Task switching in video game players: Benefits of selective attention but not resistance to proactive interference

Abstract

Research into the perceptual and cognitive effects of playing video games is an area of increasing interest for many investigators. Over the past decade, expert video game players (VGPs) have been shown to display superior performance compared to non-video game players (nVGPs) on a range of visuospatial and attentional tasks. A benefit of video game expertise has recently been shown for task switching, suggesting that VGPs also have superior cognitive control abilities compared to nVGPs. In two experiments, we examined which aspects of task switching performance this VGP benefit may be localized to. With minimal trial-to-trial interference from minimally overlapping task set rules, VGPs demonstrated a task switching benefit compared to nVGPs. However, this benefit disappeared when proactive interference between tasks was increased, with substantial stimulus and response overlap in task set rules. We suggest that VGPs have no generalized benefit in task switching-related cognitive control processes compared to nVGPs, with switch cost reductions due instead to a specific benefit in controlling selective attention.

Introduction

As playing video games has become an increasingly popular and widespread activity over the past several decades, research into the potential perceptual and cognitive effects of video game play has similarly developed. Following initial video game-related research focusing on transfer of training (e.g., Fabiani et al., 1989, Gopher et al., 1994), an increasing number of authors have become interested in investigating how expert video game players (VGPs) may differ from non-video game players (nVGPs), in terms of specific underlying mental processes. Visual perception and attention have been particularly well represented in studies to date. Superior ability has been reported for VGPs compared to nVGPs in divided visual attention (Greenfield, deWinstanley, Kilpatrick, & Kaye, 1994) and spatial attention via the useful field of view task (Feng, Spence, & Pratt, 2007). Similar findings have been demonstrated in children, including benefits in selective attention (Blumberg, 1998), and attentional capacity via multiple object tracking (Trick, Jaspers-Fayer, & Sethi, 2005).

Support for these findings can be found in a series of studies conducted by Green and Bavelier, 2003, Green and Bavelier, 2006a, Green and Bavelier, 2006b, Green and Bavelier, 2007, who have consistently demonstrated that VGPs outperform nVGPs on a variety of tasks that tap visuospatial attentional processing, and that such benefits appear to be trainable to a non-game playing population. In their earlier work, Green and Bavelier (2003) demonstrated VGP performance benefits in an attentional blink task, with better T1 identification and T2 detection compared to nVGPs. From these data, they suggested that expert video game players may have greater control over task switching in addition to better temporal attentional processing. Through their subsequent work with multiple object tracking (2006a, 2006b), and visual crowding (2007), Green and Bavelier argued that their findings indicated that VGPs’ superior performance on complex visual processing tasks was likely the result of changes in the fundamental characteristics of the visual system brought about by extensive gameplay experience, and that it remained to be determined if there were also improvements to higher-order processing and cognitive control mechanisms. Castel, Pratt, and Drummond (2005) also found performance benefits for VGPs versus nVGPs using cuing and visual search paradigms. While VGPs were faster overall, and showed some benefit for more efficient self-directed visual search, they showed very similar patterns of lower-level effects, such as cuing and inhibition of return. From these data, Castel et al. (2005) suggested that VGPs may instead have a benefit in higher-level executive control processes, allowing for more efficient control and allocation of selective attention, and the ability to more rapidly establish stimulus–response mappings.

Recently, several authors have more directly examined whether video gaming expertise may be related to differences in cognitive control, specifically processes involved in task switching. Task switching paradigms typically measure the effects of various factors on task switching cost, defined as the difference between performing a task for a second time in sequence (repeat trials) compared to performing a task for the first time in sequence following a previous different task (switch trials). Andrews and Murphy (2006) used an alternating-runs (AABB task sequence) task switching paradigm based on methods from Rogers and Monsell (1995), and demonstrated that VGPs showed smaller task switching costs than nVGPs when response-to-stimulus durations were relatively short (150 ms in their study). Boot, Kramer, Simons, Fabiani, and Gratton (2008) investigated a range of cognitive abilities in a training study comparing VGP and nVGP participants, with executive control assessments including task switching and working memory operation span. They found no difference in operation span between gaming groups or as a result of training novices on a range of video games for 20 or more hours. Expert gamers showed smaller switch costs compared to novices, but video game training did not affect switch costs.

Our present study sought to more carefully distinguish what aspects of task switching and related cognitive control processes might selectively differ between expert video game players and non-video game players. We conducted two experiments to more closely examine VGP versus nVGP differences with respect to factors known to influence task switching processes, including the amount of time and information available prior to stimulus onset during which endogenously driven task set reconfiguration can be performed, and the degree of stimulus and response overlap between tasks.

In Experiment 1 we manipulated a range of stimulus, response and cuing parameters, to parametrically vary the difficulty in preparing for and responding to a given trial, in addition to a basic task switching manipulation. Cuing and trial timing parameters were manipulated so that even when response mappings were difficult, substantial endogenous preparation for a particular trial was possible given an informative cue and a longer cue-to-stimulus interval in some conditions. Although Experiment 1 employed randomized shifts between two sets of semantically distinct stimuli (letters A, B, C, and digits 1, 2, 3), we reduced the degree of overlap of task set rules, and hence the likely extent of trial-to-trial interference, by having no stimulus or response overlap between tasks, and having a direct univalent 1-to-1 mapping of each individual stimulus to a separate manual response. In effect, this task could have been conceptualised as a single task requiring mapping of six distinct stimuli to six distinct responses, with no requirement for any real switch of task set rules. This design also allowed us to assess VGP versus nVGP differences in a range of other effortful, selective attention-demanding processes, independent of task switching behaviour.

For Experiment 2, we employed a different task switching paradigm based on Arbuthnott and Frank’s (2000) method for demonstrating additional reaction time costs for task alternation as compared with simple task switching—for example, longer reaction times on the final Task C in a C–B–C task sequence compared with an A–B–C task sequence. These findings have generally been taken as evidence for active inhibition of recently abandoned task set representations, a phenomenon that has come to be termed Backward Inhibition (Arbuthnott, 2005, Arbuthnott and Woodward, 2002, Mayr and Keele, 2000). In this experiment, we employed tasks with extensively overlapping task set rules, where the same six stimuli were remapped to two alternative responses for each task. As such, this required endogenous reconfiguration of task set based only on a pre-stimulus cue, in the face of likely substantial task switching-related proactive interference (Wylie & Allport, 2000).

Throughout this study, we were interested to see what aspects of task switching performance might differentiate video game experts from non-video game players, and whether other aspects of performance close to but separate from task switching itself might be revealed as more distinguishing of VGP and nVGP groups. To anticipate our results somewhat, the results from Experiment 1 demonstrated selectively better performance for VGP participants in a small set of conditions, reflecting a superior ability to actively prepare for an upcoming task when time and information were available, including relatively greater reductions in task switching costs under these conditions. However, while Experiment 2 showed faster performance for VGP versus nVGP groups in general, video gamers showed no selective benefit at all with regard to task switching costs. We consider these results with respect to likely component processes involved in task switching performance, including effortful selective attention-dependent preparation for upcoming task performance, and processes involved with resolving proactive interference arising from successive performance of different but overlapping tasks.