Phantom controllers: Misspecified models create the false appearance of adaptive control during value-based choice

Area of Science:

• Cognitive Psychology
• Decision Science
• Computational Neuroscience

Background:

• Decision scientists investigate adaptive control of decision-making processes.
• Evidence accumulation parameters, like decision thresholds, are influenced by external information.
• A recent 'motivated control' account proposed value influences decision parameters, predicting lower thresholds for difficult choices under time pressure.

Purpose of the Study:

• To re-evaluate evidence for the motivated control account of decision-making.
• To investigate whether observed decision threshold adjustments are robust or artifacts of model specification.
• To clarify the mechanisms underlying adaptive control in decision-making.

Main Methods:

• Reanalysis of existing experimental data using drift diffusion modeling.
• Application of a more systematic and comprehensive modeling approach.
• Comparison of different model specifications to identify potential artifacts.

Main Results:

• The previously observed decrease in decision thresholds for difficult choices was not replicated under a more complete model.
• Evidence for value-driven threshold adjustments disappeared or reversed upon reanalysis.
• Apparent evidence for motivated control was shown to arise from model mis-specification.

Conclusions:

• The motivated control account, as previously supported, does not hold up under further scrutiny.
• Apparent value-driven decision parameter adjustments can be artifacts of simpler, stimulus-driven processes.
• This work provides crucial insights into the constraints and validity of models of decision control.