Reading disability (RD), the most prevalent form of learning disorder, is recognized essentially as a significant difference between an individual’s reading ability and his or her achievement based on age, intelligence, education and social supports [1] . Although RD is one of the most common childhood disorders, prevalence is uncertain. Studies conducted in English-speaking countries show prevalence ranging from 5% to 17.5% [2] . This disorder is considered a cognitive dysfunction, itself secondary to a neurobiological dysfunction. Many researches in cognitive neuropsychology, neuroscience and genetics have resulted in suggestions for possible causes of dyslexia. Although this disorder might be secondary to more basic auditory temporal processing deficits or speech perception deficits, phonological deficit as the cognitive basis of developmental dyslexia is now widely accepted. In addition, low level visual processing disorders induced by a magnocellular dysfunction may contribute to developmental dyslexia [3] .

Modern electrophysiologic and brain imaging technologies have used to enhance diagnostic precision and illuminate the neurobiological underlying of learning disorders [4] . Electroencephalography (EEG) provides methods by which human brain activity can be assessed during cognitive processes. Computerized EEG studies including evoked potentials (EP) or event-related potentials (ERP), spectral EEG analysis, and topographic EEG brain mapping have also identified a number of brain irregularities in individuals with learning disabilities [4] . Children with reading disability, in addition to reading problems, also have a number of difficulties associated to attentional deficits. Moreover, it has been reported that children with reading disability have impaired visual-spatial attention abilities [5] . Currently, findings suggest that phonological problems and the reading impairment in dyslexia both result from poor visual coding. More precisely, the retinal cells have an important role in serial scanning and spotlighting the letters of a text [6] . Vidyasagar et al. (2009) suggested that an early perceptual defect occurring in visual pathway was the cause of reading problems in children with dyslexia. They also discussed attentional mechanisms controlled by the dorsal visual pathway having a special role in serial scanning of letters. Any deficits in this process will result in subsequence effects, including problems in visual processing of graphemes, changing them into phonemes and the enhancement of phonemic awareness [7] [8] .

Such information about visual-spatial attention mainly arises from studies using ERPs. ERPs were first recognized in 1964, and had since remained a diagnostic instrument in both the fields of psychiatry and neurology [9] . ERPs are those EEGs that directly measure the electrical response of the cortex to sensory, affective, or cognitive events [9] . In most ERP studies on visual-spatial attention, researchers often use a variant of Posner paradigm. In this paradigm, subjects should focus attention on one side and ignore events in the opposite side [10] . In this task, the subject is instructed to fixate at a central point on the screen. A target stimulus appears on either the left or right side. The subject must respond to the target immediately after detecting it on the left or right visual field. The side of target presentation is most often announced by a previous “cue” stimulus, which may be either “valid” (it appears on the correct target side) or “invalid” (it appears on the opposite side) [11] . This paradigm is used for assessing the engagement, disengagement, orientation, and covert shifting of attention. Some investigators propose the magnocellular system (M-system) as the neural basis of such deficits. This system processes the information, followed by sending the processed information into the posterior parietal cortex, which is a particular multi-modal selective spatial attention area. The M-system is crucial for smoothing the flow of attention, which plays an important role for the identification of individual letters in a text. In addition, the automatisation deficit hypothesis is in line with the M-system deficit, but it attributes that abnormal spatial and temporal processing to the output from the cerebellum rather than to the M-system input only [5] .

Neurofeedback is one of the applications of Brain―computer interface (BCI) in rehabilitation and behavioral medicine [12] . Neurofeedback is reportedly used to treat associated deficits in children with reading disability [13] and also some of other psychiatric disorders [14] [15] . Some studies reported improvements in the reading ability and phonological awareness deficit, as well as changes in the EEG in children with reading disabilities as a result of neurofeedback in a single-subject study [16] .

The theoretical basis of neurofeedback is based on a biological model of Learning Disorders (LD), considering that LD is a disorder of neural regulation and assuming under this approach that these neural deficiencies are amenable to change using the feedback method (e.g. an operant conditioning procedure, whereby an individual learns to self-regulate the electrical activity) [17] .

A protocol of neurofeedback that introduced by Hammond (2005) was used in this study. This protocol directly or indirectly affected the cerebellum and occipital areas interfering in visual process and movement analysis [18] . Accordingly, it was decided to use protocol to evaluate visouspatial attention ability by means of Posner paradigm task and ERP in children with reading disabilities.

The analysis of the behavioral data revealed an improvement in the visual-spatial attention ability by neurofeedback in the two participants. At the end of the treatment, the participants showed a reduction in the reaction time factor, and an increase in the correct responses. As demonstrated in the results, children showed a better performance on Posner paradigm task after treatment “Tables 1-3”. Moreover, the post treatment assessments displayed stable behavioral improvements during the time. The improved behavioral performance was accompanied by P3a and P3b amplitude reduction after the EEG biofeedback training.

The efficacy of neurofeedback by normalization of targeted EEG abnormalities is reported in dyslexia, with improvements in the reading ability at least 2 grade levels post-neurofeedback [20] . Some researchers investigated the effects of neurofeedback on 12 children with reading disorder who were randomly assigned to experimental and control groups. After 20 treatment sessions, children in the experimental group showed significant improvement in spelling ability. However, no significant changes in brain mapping were shown [21] . Nazari et al. (2012) demonstrated that although EEG analysis did not show notable changes in the power of the targeted bands (delta, theta, and beta), there was normalization of coherence of the theta band at T3-T4, delta band at Cz-Fz, beta band at Cz-Fz, Cz-Pz and Cz-C4 [16] .

The ERP parameters, such as amplitude and latency, are the indicators of brain functions. ERP is also related to the circumscribed cognitive processes. Predominantly, the P3 wave represents cognitive functions involved in orientation of attention, contextual updating, response modulation, and response resolution [9] . To explore the links between behavioral and EEG changes in the participants, the P3 component was analyzed. It has been demonstrated that deficits in later cognitive control and error monitoring mechanisms, indexed by the P3, are common in children with reading disorder. If ERP amplitude reflects brain activation level, increasing brain efficiency should be related to a decrease in the amplitude of ERP waves [22] . In this line, the results pointed out that after neurofeedback training the lower amplitude is related to more correct and less error response. This finding might be interpreted as a decreased attentional demand at neural level. Furthermore, P3 peak latency is proportional to stimulus evaluation timing, sensitive to task processing, and it varies with individual differences in cognitive capability [23] . We found shorter P3 latency after treatment, indicating an improvement in the timing of visouspatial processing.

The human visual system starts in the retina, and from the retina, visual information is relayed to the lateral geniculate nucleus and then onto the extra striate cortex (V1). Two distinct systems exist in visual pathway; the magnocellular or M-system, and the parvocellular or P-system. The M-system is sensitive to fast changing stimuli and moving objects and are insensitive to colors [24] . Some investigators propose the M-system as the neural basis of children with reading disorder. For example, Galaburda and Livingstone (1993), examined the brains of dyslexics post mortem, and found that the neurons in the magnocellular layers of the lateral geniculate nucleus were some 30% smaller than those of controls [25] . According to the magnocellular theory of dyslexia, impairments along the magnocelluar pathway would cause problems with reading [26] - [28] . The idea behind the role of the M-system in reading is that any weakness in M-system can cause visual confusion of letter order and

poor visual memory for written word [29] . Considering the evidence about impaired functioning of the M-sys- tem in dyslexics and the projection of this system on the posterior parietal cortex, there are some evidence that the posterior parietal cortex is involved in directing visual-spatial attention [30] , and it is also known that damage to the posterior parietal cortex affects the ability to disengage visual attention [31] .

Cells in the primary visual cortex project further to the cerebellum and superior colliculus, which are important for visuo-motor control. On the other hand, it is known that cerebellar dysfunction impairs linguistic proc- essing, acquisition and automatisation of new cognitive procedures [32] , the reason why some researchers have proposed cerebellar theory of dyslexia [33] . Problems with balancing are other manifestations of cerebellar impairments that have also been found frequently in dyslexic children [33] . Nicolson & Fawcett (1999) suggest that because the M-system projects onto the cerebellum, deviant functioning of the magnocelluar pathway could be responsible for cerebellar dysfunctions found in dyslexic children [34] .

For neurofeedback purpose in this study, the balance protocol introduced by Hammond (2005) was used. It has been hypothesized that improving function in the vicinity of primary visual cortex (i.e. Brodmann area 17 - 18) may improve the visual guidance for the cerebellum. This protocol affects directly or indirectly on cerebellum and occipital areas interfering in visual process [18] . Because of the linkage between balancing, cerebellum, magnocelluar pathway, and visual-spatial attention in dyslexic children, the results would explain the improvements in visual-spatial attention performance at behavioral and neural level. However, further controlled studies with larger sample sizes seem merited to replicate these findings and to investigate the mechanism of action by which the positive results are obtained from dyslexic children.