Attentional engagement deficits in dyslexic children

Abstract

Reading acquisition requires, in addition to appropriate phonological abilities, accurate and rapid selection of sublexical orthographic units by attentional letter string parsing. Spatio-temporal distribution of attentional engagement onto 3-pseudoletter strings was studied in 28 dyslexic and 55 normally reading children by measuring attentional masking (AM). AM refers to an impaired identification of the first of two sequentially presented masked objects (O1 and O2). In the present study, O1 was always centrally displayed, whereas the location of O2 (central or lateral) and the O1–O2 interval were manipulated. Dyslexic children showed a larger AM at the shortest O1–O2 interval and a sluggish AM recovery at the longest O1–O2 interval, as well as an abnormal lateral AM. More importantly, these spatio-temporal deficits of attentional engagement were selectively present in dyslexics with poor phonological decoding skills. Our results suggest that an inefficient spatio-temporal distribution of attentional engagement – probably linked to a parietal lobule dysfunction – might selectively impair the letter string parsing mechanism during phonological decoding.

Introduction

Developmental dyslexia (DD) is a neurobiological disorder characterized by a difficulty in reading acquisition despite adequate intelligence, conventional education and motivation (American Psychiatric Association [APA], 1994).

The prevailing view supports the hypothesis that DD results from a specific deficit of auditory-phonological perception, representation and phonological memory (see Gabrieli, 2009, Goswami, 2003, Tallal, 2004, Vellutino et al., 2004, Ziegler and Goswami, 2005, for reviews).

Children and adults with DD show, indeed, deficits in the representation and manipulation of phonological information (e.g., poor speech-sound awareness, slow lexical retrieval and poor phonological short-term memory; see Ramus, 2003, for a review). These phonological deficits could interfere with one of the most critical skills for successful reading acquisition, such as phonological decoding (Share, 1995, Ziegler and Goswami, 2005).

Phonological decoding is based on letter-sound reading and it allows children to make the connection between novel letter string sequences and words that are already stored in their phonological (spoken word) lexicon. The ability to assemble the phonological code for any string of letters allows the child to successfully decode and construct orthographic entries for thousands of new words during the first years of education (Share, 1995, Share, 1999, Share, 2004). Phonological decoding is the primary procedure used by beginning readers both for aloud and silent reading (e.g., Sprenger-Charolles, Siegel, Béchennec, & Serniclaes, 2003). A typical measure for phonological decoding is given by performance in nonword reading. Nonword reading skills are consistently impaired in DD children across different languages (Ziegler, Perry, Wyatt, Ladner, & Schülte-Korne, 2003).

Efficient phonological decoding requires accurate representations at the phoneme level (e.g., Harm and Seidenberg, 1999, Perry et al., 2007, Ziegler and Goswami, 2005). In fact, a low-level auditory processing deficit in children with DD seems to impair speech-sound perception and more specifically its sublexical processing (see Goswami, 2003, Tallal, 2004, Wright et al., 2000, for reviews), which, in turn, would affect grapheme-to-phoneme mapping and phonological short-term memory (Ramus, 2003). Goswami et al. (2002) reported, indeed, that children with DD are relatively insensitive to the rise times of amplitude envelope onsets in acoustic signals compared to normally reading children. The ability to detect this acoustic signal feature provides a non speech-specific mechanism for segmenting syllable onsets and rimes, a crucial precursor to the development of phoneme segmentation skills (Goswami et al., 2002). Experimental evidence has provided support for the idea that auditory attention is necessary for learning phonetic discriminations based on acoustic cues. This suggests that speech signal segmentation requires rapid shifting of auditory attention (Gordon et al., 1993, Francis et al., 2008). Consistently, auditory attention is impaired in children with DD (e.g., Asbjørnsen and Bryden, 1998, Facoetti et al., 2010b, Geiger et al., 2008, Renvall and Hari, 2002) as well as in children with specific language disorders (e.g., Stevens, Sanders, & Neville, 2006).

Neurobiological evidence has shown that the left temporo-parietal junction (TPJ) is crucial for auditory-phonological processing (see Pugh et al., 2000, Ramus, 2004, for reviews). However, activation of the left TPJ auditory phonological system occurs after the left occipito-temporal area's involvement, during phonological decoding (e.g., Simos et al., 2000, Simos et al., 2002). The development of the visual-orthographic system (i.e., visual word form area, VWFA) reflects a specialization of the object recognition system that is particularly suitable for letter string processing (see McCandliss, Cohen, & Dehaene, 2003 for a review). VWFA seems to be hierarchically organized for visual orthographic processing: the posterior areas (occipital lobe) are specifically involved in low-level visual features and letter shape processing, while the anterior areas (ventral temporal lobe) are linked to more abstract letter string processing (McCandliss et al., 2003).

Developmental reading disabilities could arise not only from a specific disorder of the auditory phonological system, but also from a visual-orthographic system dysfunction. In fact, a low-level visual processing (e.g., perceptual noise exclusion or attentional) deficit in children with DD seems to impair the visual-orthographic system (e.g., Bosse et al., 2007, Hari and Renvall, 2001, Hawelka et al., 2006, Martelli et al., 2009, Sperling et al., 2005).

The visual-orthographic system receives stimulus-driven (bottom-up) as well as goal-directed (top-down) attentional influence that modulates all visual processing levels from V1 to VWFA (see Corbetta and Shulman, 2002, Laycock and Crewther, 2008, Vidyasagar and Pammer, 2010a, for reviews). In particular, the letters string perceptual segmentation into its constituent graphemes (i.e., graphemic parsing) involves accurate and rapid attentional shifting (Cestnick & Coltheart 1999; Facoetti et al., 2010b, Facoetti et al., 2006; see Ans et al., 1998, Perry et al., 2007, for computational studies). Before the letter-to-sound mapping mechanism is applied, irrelevant lateral letters should be filtered out by attentional shifting. Attentional shifting improves perception in several visual tasks, such as contrast sensitivity, texture segmentation and visual search, by intensifying the signal and enhancing spatial resolution as well as diminishing the noise effect outside the focus of attention (e.g., Boyer and Ro, 2007, Carrasco et al., 2002, Dosher and Lu, 2000; see Reynolds & Heeger, 2009, for a review). Attentional shifting can be considered as the resultant of the processing resources engagement mechanism onto the relevant object (e.g., letter or grapheme that has to be mapped to its correspondent speech-sound) and the subsequent disengagement mechanism from the previous object to the next one. Spatio-temporal proximity between letters causes a reduction in the letter identification accuracy (Bouma, 1970; see Pelli, 2008, for a recent review) because of massive competition between resources’ processing (Potter, Staub, & O’Connor, 2002; see Keysers & Perrett, 2002, for a review). When the stimulus onset asynchrony (SOA) between two targets is short, the second target (T2) is often identified first (Potter et al., 2002). On the other hand, with larger SOAs, the probability that the first target (T1) is identified first increases (attentional blink, AB). Thus, a target attracts attentive processing resources rapidly, but in the first perceptual stage (i.e., at short SOAs) the attentional engagement is labile; consequently, T2 detection draws resources away from T1 (Potter et al., 2002). “Attentional masking” (AM) is described as the T1 accuracy changes in function of the SOA between targets (e.g., Kavcic and Daffy, 2003, Facoetti et al., 2008; see Fritz, Elhilali, David, & Shamma, 2007, for a recent review on auditory modality). However, almost no AM occurs if attention is rapidly engaged onto the object, whereas powerful AM ensues if attentional engagement on the object is delayed (e.g., Facoetti, 2001, van der Lubbe and Keuss, 2001; see Enns & Di Lollo, 2000, for a review). Although attentional shifting modulation in the presence of AB is still debated (see Nieuwenstein, Chun, van der Lubbe & Hooge, 2005, but Ghorashi, Enns, Spalek, & Di Lollo, 2009), attentional shifting modulation in the presence of the AM paradigm was recently shown by Corradi, Ruffino, Gori, and Facoetti (2010).

Visual attentional shifting deficit has been repeatedly described in DD (see Facoetti, 2004, Hari and Renvall, 2001, Valdois et al., 2004, Vidyasagar and Pammer, 2010a for reviews) and more specifically in dyslexics with poor phonological decoding skills (e.g., Cestnick and Coltheart, 1999, Buchholz and McKone, 2004, Facoetti et al., 2006, Facoetti et al., 2010b, Kinsey et al., 2004, Jones et al., 2008, Roach and Hogben, 2007).

Consistently with the multi-sensory “sluggish attentional shifting” hypothesis by Hari and Renvall (2001), as well as with the “perceptual noise exclusion deficit” by Sperling et al. (2005), children and adults with DD are specifically impaired from rapidly engaging their attention, showing both abnormal temporal (e.g., Di Lollo et al., 1983, Montgomery et al., 2005) and lateral masking (e.g., Geiger et al., 2008, Martelli et al., 2009, Sperling et al., 2005, Spinelli et al., 2002). Indeed, temporal and lateral masking are probably supported by common attentional mechanisms (see Enns & Di Lollo, 2000, for a review).

Evidence of sluggish attentional deployment in the visual modality for children and adults with DD is provided by AB (Buchholz and Aimola-Davies, 2007, Facoetti et al., 2008, Hari et al., 1999, Lallier et al., 2010, Visser et al., 2004), temporal order judgment (Hari et al., 2001, Jaśkowski and Rusiak, 2008, Liddle et al., 2009), rapid multi-element presentation (Hawelka et al., 2006, Bosse et al., 2007) and spatial-cueing tasks (Brannan and Williams, 1987, Facoetti et al., 2006, Facoetti et al., 2010b, Roach and Hogben, 2007) that involve efficient spatio-temporal attentional shifting to rapidly displayed stimuli.

In a cross-sectional study with typically developing children, Bosse and Valodis (2009) have shown that visual attention contributes to phonological decoding skills, independently from auditory-phonological processing in the first grade. Moreover, longitudinal studies have shown that visual attention shifting, in addition to speech-sound awareness, is one of the most important predictors of early reading abilities (e.g., Ferretti et al., 2008, Plaza and Cohen, 2006, Facoetti et al., 2010a). All these longitudinal studies involve serial attentional processing of a single target identification in cluttered conditions. Finally, reading performance has been shown to improve following specific training for visual attention engagement in dyslexic children (e.g., Facoetti et al., 2003, Geiger et al., 1994).

Thus, independently from an auditory-phonological disorder, visual attention engagement might play a critical role in the acquisition of spelling-to-sound mapping during letter string processing because it is crucially involved in parsing and identification of relevant sublexical orthographic units.

The aim of this study is to verify whether a deficit of spatio-temporal distribution of attentional engagement in children with DD (i.e., “sluggish attentional shifting”; Hari & Renvall, 2001) supports the recent proposal of “perceptual noise exclusion deficit” in DD (e.g., Geiger et al., 2008, Martelli et al., 2009, Sperling et al., 2005, Ziegler et al., 2009). In particular, a visual attentional engagement deficit would have a detrimental effect on the segmentation mechanism of the visual input (i.e., letter string) into components (i.e., graphemic parsing). Computational studies have shown that phonological assembly relies on efficient parsing and identification into grapheme units (e.g., Ans et al., 1998, Perry et al., 2007).

Neuroimaging studies have highlighted an association between the cortical regions controlling attentional engagement and both typical and atypical reading development. Several studies employing phonological decoding tasks have shown reduced task related activation in areas surrounding the bilateral TPJ in dyslexics (see Eden & Zeffiro, 1998, for a review). While the left TPJ has been linked to auditory-phonological processing (Pugh et al., 2000, Ramus, 2004), the right TPJ is a crucial component of the network subserving the stimulus-driven control of attentional engagement (Corbetta & Shulman, 2002). TPJ activation changes have been observed during reading acquisition in normally-developing children (Hoeft et al., 2006, Turkeltaub et al., 2003), whereas some studies reported a right TPJ deficiency in dyslexics (e.g., Hoeft et al., 2006, Rumsey et al., 1997, Grünling et al., 2004). More generally, during phonological decoding, the visual attention network could apply a powerful modulation on sublexical visual unit processing in occipito-temporal areas (McCandliss et al., 2003).

A specific relationship between impaired attentional engagement and letter-sound conversion in children with DD was already described in one of our previous studies (Facoetti et al., 2008). In that work, the ability to rapidly engage attention onto a visual object correlated with speed and accuracy of nonword reading (Facoetti et al., 2008). However, further investigation was required according to the following critical issues: (i) The former study investigated only the time-course (i.e., non-spatial attention), whereas the present study included also the spatial distribution of attentional engagement, testing a possible abnormal lateral AM. (ii) The former study investigated the time-course of attentional engagement in a small sample of children with DD (i.e., N = 13), making the dyslexic subtypes difference difficult to test. In the present study, a larger sample of children with DD (N = 28) was divided into two groups on the basis of their phonological decoding skills. This approach allows us to directly test the hypothesis of spatio-temporal attentional engagement deficit only in dyslexic children with poor phonological decoding skills. (iii) The attentional engagement paradigm presented by Facoetti et al. (2008) had letters as target stimuli; therefore, it could also measure a possible effect of reading disorder associated to DD. In particular, sluggish attentional engagement reported by Facoetti et al. (2008) might be explained by a primary letter identification deficit rather than by an attentional deficit per se. In the present study, we presented no-letter targets to isolate a pure perceptual bottom-up deficit. (iv) In the former study, both the attentional engagement (i.e., AM) and the attentional disengagement (i.e., AB) were measured. This latter point is particularly important in addressing the issue of the relationship between a general attentional resources deficit and perceptual noise exclusion deficit in DD (see, for example, Badcock et al., 2008, Moores et al., 2003, for a discussion). Sluggish attentional engagement and disengagement reported by Facoetti et al. (2008) might be explained by a general attentional resources deficit rather than by a specific attentional engagement deficit, given that children had to identify both the T1 and T2 in the same trial (i.e., divided attention in dual task). Thus, in the present study, we assessed the time-course of attentional engagement in dyslexic and normally reading children by measuring the identification of only the first of two sequentially presented masked objects.

Our hypothesis suggests that if visual attention is sluggishly engaged in dyslexic children, a larger and prolonged AM is expected when the two objects are displayed in the same position. Moreover, an abnormal AM when the second object is laterally displayed is also predicted. More importantly, if an efficient spatio-temporal engagement of visual attention is specifically required for an accurate letter string parsing mechanism, we predict a specific abnormal AM in dyslexics with a phonological decoding deficit.