The effect of reinforcement rate on goal-directed and habitual choices on the multiple schedule

TING HU^1*Shun Fujimaki^2,3Hiroto Kawarada¹Yutaka Kosaki²

• ¹Department of Psychology, Waseda University, Tokyo, Japan
• ²Research Institute for Letters, Arts and Sciences, Waseda University, Tokyo, Japan
• ³Department of Psychology, Tokiwa University, Mito, Ibaraki, Japan

Voluntary behaviors can either be goal-directed, sensitive to changes in its consequences, or habitual, devoid of such sensitivity. Here we conducted three experiments investigating how forced-choice training influences goal-directed and habitual processes under varying reinforcement rates. In all experiments, rats received 15 training sessions on a two-component multiple schedule with two sequentially inserted levers. In Experiment 1, identical variable interval (VI) 15-s schedules were used across components for Group Rich and VI 90-s schedules for Group Lean, yielding different outcomes. Following taste aversion for one outcome, Group Lean reduced performance (i.e., goal-directed action) during an extinction test, while Group Rich did not (i.e., habit). Experiment 2 addressed differential outcome exposure by reversing training conditions: Group Rich received numbers of outcomes equivalent to Group Lean in Experiment 1, and vice versa. The devaluation effects were evident in both groups. Using the same outcome across components, Experiment 3 trained rats on a multiple VI 15-s VI 90-s schedule to further to clarify the role of response-outcome pairings while controlling for total amount of outcome exposure. The VI 15-s component, despite producing fewer outcomes, led to stronger devaluation effects and residual responding. The most important finding of this study is that alternating R-O contingencies in a multiple schedule under lean reinforcement conditions consistently sustain goal-directed control even after extensive training, while richer conditions promote a shift to habitual control. These findings are discussed within a dual-system model framework in a molar context by hypothesizing that both goal-directed and habitual strength may grow more rapidly with higher reinforcement rates.