Fundamental to the efficacy of cognitive training (CT) is its dose. Here we used the power and breadth afforded by a large dataset to measure precisely dose-response (D-R) functions for CT and to examine the generality of their magnitude and form. The present observational study involved 107,000 users of Lumosity, a commercial program comprising computer games designed to provide CT over the internet. In addition to training with Lumosity games, these users took an online battery of cognitive assessments (NeuroCognitive Performance Test, NCPT) on two or more occasions separated by at least 10 Here we used the power and breadth afforded by a large dataset to measure precisely dose-response (D-R) functions for CT and to examine the generality of their magnitude and form. The present observational study involved 107,000 users of Lumosity, a commercial program comprising computer games designed to provide CT over the internet. In addition to training with Lumosity games, these users took an online battery of cognitive assessments (NeuroCognitive Performance Test, NCPT) on two or more occasions separated by at least 10 weeks. Changes in performance on the NCPT between the first and second assessments were examined as a function of the amount of intervening gameplay. We report here the results of a large observational study that involved approximately 107,000 users of Lumosity, a commercial program comprising computer games designed to provide CT over the internet. In addition to training with Lumosity games, these users elected to take an online battery of cognitive assessments (the NeuroCognitive Performance Test, NCPT) on two or more occasions separated by at least 10 weeks. Changes in performance on the NCPT between the first and second assessments were examined as a function of the amount of intervening gameplay. Because the amount of gameplay between assessments differed across users, it was possible to calculate dose-response functions relating amount of gameplay to change in NCPT performance. D-R functions were obtained both for overall performance on the NCPT and for performance on its individual subtests, which are more directly related to specific cognitive domains. We examined also the differences between D-R functions from different demographic groups defined by age, gender, and education. The results of this study provide precise information about the form of the D-R function for CT and how it depends on a number of germane factors, including the course of adult aging. After registering for the Lumosity program (green), participants were eligible for assessment with a neuropsychological test battery (NCPT) at two time points separated by at least 10 weeks (blue). In parallel, participants freely played games from the Lumosity CT program (yellow), with the amount of gameplay varying between participants. Changes in performance on the NCPT between assessments were examined as a function of the amount of intervening gameplay. Participants All participants were recruited from Lumosity subscribers who signed up from 3-26-2007 to 3-26-2020 and indicated English as their preferred language. There is, however, a broad consensus that the type of learning that can be expressed verbally and whose sources can be consciously recollected (declarative episodic memory) diminishes in old age, while the type of nonverbal procedural learning that underlies skills is preserved [ 40 ]. While both types of learning might mediate improved performance from direct practice, transfer might involve only the latter. In CT, procedural learning might lead to more effective use of cognitive skills that underlie performance during both training and assessment, such as those involving attention, working memory, or executive processes. If the need for consent was waived by the ethics committee, please include this information. 3. Thank you for stating the following in your Competing Interests section: "I have read the So I assume this is the just the overall global cognitive composite. That’s correct. It’s important that we be clear about this, since GI refers to a major dependent variable of the study. After reading your comment, we realized that, when we defined the Grand Index (on what was formerly line 229), we never indicated that its abbreviation was GI. This has been rectified (line 322). We also provide a reminder on line 482 that GI is a measure of overall performance on the NCPT. 6. I was glad to see Figure 5 which shows that after a while, there is diminishing return on the therapeutic benefits of these games; that parallels Lampit et al.’s work. You have responded to the following reviewer comments (comments 5, 6, and 7) that I am copying here in italics: 5. I think GI stand for Grant Index (line 229). I figured this out later. So I assume this is the just the overall global cognitive composite. That’s correct. It’s important that we be clear about this, since GI refers to a major dependent variable of the study. After reading your comment, we realized that, when we defined the Grand Index (on what was formerly line 229), we never indicated that its abbreviation was GI. This has been rectified (line 322). We also provide a reminder on line 482 that GI is a measure of overall performance on the NCPT. 6. One can now be found on lines 59-60. 4. The review and integration of the literature is on point. Thanks. We are encouraged by your feedback. 5. I think GI stand for Grant Index (line 229). I figured this out later. So I assume this is the just the overall global cognitive composite. That’s correct. It’s important that we be clear about this, since GI refers to a major dependent variable of the study. After reading your comment, we realized that, when we defined the Grand Index (on what was formerly line 229), we never indicated that its abbreviation was GI. This has been rectified (line 266). We also provide a reminder on line 401 that GI is a measure of overall performance on the NCPT. 6.

more than 130,000 individuals aged 13–89 years from a NCPT battery that includes 8 subtests and several analyses to demonstrate reliability and validity of the NCPT battery as a measure of cognitive performance. In addition, data from participants who self-reported cognitive impairment was used to test the ability of the NCPT battery to differentiate clinical populations from healthy ones, demonstrating potential utility in research and clinical settings. Methods and results Ethical statement Since these studies evaluated performance on cognitive assessments and did not include an intervention they were exempt from IRB review. Participants were informed prior to starting the assessment (see Section Participants and Normative Data below) that their data would be used for research purposes, and opted in by choosing to take the assessment or not. Data included in all analyses were de-identified and analyzed in aggregate in accordance with Lumos Labs' Privacy Policy ( www.lumosity.com/legal/privacy_policy ). Participants and normative data Data used to generate the normative database were derived in aggregate from Lumosity subscribers who took the NCPT as part of their user experience. The majority of these users had paid for a premium Lumosity subscription. Within 1 week of their initial sign-up, Lumosity users were invited via email and an in-app prompt to take the NCPT battery (time 1). Taking the NCPT was optional and not required for continued use of Lumosity. Normative data were derived from a sample of 130,140 individuals aged 13–89 years who were generally healthy (as assessed via self-report survey) and had taken the NCPT battery at least once (Normative Sample). Participants in the Normative Sample represented 187 countries, with the majority from the United States (68.0%), Canada (9.1%), and Australia (8.6%) (representation from all other countries was < 5%, or 14.3% combined). Following completion of the NCPT battery at time 1, participants were asked to