Hemispatial neglect is a heterogenous and multicomponential syndrome encompassing several attentional and spatial deficits (Husain, 2019). Neglect is often described as a directional bias in orienting attention. It is often associated to contralesional omissions occurring in contexts of competition for selection (ipsilesional stimuli are prioritized over contralesional ones; cf., extinction), as well as to non-lateralized attention deficits (Kerkhoff, 2001; Husain, 2019). Conventionally, contralesional spatial bias in visual attention has been assessed with paper-and-pencil methods (Kortte and Hillis, 2009). In particular, cancellation tasks are widely used and considered to be very sensitive (Halligan et al., 1991; Bowen et al., 1999; Ferber and Karnath, 2001; Azouvi et al., 2002; Parton et al., 2004). Patients with severe neglect typically fail to cross targets in the contralesional hemispace, even in the easiest versions of these tasks (Albert, 1973; Halligan et al., 1991). Patients with milder forms of the deficit, however, tend not to show omissions in cancellation tasks without distractors. To overcome this problem, task sensitivity has been enhanced by specifically adjusting the stimuli, for example, by increasing the target density and the relative salience of distractors (Rapcsak et al., 1989; Aglioti et al., 1997; Chatterjee et al., 1999; Sarri et al., 2009; Bickerton et al., 2011; Basagni et al., 2017; Ten Brink et al., 2020; Villarreal et al., 2020). To prevent compensation and to identify milder deficits, time limits are sometimes introduced (Priftis et al., 2019). Another possibility is to calculate indices that are more sophisticated than the simple number of omissions, such as the starting point analysis (Azouvi et al., 2002; Nurmi et al., 2010, 2018), which allows quantifying neglect patients’ tendency to start visual searching from the ipsilesional hemispace (Kinsbourne, 1987; Karnath, 1988; Olk et al., 2002).

Despite these efforts, several studies have demonstrated that computer tasks, which are complex, and require high attentional demands, are more sensitive than paper-and-pencil cancellation tasks (Deouell et al., 2005; List et al., 2008; Rengachary et al., 2009; Bonato et al., 2010, 2012; Blini et al., 2016; Andres et al., 2019). Chronic stroke patients who present perfectly normal performance even in the most difficult cancellation tasks show contralesional omissions when brief stimuli are presented concurrently with a secondary task (i.e., dual-task setting; Deouell et al., 2005; Bonato et al., 2010, 2012; Blini et al., 2016; Andres et al., 2019). Dividing attention and performing two simultaneous tasks necessitates executive control of attention (Cohen, 2014; Strobach et al., 2018). These aspects can sometimes be dramatically impaired in neglect (Mannan et al., 2005; Ronchi et al., 2009; Bonato et al., 2012). Neglect patients show perseveration errors (Mannan et al., 2005; Ronchi et al., 2009), and their executive dysfunction may also be seen as difficulty in allocating attentional resources (Cohen, 2014). This, in turn, may lead to less effective compensation in high demand dual-tasks, resulting in the emergence of otherwise “undetectable” deficits when the task is particularly demanding (Robertson and Frasca, 1992; van Kessel et al., 2013; Bonato, 2015; Strobach et al., 2018; Andres et al., 2019). This approach allows contrasting performance in a primary visuospatial task when performed alone or concurrently with another task (single vs. dual-task setting). It also allows presenting two concurrent targets (i.e., extinction setting) and overall requires higher attentional demands than cancellation tasks. However, computer-based applications have been criticized for introducing stimuli that subtend a limited visual angle. It has been claimed that tasks presented on a standard computer screen or on a sheet of paper may have poor ecological validity (Nakatani et al., 2013; Ulm et al., 2013). Advantages of assessment methods with large perceptual space have been described in studies using dual-tasks (van Kessel et al., 2013; Villarreal et al., 2020) and cancellation tasks (Tanaka et al., 2005; Tsirlin et al., 2009; Ulm et al., 2013; Knobel et al., 2020). Both the large-screen dual-tasks and the large-screen cancellation tasks appear to be more sensitive in revealing spatial bias than paper-and-pencil cancellation tasks (Ulm et al., 2013; van Kessel et al., 2013; Villarreal et al., 2020).

Past studies have highlighted the important role of unspecific attentional demands (Dawson et al., 2008; Bonato et al., 2010, 2012; Buxbaum et al., 2012; Blini et al., 2016; Ricci et al., 2016; Andres et al., 2019), large test fields (Nakatani et al., 2013; Ulm et al., 2013), or both (Villarreal et al., 2020) in determining the sensitivity of tests for neglect and extinction. In the present study, we aimed to capitalize on the previous findings by contrasting these two characteristics. We developed a new computer-based large-screen cancellation task and compared it with a computer-based dual-task approach with brief stimulus duration, which has already proven to be more sensitive than paper-and-pencil tasks in detecting contralesional omissions (Bonato et al., 2010, 2012; Bonato, 2015; Blini et al., 2016; Andres et al., 2019).

We aimed to examine whether contralesional omissions could be triggered more easily by using a cancellation task presented on a large perceptual space or by using attentionally demanding dual-tasks presented on a standard computer screen. These findings were also compared with those obtained with a paper-and-pencil version of the new cancellation task. Our hypothesis was that both the large-screen cancellation task and the dual-tasks are more sensitive than the paper-and-pencil cancellation task in revealing contralesional deficits in spatial processing.

The aim of the present study was to examine whether visuospatial attention deficits could be identified more sensitively by using the digital and enhanced versions of a cancellation task or by using attentionally demanding computer-based tasks requiring participants to divide their attention. To answer the study question, we developed a novel large-screen cancellation task, the Twinkle Task, and compared it with a more established dual-task approach with briefly presented lateralized targets. The findings of these computer methods were also compared to those obtained with a traditional paper-and-pencil version of the Twinkle Task. We predicted that both the large-screen and the dual-task computer methods should have been more sensitive than paper-and-pencil tasks in triggering contralesional omissions. Performance was compared across two groups of stroke patients (left hemisphere LH or right hemisphere RH) and one group of healthy, matched controls.

Our main finding is that RH patients showed no spatial bias in either version of the Twinkle Task but showed contralesional omissions in the dual-task regardless of whether it was visual or auditory in nature. We then reasoned that maybe more sensitive measures (i.e., the starting points and the mean position of hits of the Twinkle Task) might have revealed spatial bias also in cancellation tasks. Additional analyses on these measures did not reveal any spatial bias in cancellation tasks. The computer-based dual-task allowed to draw a different picture. RH patients missed significantly more left-sided targets than controls in both the unilateral and bilateral trials of the Visual dual-task. They also missed significantly more left-sided than right-sided targets in the Auditory dual-task but only for bilateral trials. LH patients missed more right-sided targets than controls in the large-screen Twinkle Task, but not in the paper-and-pencil version nor in the dual-tasks. Both LH patients and controls were slower in the large-screen version compared to the paper-and-pencil version of the Twinkle task. This phenomenon did not become evident in RH patients who were already slower than controls in the paper-and-pencil version.

The findings of the present study expand on those of previous studies (Bonato et al., 2010, 2012; Bonato, 2015; Andres et al., 2019; Villarreal et al., 2020), indicating that coupling a dual-task setting with brief stimulus duration is a sensitive approach, which triggers contralesional omissions in patients with chronic right hemisphere stroke. This might be since using compensatory strategies to cope with spatial attention deficits is challenging in those situations characterized by the need to simultaneously process of two/three distinct aspects, and the time allowed for target detection is limited (Russell et al., 2004; Bonato and Deouell, 2013; Andres et al., 2019). Therefore, as stated in previous studies, unspecific attentional demands appear to be among the most important factors in increasing assessment sensitivity (Robertson and Frasca, 1992; Schendel and Robertson, 2002; Bonato, 2012, 2015; Ricci et al., 2016). If attentional demands are high enough, a task presented on a standard computer screen allows identifying even subtle deficits in contralesional spatial processing. This finding is important and might positively impact clinical practice, as in hospitals and rehabilitation centers ordinary hardware for assessment purposes is more easily available than additional technology. Nevertheless, a large test field, if available, presumably increases ecological validity (Nakatani et al., 2013), and along with a dual-task setting, it provably reveals even subtle neglect (Villarreal et al., 2020).

According to the present findings, cancellation tasks would not appear to be sufficiently demanding in detecting contralesional omissions in right hemisphere stroke. RH patients did not show spatial bias in either version of the Twinkle Task, even though the increase in the visual angle was massive and performance was analyzed with multiple parameters in addition to omissions. Unlike in previous studies (Azouvi et al., 2002; Nurmi et al., 2010, 2018; Basagni et al., 2017; Priftis et al., 2019), analyzing starting points, increasing perceptual demands, or limiting performance time had no sensitivity increment effect in detecting spatial bias in the group of RH patients. Standard, paper-and-pencil approaches are maximally informative in acute stroke, and with rather severe cases. However, our study confirms that in chronic right hemisphere stroke resulting in subtle symptoms, attentionally more demanding measures than cancellation tasks are needed (Taylor, 2003; Pedroli et al., 2015; Gammeri et al., 2020).

Post-stroke spontaneous recovery, rehabilitation, and learning effects might have contributed to modulating the performance of the patients. Fifty five percent of the RH patients were diagnosed with neglect initially (13 days post stroke) by using traditional methods. However, at the time of the study (106 days post stroke), none of the patients scored below the cut-off in the Bells Test. This might be considered compatible with the idea that spontaneous neurobiological recovery is maximal in the first 2–3 months post stroke (Nijboer et al., 2013). It warrants mentioning that patients in our sample additionally received rehabilitation and that they underwent assessments with cancellation tasks already in the early stages of recovery (see Supplementary Table A; Schendel and Robertson, 2002; Parton et al., 2004; Danckert and Ferber, 2006). Therefore, patients may quickly learn to compensate for mild deficits since the test type is familiar and static, and attentional demands are low (Erez et al., 2009; Hasegawa et al., 2011; Ogourtsova et al., 2017; Spreij et al., 2020).

A rather surprising yet logical finding was that LH patients missed slightly more right-sided targets than controls in the large-screen Twinkle Task. It is possible that the Twinkle Task is visuoperceptually more challenging than some classical cancellation tasks. The relative number of targets is greater, and they more closely resemble the distraction stimuli, for example, than in the Bells Test (Gauthier et al., 1989; see also Basagni et al., 2017). Some studies have reported that patients with left hemisphere stroke show reduced perceptual capacity (Woods et al., 2006). This may reflect an attentional deficit (Schendel et al., 2016) since perceptually demanding processing necessitates the engagement of attention (Aglioti et al., 1997). Possibly as a consequence, increments in visuoperceptual demands increase the sensitivity of neglect assessment (Riddoch and Humphreys, 1987; Aglioti et al., 1997; Basagni et al., 2017). Therefore, it could be hypothesized that the visuoperceptually demanding Twinkle Task brought up LH patients’ hidden neglect. This interpretation seems unlikely, however, since LH group showed no signs of neglect nor extinction in any of the computer-based dual-tasks in the present nor the previous study (Villarreal et al., 2020). These same dual-tasks, however, brought up subtle neglect or extinction in RH group. Besides, the present dual-tasks are also known to be sensitive in revealing subtle neglect and extinction in chronic left hemisphere stroke patients (Bonato et al., 2010; Blini et al., 2016). It should also be borne in mind that omission rate in the large screen Twinkle Task was identical for LH and RH group (2%), but only LH group’s performance differed statistically significantly from controls. These facts speak against our original hypothesis of the large-screen version being more sensitive than the paper-and-pencil version (even though the statistical finding in LH patients supports it). Therefore, the possibility of type 2 error cannot be excluded.

On the other hand, at the single case level, two of the LH patients showed statistically significantly more right-sided than left-sided omissions in the dual-tasks (see Supplementary Table A). These patients were not diagnosed with neglect initially by using traditional methods indicating that their contralesional attention deficits were subtle already in the early stage of recovery. It is also possible that these deficits were initially omitted e.g., due to prominent linguistic deficits (Kleinman et al., 2007). Left hemisphere stroke patients’ neglect is often underdiagnosed and, therefore, should be assessed with sensitive methods specifically designed for these patients (Kleinman et al., 2007; Blini et al., 2016).

Right hemisphere patients performed significantly slower than controls in the paper-and-pencil version of the Twinkle Task. LH patients and controls, on the other hand, were slower in the large-screen than the paper-and-pencil version, even though only the large-screen version had a time limit. It is possible that the large-screen version was more demanding than the paper-and-pencil version and, therefore, LH patients and controls took longer in completing the task. RH patients, on the other hand, performed at a consistently slower pace since no differences in performance times between the two task versions within the RH patients were found. This finding is supported by past studies, where patients with right hemisphere damage have shown slow processing speed, presumably as a remnant of neglect (Samuelsson et al., 1998; Bartolomeo and Chokron, 1999; Buxbaum et al., 2004; Viken et al., 2014; Nurmi et al., 2018).

Two limitations of the present study were that the patients showed only very subtle deficits and that the sample size was limited. As the RH patients’ visuospatial attention deficits were only subtle, they became more clearly evident in bilateral than unilateral trials, i.e., as an extinction. While the phenomenon was seen here in the dual-tasks, RH patients also showed extinction in the single task in our previous study (Villarreal et al., 2021). Our first study (Villarreal et al., 2020) demonstrated that the RH patients (but not LH patients) showed statistically reliable contralesional omissions for unilateral targets as well, in two different dual-tasks. This was present as prominent left-sided omissions when compared to those of the controls, but also in RH patients as statistically significantly more left-sided than right-sided omissions. These findings are in line with the view of extinction being one of the several sub-symptoms of neglect or co-occurring with the syndrome (Brozzoli et al., 2006; Cohen, 2014; Husain, 2019). However, there are also studies suggesting anatomo-functional dissociations between neglect and extinction (Neppi-Modona, 1999; Karnath and Rorden, 2012). As the dual-task method offers identical stimuli across different conditions, future studies with larger sample size might allow determining whether neglect and extinction are a continuum or whether they are independent. RH patients also showed ipsilesional omissions in the dual-tasks (although not statistically significant). The phenomenon may be explained by general inattention, which commonly co-occurs with neglect (Hjaltason et al., 1996; Husain and Rorden, 2003). In fact, our previous study (Villarreal et al., 2021) explicitly demonstrated that the RH patients were suffering from general inattention. LH group was suffering from general visual inattention as well but did not show neglect or extinction (Villarreal et al., 2020, 2021). Therefore, it is likely that RH group’s statistically reliable contralesional omissions are due to some specific consequence of right hemisphere damage. These findings may be explained by characteristics of the patients’ strokes (see Supplementary Table A), and are in line with the model of Husain (2019) suggesting that neglect is a multicomponential, rather heterogenous, syndrome. As the attention deficits were only subtle, and the sample size was limited, additional studies are needed to confirm the present results. Additional studies would also be needed to answer pertinent questions related to the new cancellation task: (1) was the Twinkle Task in itself sensitive and a useful screening task for clear neglect in the acute phase of stroke; (2) could it, much like some other “gap-detection” tasks (e.g., Ota et al., 2001), be useful in differentiating between body-centered and stimulus-centered neglect; and (3) could the scanning pattern recordings enabled by the large-screen version provide insights into the abnormal scanning latency and crossing index, which are typical for neglect patients (Rabuffetti et al., 2002; Ulm et al., 2013)?

Limitations of the present study also include the fact that we solely focused on deficient processing of peripersonal and extrapersonal space. Therefore, other subtypes (e.g., neglect in other sensory modalities, personal, motor, or premotor neglect) were not assessed. The presence/absence of these other subtypes might have influenced the findings. Even though the patients showed only subtle deficits in extrapersonal/peripersonal space, and many of them had recovered with respect to initial assessment, they still might have suffered from personal neglect. In the initial assessment, several RH patients showed indications of personal neglect (Supplementary Table A), and some evidence suggest that recovery for this type of neglect can be slower than, and independent of, extrapersonal neglect. In the study of Iosa et al. (2016), recovery rate after a physiotherapeutic and a six-week neuropsychological rehabilitation was nearly 80% for extrapersonal neglect but only 58% for mild personal neglect. Recoveries from these neglect subtypes did not correlate. Finally, as the present patients were not assessed for motor or premotor neglect, or mild physical deficits, their possible effect on the cancellation task performance remains elusive given that the responses imply directional movements (Na et al., 1998). However, as the omission rate in the cancellation tasks was low, and slow performance was seen only in RH patients, whose dominant hand was largely unaffected by stroke (see Table 1), this limitation seems to be uninfluential in practical terms.

We only used the Bells cancellation task as a traditional test to assess the patients’ visuospatial attention deficits (see Villarreal et al., 2020) while often a battery of tests is used [e.g., Behavioral Inattention Test (BIT) by Wilson et al. (1987)]. Therefore, the conclusion of the computer dual-tasks being more sensitive than traditional tests is limited to cancellation tasks only. However, several previous studies (Bonato et al., 2010, Bonato et al., 2013; Bonato, 2015; Blini et al., 2016) have demonstrated with chronic unilateral stroke patients that the computer dual-tasks used in the present study are more sensitive in identifying subtle neglect than various traditional tests (e.g., conventional BIT, and even specific versions of paper-and-pencil cancellation tasks adapted to make them more demanding; Bonato et al., 2012). Therefore, as the patients showed only very subtle symptoms even in these sensitive computer tests, a very plausible possibility is that adding further traditional methods to assess visual neglect would not change the present results. Nevertheless, further studies with larger batteries of tests are needed as they might bring additional information of the sensitivity of the present computer tests in revealing very subtle visuospatial symptoms.

A characteristic of the present study is the fact that we compared targets presented in the peripersonal space (dual-tasks, paper-and-pencil cancellation task) with targets presented in the extrapersonal space (large-screen cancellation task). It is possible that RH patients’ spatial bias was limited only to peripersonal but not to extrapersonal space, and this, rather than attentional demands, would explain the dissociation we found. However, in our previous study (Villarreal et al., 2020), the same RH patients showed subtle neglect in the extrapersonal space, as assessed with large-screen dual-tasks, and both studies demonstrated that paper-and-pencil cancellation tasks were not able to reveal spatial bias in RH patients. Hence, in our present and previous studies, RH patients did not show spatial bias in peripersonal or extrapersonal space in the cancellation tasks but did in both sectors of space in the dual-tasks. The present findings are supported by the single case analyses presented in Supplementary Table A. Statistically reliably contralesional (as compared to ipsilesional) omissions were evident in eight of the patients with the dual-task method but in none of the patients with the cancellation tasks. Most patients showing deficits were initially not diagnosed with neglect by using traditional methods, indicating that their symptoms were subtle already in the early phase of recovery.

To conclude, we compared the sensitivity of two different computer-based methods (cancellation large-screen vs. dual-tasking) and of one paper-and-pencil cancellation task, in detecting subtle visuospatial attention deficits in right hemisphere stroke. The dual-task approach outperformed the cancellation task approach when using paper and pencil but also on a large screen. Therefore, attentionally more demanding methods than conventional approaches are useful in revealing mild forms of visuospatial inattention. It is important to detect even these minor deficits since they may cause important real-life problems (see, e.g., Deouell et al., 2005; Bonato et al., 2012; Sotokawa et al., 2015).

The datasets presented in this article are not readily available because of the privacy and/or ethical restrictions. Individual requests for sharing the data will be considered after additional approval for sharing by the local ethics committee. Requests to access the datasets should be directed to SV, sanna.villarreal@helsinki.fi.

The studies involving human participants were reviewed and approved by the Ethics Committee of Helsinki University Hospital. The patients/participants provided their written informed consent to participate in this study.

SV, ML, RS, HJ, and MH planned and initiated the study and planned the twinkle task. ML and RS planned and constructed the active space application, and responsible for the technical properties of the large-screen version. SV and OV constructed the paper-and-pencil version of the twinkle task and recruited participants and collected the data. MB planned and constructed the visual and auditory dual-tasks and related study-design. SV and MB planned the statistical approach. SV conducted the statistical analyses and drafted and finished the manuscript. All authors contributed to the critical revision of the manuscript and approved the final version to be published.

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.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.