<p>Our ability to envision the future is often taken to play a central role in temporally extended agency. Yet it remains unclear whether and how mental imagery (perception-like mental states that present us with sensory information without a direct external stimulus) contributes to core aspects of temporally extended agency such as how we encode, retain, and retrieve intentions to act. This paper argues that empirical progress on this requires a shift in methodology. Existing experimental studies investigate the role of mental imagery by comparing the accuracy and speed with which people retrieve intentions that have been encoded via verbal rehearsal compared to rehearsal via mental imagery. However, these studies yield mixed and conflicting results. I argue that this may be due to methodological issues that follow from the unjustified assumption that these task instructions can isolate the role of imagery and from ignoring that mental imagery is a capacity that varies across individuals. To address these issues, I propose that future empirical studies should make use of the naturally occurring extreme variation in imagery vividness that exists between aphantasics (individuals with little or no imagery) and hyperphantasics (individuals with extremely vivid imagery). Drawing on recent evidence linking lower imagery vividness to less detailed episodic thinking, I argue that imagery-dependent strategies for encoding, retaining, and retrieving intentions to act—if they exist—should be unavailable to aphantasics and readily available to hyperphantasics. I conclude by outlining testable hypotheses and some suggested tests that could help clarify whether and how mental imagery influences temporally extended agency.</p>