Demonstrating and disrupting well-learned habits

Abstract

Researchers have exerted tremendous efforts to empirically study how habits form and dominate at the expense of deliberation, yet we know very little about breaking these rigid habits to restore goal-directed control. In a three-experiment study, we first illustrate a novel approach of studying well-learned habits, in order to effectively demonstrate habit disruption. In Experiment 1, we use a Go/NoGo task with familiar color-response associations to demonstrate outcome-insensitivity when compared to novel, more flexible associations. Specifically, subjects perform more accurately when the required mapping is the familiar association of green-Go/red-NoGo than when it is red-Go/green-NoGo, confirming outcome-insensitive, habitual control. As a control condition, subjects show equivalent performance with unfamiliar color-response mappings (using the colors blue and purple mapped to Go and NoGo responses). Next, in Experiments 2 and 3, we test a motivation-based feedback manipulation in varying magnitudes (i.e., performance feedback with and without monetary incentives) to break the well-established habits elicited by our familiar stimuli. We find that although performance feedback prior to the contingency reversal test is insufficient to disrupt outcome-insensitivity in Experiment 2, a combination of performance feedback and monetary incentive is able to restore goal-directed control in Experiment 3, effectively breaking the habits. As the first successful demonstration of well-learned habit disruption in the laboratory, these findings provide new insights into how we execute and modify habits, while fostering new and translational research avenues that may be applicable to treating habit-based pathologies.

Fig 1. Go/NoGo task with familiar and novel lights.

Participants are assigned to Familiar or Novel conditions. In the Familiar condition, subjects complete two phases: one where green signals Go and red signals NoGo (“congruent” mapping) and one where red signals Go and green signals NoGo (“incongruent” mapping). In the Novel condition, participants complete two similar phases, but the colors are blue and purple, for which there should be no strong pre-existing associations with “stop” and “go” responses. We predicted more commission errors in the Familiar condition for incongruent than congruent mappings, indicating outcome insensitivity, with no such within-subject differences expected in the Novel condition. Mapping orders were counterbalanced across subjects.

Fig 2. NoGo and Go performance to familiar and novel stimuli.

(A). Familiar stimuli elicit mapping-related change in NoGo accuracy. Participants make significantly fewer errors of commission when the NoGo signal is red compared to green. There is no difference in accuracy in the Novel condition when the NoGo signal is purple vs. blue. Stim_Familiarity x Color–Response_Mapping interaction: p = .004. (B) Familiar stimuli do not significantly elicit mapping-related change in Go accuracy. There is no significant difference in Go accuracy when Go signal is red compared to green. Likewise, no differences are seen in the Novel condition, when Go signal is blue vs. purple. Stim_Familiarity x Color–Response_Mapping interaction: p = .140. Error bars depict standard error of mean (SEM). Color of bars reflects NoGo and Go stimulus colors.

Fig 3. The effects of performance feedback on NoGo and Go accuracy.

(A) Performance feedback does not significantly disrupt well-established habits. In the Familiar condition, both Feedback and No-Feedback groups display a similar mapping-related change (interaction p = .513) in NoGo accuracy. Note that the non-significant interaction indicates the comparable mapping-related accuracy change across feedback groups. (B) NoGo accuracy in the Novel condition is significantly improved by performance feedback (sig. interaction of p = .028) when controlling for Age (which was significantly different across the Feedback groups). (C) Performance feedback protects against habitual Go actions. Providing cumulative performance feedback prevented the mapping-related Go accuracy change when managing Familiar stimuli (Feedback x Color–Response_Mapping interaction p = .019). (D) Performance feedback did not significantly improve Go accuracy in the Novel condition (Feedback x Color–Response_Mapping interaction p = .05). Error bars denote SEM. Color of bars reflects NoGo and Go stimulus colors.

Fig 4. The effects of combined monetary and performance feedback on NoGo and Go accuracy.

(A) Monetary and performance feedback disrupt habits while improving goal-directed performance to newly-learned stimuli. Providing performance and monetary feedback protects against the mapping-related change when overriding well-learned associations (Feedback x Color–Response_Mapping interaction: p = .005). (B) The effect of dual feedback on goal-directed control of novel associations fell short of significance (Feedback x Color–Response_Mapping interaction: p = .066, controlling for Gender). (C) Dual feedback improves goal-directed Go accuracy. Dual feedback had a significant effect on mapping-related Go accuracy change when overriding well-learned Go responses (interaction p = .033). (D) Dual feedback improved goal-directed Go responses to novel associations (interaction p = .006). Error bars denote SEM. Color of bars reflects NoGo and Go stimulus colors.