Rehabilitative training has long been used to improve perceptual and cognitive performance in normal (Mahncke et al., 2006; Takeuchi et al., 2010) as well as clinical (learning impairement, Merzenich et al., 1996; e.g., amblyopia, Maurer and Hensch, 2012; stroke, Taub, 2012) populations. However, rehabilitative training, typically consisting of many hours of repetitive practice, is often too effortful and tedious for the intended users (Levi and Li, 2009). In contrast, computer games, which can be entertaining and easily modified to suit different users, have become a ubiquitous part of modern life. Playing fast action video games (with audio effects) has been shown to improve a wide range of visual perception and attention skills (Green and Bavelier, 2003, 2012). Auditory benefits of game play, on the other hand, have typically been studied using laboratory designed auditory games (Honda et al., 2007; Whitton et al., 2014; Wise et al., 2015) rather than entertaining video games. In one notable exception, we previously reported that playing Tetris, a popular arcade type video game involving fast visual-motor control, improved auditory perception (Amitay et al., 2006). This result raised the possibility that existing video games may be utilized for auditory rehabilitation or enhancement of auditory skills. Here we re-examine the supramodal effect of video game play on auditory learning.

In our previous report (Amitay et al., 2006), after an hour of Tetris play, either in silence or accompanied by passive presentation of tones, players improved on tone frequency discrimination (FD), and the FD improvement was correlated with Tetris learning. There are at least three possible mechanisms for Tetris play to improve auditory perception. The first account is that auditory learning of Tetris players may reflect general behavioral improvements induced by familiarization with the training procedures and environment. This is often referred to as procedural learning. It occurs at the beginning of training and saturates rapidly (Hawkey et al., 2004; Ortiz and Wright, 2009). In the previous one-session training experiment (Amitay et al., 2006), FD testing prior to training was very brief (∼2 min). Procedural learning could therefore have continued into the training session. Second, playing Tetris may improve some supra-modal cognitive functions such as attention or working memory (WM), which in turn improves auditory perception. For example, we have shown that WM training can improve FD performance (Zhang et al., 2016). A third possibility is that game play interacts across time and modality with auditory experiences, enhancing rather than generating auditory learning. Supporting this possibility, within-modal learning enhancement has been reported for contiguous auditory (Wright et al., 2010) and visual (Szpiro et al., 2014) stimulus exposure.

Here, we tested the effect of Tetris play on auditory learning using a multi-session training paradigm designed to distinguish these three possible mechanisms (Figure 1). We trained FD with a roved standard frequency (FD-rove; Figure 1B), a task that we have previously shown to produce learning over multiple sessions (Amitay et al., 2005) and transfer to auditory WM (Zhang et al., 2016). After a 90-min pretest, four training groups practiced for four daily sessions on either FD, Tetris in silence, FD interleaved with Tetris, or FD interleaved with another mental rotation task. All groups were compared with an untrained control group for evaluation of training. The procedural learning hypothesis predicts no Tetris effect because the pretest session was long enough to saturate procedural learning (Hawkey et al., 2004). The cognitive learning hypothesis predicts auditory benefits for practicing Tetris only. In contrast, the learning enhancement hypothesis predicts auditory benefits for Tetris interleaved with FD, but not for Tetris only. Finally, if the Tetris effect arises from the visual spatial skills involved in Tetris play, a similar effect should be observed with the mental rotation task.

The results demonstrate that game play enhances perceptual learning of contiguous auditory experiences. Confirming and extending our previous report of auditory benefits of Tetris play in brief, single-session training (Amitay et al., 2006), we showed in a multi-session training paradigm that Tetris play in isolation produced no benefits for an FD-rove task. This indicated that the previously observed auditory benefits did not result from improved supra-modal cognitive function. However, Tetris play did enhance this form of FD learning and its transfer to WM when mixed with auditory training. The learning enhancement cannot be attributed to procedural learning or familiarization with the testing environment, as the pretest session was long enough to saturate learning on the easier FD-fixed task. Nor can the enhancement be attributed to the extended training duration or visual spatial training brought by Tetris play, as the enhancement was gone when Tetris was replaced by a visual mental rotation task with similar stimuli.

The supramodal nature of the observed learning enhancement challenges existing theories of perceptual learning. In training, mixing two perceptual tasks has been shown to produce transfer between the tasks both in vision (Zhang et al., 2010) and in audition (Wright et al., 2010). It was suggested that this intramodal transfer results from the application of neural computation (Zhang et al., 2010) or resources (Wright et al., 2010) required for learning the target task to sensory inputs of the non-target task. In both views, the role of the non-target task is to provide exposure to the stimulus feature to be learned. However, in the current study, no such sensory stimulation for auditory learning was provided by Tetris, which was played in silence. Another example of enhanced learning through a non-target task is an increase of sensitivity to background stimuli while performing a foreground task (Watanabe et al., 2001). Seitz and Watanabe (2005) proposed that diffuse reinforcement signals produced by a foreground (target) task enhanced processing of concurrent background (non-target) stimuli. However, the task-irrelevant learning occurred only when the background stimuli were subliminal and coincident with the reinforced targets of the foreground task (Choi and Watanabe, 2009). In the current study, learning enhancement violated both of these requirements.

The game training literature provides no clue that we can find to the nature of the supramodal enhancement of auditory learning and transfer by Tetris play. Though playing action games improves visual perception and attention, Tetris training alone does not improve unrelated perceptual or cognitive tasks (Sims and Mayer, 2002; Green and Bavelier, 2003). Consistently, in the current study, we found no evidence that playing Tetris had any auditory benefits beyond the pretest. Indeed, auditory benefits of game play have only been reported for games specially designed to emphasize the role of auditory stimuli (Honda et al., 2007; Whitton et al., 2014). In contrast, the game-enhanced perceptual and cognitive learning observed here required no auditory stimuli during game play and must therefore involve different mechanisms from those induced by game play alone.

The results indicate that auditory training and transfer was enhanced by some factors that were provided by Tetris play but not by the visual-spatial stimulation involved in the game. One candidate of such learning-enhancing factors is reward signals. Playing video games, even the arcade-style ones, can lead to release of reward signals such as dopamine (Koepp et al., 1998), which promotes synaptic plasticity (Buchanan et al., 2010; Giessel and Sabatini, 2010) and contributes to experience-dependent learning (for review, see Harley, 2004; Glimcher, 2011). Conventional behavioral training, like the auditory discrimination training examined in the current study, involves extensive repetition that can be far from rewarding or motivating. Compared with game play, such training may be accompanied by a low level of reward signals, making it sensitive to the learning-boosting effects brought by an increase of reward signals. For example, coupling auditory stimuli with electrical stimulation of the dopamine network has been shown to enable and enhance cortical reorganization in primary as well as association cortices (Bao et al., 2001). Game play could function as a behavioral stimulator of the reward system, promoting learning of contiguous sensory experiences in a similar way as stimulating the dopamine network enhances reorganization of cortical responses to accompanied auditory stimuli (Bao et al., 2001).

Note that the learning-enhancing effect of game play requires no temporal coupling of the rewarding event (game play) with the target training (auditory discrimination), unlike the aforementioned task-irrelevant learning, which occurs only to the subliminal background stimuli that are synchronous with the presumably reward-producing target (Watanabe et al., 2001). As proposed for the task-irrelevant learning, in conventional training reward signals are considered to play a reinforcing role, in that their fluctuations in level allowed trial-by-trial feedback to modulate behavior (Watanabe et al., 2001). In contrast, by mixing game play with training, we seek to increase systematically the level of reward signals throughout training. The reward signals produced by game play may not be informative in improving performance on the perceptual task on a trial-by-trial level. Rather, reward signals may purportedly gate neural plasticity by encoding the ecological significance of the experiences (Gee and Oertner, 2016). Consistent with the plasticity modulating role, dopamine activity has been shown to promote subsequent associative learning (Popescu et al., 2016) and consolidation of associative memory (Schicknick et al., 2012) in animals. In humans, orally taken dopamine precursors, which would increase plasma dopamine level for tens of minutes to hours, have been reported to enhance motor (Floel et al., 2005) and word (Knecht et al., 2004) learning. In this view, the proximity of game play and perceptual training could enhance the significance, and hence the induced plasticity, of the training. Consequently, the benefits may not depend on task or stimulus configuration, making training-while-playing widely and conveniently applicable in educational and rehabilitative settings.

This study was carried out in accordance with the recommendations of the Nottingham University Hospitals Research Ethics Committee with written informed consent from all subjects.

Y-XZ and SA conceived and designed the study. Y-XZ performed the experiments, Y-XZ and D-LT analyzed the data, and prepared the manuscript with contributions from SA and DM.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer SA and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.