Neurocognitive frailty is the biggest threat to successful aging [1]. There is overwhelming evidence that the number of inflammatory mediators increases vastly in pathological aging [2] and in healthy aging as well [3]. This inflammation has a negative impact on cognition [4]. C-reactive protein (CRP), an acute-phase protein of hepatic origin that increases following interleukin-6 secretion by macrophages and T cells, has been identified as a hallmark of systemic inflammation, and it can be assessed by noninvasive methods in saliva [5]. A key challenge that faces the aging society is to find new approaches to understand the aging mind and brain that could enhance successful, optimal aging [6]. Another major challenge is to understand the factors that could reduce or delay the negative consequences of age-related cognitive decline [7]. The idea behind this randomized controlled trial (RCT) is that cognitive abilities could be strengthened by interventions that promote positive brain plasticity [8].

Normal aging is associated with gray and white matter shrinkage. The lateral prefrontal cortex, cerebellum, and medial temporal lobe system including hippocampus are affected. Minimal reduction occurs in the entorhinal and occipital cortices [1,9,10]. These brain changes are associated with declines in several important cognitive functions, including processing speed, executive functions, working memory, and episodic memory [11-14]. However, other crystallized abilities such as general knowledge, verbal abilities [15-17], and implicit memory [18,19] are mostly preserved. Older adults with mild cognitive impairment [20] and Alzheimer disease patients have shown preserved implicit memory despite huge deteriorations in episodic memory [21,22]. Although behavioral priming for repeated pictures is spared in older adults, reduced neural activation, which is a signature of implicit memory, is affected. The relationship between brain function and behavior found in young adults is altered in older adults, although these age-related changes do not affect behavioral facilitation. These findings have implications for the notion that automatic processes, previously considered preserved with age, are susceptible to the effects of aging at the neural level. The age-invariant behavioral facilitation with stimulus repetition is observed as a result of more sustained neural processing of visual stimuli in older adults as a form of compensatory neural activity [23].

Electrophysiological studies have also found significant age-related event-related potential (ERP) changes in brain activity associated with memory retrieval in spite of similar performance in young and older adults [24]. Elders compensate for their lower level of parieto-occipital functioning as shown by smaller P300 amplitude at posterior sites by recruiting frontal sites as a mode of brain adaptation [25]. In touch, despite similar behavioral performance in a recognition memory task involving familiar 3D objects, young and older adults recruited different neural resources to perform the task. Age-related differences were found in brain oscillations, as measured by event-related spectral perturbations (ERSPs), suggesting the recruitment of additional neural resources as shown by greater alpha and beta power reductions in older adults [26]. Despite similar behavior performance, normal aging was found to affect ERPs and oscillatory brain activity during an incidental speeded haptic symmetry detection priming task. However, the young adults showed more positive amplitude than the older adults in the alpha and beta bands from stimulus onset [27]. All these neural results suggest that the older brain adapts in order to maintain the same level of performance as the younger brain both in explicit and implicit memory tasks.

Neural plasticity exists at several levels of the neural substrate [28,10], although not to the same degree in older as in younger adults [15,29,30]. These findings suggest that the aging brain retains some neuroplasticity, which can be influenced by the individual’s behavior. The prolonged mismatch between functional organismic supplies and environmental demands produces cognitive plasticity and underlines the capacity of the brain to implement behavioral flexibility [31]. Although there are personal factors, human behavior can change with appropriate training as the brain reorganizes in response to the environment [32]. Based on the idea of neuroplasticity, different types of interventions have been developed to ameliorate cognitive and functional declines by strengthening social networking using new information and communication technology (ICT) [33-35] and cognitive skills training [36,37] and promoting an active lifestyle [38,39] and physical activity training [40,41].

As people age, they experience declines in attentional control, mediated by the dorsolateral prefrontal cortex, and in long-term memory functions, mediated by the medial temporal lobe and the hippocampus. These areas suffer the highest degree of age-related atrophy [9]. Moreover, the prefrontal cortex facilitates the organization and contextualization of incoming information and interacts with the hippocampus during working memory implementation [42,43]. This relationship is strengthened with age [44]. These findings are especially relevant due to the existing links between these basic cognitive abilities and everyday functioning. The malfunction of these basic abilities is a significant predictor of older adult difficulties with the instrumental activities of daily living, leading to loss of independence [45,46]. Therefore, it is of vital importance to pursue the question of whether cognitive decline can be reversed or delayed through cognitive training interventions [47].

Video games and other computerized programs are receiving increasing interest from cognitive and neuroscientists to investigate the possibility of transfer to untrained tasks [48-50]. Researchers are increasingly using new technology, including cognitive training platforms and video games, to investigate their impact on cognition [48-52]. Computer-based interventions have the advantage that they can be easily used by elderly people living either at home or in nursing homes and could be a good alternative to traditional training programs [53]. This RCT aims to assess the effectiveness of training older adults with digital nonaction video games.

Two challenges must be overcome in order to develop an effective cognitive aging intervention [54-55]: (1) transferring training gains to untrained tasks and (2) designing interventions that encourage compliance. Older adults prefer games that involve mental challenge [56-57], while studies with young adults have shown that fast-paced action games result in broader transfer effects. A “first-person shooter” video game is not an appropriate choice for older adults [58]. Nonaction video games have potential advantages as they are enjoyable, low-cost, and can be self-administered. Recent results from a systematic review [53] and meta-analyses [59-61] suggest that training elders with video games improves information processing, with unexpected larger effects in old-older adults than in young-older adults. Moreover, the results of a recent meta-analysis of action video game training showed that healthy adults benefit from training in overall and specific cognitive domains, but young adults benefit more than older adults [62]. These findings underscore the potential of video game training as an effective intervention tool for cognitive improvement. Results of studies on executive function and working memory are less consistent. It is possible that nonaction video games are not an effective means of improving or maintaining these functions in older adults. Games provide an enjoyable way of passing the time and of giving meaning to the day [56]. In short, video games can offer important benefits to older adults, bearing in mind that intervention compliance is a key factor in longitudinal training studies [54].

A previous RCT conducted in our laboratory investigated the effects of nonaction video game training on a series of cognitive functions that decline with age and subjective well-being. Two groups of older adults participated in the study: an experimental group that received the training and a noncontact control group. Groups were similar at baseline on demographics, vocabulary, global cognition, and depression status. The results showed improvements in the video game–trained group and no change in the passive control group in processing speed, attention, and immediate and delayed visual recognition memory and a trend to improve in 2 dimensions of the Well-being Scale (affection and assertivity) [63]. However, visuospatial working memory and executive control (shifting strategy) functions did not improve [49,64,65].

A recent longitudinal intervention study conducted to investigate the effects of video game training in healthy older adults showed that 15 1-hour training sessions with 6 nonaction video games produced significant improvements on 2 visuospatial working memory tasks and on episodic and short-term memory tasks. Some of these gains were maintained over a 3-month follow-up period [52]. In both previous studies, experimental groups were compared with passive control groups. To better attribute training-related improvement to the intervention and to avoid placebo effects [66], the current RCT compares performance on a series of attentional and visuospatial working memory tasks of an experimental group trained with selected nonaction video games from the commercial Lumosity computerized training program (www.lumosity.com) with that of an active control group carrying out the same number of training sessions with The Sims, a simulation strategy game in which the player takes control of the life of a character in everyday activities, and SimCity, a life simulation game in which the player is the mayor of a city that he or she must develop. Both groups used a mobile tablet device during the training sessions. These video games were chosen to train the active control group because this group should also play with engaging and challenging video games the same training hours, potentially equating the social contact experienced during training by experimental and active-control groups.

Training expectation is an important issue in cognitive training studies [51,67]. To address this issue, participants report their expectations (increase or decrease) regarding their task performance for any assessment task on a 5-point Likert-type scale. After training sessions 1, 8, and 16 (final), participants respond to training feedback questions about engagement and motivation for each training game. The objective was to find out whether both groups were similarly engaged and motivated during training.

In sum, the growth of ICT-based tools, along with their progressive implementation in all social groups, offers a great opportunity to develop low-cost preventive intervention programs able to reach the entire population. This can contribute to improving the quality of life of elderly people, with a substantial reduction in health care costs related to the loss of autonomy and independence in this population group [68]. In this RCT we investigate not only possible cognitive improvement in healthy older adults in selected cognitive domains that decline with aging but also possible neural changes of training using electrophysiological methods at pre- and posttraining and 6-month follow-up. The study also explores neuroinflammatory mechanisms underlying possible cognitive training-induced effects using noninvasive saliva-testing methods.

The study has 3 main complementary objectives:

1. To examine the behavioral and electrophysiological effects of training older adults with nonaction video games on a series of cognitive tasks designed to assess attentional functions (mainly response inhibition, distraction, and alertness) and to maintain and update verbal and visuospatial working memory. We are also interested in investigating the sustainability of the transfer effects after a 6-month no-contact period.

2. To explore whether motivation, engagement, or expectations account for possible training-related improvements.

3. To examine whether inflammatory mechanisms assessed with noninvasive measurement of CRP in saliva impair cognitive training-induced effects. Inflammatory parameters in the participants obtained by noninvasive saliva-testing methods will be analyzed in biochemical assays to study their possible correlations with cognitive abilities.

We hypothesize that participants in the Lumosity group will improve selective attention assessed with oddball and Stroop tasks and working memory assessed with Corsi blocks and n-back tasks in comparison with participants in the active control group. From an electrophysiological perspective, we also hypothetize that in the oddball task the cognitive video game–trained group will improve both the ability to return attention to the relevant task after having been distracted by the novel stimulus (reorientation negativity [RON]) and the ability of the brain to pay attention to the novel stimulus (P300).

Intervention studies conducted with older adults have focused mostly on training a given cognitive function and then measuring the improvement in that function [69,29] and less often on possible changes in neural activity after training [70,71]. Based on the idea that the older human brain still has some plasticity (the ability to change in response to external stimulation), our main hypothesis is that training older adults with appropriate, adapted, and motivating nonaction video games will transfer to cognitive functions that decline with age, such as attention and working memory, by improving these pillars of cognition, which are crucial to maintaining independent living and cognitive health in old age.

A 2-arm, parallel-design RCT [72] was designed to evaluate the effectiveness of video game training to promote cognitive and brain health in older adults. Figure 1 shows the Consolidated Standards of Reporting Trials flow diagram of the study. This RCT is being conducted in Madrid, Spain, with an allocation ratio of 1:1 to evaluate the influence of training with nonaction video games on attention and memory functions. The design was a single-blind design. Eligible participants were randomized using a computer algorithm into intervention (cognitive nonaction video game training) or active control (life-simulation games) groups. Researchers were not blind to treatment allocation. The Universidad Nacional de Educación a Distancia (UNED) Institutional Review Board approved the study protocol. All participants provided written informed consent.

All participants live independently with normal or corrected-to-normal hearing and vision and are free of neurological or psychiatric disorders or traumatic brain injury. To determine their eligibility, each participant completed a screening battery consisting of the Mini-Mental State Examination (cut-off score 26) [73] to rule out possible cognitive impairment, the Yesavage Depression Scale (cut-off score less than 5) [74], and the Information subscale of the WAIS-III scale (normal score) [75]. Exclusion criteria were a diagnosis of dementia, cognitive impairment (score of <26 on the Mini-Mental State Examination), depression, less than 20/60 vision with or without correction, inability to complete the study activities, communication problems, or current plans to move to another city. All participants were healthy volunteers.

The sample size is based on the number of observations needed to compare the 2 groups of gamers, those trained with Lumosity and those trained with The Sims. As we do not have standard deviations of the scores in our main dependent variables, we used G*Power 3.1 (Department of Psychology) to compute the number of participants, using analysis of variance as a statistical test. For this, using a power of .80, effect size of 0.50, an alpha level of .05, and 2 groups within the F test family, we obtained a required total sample size of 57. With a dropout rate of 10%, a total of 63 participants would be required. Rounding up the numbers, we would need more than 64 participants (32×2). As the flow diagram shows, the final number of participants matches these a priori computations closely: 38 participants were assigned to the experimental group (8 lost at posttest) and 37 to the control group (12 lost at posttest).

Participants were randomly assigned to 1 of the 2 parallel groups, either the intervention video game training group or the active control group (see Figure 1). The pretest, posttest, and 6-month follow-up assessments are conducted at the UNED Department of Basic Psychology II by one of the authors with the assistence of other members of the group. The training sessions were carried out at the UNED Associate Center, Escuelas Pías, Madrid, and lasted 8 to 10 weeks. A total of 138 participants were recruited from flyers, advertisements, by word of mouth, and from UNED’s special program for the elderly screened for eligibility. Finally, 75 older adults (age range 55 to 84 years) met the requirements for participation in the study and signed the consent form and were randomly assigned to the experimental or the active control groups before carrying out the laboratory tasks. The 2 groups do not differ significantly in age, years of education, global cognition, depression, memory, or verbal abilities.

While performing the experimental tasks, continuous electroencephalograph (EEG) activity is recorded using a NuAmps amplifier (Neuroscan Inc) located inside a soundproof, electrically shielded room. A 34-channel elasticized Quik-Cap with Ag/AgCl sintered electrodes (Neuroscan Inc) is used to record EEG data from scalp electrodes positioned according to the extended international 10-20 system (American Medical EEG Association, 1991). To control for the influence of ocular artifacts, vertical and horizontal electrooculograms are recorded in 2 bipolar channels. Eye blinks and vertical eye movements are monitored via electrodes located below and on the supraorbital ridge of the left eye. Horizontal artifacts are monitored via electrodes on the outer canthus of each eye. Linked mastoids (A1, A2) are used as reference, and participants are grounded to the AFz electrode. All data will be digitized using a NuAmps amplifier in continuous recording mode. Sampling rate will be 250 Hz, and all channels will be online band-pass filtered (0.1-70 Hz) and notch filtered (50 Hz) to eliminate power line artifacts. Analysis of the EEG recordings will be performed using the EEGLAB toolbox [76] and ERPLAB plugin for EEGLAB [77] both running under MATLAB environment (The MathWorks Inc). Continuous data will be filtered offline using a digital Butterworth filter (0.1-40 Hz; 12 dB per octave roll-off), an infinite impulse response filter that achieves a given filtering characteristic using less memory and fewer calculations than a similar finite impulse response filter. After filtering, data will be separated into baseline corrected and nonoverlapping epochs time-locked to the target onset (in both Stroop and oddball tasks) ranging from 200 ms before to 1000 ms after the onset of the target with the prestimulus interval (200 ms) as baseline period. Epochs containing high amplitude/frequency and muscle or other irregular artifacts will be removed by visual inspection. Only artifact-free epochs from correct trials will be selected for averaging. The existence of blinks and other ocular movements will not be a criterion for epoch rejection. This kind of artifact will be eliminated using Infomax Independent Component Analysis (ICA) decomposition [78-82]. We will use the default extended-mode runica training parameters [83], an extension of the original algorithm [78]. This extended mode allows a wider range of source signals (both super- and sub-Gaussian) while maintaining simplicity. Stopping weight change will be set to 1e–7. This rather conservative criterion for stopping learning lengthens ICA training, enabling cleaner and more reliable decompositions, particularly with more than 33 channels and limited number of epochs. After submitting epochs to ICA decomposition, artifactual components will be removed by inspection of their scalp topography and spectral power [84,27].

Participants in the study were randomly assigned to the cognitive nonaction video game training group or the noncognitive game training active control group. The introduction of an active control group is necessary to avoid placebo effects [54]. Based on the results of our meta-analytic study [61], we use training regimes that are not too long to avoid loss of motivation, and simple, nonaction games that appeal more to elders. According to the temporal discount hypothesis [85], future rewards are less valuable than inmediate rewards. In each session, the trainees play 10 nonaction video games from Lumosity, described below.

The main question examined in our study is whether group (experimental group, active control group) interacts with testing session (pretest, posttest, and follow-up) with regard to performance on a series of cognitive tasks. At least 2 measures for each cognitive domain are acquired to examine transfer of training to untrained tasks demanding attention and working memory. We selected these cognitive domains because these functions deteriorate with aging and tax functions that are critical for independent living. To explore the underlying neural mechanisms of successful transfer of training gains to attentional mechanisms (alert, distraction, and inhibitory effects) and executive control (spatial working memory), electrophysiological data (ERPs) are recorded at pretraining (T1), posttraining (T2), and 6-month follow-up (T3).

All the methodological designs of the primary outcome measures have been constructed using the rules of counterbalancing and stimulus rotation. Response keys are counterbalanced across conditions. The computerized tasks have been programmed using E-Prime 2.0 (Psychology Software Tools Inc). All the statistical analyses of the behavioral results will be performed using SPSS (IBM Corp) and results will be considered significant at P<.05, with Bonferroni-corrected post hoc tests performed as appropriate. Continuous EEG activity are recorded in our laboratory with thin electrodes from 34 scalp sites using a NuAmps amplifier as mentioned before.

After randomization and initial assessment (baseline), participants were invited to attend the University Center of Escuelas Pias in Madrid to perform the training sessions in small groups. Trained and active control groups followed 16 training sessions carried out on different hours and days that lasted approximately 10 to 12 weeks.

Table 1 presents a short summary of the video games played by the experimental (trained) group. During the training sessions, participants in the experimental group play 10 nonaction video games selected from Lumosity.

The active control group carried out 16 training sessions with The Sims and SimCity BuildIt. Table 2 presents a short summary of these video games.

This clinical trial is registered on the ClinicalTrials.gov database [NCT02796508] [72]. The UNED Ethical Review Board approved the trial. All the participants gave their written informed consent before the study started and were informed of their right to terminate participation at any time. The work described has not been published previously.

Several parameters influencing inflammation (eg, infections, diabetes) are obtained from the participants: height and weight are recorded to obtain the body mass index (gross obesity could increase inflammation), drugs ingested (anti-inflammatories, antibiotics, antioxidants), possible infections and when they happened (last month, last week, or today), smoking habit (which may increase CRP levels in saliva), and defective oral health (eg, bleeding gums). Participants are asked not to eat or smoke at least 30 minutes before saliva collection. Following a mouth rinse with water, 1 mL of saliva is collected with a straw into a tube containing 2% ethylenediaminetetraacetic acid as preservative. Samples are stored at −80ºC until assayed. Assessment of CRP levels is performed with an enzyme-linked immunosorbent assay (ELISA) obtained from Salimetrics according to the manufacturer instructions. A saliva sample collected from participants at pretest, posttest, and follow-up assessments will be analyzed with ELISA in order to correlate the quantity of CRP with task performance.

Improved game performance by the experimental group and the active control group between training session number 1 and 16 (final) is a secondary outcome measure. Efficacy of training will be shown by better performance in the trained video games across the training sessions. Another secondary outcome measure, maintained level of motivation and engagement, will be assessed with a questionnaire administered at training sessions 1, 8, and 16.