​Year One & Year Two (Haskins Laboratories)

 

Behavior

 

Based on previous work (e.g. Pavlidou et al., 2009; 2010; 2014) suggesting differences in implicit learning performance between poor and good readers, we explored whether young children with reading problems also show difficulties with implicit learning and whether those difficulties are linked with the way specific brain networks are formed and operate. For that reason, we used both behavioral and neuroimaging methods to examine the interplay between behavior and neurobiology of the brain.

 

More specifically, to test whether implicit learning is impaired or not in developmental dyslexia, we recruted children and assigned them to two groups: good readers and poor readers. Then, we trained children on a visual (i.e. unfamiliar shapes) implicit learning task [called artificial grammar learning (AGL) task where participants are passively exposed to stimuli that follow a set of complicated rules unbeknown to them; they are then informed about the existense of such rules and are asked to decide from new items those that they think follow the same rules] and measured their sensitivity to novel sequences of unfamiliar shapes to test whether they would preferentially learn the specific features of the items or the rules that generated these items. So far, analysis of the data indicated that good readers show learning of both structure (measured by the number of items children identify as conforming or not to the rules of grammar they were trained on) as well as stimulus-specific information (measured by the number of novel items that resemble more closely to training items children identified) whereas poor readers do not: children with poor reading skills show a difficulty in learning the complex relationships in the visual stimuli. Subsequently, to test whether this difficulty applies to other modalities, children were tested on auditory (sound) and tactile (touch) tasks. Poor readers show a general difficulty in performing implicit learning tasks irrespective of the modality of the task. 

 

Brain 

 

For the neuroimaging component of the project, we used a 3T MRI scanner based at Yale Magnetic Resonance Research Center (MRRC). We have so far acquired functional images (fMRI) from both good readers and poor readers while they performed a visual implicit learning task (AGL). During the task, children completed four runs of a passive learning phase (training) and two runs of an active testing phase. During training, children were shown both structured items (i.e. unfamiliar shapes that follow the rules of an artificial grammar) and random items (i.e. Greek letters that do not obey to any rules). The testing phase runs contained rapidly sequentially presented stimuli and the children were asked to indicate, with button presses, whether or not the stimuli looked 'familiar' or 'similar' to the ones they were trained on. Preliminary analysis of the passive learning phase shows that while children with good reading skills are showing increased activation for items that follow the rules suggesting creation of new representations of the structured items (i.e. learning) children with poor reading skills do not show the same pattern of activation. However, we expect brain maps to change as we increase the poor readers sample size and continue to clean up the data.  Stronger conclusions can be reached once data collection and analysis of both passive learning and active testing phases are complete.