Elsevier

Journal of Neuroscience Methods

Volume 289, 1 September 2017, Pages 99-109
Journal of Neuroscience Methods

Research Article
The AttentionTrip: A game-like tool for measuring the networks of attention

https://doi.org/10.1016/j.jneumeth.2017.07.008Get rights and content

Highlights

  • The measurement of the components of attention, using the Attention Network Test (ANT) has burgeoned in the last decade.

  • The test is unengaging which poses a problem for some groups (e.g. children) or when multiple test administrations is required.

  • Our development of a more engaging, game-like task, the AttentionTrip@ is described.

  • Encouraging results demonstrate that the AttentionTrip@ can be used to measure alerting, orienting and executive control.

Abstract

Background

Recognizing that attention is not a unitary system, the Attention Network Test (ANT) and its variants were developed to measure the efficacy of the multiple components of attention. One potential weakness of these tests (ANTs) is that they are unengaging. This poses a problem when particular groups are tested (e.g., young children), when more stable measures of performance are desirable (and can only be achieved in longer testing sessions) and when repeated testing is necessary.

New method

Here we describe the evolution of a game-like tool, which we call the AttentionTrip©, that is suitable for investigating three isolable attentional networks (alerting, orienting, and executive functions).

Results

Utilizing this tool we were able to generate reasonable network scores for alerting, executive control (from both the flanker and Simon effects), endogenous orienting and, after some motivated modifications, exogenous orienting.

Comparison with existing method(s)

Split-half reliabilities of the alerting and executive (flanker) network scores were considerably higher than those reported by MacLeod et al. (2010) in their psychometric review of the ANT. Informal observations (e.g., some participants asking if they could keep doing the task when their session was over) suggesting that the AttentionTrip is considerably more engaging than the traditional ANT have been confirmed in a head-to-head comparison (Vallis & Klein, 2016).

Conclusions

The AttentionTrip@ is available now for research purposes. A tablet version, which will have greater clinical utility, is under development.

Introduction

Attention is a multidimensional construct at the heart of cognitive functioning. It has been proposed to be composed of multiple separate, yet interacting neural systems that underlie functions related to alerting, orienting and executive control (Posner and Petersen, 1990, Petersen and Posner, 2012). Neurological and behavioural research has supported Posner’s concept of a three component model of attention, whereby separate yet interacting attentional brain networks serve the underlying functions related to alerting, orienting and executive control (Posner and Boies, 1971, Posner and Petersen, 1990, Posner and Raichle, 1994). Moreover, neuroimaging research (e.g., Fan et al., 2005) and research with patient populations has supported this 3 component model, mapping these attentional systems to specific anatomical areas of the brain.

In this paper, after describing these three networks of attention and traditional tests that were developed to assess their efficacy, we will present the evolution of our game-like task, the AttentionTrip, designed with the same underlying purpose. Our primary rationale for developing this game-like assessment task was to be able to offer to scholars a more engaging tool for measuring the networks of attention than the standard Attention Network Tests (ANTs). A recent meta-analytic review (Lumsden et al., 2016) concluded that our motivation (increasing engagement) was by far the most frequent motivation they uncovered for the gamification of tools aimed at cognitive assessment. Although this indicates that by our motivation we are in good company (e.g., see Abeele et al., 2015, Belmonte, 2008), we are not aware of any similar effort to ours for measuring the networks of attention as described by Posner and colleagues.

Alerting is our ability to develop and maintain an alerted state over short periods of time and is usually explored by providing a warning signal some time before a signal requiring a speeded response. The right cerebral hemisphere in the frontal and dorsal parietal regions has been shown through neurophysiological research to be central to this alerting component of attention (Heilman et al., 1985, Posner and Petersen, 1990, Petersen and Posner, 2012). Orienting entails selective adjustments that direct one’s attention toward a particular modality, spatial location or stimulus feature. These adjustments may controlled in two different manners (Jonides, 1981, Klein, 2009): “top down” by the individual’s intentions and habits (e.g., when we try to keep our eyes on the road while occasionally checking the rear-view mirror or road signs to ensure we get to our destination) or “bottom-up” by sudden and/or unexpected stimuli (e.g., a door opening or child unexpectedly darting out in traffic). In both situations, orienting necessitates that attention be disengaged from its present focus and re-engaged on the new pathway of interest. Recent evidence suggests two interacting networks in the frontal and posterior parietal areas of the brain are involved with spatial orienting in the visual modality (Corbetta and Shulman, 2002, Petersen and Posner, 2012). Executive control is a complex function that has been closely associated with monitoring for and resolving conflicting mental states. Executive control is thought to allocate resources during a wide range of situations that can include everything from learning novel sequences of actions, to holding or manipulating information, to dealing with errors/conflicting responses, etc. The executive control network has been conceptualized as two isolable systems within the medial frontal cortex and anterior cingulate cortex (Petersen and Posner, 2012).

Whether or not one accepts this particular taxonomy of attention (see Klein and Lawrence, 2011, for one effort to improve upon it), these components, or networks, of attention play a critical role in many of our daily activities and one or more of them can be severely impaired in several neurological conditions. To effectively help any person exhibiting any attentional deficits due to brain injury or disease, one needs to successfully measure the efficiency of these attentional abilities. Thus, in 2002, Michael Posner and colleagues (Fan et al., 2002) put forward a simple tool, the Attention Network Test (ANT) which, through the thoughtful application of mental chronometry (Posner, 1978, Klein, 2003) could be used to simultaneously measure these three isolable components of attention.

In subsequent studies this tool was combined with genetic, neuroimaging, neuropsychological and purely behavioural methods in explorations of how these networks are influenced by nature and nurture in their development, how they breakdown in psychiatric disorders and following brain damage, how they are represented in the brain, and how they might be influenced by training and meditation (Tang et al., 2007, Fan et al., 2005, Ishigami and Klein, 2009, Posner et al., 2002, Rueda et al., 2005).

The ANT and its several variants (ANTs [including: child ANT (Rueda et al., 2004), ANT-Interaction (Callejas et al., 2004), lateralized ANT (Greene et al., 2008), auditory ANT (Roberts et al., 2006), Combined Attention Systems Test (Lawrence et al., 2011), ANT-R (Fan et al., 2009), etc., have since become widely used as cognitive neuropsychological and cognitive neuroscientific tools (in the last 5 years almost 300 studies using at least one of these tests were published). During the ANT task (as developed by Fan et al., 2002), participants are required to respond during each trial to whether a target an arrow points leftward or rightward (see Fig. 1a). The target arrow can be accompanied by irrelevant flanking arrows that point in the same direction as the target (congruent) or in the opposite direction (incongruent). The target array is preceded by one of 3 cue types or no cue. When a single asterisk appears above or below the fixation point (the spatial cue), this indicates with 100% probability where the target array would be presented. Neither the single cue presented at fixation nor the double cue provide information about the upcoming target's location, but compared to the no-cue condition, these cues do provide a warning signal letting the participant know when the target array would be presented. Following a strategy first put forward by Donders (1868), see Klein, 2003, for a modern description), the subtractions described in Table 1 are used to assess the efficacy of three isolable components of attention.

The ANT-Interaction (ANT-I) introduced two changes to the ANT. A short duration tone was introduced as an alerting signal and the peripheral cues were made uninformative. The presence versus absence of the alerting tone allows experimenters to measure alerting and was intended to allow measurement of the interaction between alerting and orienting (an interaction that could not be assessed with the original ANT). Because the peripheral cues in the original ANT were 100% valid (predicted the location of the target perfectly) the orienting generated by these cues could be a hybrid of exogenous and endogenous control (Klein, 2009; but see discussion). In the ANT-I the use of uninformative peripheral cues enables the measurement of a pure form of exogenous orienting.

One potential weakness of all the traditional ANTs (including the child ANT and ANT-I) is that they are unengaging. This poses a problem when more stable measures of performance are desirable (and can only be achieved in longer testing sessions), when repeated testing is necessary (as in longitudinal studies or studies tracking the effectiveness of a training or remediation program) and when easily bored participants are tested (e.g., Ishigami and Klein, 2015).

We had previously developed a game-like driving task (SimonShip) to measure a spatial-compatibility effect (Klein et al., 2011). When participants are performing a 2-choice discrimination of a non-spatial feature of a target, their response times are affected by the spatial correspondence between the location of the responding effector and the task-irrelevant location of the target, with response times increasing as this spatial correspondence decreases. First observed by Dick Simon in the mid-1960s (Simon and Rudell, 1967), this has since been called the Simon effect. Our rationale was rooted in a skeptical comment made by a reviewer of an eventually published manuscript (Klein et al., 2006) in which it was demonstrated that an approximately 25 ms Simon effect was robust regardless of the placement of the responding fingers. The reviewer scoffed at the relatively small magnitude of the effect wondering what could be the impact of a ∼20 ms effect outside the laboratory. We developed the SimonShip “game” (see Fig. 1b–d) as a simulation of two “real-world” situations (first described in Klein et al., 2011). The task required different (speeded) responses to green and blue targets and a story line describing steering “a spaceship through a worm-hole” or “a miniaturized medical repair-ship through a blood vessel”. Although we have tended to use these story lines interchangeably, based on the suggestion of an anonymous reviewer we think in the future it might be wise to allow the participant to select which story line they prefer.

We conducted several experiments using the SimonShip task. In all of them we obtained substantial Simon effects which are presented in Table 2 along with data from Klein et al. (2006). With this program providing the foundation, we developed “AttentionTrip©,” a tool that was designed to allow us to measure the components of attention while participants perform the same, engaging, task.1 As will be described below, to do this we added visual and auditory cues that were presented before the target and we added congruent and incongruent flankers that could be displayed along with the target. To assist the reader in appreciating how this task appears to participants we have created a demonstration video which is posted here:

https://youtu.be/T5CuMRKTUHw

Section snippets

Experiment 1: endogenous orienting

Four females and five males participated, all Dalhousie University undergraduates with normal or corrected to normal vision. The average age was 21.9, with participants ranging from 19 to 30 years of age.

Experiment 2: exogenous orienting

Eleven females and one male participated, all Dalhousie University undergraduates with normal or corrected to normal vision. The average age was 20.7, with participants ranging from 17 to 25 years of age.

Experiment 3: exogenous orienting with modified cues

Nine females and three males participated, all Dalhousie University undergraduates with

Network scores

The network scores from Experiment 1 are presented in the upper panel of Fig. 4. These were all significantly greater than zero and the magnitudes of the standard alerting, orienting and executive control (from flankers) scores are within the range of what we expected based on a previous psychometric study of the ANT (MacLeod et al., 2010, Fig. 3).

The network scores from Experiment 2 are presented in the middle panel of Fig. 4. With the exception of orienting these were all significantly

Discussion

The primary aim of this ongoing project is to develop an engaging tool, that we call the AttentionTrip, that can be used by researchers to assess the components of attention in their participants by generating attention network scores. The primary aim of the present manuscript is to describe the evolution of this tool and report on our efforts so far. Our development efforts are not complete, and toward the end of the paper we will describe some of the further avenues we are exploring or plan

Next steps

One motivation for the development of the AttentionTrip was our perception of the need for a more engaging task for measuring the efficacy of the attention networks. A reader might very well wonder whether we have accomplished this goal; and the answer is “yes”. In a separate project (Vallis and Klein, 2016) the AttentionTrip and the classic version of the ANT were compared on several measures of engagement (derived from the work of O’Brien and Toms, 2013). We found that the endogenous version

Acknowledgment

The research reported here was made possible by grants from the NSERC of Canada awarded to RMK.

References (41)

  • M. Corbetta et al.

    Control of goal-directed and stimulus-driven attention in the brain

    Nat. Rev. Neurosci.

    (2002)
  • F.C. Donders

    On the speed of mental processes

  • J. Fan et al.

    Testing the efficiency and independence of attentional networks

    J. Cogn. Neurosci.

    (2002)
  • K.M. Heilman et al.

    Attention: behavior and neural mechanisms

    Compr. Physiol.

    (1985)
  • Y. Ishigami et al.

    Are individual differences in absentmindedness correlated with individual differences in attention?

    J. Individ. Diff.

    (2009)
  • Y. Ishigami et al.

    Repeated measurement of the components of attention of young children using the Attention Network Test (ANT): Stability, isolability, robustness, and reliability

    J. Cogn. Dev.

    (2015)
  • J. Jonides

    Voluntary versus automatic control over the mind’s eye’s movement

    (1981)
  • R.M. Klein et al.

    The modes and domains of attention

  • R.M. Klein et al.

    Parametric exploration of the Simon effect across visual space

    Can. J. Exp. Psychol./Revue canadienne de psychologie expérimentale

    (2006)
  • R.M. Klein et al.

    The AttentionTrip: a tool for exploring the components of attention during a simulated driving task

  • Cited by (16)

    • Thinking about attention: Successive approximations to a productive taxonomy

      2022, Cognition
      Citation Excerpt :

      The ANT-I, in contrast, generates a purer form of exogenous orienting. Among the many variants, we are particularly excited about our game-like variant, the AttentionTrip that we developed anticipating that it would be more engaging than the typical ANTs (Klein, Hassan, Wilson, Ishigami, & Mulle, 2017). The need for engagement was encouraged by the possibility that boredom might compromise the value of the computed network scores (e.g., Matthews, Warm, Reinerman, Langheim, & Saxby, 2010) and the fact that this problem would be exacerbated in studies that required assessing the networks of attention in multiple sessions (e.g., when testing the effectiveness of treatment for attention problems or a method for improving attention).

    • Comparing saccadic and manual responses in the attention network test

      2021, Cortex
      Citation Excerpt :

      As proposed by Posner (1994), the literature substantiates that the alerting network produces an inhibitory effect on the executive network resulting in a larger congruency effect in trials with alerting sound Fan et al., 2009; Callejas et al., 2004, 2005 - Exp.1; Ishigami & Klein, 2010; Asanowicz et al. (2017; 2019). This interaction, again, is not universal with a variety of studies reporting this interaction as nonsignificant (Fan et al., 2002; Fuentes & Campoy, 2008; Klein et al., 2017) or accelerating (Funes & Lupiáñez, 2003). The alerting and executive control interaction is supported by fMRI data showing an overlap in activation (Fan et al., 2005; Xuan et al., 2016).

    • On the origins and evolution of the Attention Network Tests

      2021, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      Because of its game-like nature, the AttentionTrip doesn’t have a “trial” structure and the timing of the delivery of targets depends on the steering performance of the participants. With that noted, the timing of events as used in Experiment 1 of Klein et al. (2017) can be found in the “AttentionTrip” row of Table 2. The AttentionTrip succeeded in its main goal: it was capable of measuring the attentional networks while being more engaging task than the ANT in both adults (Klein et al., 2019) and in children (against which it was compared to the child-ANT, Arora et al., 2021).

    View all citing articles on Scopus
    View full text