Abstract
In experimental psychology different experiments have been developed to assess goal–directed as compared to habitual control over instrumental decisions. Similar to animal studies selective devaluation procedures have been used. More recently sequential decision-making tasks have been designed to assess the degree of goal-directed vs. habitual choice behavior in terms of an influential computational theory of model-based compared to model-free behavioral control. As recently suggested, different measurements are thought to reflect the same construct. Yet, there has been no attempt to directly assess the construct validity of these different measurements. In the present study, we used a devaluation paradigm and a sequential decision-making task to address this question of construct validity in a sample of 18 healthy male human participants. Correlational analysis revealed a positive association between model-based choices during sequential decisions and goal-directed behavior after devaluation suggesting a single framework underlying both operationalizations and speaking in favor of construct validity of both measurement approaches. Up to now, this has been merely assumed but never been directly tested in humans.
Highlights
Habitual decisions arise from the retrospective, slow accumulation of rewards via iterative updating of expectations
In the present study, we used two reinforcement learning tasks in the same participants, selective devaluation and sequential decision-making, which are frequently used in human research
We aim to assess the construct validity of these two measurements which have both been suggested to capture the dichotomy of goal-directed or model-based vs. habitual or model-free control, respectively
Summary
Habitual decisions arise from the retrospective, slow accumulation of rewards via iterative updating of expectations. If outcome values change suddenly e.g., after devaluation (i.e., satiety), the goal-directed system enables quick behavioral adaptation, whereas the habitual system requires new reward experience before it can alter behavior (Balleine and Dickinson, 1998) This dual system theory has been advanced by the use of computational models of learning which either purely update reward expectations based on reward prediction errors (“model-free”) or aim to map possible actions to their potential outcomes (“modelbased”; Daw et al, 2005). In their comprehensive review Dolan and Dayan (2013) subsume both concepts (goal-directed/habitual and model-based/model-free) under a single framework of reflective vs reflexive decision making. This longstanding dichotomy has been described as goal-directed vs. habitual behavior by experimental psychologists while the model-free vs. model-based theory provides a computational account of the same construct
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
