The Connectivity Fingerprints of Highly-Skilled and Disordered Reading Persist Across Cognitive Domains.

The capacity to produce and understand written language is a uniquely human skill that exists on a continuum, and foundational to other facets of human cognition. Multivariate classifiers based on support vector machines (SVM) have provided much insight into the networks underlying reading skill beyond what traditional univariate methods can tell us. Shallow models like SVM require large amounts of data, and this problem is compounded when functional connections, which increase exponentially with network size, are predictors of interest. Data reduction using independent component analyses (ICA) mitigates this problem, but conventionally assumes linear relationships. Multilayer feedforward networks, in contrast, readily find optimal low-dimensional encodings of complex patterns that include complex nonlinear or conditional relationships. Samples of poor and highly-skilled young readers were selected from two open access data sets using rhyming and mental multiplication tasks, respectively. Functional connectivity was computed for the rhyming task within a functionally-defined reading network and used to train multilayer feedforward classifier models to simultaneously associate functional connectivity patterns with lexicality (word vs. pseudoword) and reading skill (poor vs. highly-skilled). Classifiers identified validation set lexicality with significantly better than chance accuracy, and reading skill with near-ceiling accuracy. Critically, a series of replications used pre-trained rhyming-task models to classify reading skill from mental multiplication task participants' connectivity with near-ceiling accuracy. The novel deep learning approach presented here provides the clearest demonstration to date that reading-skill dependent functional connectivity within the reading network influences brain processing dynamics across cognitive domains.

Corrigendum: Children With Reading Difficulty Rely on Unimodal Neural Processing for Phonemic Awareness.

[This corrects the article DOI: 10.3389/fnhum.2019.00390.].

Linking ADHD and Behavioral Assessment Through Identification of Shared Diagnostic Task-Based Functional Connections.

A mixed literature implicates atypical connectivity involving attentional, reward and task inhibition networks in ADHD. The neural mechanisms underlying the utility of behavioral tasks in ADHD diagnosis are likewise underexplored. We hypothesized that a machine-learning classifier may use task-based functional connectivity to compute a joint probability function that identifies connectivity signatures that accurately predict ADHD diagnosis and performance on a clinically-relevant behavioral task, providing an explicit neural mechanism linking behavioral phenotype to diagnosis. We analyzed archival MRI and behavioral data of 80 participants (64 male) who had completed the go/no-go task from the longitudinal follow-up of the Multimodal Treatment Study of ADHD (MTA 168) (mean age = 24 years). Cross-mutual information within a functionally-defined mask measured functional connectivity for each task run. Multilayer feedforward classifier models identified the subset of functional connections that predicted clinical diagnosis (ADHD vs. Control) and split-half performance on the Iowa Gambling Task (IGT). A sample of random models trained on functional connectivity profiles predicted validation set clinical diagnosis and IGT performance with 0.91 accuracy and d' > 2.9, indicating very high sensitivity and specificity. We identified the most diagnostic functional connections between visual and ventral attentional networks and the anterior default mode network. Our results show that task-based functional connectivity is a biomarker of ADHD. Our analytic framework provides a template approach that explicitly ties behavioral assessment measures to both clinical diagnosis, and functional connectivity. This may differentiate otherwise similar diagnoses, and promote more efficacious intervention strategies.

Integrating functional connectivity and MVPA through a multiple constraint network analysis.

Traditional general linear model-based brain mapping efforts using functional neuroimaging are complemented by more recent multivariate pattern analyses (MVPA) that apply machine learning techniques to identify the cognitive states associated with regional BOLD activation patterns, and by connectivity analyses that identify networks of interacting regions that support particular cognitive processes. We introduce a novel analysis representing the union of these approaches, and explore the insights gained when MVPA and functional connectivity analyses are allowed to mutually constrain each other within a single model. We explored multisensory semantic representations of concrete object concepts using a self-paced multisensory imagery task. Multilayer neural networks learned the real-world categories associated with macro-scale cortical BOLD activity patterns from the task, with some models additionally encoding regional functional connectivity. Models trained to encode functional connections demonstrated superior classification accuracy and more pronounced lesion-site appropriate category-specific impairments. We replicated these results in a data set from the openneuro.org open fMRI data repository. We conclude that mutually constrained network analyses encourage parsimonious models that may benefit from improved biological plausibility and facilitate discovery.

A longitudinal neuroimaging dataset on multisensory lexical processing in school-aged children.

Here we describe the open access dataset entitled "Longitudinal Brain Correlates of Multisensory Lexical Processing in Children" hosted on OpenNeuro.org. This dataset examines reading development through a longitudinal multimodal neuroimaging and behavioral approach, including diffusion-weighted and T1-weighted structural magnetic resonance imaging (MRI), task based functional MRI, and a battery of psycho-educational assessments and parental questionnaires. Neuroimaging, psycho-educational testing, and functional task behavioral data were collected from 188 typically developing children when they were approximately 10.5 years old (session T1). Seventy children returned approximately 2.5 years later (session T2), of which all completed longitudinal follow-ups of psycho-educational testing, and 49 completed neuroimaging and functional tasks. At session T1 participants completed auditory, visual, and audio-visual word and pseudo-word rhyming judgment tasks in the scanner. At session T2 participants completed visual word and pseudo-word rhyming judgement tasks in the scanner.

Children With Reading Difficulty Rely on Unimodal Neural Processing for Phonemic Awareness.

Phonological awareness skills in children with reading difficulty (RD) may reflect impaired automatic integration of orthographic and phonological representations. However, little is known about the underlying neural mechanisms involved in phonological awareness for children with RD. Eighteen children with RD, ages 9-13, participated in a functional magnetic resonance imaging (fMRI) study designed to assess the relationship of two constructs of phonological awareness, phoneme synthesis, and phoneme analysis, with crossmodal rhyme judgment. Participants completed a rhyme judgment task presented in two modality conditions; unimodal auditory only and crossmodal audiovisual. Measures of phonological awareness were correlated with unimodal, but not crossmodal, lexical processing. Moreover, these relationships were found only in unisensory brain regions, and not in multisensory brain areas. The results of this study suggest that children with RD rely on unimodal representations and unisensory brain areas, and provide insight into the role of phonemic awareness in mapping between auditory and visual modalities during literacy acquisition.

Dyslexia on a continuum: A complex network approach.

We investigated the efficacy of graph-theoretic metrics of task-related functional brain connectivity in predicting reading difficulty and explored the hypothesis that task conditions emphasizing audiovisual integration would be especially diagnostic of reading difficulty. An fMRI study was conducted in which 24 children (8 to 14 years old) who were previously diagnosed with dyslexia completed a rhyming judgment task under three presentation modality conditions. Regression analyses found that characteristic connectivity metrics of the reading network showed a presentation modality dependent relationship with reading difficulty: Children with more segregated reading networks and those that used fewer of the available connections were those with the least severe reading difficulty. These results are consistent with the hypothesis that a lack of coordinated processing between the neural regions involved in phonological and orthographic processing contributes towards reading difficulty.

Longitudinal Task-Related Functional Connectivity Changes Predict Reading Development.

Longitudinal studies suggest developmentally dependent changes in lexical processing during reading development, implying a change in inter-regional functional connectivity over this period. The current study used functional magnetic resonance imaging (fMRI) to explore developmental changes in functional connectivity across multiple runs of a rhyming judgment task in young readers (8-14 years) over an average 2.5-year span. Changes in functional segregation are correlated with and predict changes in the skill with which typically developing children learn to apply the alphabetic principle, as measured by pseudoword decoding. This indicates a developmental shift in the proportion of specialized functional clusters is associated with changes in reading skill and suggests a dependency of reading development on changes of particular neural pathways, specifically decreases in transitivity is indicative of greater network integration. This work provides evidence that characteristics of these pathways, quantified using graph-theoretic metrics, can be used to predict individual differences in reading development.

Aperture extent and stimulus speed affect the perception of visual acceleration.

Humans are generally poor at detecting the presence of visual acceleration, but it is unclear whether the extent of a field of moving objects through an aperture affects this ability. Hypothetically, the farther a stimulus can accelerate uninterrupted by an aperture's physical constraints, the easier it should be to discern its motion profile. We varied the horizontal extent of the aperture through which continuously accelerating or decelerating random dot arrays were presented at different average speeds, and measured acceleration and deceleration detection thresholds. We also hypothesized that manipulating aperture extent at different speeds would change how observers visually pursue acceleration, which we tested in a control experiment. Results showed that, while there was no difference between the acceleration and deceleration conditions, detection was better in the larger than small aperture conditions. Regardless of aperture size, smaller acceleration and deceleration rates (relative to average speed) were needed to detect changing speed in faster than slower speed ranges. Similarly, observers tracked the stimuli to a greater extent in the larger than small apertures, and smooth pursuit was overall poorer at faster than slower speeds. Notably, the effect of speed on pursuit was greater for the larger than small aperture conditions, suggesting that the small aperture restricted pursuit. Furthermore, there was little difference in psychophysical and eye movement data between the medium and large aperture conditions within each speed range, indicating that it is easier to detect an accelerating profile when the aperture is large enough to encourage a minimum level of pursuit.

Effects of Vertical Direction and Aperture Size on the Perception of Visual Acceleration.

It is not well understood whether the distance over which moving stimuli are visible affects our sensitivity to the presence of acceleration or our ability to track such stimuli. It is also uncertain whether our experience with gravity creates anisotropies in how we detect vertical acceleration and deceleration. To address these questions, we varied the vertical extent of the aperture through which we presented vertically accelerating and decelerating random dot arrays. We hypothesized that observers would better detect and pursue accelerating and decelerating stimuli that extend over larger than smaller distances. In Experiment 1, we tested the effects of vertical direction and aperture size on acceleration and deceleration detection accuracy. Results indicated that detection is better for downward motion and for large apertures, but there is no difference between vertical acceleration and deceleration detection. A control experiment revealed that our manipulation of vertical aperture size affects the ability to track vertical motion. Smooth pursuit is better (i.e., with higher peak velocities) for large apertures than for small apertures. Our findings suggest that the ability to detect vertical acceleration and deceleration varies as a function of the direction and vertical extent over which an observer can track the moving stimulus.

Skill dependent audiovisual integration in the fusiform induces repetition suppression.

Learning to read entails mapping existing phonological representations to novel orthographic representations and is thus an ideal context for investigating experience driven audiovisual integration. Because two dominant brain-based theories of reading development hinge on the sensitivity of the visual-object processing stream to phonological information, we were interested in how reading skill relates to audiovisual integration in this area. Thirty-two children between 8 and 13 years of age spanning a range of reading skill participated in a functional magnetic resonance imaging experiment. Participants completed a rhyme judgment task to word pairs presented unimodally (auditory- or visual-only) and cross-modally (auditory followed by visual). Skill-dependent sub-additive audiovisual modulation was found in left fusiform gyrus, extending into the putative visual word form area, and was correlated with behavioral orthographic priming. These results suggest learning to read promotes facilitatory audiovisual integration in the ventral visual-object processing stream and may optimize this region for orthographic processing.

Task dependent lexicality effects support interactive models of reading: a meta-analytic neuroimaging review.

Models of reading must explain how orthographic input activates a phonological representation, and elicits the retrieval of word meaning from semantic memory. Comparisons between tasks that theoretically differ with respect to the degree to which they rely on connections between orthographic, phonological and semantic systems during reading can thus provide valuable insight into models of reading, but such direct comparisons are not well-represented in the literature. An ALE meta-analysis explored lexicality effects directly contrasting words and pseudowords using the lexical decision task and overt or covert naming, which we assume rely most on the semantic and phonological systems, respectively. Interactions between task and lexicality effects demonstrate that different demands of the lexical decision and naming tasks lead to different manifestations of lexicality effects.

Multimodal lexical processing in auditory cortex is literacy skill dependent.

Literacy is a uniquely human cross-modal cognitive process wherein visual orthographic representations become associated with auditory phonological representations through experience. Developmental studies provide insight into how experience-dependent changes in brain organization influence phonological processing as a function of literacy. Previous investigations show a synchrony-dependent influence of letter presentation on individual phoneme processing in superior temporal sulcus; others demonstrate recruitment of primary and associative auditory cortex during cross-modal processing. We sought to determine whether brain regions supporting phonological processing of larger lexical units (monosyllabic words) over larger time windows is sensitive to cross-modal information, and whether such effects are literacy dependent. Twenty-two children (age 8-14 years) made rhyming judgments for sequentially presented word and pseudoword pairs presented either unimodally (auditory- or visual-only) or cross-modally (audiovisual). Regression analyses examined the relationship between literacy and congruency effects (overlapping orthography and phonology vs. overlapping phonology-only). We extend previous findings by showing that higher literacy is correlated with greater congruency effects in auditory cortex (i.e., planum temporale) only for cross-modal processing. These skill effects were specific to known words and occurred over a large time window, suggesting that multimodal integration in posterior auditory cortex is critical for fluent reading.

A connectionist approach to mapping the human connectome permits simulations of neural activity within an artificial brain.

Data-driven models drawn from statistical correlations between brain activity and behavior are used to inform theory-driven models, such as those described by computational models, which provide a mechanistic account of these correlations. This article introduces a novel multivariate approach for bootstrapping neurologically-plausible computational models that accurately encodes cortical effective connectivity from resting state functional neuroimaging data (rs-fMRI). We show that a network modularity algorithm finds comparable resting state networks within connectivity matrices produced by our approach and by the benchmark method. Unlike existing methods, however, ours permits simulation of brain activation that is a direct reflection of this cortical connectivity. Cross-validation of our model suggests that neural activity in some regions may be more consistent between individuals, providing novel insight into brain function. We suggest this method to make an important contribution toward modeling macro-scale human brain activity, and it has the potential to advance our understanding of complex neurological disorders and the development of neural connectivity.

Cross-modal integration in the brain is related to phonological awareness only in typical readers, not in those with reading difficulty.

Fluent reading requires successfully mapping between visual orthographic and auditory phonological representations and is thus an intrinsically cross-modal process, though reading difficulty has often been characterized as a phonological deficit. However, recent evidence suggests that orthographic information influences phonological processing in typical developing (TD) readers, but that this effect may be blunted in those with reading difficulty (RD), suggesting that the core deficit underlying reading difficulties may be a failure to integrate orthographic and phonological information. Twenty-six (13 TD and 13 RD) children between 8 and 13 years of age participated in a functional magnetic resonance imaging (fMRI) experiment designed to assess the role of phonemic awareness in cross-modal processing. Participants completed a rhyme judgment task for word pairs presented unimodally (auditory only) and cross-modally (auditory followed by visual). For typically developing children, correlations between elision and neural activation were found for the cross-modal but not unimodal task, whereas in children with RD, no correlation was found. The results suggest that elision taps both phonemic awareness and cross-modal integration in typically developing readers, and that these processes are decoupled in children with reading difficulty.

Reading acquisition reorganizes the phonological awareness network only in alphabetic writing systems.

It is unknown how experience with different types of orthographies influences the neural basis of oral language processing. In order to determine the effects of alphabetic and nonalphabetic writing systems, the current study examined the influence of learning to read on oral language in English and Chinese speakers. Children (8-12 years olds) and adults made rhyming judgments to pairs of spoken words during functional magnetic resonance imaging (fMRI). Developmental increases were seen only for English speakers in the left hemisphere phonological network (superior temporal gyrus (STG), inferior parietal lobule, and inferior frontal gyrus). The increase in the STG was more pronounced for words with conflicting orthography (e.g. pint-mint; jazz-has) even though access to orthography was irrelevant to the task. Moreover, higher reading skill was correlated with greater activation in the STG only for English speaking children. The effects suggest that learning to read reorganizes the phonological awareness network only for alphabetic and not logographic writing systems because of differences in the principles for mapping between orthographic and phonological representations. The reorganization of the auditory cortex may result in better phonological awareness skills in alphabetic readers.

A meta-analytic review of multisensory imagery identifies the neural correlates of modality-specific and modality-general imagery.

The relationship between imagery and mental representations induced through perception has been the subject of philosophical discussion since antiquity and of vigorous scientific debate in the last century. The relatively recent advent of functional neuroimaging has allowed neuroscientists to look for brain-based evidence for or against the argument that perceptual processes underlie mental imagery. Recent investigations of imagery in many new domains and the parallel development of new meta-analytic techniques now afford us a clearer picture of the relationship between the neural processes underlying imagery and perception, and indeed between imagery and other cognitive processes. This meta-analysis surveyed 65 studies investigating modality-specific imagery in auditory, tactile, motor, gustatory, olfactory, and three visual sub-domains: form, color and motion. Activation likelihood estimate (ALE) analyses of activation foci reported within- and across sensorimotor modalities were conducted. The results indicate that modality-specific imagery activations generally overlap with-but are not confined to-corresponding somatosensory processing and motor execution areas, and suggest that there is a core network of brain regions recruited during imagery, regardless of task. These findings have important implications for investigations of imagery and theories of cognitive processes, such as perceptually-based representational systems.

Prediction of reading skill several years later depends on age and brain region: implications for developmental models of reading.

We investigated whether brain activity was predictive of future reading skill and, if so, how this brain-behavior correlation informs developmental models of reading. A longitudinal study followed 26 normally developing human children ranging in age from 9 to 15 years who were initially assessed for reading skill and performed a rhyming judgment task during functional magnetic resonance imaging. Patterns of brain activation in this task predicted changes between initial and a follow-up assessment of nonword reading skill administered up to 6 years later. Brain activity in areas typically active during imaging studies of reading was found to predict future nonword reading ability, but the predictive ability of these areas depended on age. Increased activity relative to peers in neural circuits associated with phonological recoding (i.e., inferior frontal gyrus and basal ganglia) was predictive of greater gains in reading fluency in younger children, whereas increased activity relative to peers in orthographic processing circuits (i.e., fusiform gyrus) was predictive of smaller gains in fluency for older children. Interpreted within the context of a connectionist model of reading, these results suggest that younger children who are more sensitive to higher-order phonological word characteristics (e.g., coarticulations) may make greater reading proficiency gains, whereas older children who focus more on whole-word orthographic representations may make smaller proficiency gains.

Integrating conceptual knowledge within and across representational modalities.

Research suggests that concepts are distributed across brain regions specialized for processing information from different sensorimotor modalities. Multimodal semantic models fall into one of two broad classes differentiated by the assumed hierarchy of convergence zones over which information is integrated. In shallow models, communication within- and between-modality is accomplished using either direct connectivity, or a central semantic hub. In deep models, modalities are connected via cascading integration sites with successively wider receptive fields. Four experiments provide the first direct behavioral tests of these models using speeded tasks involving feature inference and concept activation. Shallow models predict no within-modal versus cross-modal difference in either task, whereas deep models predict a within-modal advantage for feature inference, but a cross-modal advantage for concept activation. Experiments 1 and 2 used relatedness judgments to tap participants' knowledge of relations for within- and cross-modal feature pairs. Experiments 3 and 4 used a dual-feature verification task. The pattern of decision latencies across Experiments 1-4 is consistent with a deep integration hierarchy.

Feature-feature causal relations and statistical co-occurrences in object concepts.

Influences of feature-feature statistical co-occurrences and causal relations have been found in some circumstances, but not others. We hypothesized that detecting an influence of these knowledge types hinges crucially on the congruence between the task and type ofknowledge. We show that both knowledge types influence tasks that tap feature relatedness. Detailed descriptions of causal theories were collected, and co-occurrence statistics were based on feature production norms. Regression analyses tested the influences of these knowledge types in untimed relatedness ratings and speeded relatedness decisions for 65 feature pairs spanning a range of correlational strength. Both knowledge types influenced both tasks, demonstrating that causal theories and statistical co-occurrences between features influence conceptual computations.