Keywords

1 Introduction

A major focus of research has been the degree to which certain activities can reduce or reverse the perceptual and cognitive declines associated with advancing age [1]. More specifically, over the past decade the potential for commercial [e.g., 2, 3] and custom [e.g., 4] digital game interventions to improve cognition has generated a great deal of excitement. While initial results appear promising, the idea that digital games and commercial brain training software packages can meaningfully improve the cognition of older adults remains controversial [5, 6]. Relatedly, others have investigated whether digital games, specifically exergames that involve physical activity, might be used to improve physical fitness and health [e.g., 7, 8].

A separate issue is the degree to which digital game-based interventions intended to improve cognitive or physical health are motivating to older adults and encourage intervention adherence. Effective digital game-based interventions are not possible unless older adults are willing and able to engage with these games for an extended period of time. While much work has explored the potential benefits of digital game-based interventions, far fewer studies have attempted to uncover the general principles related to older adults’ motivation to engage in digital game-based interventions and factors related to intervention adherence. This is crucial information needed in order to maximize benefits should these types of interventions prove effective.

Digital games are a form of technology, and technology adoption models exist that help guide predictions related to the factors that influence the intention to use digital game-based interventions by older adults. For example, the Unified Theory of Acceptance and Use of Technology (UTAUT, [9]), and the earlier Technology Acceptance Model (TAM, [10]), propose that the perceived benefits (e.g., usefulness) and perceived costs (e.g., ease of use) from technology influence both intention to use technology and subsequent use. An older adult who believes strongly that digital game interventions are effective at improving cognition is predicted, according to these models, to be more likely to adopt this technology. Perception of effort required to learn and use a new technology also plays an important role. In this case, effort expectancy may be partly shaped by previous experience with technology platforms on which these interventions are typically delivered (personal computer, tablet, smartphone). If an older adult has little or no previous experience with these technologies, he or she may perceive substantial costs with respect to the adoption of digital game-based interventions, or may not have the prerequisite knowledge to engage in such interventions. Experience successfully using similar technology may influence self-efficacy, which has also been found to influence technology adoption [11].

Fast-paced action video games have been promoted as one of the most effective types of digital games with respect to improving cognition [12]. Unfortunately, these games are among the games that older adults are least interested in playing [13]. Boot and colleagues [14] attempted to assess whether a fast-paced action game might improve the perceptual and cognitive abilities of an older adult sample. The action game chosen was a popular racing game on the Nintendo DS system (Mario Kart). A control group played the brain fitness focused game Brain Age 2. Participants in each group were asked to play the game they were assigned for a total of 60 h (5 h a week over 3 months). However, the effect of the action game could not be assessed because of the extremely low rate of adherence in this condition. The intervention having stronger evidence in support of it (action game) was adhered to significantly less by older adults (M = 22 h for Mario Kart vs. 56 h for Brain Age). Various analyses explored reasons for this lack of adherence and found that, compared to participants in the brain fitness game group, participants in the action game intervention were significantly less likely to believe the intervention they received would have a meaningful impact on their ability to perform everyday tasks. Furthermore, older adults found this intervention less enjoyable. In subsequent analyses, individual differences in enjoyment and perceived benefit predicted adherence as well as motivation to perform well on training tasks. A follow-up study of shorter duration (10 days) found much higher levels of adherence for Mario Kart, suggesting that intervention duration may play an important role [15]. These initial results highlight the importance of understanding adherence, but also provide suggestions as to the important factors that determine individual differences in adherence and motivation.

At this point, it is clear that the games that older adults prefer are not the same types of games that are most popular among younger, more active gamers, and game-based cognitive interventions need to take this into account. A number of studies have explored both older adult digital game preferences and the reasons older adult gamers play. For example, although violent first-person shooters are generally a popular game genre, older adult focus group research suggests a general aversion toward games with violent content [16]. This research also indicates a perception by older adults that games provide stimulation that might serve as a form of “brain training.” Motivation for gaming among older adults may also be shaped by feelings of wellbeing associated with gameplay [17]. Casual games, including puzzle games and computerized versions of non-digital games (e.g., card and board games) were among the most popular games reported by a Dutch older adult gamer sample, with the need for challenge reported as the primary appeal of gaming [18]. Similarly, in a sample of both gaming and non-gaming older adults, puzzle and educational games were rated as the most interesting genres, and challenge and intellectual stimulation were rated as being among the most important features of a game [13]. In sum, the types of games that older adults appear to have the most interest in may be different from those most enjoyed by younger gamers, and a variety of factors (challenge, perceived benefits to cognition and wellbeing) appear to motivate game play in this population.

The current paper attempts to replicate previous relationships observed between intervention motivation and adherence and various perceptions, attitudes, and individual difference factors in two different digital game conditions to explore the generalizability of previous findings. Specifically, our primary hypotheses predicted that adherence (number of sessions completed) and self-reported motivation to perform well while playing the intervention games would be related to self-reported perceived benefits to cognition from the intervention and intervention enjoyment. This would be consistent with our previous findings, and also consistent with models of technology adoption. Participants were randomly assigned to play 30 sessions of gameplay over the course of one month. One group received a brain fitness game made up of gamified versions of interventions that have been reported in the literature as improving cognition (e.g., N-back training, memory updating training). The control group received word and number puzzle games. After initial training in the laboratory on how to access their assigned intervention on a tablet computer, and how to play the digital games that they were assigned, participants completed training at home and used a diary to keep track of their gameplay. Upon returning to their lab after this one-month period, their attitudes toward the game they were assigned to play and their perceptions of game benefits were assessed.

2 Method

Our goal was to have 60 U.S. participants (ages 65 +) randomly assigned to either the intervention brain training group or a control game condition (N = 30 each), complete the digital game-based condition they were assigned, and complete a battery of perceptual and cognitive tasks before and after training. In total, 78 participants were recruited and randomly assigned to meet this goal, with 18 participants dropping out at various points during the study. Because game perceptions and attitudes were assessed during the second lab visit, no data were available for these measures for participants who dropped out. Additionally, some participants had incomplete survey data (2), and some participants did not record adherence as requested (3). As a result, analyses reported here included a minimum of 55 participants.

Both the experimental and control interventions were tablet-based (Acer Iconia A700 10 inch). The brain training intervention consisted of gamified versions of task-switching, N-back, memory updating, reasoning, planning, and spatial reasoning tasks that were adaptive in nature (i.e., if the participant was successful on a level, the tasks were made more difficult). These were gamified in the sense that appealing graphics were added, goals were set for players, and motivating feedback was given in the form of game scores. Progress was graphed over time. A “Wild West” theme connected these games (see [19] for a more detailed description). The control intervention consisted of three common puzzle games (crossword, Sudoku, and word search) and followed the same structure of the intervention condition. Participants in both groups were asked to play seven sessions of gameplay each week for about forty-five minutes each session for one month, and were given a diary to keep track of their gaming sessions. Both groups were asked to divide their time during each gaming session between three different games within their intervention. For the control group, these three games were crossword, Sudoku, and word search. In the brain training intervention, this was a balanced subset of three games out of a total of seven that varied from session to session.

Perceptions and attitudes toward the game participants were assigned to play were assessed with surveys at the end of the intervention. Participants rated the degree to which they thought games like the ones they were assigned to play might improve various abilities: vision, reaction time, memory, hand-eye coordination, reasoning ability, multi-tasking ability, and the ability to perform everyday tasks such as driving. Participants also rated the degree to which they found their intervention enjoyable, challenging, and frustrating. Motivation was assessed with a question that asked participants to rate the following statement: “I was motivated to perform well on the games I was given to play.” All of these questions were rated on a 1–7 scale, from very strongly disagree to very strongly agree.

3 Results

Adherence.

Our primary measure of adherence was the number of sessions (out of 30) that each group completed according to their diary records. Adherence was generally good for both interventions (> 70 % of sessions completed), though slightly below the 80 % figure generally judged acceptable for medication adherence. The control group completed 22 sessions on average (SD = 9.02) and the brain training games group completed 23 (SD = 7.67).

Game Perceptions.

Participants were asked to rate how enjoyable, challenging, and frustrating the intervention they were asked to engage in was (Fig. 1). These questions were phrased in the manner of “I found the games I was given to play enjoyable.” These data were entered into an ANOVA and a significant game group by game perception interaction was observed (F(2, 112) = 3.87, p < .05). As depicted in Fig. 1, participants assigned to the brain training digital games group found their intervention less enjoyable and more frustrating compared to the control group, and both groups found their games equally challenging. Potential reasons for these differences in ratings are discussed later.

Fig. 1.
figure 1

Game perceptions as a function of group assignment. Error bars represent ±1 SEM. * = p < .05 between groups. 4 = Neutral (Neither agree nor disagree).

Full size image

Intervention Expectations.

Participants were asked to rate how likely the intervention they were asked to engage in would improve a variety of perceptual and cognitive abilities (Fig. 2). These questions were phrased in the manner of “Digital games like the ones I was given to play have the potential to improve vision.” Expectation data were entered into an ANOVA and a significant game group by ability interaction was observed (F(6, 342) = 3.87, p < .01). As depicted in Fig. 2, participants in general expected the intervention they were assigned would result in broad improvement, but participants in the control group believed that their intervention was more likely to improve vision and reaction time compared to the brain training intervention.

Fig. 2.
figure 2

Improvement expectations as a function of group assignment. Error bars represent ± 1 SEM. * = p < .05 between groups. 4 = Neutral (Neither agree nor disagree).

Full size image

Intervention Motivation.

Intervention motivation was assessed with a single item: “I was motivated to perform well on the games I was given to play.” No difference in motivation was observed between the two interventions (M = 5.82, SD = 1.14 for Control; M = 5.28, SD = 1.19 for Brain Training; t(56) = 1.80, p = .08).

Next we turned to the best predictors of intervention motivation. A previous study [14] found that digital game enjoyment and perceived benefits to perceptual and cognition abilities predicted motivation. Since motivation was similar for both groups, these data were collapsed. A Principal Components Analysis reduced perceived benefits to a single score for analysis. Similar effects were observed in the current data set, with enjoyment, challenge, and perceived cognitive benefits positively related to motivation, and frustration negatively related to motivation (Table 1). Interestingly, higher perceptions of intervention challenge were associated with a greater belief that the intervention would improve cognition.

Table 1. Correlations with Intervention Motivation.
Full size table

When all predictors were induced (including group as a control variable) in a multiple regression analysis, enjoyment appeared to have the largest independent effect on motivation to do well in the intervention (Table 2). There was also still a trend for perceived benefits to perceptual and cognitive abilities to be predictive of motivation, though this effect was no longer significant.

Table 2. Regression model for intervention motivation.
Full size table

Predictors of Adherence.

We examined whether we could predict intervention adherence (number of sessions completed) with similar factors that predicted intervention motivation, and also examined a potential relationship between adherence and motivation itself. Unfortunately, no significant predictors of intervention adherence were observed (Table 3). There was a trend for a positive relationship between intervention adherence and motivation (r = .24, p = .08).

Table 3. Correlations with Intervention Adherence
Full size table

Tablet Computer Proficiency.

Finally, given the technological nature of the intervention we examined whether tablet proficiency at the beginning of the trial predicted either motivation or adherence. We used a measure under development called the Mobile Device Proficiency Questionnaire to assess tablet proficiency, based largely on the Computer Proficiency Questionnaire [20]. The average score on this measure was 18.87 (SD = 9.74) out of 40, consistent with very low proficiency. Floor on this scale is a score of 8. Tablet proficiency was unrelated to intervention motivation (r(55) = .02, p = .89) and intervention adherence (r(53) = -.09, p = .52). This can be interpreted positively in that the amount of tablet training provided in the laboratory was sufficient to allow participants to engage in the intervention for an extended period of time.

4 Conclusion and Discussion

Hypotheses with respect to adherence and motivation were developed based on previous adherence findings of a similar study and models of technology adoption. We found that several factors predicted intervention motivation (but not adherence). Perhaps not surprisingly, enjoyment of the intervention assigned was most strongly related to motivation. However, perceived benefits to perceptual and cognitive abilities also tended to be related to motivation. These findings largely replicated patterns observed in a previous study evaluating different games [14]. To the extent possible, when designing digital game-based interventions, these games should be fun and enjoyable. However, this might not always be possible. In order for cognitive training to be effective, it may need to push the limits of an individual’s abilities and require effortful engagement that might not be perceived as pleasant. There may be a tension between what produces cognitive benefits and what results in an enjoyable game experience. Thus, it is important to uncover factors other than enjoyment that might contribute to motivation. Our previous findings, along with the findings reported here, suggest that belief in the efficacy of digital game-based cognitive interventions may also play a role.

The brain fitness game condition was perceived as more frustrating and less enjoyable. This again may be attributed to the fact that these games were designed to challenge participants’ abilities and weren’t designed specifically for entertainment purposes (as the word and number puzzle games in the control condition were). However, that is not to say that the gamification of these brain training tasks was not effective at increasing enjoyment. In order to test this hypotheses, gamified and non-gamified versions of the same cognitive training tasks would need to be compared head-to-head.

Interestingly, participants expected greater benefits in the control condition compared to the intervention condition. This is ideal with respect to being able to eliminate placebo effects as a potential explanation for differences in performance on cognitive outcome measures. If participants demonstrate greater cognitive improvement in the brain training digital game group compared to the control group, despite lesser expectation for improvement, the presence of a placebo effect is unlikely [21].

Limitations.

Our analyses focused on participants who completed the intervention (in the sense that they came back to the lab approximately one month later to complete surveys and the cognitive assessment battery). We did not have perception and motivation data for participants who did not return to the lab to complete this post-training session. Our analyses excluded participants who may be the least adherent (participants that never played their assigned game or had extreme difficulty doing so, and thus dropped out of the study without competing any additional study components). Hence, the results presented here may represent a biased view of the factors that influence motivation and adherence.

This intervention was also relatively short with respect to duration (30 days). Ideally we would like to understand the factors that relate to long-term intervention adherence as it is assumed that greater intervention dose may result in larger benefits. Due to the limited duration of the study, and the knowledge that they were part of a study and they were expected to follow a specific training schedule and record their adherence, intervention adherence may be overestimated in the current study. A similar intervention that asked participants to participate for 3 months found much lower rates of adherence [14].

Future Directions.

Future studies of adherence would likely benefit from a longer study duration (e.g., 6 months or more). Without a greater amount of variability in adherence data it may be difficult to uncover predictive individual difference factors. Future studies might also consider imposing a less strict training schedule, allowing participants more choice with respect to how often to engage with the intervention [e.g., 22]. In our study the imposed training schedule may have largely overridden what an individual might have naturally done. To the extent possible, future studies should also rely on automatic recording of adherence rather than diary data to make the adherence aspect of the study less salient to participants, and potentially, more accurate.