There has been recent excitement over the potential for commercial and custom digital games to reverse age-related perceptual and cognitive decline. The effectiveness of digital game-based brain training is controversial. However, a separate issue is, should digital game-based interventions prove effective, how best to design these interventions to encourage intervention engagement and adherence by older adults (ages 65 +). This study explored older adults' perceptions and attitudes toward game-based interventions after they were asked to play digital games (experimental or control games) for a month-long period. Clear differences in attitudes toward game-based interventions were observed, as assessed by post-intervention surveys, with older adults finding games in the control condition (word and number puzzle games) more enjoyable and less frustrating compared to a digital game that consisted of gamified brain training interventions that have demonstrated some degree of success in the literature. Interestingly, older adults perceived the control condition as more likely to boost perceptual and cognitive abilities (e.g., vision, reaction time), as assessed by a post-intervention survey of expectations. Although predicting intervention adherence was challenging, overall motivation to do well in the intervention was significantly related to perceptions of cognitive benefit. Not surprisingly, game enjoyment also predicted motivation. Finally, older adults who perceived the game they were assigned to play as more challenging were more likely to believe the game would boost cognition. These findings identify attitudes and beliefs that could be targeted to motivate older adults to adhere to digital game-based interventions found to boost cognition. To better explore factors related to intervention adherence in the future we propose studies of longer duration (e.g., 6-12 months) and studies that allow more flexibility and choice with respect to amount of gameplay (instead of gameplay being dictated by a fixed schedule determined by the experimenter, leaving less variability to be explained by individual difference factors).