Cognitive control and cognitive interventions

The development of new cognitive interventions that can be effectively applied to compensate for age-related cognitive decline is of key concern to contemporary aging research (see Green and Bavelier, 2008; Hertzog et al., 2008; Papp et al., 2009 for reviews). Early cognitive interventions demonstrated improved performance on the previously trained task but many failed to demonstrate transfer of learning to other laboratory tasks or real-world activities. For example, a meta-analysis of 33 studies of cognitively healthy older adults trained in the use of some mnemonic technique (such as the method of loci or imagery) concluded that performance improvements were largely specific to the previously trained mnemonic technique (Verhaeghen et al., 1992). Later cognitive interventions trained cognitive control and more successfully demonstrated transfer of learning to other laboratory tasks (Kramer et al., 1995; Bherer et al., 2005, 2008) and real-world activities (Willis et al., 2006). For example, training younger and older adults to control attention in one dual-task situation improved their ability to control attention in another dual-task situation (Kramer et al., 1995).

Cognitive control and computer-based games

Training cognitive control within the context of complex computer-based games may be particularly important for demonstrating transfer of learning to a number of different tasks. For example, a recent cognitive intervention that involved training older adults to play an off the shelf computer-based strategy game called Rise of Nations demonstrated improved performance on a number of different tasks including task-switching, working memory, visual short-term memory, and reasoning (Basak et al., 2008; for a similar finding in older adults see Drew and Waters, 1986; for similar findings on computer-based games in younger adults see Frederiksen and White, 1989; Gopher et al., 1989, 1994; Green and Bavelier, 2003, 2006, 2007). Cognitive interventions that utilize less complex computer-based games such a Super Tetris have also demonstrated transfer of learning but performance improvements seem to be limited to response times (Goldstein et al., 1997; Salthouse and Somberg, 1982; Clark et al., 1987). Thus, the complexity and dynamics of the computer-based game – and its overall demands for cognitive control – may be critical for demonstrating transfer of learning to a number of different tasks and real-world activities. Complex computer-based games also lend themselves to training programs that extend over weeks and or months, another factor that may be important for the development of more enduring and stable capabilities that can transfer to performance on real-world activities.

Cognitive control and the space fortress game

In the current study, we trained cognitively healthy older adults to play a complex computer-based action game, called the SF game. The SF game was originally developed to study complex skill acquisition (Gopher et al., 1989; Mané and Donchin, 1989) and was recently revised to accommodate contemporary computer environments (Shebilske et al., 2005, based on a version developed by Gopher et al., 1994 to train flight trainees). The primary goal of the SF game is to shoot missiles and destroy a space fortress while protecting your spaceship against damage. The spaceship and the missiles are controlled with a joystick. In addition to defending your spaceship against missiles shot by the space fortress the player has to defend herself against mines that intermittently appear on the screen. The player is also asked to monitor symbols that randomly appear on the screen and add to the participants’ total score. Thus, the SF game is a fast-paced and dynamic computer game that requires divided attention, continuous and discrete motor control, visual search, working memory, long-term memory, resource management as well as decision-making processes. An individual's ability to control or switch their attention to these different task requirements is considered vital for successful SF game performance (Gopher et al., 1994; Arthur Jr. et al., 1995; Gopher, 2006). In other words, the SF game demands a considerable amount of cognitive control.

To date, older adults have not been trained to play the SF game. Thus, it remains unclear if and how older adults would learn to play this highly complex computer-based action game. The current study examined if and how cognitively healthy older adults learn to play the SF game. More specifically, we wanted to identify and determine the efficiency of the strategies that older adults adopt during training with the SF game – as a function of training instructions. Such an examination is critical for the further development of this training program as a cognitive intervention that can potentially compensate for age-related cognitive decline. In younger adults, training with the SF game has been shown to improve performance on other laboratory tasks (Fabiani et al., 1989; Frederiksen and White, 1989) as well as real-world tasks such a flight performance (Gopher et al., 1994). These findings from younger adults suggest that training cognitive control with the SF game can potentially be developed into a cognitive intervention that can compensate for age-related cognitive decline.

The space fortress game and emphasis change instructions

In younger adults, EC instructions – play the whole game but change your emphasis to different sub-components of the game at different points during training – are particularly beneficial to SF game performance. For example, training with EC instructions leads to better SF game performance than training with standard game instructions (i.e., play the whole game and develop learning strategies ad lib) in younger adults (Gopher et al., 1989). EC instructions are argued to encourage individuals to explore alternative learning strategies and further increase the demand for cognitive control. More specifically, EC instructions encourage individuals to concentrate on gaining control of the ship (ship handling emphasis), efficiently managing friend and foe mines (mine handling emphasis) or both. More specifically, controlling the ship in order to shoot and destroy the space fortress is expected to improve during ship handling emphasis while defending the ship against mines is expected to improve during mine handling emphasis.

One study suggest that hierarchical part-task instructions – successively learning to play the SF game by practicing tasks that engage only one, or a few different SF task requirements – can lead to better SF performance than EC instructions (Fabiani et al., 1989). However, in this study SF performance of individuals trained with EC instructions were also more resistant to interference when a secondary task was introduced compared to the SF performance of individuals trained with hierarchical part-task instructions. This finding suggests that EC instructions increases the demand for cognitive control processes and may be particularly important for incorporating a secondary task and demonstrating transfer of learning (Gopher et al., 1989, 1994; Kramer et al., 1995; Gopher, 2007).

Additional evidence for the above proposition is obtained in a recent imaging study of younger adults trained to play the SF game with standard and EC instructions (Erickson et al., 2010). In this study, the volume of both the dorsal and the ventral striatum predicted initial SF acquisition but only the volume of the dorsal striatum predicted later improvements in SF performance. More importantly, the volume of the dorsal striatum predicted later improvements in SF performance following EC instructions but not following standard instructions. These findings are consistent with suggestion that the ventral striatum is involved in early reinforcement-related learning while the dorsal striatum is involved in both early and later motor-related and association-related learning and cognitive control (Desmond et al., 1998; Marie et al., 1999; Lawrence et al., 2000; Belin and Everitt, 2008). Thus, it seems reasonable to conclude that EC instructions during SF game learning place a particularly high demand on cognitive control. In the current study, we addressed this issue in cognitively healthy older adults by comparing SF game learning following EC instructions and standard game instructions. Although older adults have not been trained to play the SF game, they have been shown to benefit from EC instructions in a dual-task situation (Kramer et al., 1995).

Apparatus

The revised version of the SF game (Shebilske et al., 2005) was installed on three networked PC computers. The primary goal of this game was to shoot missiles and destroy a space fortress while protecting your spaceship against damage. The spaceship and the missiles were controlled by the player with a joystick (Flightstick Pro) and the spaceship moved in a frictionless environment. Any forward movement of the joystick made the spaceship accelerate and any lateral movement of the joystick made the spaceship rotate. If the ship was steered off the screen, it would reappear (or wrap) to the other side of the screen. If a button on top of the joystick was pressed, a missile was fired in the direction of the spaceship. In addition to defending the spaceship against missiles shot by the space fortress, the player had to defend themselves against mines that appeared on the screen every 4s. There were two types of mines: friend mines and foe mines. When friend mines were shot they were energized, ran into the space fortress and scored a hit. When foe mines appeared, the player had to switch the weapon system by pressing the right mouse button twice and then eliminate the mine with a missile. Prior to each trial, the player was told what (3) letters signified foe mines. The space fortress had to be shot 10 times with more than 250ms between each shot before it became vulnerable, then the space fortress could be destroyed with a double shot (i.e., two shots against the space fortress administered with no more than 250ms between each shot). Bonus points could also be obtained by monitoring symbols that regularly popped up on the screen. Whenever the player saw 2$ signs in a row, she could press the left mouse button and choose to receive bonus points or additional missiles. An approximate sketch of the SF game environment and its components are provided in Figure ​Figure1.1. One particularly nice feature of the SF game is that it can be programmed to compute sub-scores that reflect the different task requirements, in addition to overall or Total SF performance. These sub-scores can then be used to more closely analyze the different task requirements. A breakdown and explanation of the different sub-scores that were computed in our version of the SF game are provided in Table ​Table2.2. Note that some of these sub-scores (Points, Control, Velocity, and Speed) also appear at the information panel at the bottom of the SF game environment, together with counters referring to the vulnerability of the space fortress, the letter associated with a mine, the interval between shots and the number of shots administered.

Open in a separate window

Figure 1

A Schematic of the revised Space Fortress game environment.

Total score

The first GLM revealed an effect of Week, F(11, 341)=15.58, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.34;$ Condition, F(1, 31)=4.53, p<0.05, ${\eta }_{\text{p}}^{\text{2}}=0.13;$ and Aiming F(1, 31)=61.51, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.67.$ However, there was also an interaction between Week and Condition, F(11, 341)=2.71, p<0.05, ${\eta }_{\text{p}}^{\text{2}}=0.08$ such that participants in the AC and the EC condition started off with similar performance during the first weeks of training [F(1, 32)=1.82, p>0.05] but the AC condition (improved more or) did better than the EC condition during the last week of training [F(1, 32)=8.06, p<0.01, ${\eta }_{\text{p}}^{\text{2}}=0.20;$ see Figure ​Figure2].2]. There was also an interaction between Week and Aiming, F(11, 341)=8.08, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.21$ such that aiming had more of an effect during the later weeks of training. The three-way interaction between Week, Condition, and Aiming was not significant, F(1, 31)=1.10, p>0.05.

Open in a separate window

Figure 2

Total SF performance as a function of training instruction (EC vs. Standard), adjusted by subject-specific aiming scores.

Taken together, these Total Score analyses suggest that older adults learned to cope with the high demands of the game, improved with practice and that performance changed as a function of the instructions provided during training as well as with subject-specific aiming scores. Moreover, older adults in the AC condition performed better than older adults in the EC condition.

Main sub-measures

Main sub-measures of SF performance include Points, Control, Velocity, and Speed (see Table ​Table2).2). The second GLM included these measures into a repeated-measures MANCOVA and revealed a main effect of Week F(44, 1295)=4.44, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.15$ and Aiming F(4, 28)=18.87, p>0.05, ${\eta }_{\text{p}}^{\text{2}}=0.73,$ but no effect of Condition, F(4, 28)=1.22, p>0.05. There was a two-way interaction between Week and Condition, F(44, 1295)=1.99, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.06$ and Week and Aiming F(44, 1295)=4.14, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.12$ as well as a three-way interaction between Week, Condition, and Aiming, F(44, 1295)=1.48, p<0.05, ${\eta }_{\text{p}}^{\text{2}}=0.05.$

Follow-up univariate models suggest that the Control and the Speed measure were influenced by subject-specific Aiming scores, F(1, 31)=37.23, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.55$ and F(1, 31)=31.73, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.51,$ respectively. In the Speed measure, there was also an interaction between Week and Aiming, F(11, 341)=2.36, p<0.05, ${\eta }_{\text{p}}^{\text{2}}=0.06,$ such that aiming had more of an effect during the later weeks of training. An interaction between Week and Aiming was also observed in the Points measure, F(11, 341)=7.32, p<0.001, ${\eta }_{\text{p}}^{\text{2}}=0.19,$ such that Aiming had more of an effect during the early weeks of training than during the later weeks of training. Note that this is the opposite pattern to the interactions between Week and Aiming that were observed in Speed and Total Score, where aiming had more of an effect during the later weeks of training. In the Velocity measure, there was a three-way interaction between Week, Condition, and Aiming, F(11, 341)=3.62, p<0.05, ${\eta }_{\text{p}}^{\text{2}}=0.10.$ It is difficult to visualize a three-way interaction between two continuous variables. For illustrative purposes, we therefore visualized the nature of this three-way interaction by dichotomizing Aiming in terms of high (>450) and low (<450) aiming scores and graphing velocity for high and low aimers in the EC and AC condition separately (see Figures ​Figures3A,B).3A,B). As is evident from these graphs, velocity in participants with low aiming scores were greater among participants assigned to the AC condition during the early weeks of training, but similar or worse than among participants assigned to the EC condition during the later weeks of training. Velocity in participants with high aiming scores did not differ as a function of training instructions. These main sub-measures analyses suggest that flying the space ship within the hexagon (Control) is influenced by the subject-specific aiming scores of older adults. The speed or efficiency to detect and destroy mines (Speed) is also influenced by subject-specific aiming scores, particularly during the later weeks of training. The numbers of Points collected from the different task requirements, on the other hand, are influenced by subject-specific aiming scores during the early weeks of training but less so during later the later weeks of training. In order to examine why this is the case, we analyzed each measures that contributes to the Points measure in the upcoming section. Finally, flying the ship slowly within the SF game environment (Velocity), which is related to ship handling and has been shown to be a good learning strategy among younger adults (Frederiksen and White, 1989), improved more in the EC condition for individuals with low subject-specific aiming scores – but not for those with high subject-specific aiming scores. This pattern is consistent with what is observed in younger adults with high basic motor control – whose velocity typically improve more when trained with EC instructions compared to when trained with standard game play instructions (e.g., Gopher et al., 1989). Taken together, these main sub-measure analyses of SF game learning in older adults suggest that training instructions and basic motor control abilities influence the strategies that older adults use when they learn to play the SF game.

Open in a separate window

Figure 3

(A) Velocity as a function of training instructions (EC vs. Standard) in participants with low (<450) subject-specific aiming scores. (B) Velocity as a function of training instructions (EC vs. Standard) in participants with high (>450) subject-specific aiming scores.

We would like to thank Alexis Richards and Leigh Rich for their research assistance and Brian C. Rakitin for his statistical advice. We would also like to thank the Taub Foundation for providing funding for this research.