Critics of the benefits of cognitive training often point to the failure of intervention studies to evaluate the effect of participants’ expectations; however, there is mixed evidence as to whether differential expectations produce differential outcomes. To investigate this question, using an existing dataset, we explored whether participants regularly exposed to positive messages about cognitive training efficacy improved more on self-reported cognitive ability, specifically with respect to their perceived cognitive deficit (PCD). Participants (N = 104, age 64–84) were instructed to engage in computerized cognitive training at home for a total of 40 h over 8 weeks. Groups of participants received different messages during training intended to improve intervention adherence, but only one group experienced messages related to the benefits of cognitive training. Contrary to our prediction that perceived cognitive functioning would change as a function of exposure to the different message types, participants overall demonstrated a significant decrease in PCD after training, and this effect did not interact with message type. Additional exploratory analyses found that positive messages did not differentially impact self-efficacy or objective cognitive performance. Contrary to our expectations, perceived efficacy of cognitive training did not change as a function of condition, and on average perceived training efficacy significantly decreased post-intervention. Results advance understanding of how messages delivered during training may or may not impact beliefs in cognitive training efficacy and subjective and objective outcome measures.