One claim of Robinson’s Cognition Hypothesis predicts that more cognitively complex tasks will promote greater uptake and retention of linguistic forms enhanced by interactional feedback such as recasts. The present study tested this claim by examining whether the task design variable +/- visual support affected the amount of uptake produced by learners in response to recasts and the relationship between uptake and L2 development. The study employed a pretest-posttest-delayed posttest design, with three treatment sessions. The participants were 54 adult EFL learners, randomly assigned to two experimental and a control group. Both experimental groups performed picture description tasks and received recasts targeting the past progressive. In the complex task condition, participants could view the pictures while describing them, whereas, in the simple condition, they could not do so. The pre- and post-tests included two oral production tasks designed to assess any changes in the learners’ ability to use the past progressive. Multiple regression analyses revealed that although task complexity did not affect the rate of learners’ uptake, it did modulate the relationship between uptake and L2 development. Uptake was a strong positive predictor of development when learners carried out less complex tasks, but was negatively associated with development when learners carried out more complex tasks.