Is excessive visual crowding causally linked to developmental dyslexia?
Introduction
Reading is a unique cognitive human skill crucial to life in modern societies, but for about 10% of children, learning to read is extremely difficult. These children are affected by developmental dyslexia (DD) and they have difficulties with accurate or fluent word recognition and spelling despite adequate instruction, intelligence and sensory abilities. DD is defined by difficulties with phonological decoding, whereas comprehension is more intact (American Psychiatric Association, 2013; Gabrieli, 2009; Peterson and Pennington, 2012). Several longitudinal studies have shown that auditory-phonological processing is already impaired at the pre-reading stage in children who eventually develop DD (e.g., Carroll et al., 2016; Franceschini et al., 2012; Black et al., 2017). Although the most common explanation of DD suggests a specific disorder in auditory and phonological processing (Hornickel and Kraus, 2013; Peterson and Pennington, 2015), several studies show that also difficulties in visual crowding and spatial attention could be core deficits in DD (Bosse et al., 2007; Zorzi et al., 2012; Facoetti et al., 2010a, Facoetti et al., 2010b; Franceschini et al., 2012, 2013; Stein, in press in this issue), impairing orthographic development (Vidyasagar and Pammer, 2010; Stein, 2014; Grainger et al., 2016).
Crowding is a universal phenomenon that limits our ability to identify individual stimuli when multiple objects are displayed in their vicinity (see Pelli, 2008; Pelli and Tillman, 2008; Whitney and Levi, 2011; Gori and Facoetti, 2015; Rosenholtz, 2016, for reviews). Crowding selectively impairs the discrimination and the ability to recognize stimuli in clutter (Whitney and Levi, 2011). Some neuroimage studies have shown that the strongest effects of crowding occur in the earliest stages of cortical processing in V1 (Chen et al., 2014; Millin et al., 2014), whereas other studies showed that it could arise at later stages in the visual processing hierarchy (Chicherov et al., 2014; Ronconi et al., 2016a, Ronconi et al., 2016b; Ronconi and Bellacosa Marotti, 2017). Crowding depends on the critical spacing between target and flankers, which is defined as the minimal distance between the target and the flankers that is necessary to accurately recognize the target comparable to when the flankers are absent (Yashar et al., 2015). Bouma's law states that critical spacing is proportional to target eccentricity: the higher the target eccentricity the larger the critical spacing for correctly discriminating the target (Bouma, 1970; Whitney and Levi, 2011). Crowding can occur with simple objects such as oriented gratings (e.g., Greenwood et al., 2012), and with complex objects such as faces and letters (Pelli and Tillman, 2008; Freeman et al., 2012; Whitney and Levi, 2011). In the periphery of the visual field, many letters printed at fixed spacing and embedded within a word are unrecognizable because of crowding (Bouma, 1970; Martelli et al., 2009).
Letter identification is a fundamental stage in visual word recognition and reading (McClelland and Rumelhart, 1981; Pelli et al., 2003; Perry et al., 2007). During reading acquisition the analysis of the graphemes that compose the letters string is a fundamental component of phonological decoding, i.e. the translation of the orthographic code into its phonological counterpart (Perry et al., 2007; Goswami, 2003; Ziegler and Goswami, 2005). Phonological decoding is also fundamental for a fast access to semantics from print during reading acquisition (Share, 1995). Recently, Grainger et al. (2016) described a specialized system for parallel letter processing that assigns letter identities to different locations along the horizontal meridian within the limits mainly imposed by crowding, in which spatial attention is used to set up this system during reading development. In particular, efficient development of reading skills involves the use of visuo-spatial attention to implement parallel letter processing. Developing a mechanism of spatial attention to process letter identities, their location and their position within a word, is one of the keys to becoming a skilled reader (Grainger et al., 2016).
Although some studies showed no or small effects of spatial attention on crowding ( Joo et al., 2018; Nazir, 1992; Wilkinson et al., 1997), other studies suggest that crowding could be the result of a limit in the resolution of spatial attention (He et al., 1996; Intriligator and Cavanagh, 2001; Strasburger, 2005; Yeshurun and Rashal, 2010; Grubb et al., 2013). Indeed, a spatial cue that orients attention on the target position before the array of stimuli (target and flankers) reduces crowding (Huckauf and Heller, 2002; Scolari et al., 2007; Franceschini et al., 2012), decreasing the critical spacing (Yeshurun and Rashal, 2010).
People with DD appear to suffer from an excessive crowding as compared to typical readers (e.g., Geiger and Lettvin, 1987; Moores et al., 2011; Callens et al., 2013; Moll and Jones, 2013; see Gori and Facoetti, 2015 for a review; but Doron et al., 2015; Sacchi et al., 2018). An excessive crowding in individuals with DD could be due to sluggish orienting of their spatial attention (Facoetti et al., 2000, 2001, 2008, 2010, 2010; Lallier et al., 2009, 2010; Ding et al., 2016; see Hari and Renvall, 2001; Gori and Facoetti, 2014; Krause, 2015; Grainger et al., 2016 for reviews).
However, the causal link between an excessive crowding and DD is not yet clearly established because group differences between individuals with and without DD in crowding might be a simple effect of the reduced reading experience associated to DD (Goswami, 2003, 2015). Some studies showed that extra-large interletter spacing enhances their reading efficiency on the fly, suggesting a possible causal link (e.g., Spinelli et al., 2002; Zorzi et al., 2012; but Schneps et al., 2013). Here, we employ a comprehensive approach incorporating all causal experimental designs to test the relationship between an excessive crowding and DD.
In particular, after testing the efficacy of our task to capture an excessive crowding in children with DD, we measured the crowding to show the possible causal link in: (i) a study in which we manipulated directly the interletter and interline spacing; (ii) two interventions studies; and (iii) a longitudinal study. The intervention and longitudinal studies are the main experimental design to demonstrate whether an excessive crowding has a pivotal role in a more specific-domain skill (i.e., reading).
In Experiment 1, we measured crowding in two age-matched groups of children with and without DD manipulating spatial attention. Results show an excessive crowding in children with DD only for unattended location. In Experiment 2, we experimentally decreased crowding through an extra-large spaced text, showing that reading accuracy was increased in children with DD. In addition, two intervention studies (Experiment 3 and 4) and a longitudinal study (Experiment 5) were conducted to directly establish a causal link between crowding and reading skills development. In the intervention studies, children with DD were trained with action video games (AVG), which have repeatedly shown to reduce crowding (Green and Bavelier, 2007; Franceschini et al., 2013, 2017b) and improve spatial attention (see Green and Bavelier, 2012 for a review and Bediou et al., 2018 for a recent meta-analysis). AVG were also found to improve reading efficiency in individuals with DD (Franceschini et al., 2013, 2017a, 2017b; Gori et al., 2016; Łuniewska et al., 2018; see Franceschini et al., 2015 for a review) and other visual disorders (Vedamurthy et al., 2015; Gambacorta et al., 2018), possibly through the amelioration of the dorsal fronto-parietal pathway efficiency (Bavelier et al., 2012; Gori et al., 2016; Föcker et al., 2018a, 2018b). Thus, in Experiment 3 and 4 we used AVG training to reduce crowding and to improve reading efficiency in two groups of unselected children with DD. For unselected group we mean children with diagnosed DD that are not selected by subtype of DD or for different neurocognitive deficits. Finally, in Experiment 5, we used a longitudinal approach where crowding was measured in pre-readers and its predictability for future reading development is prospectively investigated.
Section snippets
Participants
Thirteen children (5 female) with DD, and twenty-two children (11 female) who were typical readers (TR) took part in the experiment. Children received the diagnosis of DD by the Italian National Health Service, based on standard exclusion and inclusion criteria (APA, 2013). The reading performance of each child with DD was at least −1 SDs below the age-standardized norm in the average score of the 4 clinical measures (Sartori et al., 2007). Other inclusion criteria for this study were normal IQ
Participants
Eighteen children (14 female) with DD, and thirty-two TRs (11 female) took part in this experiment. The same DD diagnostic criteria of Experiment 1 were used. The two groups (DD and TR) did not differ (t(48) = -0.439, p > 0.662) for chronological age (TR mean = 11.6, SD = 20 and DD mean = 11.8, SD = 23) and IQ (Wechsler, 2003; all ps > .38), whereas they differed (t(48) = 2.93, p = 0.005) both in words reading time (TR: mean = 105 s, SD = 40; DD: mean = 146 s, SD = 56) and errors (t(48) = 4.55,
Results
The reading performance (errors and reading time) was analysed by two separate ANOVAs.
The within-subject factor was the spacing condition (extra-small and extra-large), while the between-subject factor was group (children with DD and TR).
The ANOVA on errors showed a spacing condition main effect (F(1,48) = 16.132, p = 0.0001, η2 = 0.252), a group main effect (F(1,48) = 22.919, p = 0.0001, η2 = 0.323), and a significant spacing condition X group interaction (F(1,48) = 4.488, p = 0.039 η2
Participants
The participants were fourteen children (6 female; mean age = 10.1 years, SD = 1.6) with DD that agreed to participate in an experimental video game training. The same DD diagnostic criteria of Experiment 1 and 2 were used. Information about video game experience were collected in interviews with parents during pre-informative briefing about the experimental training. Children with DD did not know the aim of the training and they declared that in the previous six months they did not play AVG
Participants
The participants were eighteen children (8 female; mean age = 9.79 years, SD = 1.33) with DD that agree to participate to a clinical AVG training. The same DD diagnostic criteria of Experiment 1, 2 and 3 were used. As in Experiment 3, information about video game experience were collected in interviews with parents during pre-informative briefing about the clinical training. Children with DD did not know the aim of the training and they declared that in the previous six months they did not play
Participants
In Experiment 5, we longitudinally investigated the causal link between crowding and learning to read. Sixty-four (33 female), 5-year-old pre-reading children attending the last year of kindergarten in Northern Italy, were selected by a larger sample and took part in our longitudinal study. In the Italian school system, formal reading instruction starts in grade 1. Consequently, Italian pre-schoolers are also pre-readers. We excluded the few children that were able to read at the kindergarten
Discussion
In Experiment 1, the main effect of target-to-flankers spacing was absent when attention was preallocated to the target position, demonstrating that crowding can be nullified by an efficient attentional orienting and zooming (e.g., Turatto et al., 2000; Facoetti and Molteni, 2000; Ronconi et al., 2012, 2016, 2018). Although Joo et al. (2018)) did not find any correlation between spatial attention and crowding, our finding confirm several studies showing a direct link between attentional
CRediT authorship contribution statement
Sara Bertoni: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Validation, Writing - original draft, Writing - review & editing. Sandro Franceschini: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Validation, Writing - original draft, Writing - review & editing. Luca Ronconi: Conceptualization, Methodology, Validation, Writing - review & editing. Simone Gori: Conceptualization, Methodology, Supervision,
Acknowledgements
This work was funded with grants from the CARIPARO Foundation (Borse di Dottorato CARIPARO 2015, to S.B.; Progetti di Eccellenza CARIPARO, 2011–2012 to A.F.), and a grant from MIUR (Dipartimenti di Eccellenza DM 11/05/2017 n. 262) to the Department of General Psychology, University of Padua. Progetto di ateneo anno 2013 - CPDA130437.
References (93)
- et al.
Neural bases of selective attention in action video game players
Vis. Res.
(2012) - et al.
Developmental dyslexia: the visual attention span deficit hypothesis
Cognition
(2007) - et al.
Neural correlates of visual crowding
Neuroimage
(2014) - et al.
Is attentional focusing an inhibitory process at distractor location?
Cogn. Brain Res.
(2000) - et al.
Visual-spatial attention in developmental dyslexia
Cortex
(2000) - et al.
Sluggish engagement and disengagement of non-spatial attention in dyslexic children
Cortex
(2008) - et al.
A causal link between visual spatial attention and reading acquisition
Curr. Biol.
(2012) - et al.
Action video games make dyslexic children read better
Curr. Biol.
(2013) - et al.
An action video game for the treatment of amblyopia in children: a feasibility study
Vis. Res.
(2018) - et al.
Perceptual learning as a possible new approach for remediation and prevention of developmental dyslexia
Vis. Res.
(2014)