learned Value of Control (LVOC) model proposes feature-based learning for control decisions.
• Generalization of learned control values across situations is a key prediction.

Purpose of the Study:

• To investigate if learned control allocation strategies generalize to new situations.
• To test the LVOC model's prediction of potential overexertion of control due to feature-based generalization.
• To examine how stimulus features influence the decision to engage cognitive control.

Main Methods:

• Participants performed a modified Stroop task with choices between a control-demanding (color naming) and automatic (word reading) task.
• Stimulus features (color, word) predicted task-specific rewards.
• Transfer learning was assessed using novel stimuli with overlapping features to probe misestimations of control value.

Main Results:

• Learned stimulus-task-reward associations were successfully acquired.
• Participants exhibited overexertion of cognitive control in transfer conditions where learned associations incorrectly predicted a higher-reward, control-demanding task.
• Performance was sometimes impaired due to this inappropriate control engagement.

Conclusions:

• Findings support the LVOC model's mechanism of feature-based learning and control generalization.
• Learned associations can lead to maladaptive overexertion of cognitive control in novel contexts.
• Situational features play a significant role in learned control allocation strategies.