Research Day 2024 Program Book

Dr. Natasha Rajah received her Ph.D. from the University of Toronto, St. George Campus; and did her postdoctoral training at U.C. Berkeley. She was a Professor at McGill University from 2005 to 2023 before joining the Department of Psychology at Toronto Metropolitan University (formerly Ryerson) in Fall 2023 as a tenured Full Professor and Tier 1 Canada Research Chair in Sex, Gender, and Diversity in Brain Health, Memory and Aging. Dr. Rajah uses multidisciplinary methods, including MRI, to investigate how sex, gender, and social determinants of health contribute to individual differences in brain and cognitive function in aging. She is an Editor-in-Chief for Aging, Neuropsychology and Cognition, Senior Editor for Brain Research and serves as Vice-Chair of the CIHR’s Institute of Aging Advisory Board; and is the EDI Lead for the Canadian Consortium on Neurodegeneration in Aging (CCNA) in Phase III. 

Title: The effect of biological sex and menopause status on episodic memory and related brain function at midlife: The BHAMM Study

Abstract: Our ability to encode, store and retrieve past experiences in rich contextual details (episodic memory) declines with age and can be an early sign of Alzheimer’s disease (AD) – a disease that is more prevalent in females than males. Recent evidence also suggests that midlife is a critical period in adulthood when episodic memory decline and changes in brain regions affected by AD arise. Midlife is also the time that females with ovaries experience spontaneous menopause. Yet, little is known about how biological sex, menopause, and age affect episodic memory and related brain systems at midlife; and how this may differ in adult with risk factors for AD. In this talk I will present results from the ongoing Brain Health at Midlife and Menopause (BHAMM) study which aims to fill this knowledge gap and determine what factors support and/or hinder memory and brain function in males and females at midlife. Our initial results suggest that middle-aged females, but not males, exhibited age-related episodic memory decline, and that this effect was observed at post-menopause. Task-based fMRI analysis shows that memory decline at post-menopause was associated with altered activity in parahippocampal, posterior parietal and occipital cortices. Together these studies highlight the importance of considering biological sex and endocrine aging in the cognitive neuroscience of aging and dementia prevention.

Dr. Stephanie Aghamoosa is a clinical neuropsychologist and Assistant Professor at the Medical University of South Carolina with expertise in using cognitive assessments and fMRI to study aging and Alzheimer's disease. She obtained her PhD at Ohio State in 2020 under the mentorship of Dr. Ruchika Prakash, conducting research on characterizing age-related neurocognitive changes and testing behavioral interventions to enhance cognitive and emotional function in older adults. She then completed a postdoctoral fellowship specializing in the clinical neuropsychology and cognitive neuroscience of early Alzheimer’s disease at the Medical University of South Carolina, where she is now a tenure-track faculty member. Her ongoing research is focused on developing non-pharmacological interventions to slow the clinical progression of Alzheimer’s disease. She is conducting technology-leveraged clinical trials of non-invasive brain stimulation and cognitive training designed to maximize neuroplasticity, using fMRI and digital health technology to sensitively measure intervention effects.  

Title: Leveraging Functional Connectivity to Develop Effective Interventions for Early Alzheimer’s Disease

Abstract: Alzheimer’s disease (AD) is characterized by progressive cognitive impairments that are accompanied by widespread alterations in brain structure and function. The early stages of AD represent a critical window of opportunity for disease detection and intervention, but changes in cognition and brain function during this period are subtle and difficult to detect. However, advanced fMRI analysis techniques offer a promising avenue for sensitively characterizing and tracking early AD-related network alterations. In this talk, I will review my work using functional connectivity to reveal brain network disruptions that relate to cognitive deficits in preclinical AD. I will then discuss our clinical trials of non-invasive brain stimulation for improving cognition in mild cognitive impairment due to AD that integrate functional connectivity as an outcome measure to capture treatment effects, investigate neural mechanisms, and guide intervention development. I will discuss how recent AI advances have enabled high-quality reconstructions of seen images from solely fMRI brain activity. Specifically, we leveraged large-scale fMRI data to align brain activity to the pre-trained embedding space of CLIP via diffusion models and contrastive learning. Subsequently we developed a simple approach to achieve high-quality reconstruction results with just 1 hour of fMRI training data by aligning multiple participants to a shared latent space.
 

Charlie Ferris is a postdoctoral fellow at McGill University where he researches neural mechanisms supporting memory in humans. Charlie earned his Ph.D. at Emory University in Georgia working with Dr. Stephan Hamann, and then completed his first post-doc at The Ohio State University under the supervision of Dr. Baldwin Way. By combining fMRI, brain stimulation, and behavioral approaches, Charlie tests how brain networks dynamically process different types of memory content. As Charlie’s current work investigates episodic memory formation, consolidation, and retrieval using naturalistic stimuli in humans, he recommends you read this bio carefully as there may be a quiz on this information during his talk.

Title: Hippocampal-cortical networks predict conceptual versus perceptually guided narrative memory

Abstract: If the same basic story is told with a focus on different details, is the memory representation for the core elements of the story the same? Or does the underlying memory trace shift to match the surrounding details? Current theories of event memory propose distinct connections between the hippocampus and neocortical regions to support processing different types of content in memory. Additionally, it has been established that hippocampal connectivity supports integrating disparate content into unified event memories. This suggests that changing the way that an event is described could change the underlying neural representation of integrated event memories. We tested this proposition by developing event narratives that described the same core story with identical central story details (e.g., grocery shopping), but described with additional descriptive details related to the story that were conceptual or perceptual in nature. Using fMRI, we established hippocampal connectivity patterns as a group of human participants encoded these narratives, and then related these patterns to later memory for the narrative details. We found distinct patterns of hippocampal-cortical connectivity during encoding of the identical central event details when they were embedded between conceptual vs. perceptual details.

Dr. Paul Scotti is Head of Neuroimaging & AI at Stability AI and a visiting scientist at Princeton University, where he did his postdoc with Dr. Ken Norman. Paul leads the MedARC Neuroimaging & AI Lab which adopts an open lab concept where the research team is almost entirely composed of volunteers working together on open-source neuroAI projects. Paul did his Ph.D at The Ohio State University co-advised by Dr. Julie Golomb and Dr. Andy Leber, where he worked on projects including attraction/repulsion dynamics surrounding visual mnemonic representations and novel improvements to the inverted encoding model technique used in neuroimaging.

Title: Reconstructing seen images from fMRI brain activity

Abstract: I will discuss how recent AI advances have enabled high-quality reconstructions of seen images from solely fMRI brain activity. Specifically, we leveraged large-scale fMRI data to align brain activity to the pre-trained embedding space of CLIP via diffusion models and contrastive learning. Subsequently we developed a simple approach to achieve high-quality reconstruction results with just 1 hour of fMRI training data by aligning multiple participants to a shared latent space.

Examining Developmental Variability in Functional Selectivity and Spatial Location of High-Level Visual Regions

Kelly J. Hiersche, Anna L. Quatrale, Zeynep M. Saygin

Department of Psychology, The Ohio State University

Introduction: The brain is a patchwork of regions, each devoted to different mental functions. Some of the most robust and replicable regions include high-level visual areas within ventral temporal cortex (VTC), such as the word-selective visual word form area (VWFA), face-selective fusiform face area (FFA), and object-selective posterior fusiform sulcus (PFS). Face and object regions show category selectivity very early in development, whereas word regions develop after literacy. These regions show high inter-subject variability in adults, but how does the variability of the selectivity and location of these regions change with age? 

Methods: We collected structural data, and two runs of a high-level visual localizer from 112 children (including repeated timepoints) and 64 adults. From this sample, children were grouped by age and motion matched across groups (24 per group, ages 3-6, 6-9, and 9-12 years), and to an adult reference group (N=17). We defined subject specific functional regions of interest (for VWFA, FFA, and PFS), used independent data to calculate selectivity to the preferred category, and determined the location of the fROI along the VTC based on each subject's parcellated brain. Using the adults as a reference, we calculated the coefficient of variation, a measure of variability, for the selectivity and location of each fROI for each child and examined age related changes. 

Results: We found variability of selectivity in the LH VWFA increases with age, whereas variability of selectivity decreases in bilateral FFA. Further, location variability of the LH VWFA increases with age, and LH VWFA shows a medial shift across development. This may be due to brain growth, as the youngest children also show significantly greater cortical thickness in most lateral sections of the VTC; however, cortical thickness of an fROI did not explain its location, selectivity, or the variability in either of those measures. 

Discussion: Increases in variability of the LH VWFA suggest that word specialization diverges over development and as literacy develops, whereas variability of face specialization converges towards adults over development. While cortical thickness is likely important for large-scale organization and development, it does not account for the variability of fROI location or responses. We next plan to examine if functional connectivity to functional-defined networks (e.g., language) may explain these variability differences across ages and individuals.

Keywords: Development, functional regions of interest, variability

How do we store Kanizsa illusions in visual working memory?

Jenny Lin, Lisa M. Heisterberg, Mackenzie J. Siesel, and Andrew B. Leber

Department of Psychology, The Ohio State University

Introduction: Visual working memory is limited in capacity, but we can use various strategies to improve performa2 items that form an illusory shape (Kanizsa illusion) versus items that are randomly oriented (Allon et al., 2019). However, the mechanism behind this benefit is still unclear. Some have proposed that Kanizsa illusions improve performance by reducing the number of items stored in working memory. Previous studies tested this idea using neuroimaging methods to indirectly measure working memory load during task performance. Specifically, researchers have measured the contralateral delay activity (CDA), an event-related potential component shown to track working memory load. Thus far, there is mixed evidence that Kanizsa illusions reduce storage demands (Heisterberg, 2021; McCollough, 2011; Ping, 2020; see also Diaz et al., 2021). Here we further explore the role of encoding time in the Kanizsa benefit. Specifically, we examine whether participants require longer encoding time to integrate the illusory shape and reduce working memory load. 

Methods: We had participants perform an orientation change detection task with EEG recording (N = 39; this preregistered study is in progress with a total planned sample of N = 80). On each trial, an arrow cued participants to attend to the left or right side of the screen. Next, participants saw memory stimuli consisting of either one “pacman” stimulus (single), three randomly oriented stimuli that form a triangle-like configuration (proximity), or three stimuli that form an illusory triangle (Kanizsa). Following the retention delay, participants responded whether the orientation of the test pacman is same or different as the memorized item. Critically, we manipulated encoding time of the memory display (short vs. long) between subjects. The encoding time was 200 ms in the short encoding group and 1000 ms in the long encoding group. 

Results: Preliminary results show a Kanizsa benefit for both short and long encoding groups: memory accuracy was higher for the Kanizsa versus proximity condition. There was a greater CDA amplitude for proximity versus single condition for both the short and long encoding group. This replicates the finding that CDA is sensitive to working memory load. Critically, we found a lower CDA amplitude for the Kanizsa compared to the proximity condition in the long encoding group.However, we did not find a statistically reliable difference between these two conditions in the short encoding group. 

Discussion: These findings might suggest that Kanizsa illusions reduce storage demands when there is sufficient encoding time for participants to use chunking strategies.

Keywords: visual working memory, strategy use, contralateral delay activity

Investigating Cerebellar Grey-Matter Volume Differences Across Brain Regions in Schizophrenia Versus Bipolar Disorder

Jayden Morales, Jessica Turner, Mahmoud Rashidi

The Ohio State University

Introduction. Schizophrenia (SZ) and bipolar disorder (BP) are debilitating mental illnesses that pose significant challenges for affected individuals. Symptoms of these conditions are associated with shared abnormalities in brain structure. Previous studies using neuroimaging techniques have revealed patterns of volume reduction in cortical regions, relative to healthy controls. This study examined volume differences across cerebellar gray- matter regions to identify distinctions between patients with bipolar disorder, patients with schizophrenia, and healthy controls, 

Methods. Participants were N=289 adults-subjects, including 52 patients with bipolar disorder, 89 patients with schizophrenia, 34 first-degree relatives of a patient with bipolar disorder, 56 first-degree relatives of a patient with bipolar disorder, and 51 healthy controls. All patient data was collected at the University of Minnesota. The automatic cerebellum anatomical parcellation using U-Net with locally constrained optimization (ACAPULCO) algorithm was used to perform a segmentation of the cerebellar lobules across 28 brain regions. Volumes associated with segmented cerebellar lobules were analyzed using linear regression models to measure individual differences between participant groups. Covariates including age, gender, and intracranial volume (ICV) were included in the models to account for the variability they bring forth. 

Results. Participants with Schizophrenia show reduced total cerebellar gray-matter volume relative to controls (p = 0.038). This effect no longer remained significant when gender, age, and ICV were uncontrolled. When SZ patients were compared to controls, the largest effects were found in the left Crus I, right Crus II, right VIIB, and left VIIB, with respect to order. All of these brain regions showed statistically significant reductions in volume. When BP patients were compared to controls, the largest effects were found in the left Crus I, left VIIA, left VIIB, and left V, with respect to order. None of these brain regions showed statistically significant reductions in volume. When SZ patients were compared to BP patients, the largest effects were found for total cerebellar volume (TCV), left Crus II, right Crus II, and right VIIB, with respect to order. All of these regions showed statistically significant reductions in volume. Interestingly, there were four brain regions where BP patients showed a larger magnitudinal effect in volume compared to SZ patients, relative to controls. These regions included the Vermis X, right VIIIB, left VIIIB, and left Crus II.

Discussion. Our results suggest that individuals with schizophrenia have notable volumetric reduction patterns across cerebellar regions, relative to both controls and patients with bipolar disorder. This indicates that there is in-fact a difference between patients with schizophrenia and patients with bipolar disorder in-terms of their cerebellar volume. None of the effects demonstrated by the BP group showed statistical significance, further analysis of this group in the future is needed. Furthermore, similar volume reduction patterns are found between patients with schizophrenia and patients with bipolar disorder relative to controls in the left Crus I, left VIIIA, left VIIB, left V, Vermis IX, Vermis VI, and a few other brain regions. While these volumetric reduction patterns are similar, they are significantly more apparent in the SZ group compared to the BP group.

Keywords. schizophrenia, bipolar disorder, cerebellar volumes, cerebellum, ACAPULCO, brain lobules

Connectomic reorganization following an 8-week mindfulness-based intervention in older adults

James Teng¹, Anita Shankar¹, Megan E. Fisher¹, Nathan R. McPherson¹, Madhura Phansikar¹, Rebecca Andridge^1,2, and Ruchika S. Prakash¹.

¹Department of Psychology, The Ohio State University, ²Division of Biostatistics, The Ohio State University

Introduction. Aging is associated with decreases in a multitude of cognitive and neural domains (Fortenbaugh et al., 2015; Koen and Rugg, 2019). To mitigate these age-related declines, behavioral interventions such as mindfulness have shown initial promise that must be investigated with methodological rigor, as the neural reorganization that accompanies these improvements remain poorly understood (Cotier et al. 2017). Entropy, a measure of network uniformity and specialization (Wang et al., 2014), can be analogous to neural dedifferentiation, and recent work has demonstrated increases in entropy with advancing age, which moderated cognitive declines (Shankar et. al, 2024). The current study examined within-sample entropy changes over a one-year period in older adults (Prakash et al., 2022) and examine the short-term (8 weeks) and long-term (one year) ability of mindfulness training to alter trajectories of network reorganization, at nodal, network, and whole-brain levels. Especially pertinent are the Default Mode network (DMN; Kral et al., 2019) and its posterior cingulate cortex (PCC) nodes, and the Frontoparietal Network (FPN) with its Inferior Frontal Gyrus (IFG) nodes implicated in mindfulness-related improvements in cognition (Taren et al., 2017).

Methods. 107 Older adults (aged 65 – 85) participated in the 8-week HealthyAgers randomized clinical trial with two intervention arms: Mindfulness-based stress reduction (N = 52) and lifestyle education (N = 55). Functional MRI (fMRI) data was collected at three timepoints (pre-intervention, post-intervention, and at 12-month follow-up) during rest and during a gradual-onset continuous performance task (gradCPT, Rosenberg et al., 2013). Functional connectivity was calculated from time-series Pearson's correlations for each node-pair of the Shen 268 atlas. From these, edge time series was derived by multiplying element-wise z-scored parcel time series for each node-pair in the connectivity matrix (Faskowitz et al., 2020). Next, k-means clustering was performed on the edge time series to assign a community label for each edge, before the distribution of edge community participation is calculated. Entropy is thus calculated as:

Where h_i is the entropy of the i^th node and p_ic is the fraction of the connections in the i^th node assigned to edge community c. Average entropy scores were calculated for the whole brain, the DMN and FPN, and nodes of the PCC and IFG. 

Results. Task fMRI showed no significant short- or long-term decreases in entropy across the whole-brain, nor within networks and nodes. However, across all participants, entropy was significantly reduced in resting state fMRI at a whole-brain level (F(2,144.4)=5.52, p=.0049), as well as within the DMN (F(2,147.9)=4.40, p=.0139), and FPN (F(2,154.1)=5.52, p=.0035). Nodal entropy did not significantly decrease, nor were there differential effects of the intervention on entropy. 

Discussion. While entropy was significantly improved at both a whole-brain and network level during resting state, improvements in task performance were only seen at a nodal level. This suggests that age-related resting entropic changes may be enhanced with standardized 8-week mind-body interventions, though the granularity of these effects suggests the benefits may not be uniform. Nevertheless, our results offer potent intervention candidates for addressing the deleterious effects of aging on cognitive function. 

Keywords: Network Neuroscience, Neural Reorganisation, fMRI, Entropy, Aging

Links among subcortical brain volume and social and emotional outcome in pediatric congenital heart disease

Declan E. Alford¹, Florencia Ontiveros¹, Aaron McAllister, M.D.^2,3, May Ling Mah, M.D.^2,3, Kathryn Vannatta, Ph.D.^1,3, Kristen R. Hoskinson, Ph.D.^1,3

¹Abigail Wexner Research Institute at Nationwide Children’s Hospital, ²Nationwide Children’s Hospital, ³The Ohio State University College of Medicine

Introduction: Across specific diagnoses, pediatric congenital heart diseases (CHD) can disrupt the oxygenation of blood circulating in the central nervous system, which could impact subcortical brain areas, including through volume reduction. This may contribute to comorbidities faced by CHD patients, including neurodevelopmental disruptions to behavioral and emotional function. Examining links among neural and behavioral sequelae of childhood CHD is essential, as a growing population survives into adolescence and into adulthood. This project explores the relationship among various subcortical volumes and behavioral implications in youth with CHD relative to healthy children (HC).

Methods: This project included children with CHD (n=23, M_age=11.71yr, Male=13), and HC (n=14, M_age=11.47yr, Male=12). Participants completed functional and structural MRI in a 3T Siemens Prisma Scanner. The participants’ parents completed the Child Behavior Checklist (CBCL) and youth completed the Youth Self Report (YSR), measuring emotional and behavioral function. Subcortical volumes, including bilateral basal ganglia, thalami, amygdale, and hippocampi, were segmented and quantified using Freesurfer and extracted for further analysis into IMB SPSS. Due to our small sample size, we focused on medium-to-large effect sizes (n_p²>.06, and Cohen’s d>0.50), and moderate-to-strong correlations (r>.30). 

Results: General linear models, each accounting for participant age and sex, identified differences in left thalamic volume (n_p²=.072), indicating that youth with CHD has a smaller volume compared to HC. Independent samples t-tests, one-tailed due to directional hypotheses, showed that mothers of youth with CHD reported reduced social competence (d=1.02), and elevated affective (d=-0.76) and anxiety problems (d=-0.61) relative to parents of HC, showing children with CHD are more prone to anxiety problems and socially difficulties. Partial correlations, accounting again for age and sex, showed that reduced thalamic volume was correlated with poorer mother-rated social competence (r=.41), affective problems (r=-.47) and anxiety problems (r=-.33). 

Discussion: Our findings suggest a potential relationship among brain volume and behavioral outcomes that clarify how the neural ramifications of CHD may contribute to observed functional morbidities in youth with critical forms of CHD. This suggests that providers may need to pay particular attention to behavioral outcomes for those with documented subcortical abnormalities as they enter middle childhood.

Keywords: Childhood congenital heart disease, subcortical brain regions, volume reduction, thalamus, behavioral outcomes

Preprocessing and reproducibility: Different artifact removal methods generate conflicting model conclusions

Cindy An, Kami Pearson, Katrina Aberizk, and Jessica Turner

Department of Psychiatry and Behavioral Health, The Ohio State University

Introduction: Reproducibility is essential to produce valid and stable research. As we currently face a crisis of reproducibility, a large focus has shifted to creating standardized data processing methods. Previously, as technology of neuroimaging developed, various tools emerged to prepare the data, including many which remove artifacts from head motion. The variety of the selection of tools and inner-customizability resulted in a damaging lack of reproducibility. 

Methods: In our study, we analyze the differences between Group Iterative Multiple Model Estimation (GIMME) networks created from data denoised using anatomical Component Based Correction (aCompCor) compared to Independent Component Analysis-based Automatic Removal of Motion Artifacts (ICA-AROMA). Data was acquired from the public COBRE dataset (n=157) which contained both healthy controls (HC) (n=90) and patients with schizophrenia (SZ) (n=67). Resting state fMRI data was preprocessed using the Harmonized Analysis of Functional MRI pipeline (HALFpipe), producing data different only by the specified denoising method. 

Results: This variation resulted in significantly different models generated by GIMME analyses despite usage of the same dataset and purportedly similar tools. Data denoised using ICA-AROMA resulted in no significant differences between HC and SZ. Comparatively, data denoised using aCompCor found significant differences between HC and SZ. Further, after removing comparisons between the clinical groups, whole study networks were still significantly different between the methods. 

Discussion: Our findings suggest that even minute differences in artifact removal may result in vastly different findings, even in studies with identical hypotheses.

Keywords: fMRI preprocessing, artifact removal, reproducibility, GIMME, ICA-AROMA, aCompCor

Examining structural and functional changes within the multiple demand network following a season of youth tackle football

Nii-Ayi Aryeetey¹, Kelly Hiersche¹, Jin Li¹, Jeff Pan¹, Ginger Yang², Sean Rose², James Onate¹, Jaclyn Caccese¹, Zeynep M. Saygin¹

¹The Ohio State University, ²Nationwide Children’s Hospital, Columbus, Ohio

Introduction: Should parents allow their children to play tackle football? Youth tackle football usually begins between ages 8-12, a period of rapid brain development. Specifically, children experience rapid gains in executive function in this age range. Executive function is supported by the frontoparietal multiple demand (MD) network, which is activated by cognitively demanding tasks that involve various aspects of executive function. Prior work suggests long-term disturbances in executive dysfunction in former tackle football players compared to non-football players. Further, working memory deficits are some of the most common in pediatric traumatic brain injury. Given the rapid growth of the MD network between ages 8-12 and the vulnerability of this network to repetitive neurotrauma, we asked whether the MD network showed blunted gray matter development due to football-related neurotrauma, and whether the MD network exhibited weaker functional connectivity in football players compared to controls. 

Methods: We scanned boys between 8 and 12 years of age (n=26) using fMRI before their initial season of tackle football to establish a baseline. We then scanned them post-season and compared their neurodevelopment to an age- and motion-matched control cohort (n=15) scanned within a similar timeframe. Participants were scanned while performing a spatial working memory task with hard and easy conditions, using one run to define the frontal and parietal functional regions of interest (fROIs) of the MD network and the other run to extract load-based activation (Hard > Easy). Participants were also scanned using resting state fMRI at each timepoint to determine their functional connectivity while not performing a task. We created connectivity matrices based on this data, identifying general differences in the full matrix between groups and across timepoints. We then employed machine learning algorithms to distinguish football players’ connectivity patterns from controls. Lastly, we conducted surface group analyses on sulcal depth and cortical thickness between groups and across time. 

Results: We found significant differences in changes to the working memory response across season between football players and controls within frontal and parietal fROIs of the MD network bilaterally. In these fROIs, football players showed a greater increase in load-based working memory activation as compared to controls. However, improvement on the task across time was similar between groups. The surface group analysis indicated greater sulcal depth longitudinally in a cluster within the left superior parietal sulcus in football players compared to controls. Further, we observed weaker functional connectivity within the MD network in football players compared to controls. 

Discussion: Our results indicate that after a single season of tackle football, children exhibit greater effort on a working memory task despite similar performance on the task between groups. This increased effort may be related to increased sulcal depth longitudinally in football players compared to controls. Ongoing longitudinal investigations will further explore the structural and functional relationships within the MD network and dose-response relationship of head impacts on cognitive outcomes.

Keywords: brain injury, multiple demand, executive functioning, working memory, fMRI, gray matter, connectivity

Resting state functional connectivity predicts individual differences in negative urgency in youth

Sophia A. Bibb and Zeynep M. Saygin

Neuroscience Graduate Program, Department of Psychology, Department of Psychiatry and Behavioral Sciences; Department of Psychology

Introduction: Adolescence is a critical period in the development of problem behaviors and psychopathology, including alcohol use disorder (AUD). Impulsivity is one neurobehavioral trait that presents strongly during adolescence and is robustly related to alcohol use (1,2). There are different components of impulsivity, one of which is the tendency to act impulsively when in a negative mood (i.e., negative urgency). The drinking-to-cope hypothesis posits that many individuals consume alcohol to alleviate distress, and previous work has demonstrated that negative urgency is related to increased binge drinking (3,4); thus, predicting negative urgency using brain connectivity data from pre-adolescent youth may provide insight into which individuals are most likely to initiate problem drinking later in adolescence. The goal of this project was to assess whether we can predict negative urgency scores using resting state functional connectivity (rsFC) data. 

Methods and Results: All data was collected through the Adolescent Behavioral Cognitive Development study and pre-processed by DCAN Labs. The rsFC matrices contained the Pearson’s correlations (Fisher’s r-to-z transformed) between the BOLD signal time courses of every pair of cortical regions in the 333-node Gordon atlas, plus 19 subcortical regions. We split our data into two equal training and a test groups (n1=3989 and n2=3980), matched for age and sex. We then trained a linear machine learning model on the training set of the rsFC to predict negative urgency data. The resulting model was then applied to the test subjects’ rsFC data to predict their negative urgency scores. Predicted negative urgency scores were significantly correlated with actual negative urgency scores. Nonparametric randomization tests on negative urgency scores will determine significance, and the model will reveal resting state correlations that are most predictive of negative urgency.

Discussion: The present work suggests that negative urgency, an important component of trait impulsivity, can be predicted using an individual’s functional connectome. This work has important implications for understanding and predicting the onset of problem alcohol use in youth. Future work will investigate whether a machine learning classifier can identify alcohol sipping behavior using rsFC, and whether negative urgency moderates the relationship between intrinsic functional connectivity and alcohol consumption in youth.

Keywords: resting state functional connectivity, impulsivity, alcohol, ABCD 

Interhemispheric Differences in Sulcal Depth Across Development

Laura Bradley, Kelly Hiershe, Anna Quatrale, Zeynep Saygin

Department of Psychology, The Ohio State University

Introduction: Numerous studies underscore the significance of sulcal pits as anatomical features closely linked to human brain function. Increased cortical folding (e.g. greater sulcal depth) may allow the human brain to achieve greater processing capacity and has been associated with higher-order cognitive abilities; as such, there may be hemispheric asymmetries in the sulcal pits around putative language areas and these asymmetries may even be precursors to the development of language specialization. However, language and other functional activation is highly variable in spatial location across subjects; there is a paucity of studies that map functional activation and link it to structure in individual participants, and so these hypotheses have not been formally tested. Here, we look at our large database of adult (n = 63) and child (n = 58) participants (some who have been scanned at multiple timepoints) and chart the development of sulcal depth asymmetries and link this to functional development. 

Methods: T1-weighted and task-based fMRI images were acquired in participants. We registered all subjects to an interhemispheric standard template in Freesurfer (using cortical folding to guide registration), concatenated LH and RH into a single template. We used a general linear model to examine variations in sulcal depth between the left and right hemispheres within and across age groups. Functional data for the language localizer (comparing blocks of auditory-presented meaningful sentences to matched nonsense sentences) were analyzed in native space and projected to the same template as the structural data, so that we could perform analyses at the group level. Individual subject analyses will also be performed to link function with structural asymmetries. 

Results: Adults showed 28 significant clusters that were asymmetric in sulcal depth, located in frontal (medial orbitofrontal, lateral orbitofrontal, superior frontal), temporal (superior temporal, middle temporal), parietal (inferior parietal, superior parietal), and other key regions (e.g., insula, precentral, lingual). Children showed a similar profile with age-based increases in asymmetry in these regions. Ongoing analyses explore longitudinal changes and the relationship of these asymmetries in functional development of the language network. 

Discussion: Our findings indicate maturation-related structural specialization, suggesting that enhanced sulcal depth reflects cognitive refinement throughout development.

Keywords: sulcal morphology, structural asymmetries, development, language 

Voxel-Wise Predictive Encoding Models Reveal Evidence for Pre-Saccadic Remapping in the Human Visual Cortex

Yong Min Choi, Julie D. Golomb

Department of Psychology, The Ohio State University

Introduction. Visual inputs shift drastically across saccadic eye movements. To align pre- and post-saccadic visual information and aide perceptual stability, neurons’ receptive fields (RFs) are remapped to new spatial locations during saccade preparation. Despite extensive evidence of pre-saccadic RF remapping in non-human primates, there remain open debates about the nature of remapping in different brain areas, and more generally it remains unclear how remapping operates in the human visual system. In the current study, we developed a novel fMRI paradigm to investigate pre-saccadic RF remapping across the human brain using a hypothesis-driven, voxel-wise predictive encoding model approach. 

Methods. Five participants completed two fMRI sessions: (1) a static population receptive field (pRF) mapping session to estimate voxel-wise static pRFs, and (2) a main experiment session presenting noise-like visual stimuli either during stable fixation or during saccade preparation. Using each voxel’s estimated static pRF, we constructed different hypothetical models of pre-saccadic remapping (e.g., “no predictive remapping”, “forward remapping”, “convergent remapping”) and evaluated how well each model predicted actual neural responses in the main experiment session. 

Results and Discussion. During stable fixation, the neural activity of most voxels in visual areas was best predicted by models assuming the current static pRF. Critically, during saccade preparation, many voxels - especially in later visual areas - showed improved predictability for models incorporating either forward or convergent RF remapping. These preliminary findings, consistent across subjects, provide strong evidence for pre-saccadic remapping in the human brain, with this new approach offering exciting potential to further broaden our understanding of how RF remapping operates across the cortex.

Keywords: fMRI, eye movement, pre-saccadic remapping, population receptive field, predictive encoding model 

Cardiorespiratory Fitness is Differentially Associated with Laterality of Motor Cortex fMRI Activation in Middle-Aged and Older Adults

Jessica A Cloud¹ Jasmeet P Hayes^1,2 Scott M Hayes^1,2

¹Department of Psychology, The Ohio State University; ²Chronic Brain Injury Initiative, The Ohio State University

Introduction: Unilateral patterns of BOLD activation (hemispheric asymmetry) during tasks in younger adults often shift to a more bilateral pattern of activation in older adulthood (reduction in hemispheric asymmetry). Individual differences (e.g., sex) are related to this reduction in hemispheric asymmetry. Modifiable fitness variables, including cardiorespiratory fitness, have been linked to brain measures in older adults, but limited research to date has explored the impact of modifiable fitness variables on task-related hemispheric asymmetry. In the current study, we examined the extent to which cardiorespiratory fitness, a modifiable lifestyle variable considered to support brain and cognitive maintenance, is associated with hemispheric asymmetry and task performance in middle-aged and older adults during a visual motor task. 

Methods: Adults (n = 80; 35-85 years; mean age (SD) =62.3 (10.9) years) from the Fitness, Aging, Stress, and TBI Exposure Repository (FASTER) were included in the analyses. Participants completed a graded maximal exercise test on a cycle ergometer to assess cardiorespiratory fitness (peak VO2). Participants completed a brain MRI using a Siemens 3T Prisma scanner at a separate appointment. Functional MRI scanning included a visual checkerboard task with alternating blocks of flickering checkerboard and rest. During checkerboard blocks, participants pressed buttons with the right hand to indicate the location of a periodic stimulus (left or right side of screen). fMRI data were processed using the FMRIB Software Library (FSL), identifying fMRI activation during the checkerboard blocks. Multiple regression assessed the relationship between laterality, peak VO2, and age, covarying for sex and education. 

Results: Group fMRI analyses identified a region in the right (ipsilateral) motor cortex showing age-related increases in BOLD activation (p < 0.001). Beta values from this region and a corresponding contralateral region were extracted and an asymmetry index calculated. Age modified the relationship between VO2peak and asymmetry such that with increasing age, the relationship between VO2peak and asymmetry became more negative (partial p < 0.001; model F(5, 74)= 2.85, p = 0.02). Among middle-aged adults greater VO2peak was associated with more asymmetry (p < 0.001); among older adults, greater VO2peak showed a trending association with reduced asymmetry (p = 0.11). More asymmetry was found to predict better task performance (p = 0.04), as assessed by a combination of accuracy and response time, after accounting for age, sex, and education. 

Discussion: Cardiorespiratory fitness was differentially associated with fMRI activation during a visual motor task in middle-aged and older adults. In middle-aged adults, greater cardiorespiratory fitness was associated with brain function patterns typically observed in young adults (unilateral activation); as age increased, however, this relationship became more negative, such that greater cardiorespiratory fitness was associated with a reduction in hemispheric asymmetry (more bilateral pattern of activation). These findings suggest that the role of cardiorespiratory fitness in supporting brain function changes with age and may suggest that decreased asymmetry reflects neural degeneration and dedifferentiation in aging.

Keywords: functional MRI, aging, cardiorespiratory fitness

Development of Visual Word Form Area (VWFA) Functional Laterality During Early Childhood and Relevance for Reading Behavior

Leah DiRubio, Jin Li, Zeynep Saygin

The Ohio State University

Introduction: Many functional regions in the human brain are lateralized and the Visual Word Form Area (VWFA) is one such region. The VWFA is an area of ventral temporal cortex (VTC) specialized for written scripts, and develops category selectivity only after literacy. In adults, this word selectivity is dominant on the left. What drives the emergence of this asymmetry over development? We tested three potential sources of this word laterality: 1) ipsilateral structural connectivity of the VWFA with the high-level language network in frontal and temporal cortices, 2) activation in these language regions, and 3) cross-hemispheric connectivity. We also explored the potential relevance of VWFA laterality for reading behavior in young children.

Methods: We scanned children (3-13 years, prereaders and readers) on a visual fMRI task to extract functional activation to visual words, scrambled words, line drawings of faces, and objects, as well as a separate auditory fMRI task with conditions for meaningful sentences, nonword phrases, and textured sound, and collected diffusion-weighted imaging (DWI) to examine white matter connectivity. We defined face, word, and (oral) language regions for each individual child and independent runs were used to extract activation to each condition to calculate selectivity and laterality; probabilistic tractography was used to quantify connections between these regions. 

Results: 1) Given that the VWFA’s instantiation is driven by connectivity with the language network, we hypothesize that structural connectivity between the VWFA and ipsilateral high-level language may relate to word laterality. 2) In line with the Interactive Specialization Hypothesis, which predicts that the functional specialization of a region relates to specialization of regions it is connected to, we suspect that language selectivity may also relate to word laterality. 3) We explored the potential role of cross-hemispheric VWFA connectivity in shaping laterality. Interestingly, in preliminary results, we found no relationship between cross-hemispheric VWFA connectivity and its laterality. 4) Finally, our preliminary results suggest that stronger VWFA laterality is associated with stronger reading ability. 

Discussion: Our results will elucidate how functional specialization of high-level language network and its connectivity with visual cortex may contribute to laterality of the VWFA. Additionally, exploring the development of VWFA laterality and how it relates to reading behavior is relevant for understanding the neural correlates of reading acquisition and can pave the way for future research related to education and reading impairment.

Keywords: Functional Laterality, Structural Connectivity, VWFA, Language

Mindfulness training effects on attentional control in older adults: A neurobehavioral randomized controlled trial

Megan Fisher, Madhura Phansikar, James Teng, Nathan McPherson, Rebecca Andridge, & Ruchika S. Prakash

Department of Psychology, The Ohio State University; College of Public Health, The Ohio State University; Center for Cognitive and Behavioral Brain Imaging, The Ohio State University

Introduction: Older adults exhibit age-related declines in processes of attentional control, prompting a growing need to identify mind-body interventions that support healthy neurocognitive aging. Mindfulness training is one such intervention showing promise for improving attentional control; however, most prior trials suffer from methodological limitations including passive control groups, small sample sizes, and lack of long-term assessments. A rich literature has also demonstrated that mindfulness training reliably alters functional connectivity patterns of select networks implicated in attentional control, including the default mode, frontoparietal, and salience networks. To-date, surprisingly few studies have employed network neuroscience approaches within the context of mindfulness training. Thus, our study leveraged whole-brain networks of sustained attention—a key component of attentional control—as one of our intervention targets given prior work showing that these networks are reliable, generalizable, and sensitive to interventions. 

Methods: This pre-registered phase-II randomized controlled trial (RCT) of 150 older adults examined the effects of 8-weeks of Mindfulness-based Stressed Reduction (MBSR) training compared to 8-weeks of an active control lifestyle education (LifeEd) training on neurobehavioral measures of attentional control. Participants completed two behavioral tasks of sustained attention (Go/No-Go with embedded mind- wandering thought probes and Conners Continuous Performance Test) at pre-training, post-training, 6-month, and 12-month follow-up. Functional magnetic resonance imaging (fMRI) assessments (N = 110)—during which participants completed the Gradual-Onset Performance Task—were completed at pre-training, post-training, and 12-month follow-up. Pre-specified contrasts were used to examine the neurobehavioral effects of MBSR versus an active control training over the short-term (pre-training to post-training) and long-term (pre-training to average of 6- and 12-month follow-ups). 

Results: Linear mixed effects models revealed that MBSR, relative to LifeEd, did not significantly improve behavioral measures of attentional control over the short-term and long-term. Similarly, MBSR, compared to LifeEd, did not significantly alter functional connectivity of whole-brain networks of sustained attention over the short- or long-term. Post-hoc exploratory analyses were conducted to examine potential main effects of training group and assessment time point on neurobehavioral measures of attentional control. 

Discussion: Our findings suggest that the idea that mindfulness training improves neural and/or behavioral measures of attentional control in older adults above-and-beyond active control training may be premature and warrants further investigation. Although our findings were unexpected, future studies may benefit by seeking ways to optimize mindfulness training for older adults and by identifying neural indices that may be more sensitive to the effects of mindfulness training. Doing so would allow researchers to better examine the impacts of mindfulness training on functional brain reorganization and the neural mechanisms through which cognitive changes may occur.

Keywords: aging; mindfulness training; fMRI; connectome-based predictive modeling; attentional control

Multidimensional default-mode network architecture in pediatric traumatic brain injury and its social correlates

Sophia M. Jeng¹, Florencia Ontiveros¹, William A. Cunningham, PhD², Warren D. Lo, MD^3,4, Kathryn Vannatta, PhD^1,3, Elisabeth A. Wilde, PhD⁵, Keith O. Yeates, PhD^6,7, and Kristen R. Hoskinson, PhD¹

¹Abigail Wexner Research Institute at Nationwide Children’s Hospital; ² University of Toronto, ON; ³The Ohio State University College of Medicine; ⁴Nationwide Children’s Hospital; ⁵University of Utah; ⁶University of Calgary, AB; ⁷Alberta Children’s Hospital Research Institute

Introduction: The default mode network (DMN) is a major brain network that is active during rest, and is particularly implicated in self-referential thought and social cognition. Traumatic brain injury (TBI) has been shown to negatively impact within-DMN connectivity which may contribute to persistent social morbidities even after acute recovery. As a growing number of youth survive even severe injury, understanding brain-behavior mechanisms is increasingly important. This study explores the relationship between multidimensional metrics reflective of DMN architecture and their associations with social outcomes in individuals with TBI. 

Methods: Fifty-eight 8-16 year old youth with orthopedic injury (OI; n=26; Mage=11.60y; 17 male), complicated- mild TBI (cmTBI; n=13; Mage=12.40y; 9 male), and moderate-to-severe TBI (msTBI; n=19; Mage=11.46y; 13 male) underwent MRI in a 3T Siemens scanner, including structural (MPRAGE) and resting state functional sequences. DMN cortical thickness, volume, surface area, and mean curvature were quantified using Freesurfer and within-DMN connectivity was measured using ROI-to-ROI analysis in CONN Toolbox. Parents rated their child’s Social Competence and Social Problems on the Child Behavior Checklist (CBCL) and Social adaptive skills on the Adaptive Behavior Assessment System-3 (ABAS-3). Group differences in DMN architecture and social outcome were assessed using one-way ANOVA and partial correlations determined links between DMN and social function in IBM SPSS. Due to limited sample size, we highlighted group differences that met or exceeded a medium effect size (Eta2³.06). 

Results: Groups differed in CBCL Social Competence (Eta2=.22) and ABAS-3 Social adaptive skills (Eta2=.23), showing that msTBI group were rated as having poorer social function compared to OI and cmTBI. Right and left DMN surface area (Eta2=.12 and .10, respectively) and right DMN cortical thickness (Eta2=.11) and mean curvature (Eta2=.11) also differed, with msTBI having lower surface area, cortical thickness, and mean curvature compared to cmTBI and OI. DMN functional connectivity between the medial prefrontal cortex and left lateral parietal region was reduced in youth with msTBI compared to OI. DMN surface area and CBCL Social Competence were significantly correlated (r=.287, p=.031). 

Discussion: TBI, particularly more severe injuries, has documented functional and structural brain changes, even after medical recovery. Greater DMN surface area was related to better social outcomes following pediatric TBI, meaning that those with more extensive structural damage after TBI are at especially high risk for poor social outcome. This improves our understanding of potential brain mechanisms underlying social morbidities that may follow a more severe TBI.

Keywords: Traumatic brain injury Default-mode network Social behavior

White Matter and Cognitive Changes due to an Initial Season of Tackle Football

Mei Jira, Dr. Saygin Zeynep

Department of Psychology, The Ohio State University

Introduction: American football is one of the most popular sports in the United States among youth. However, it has consistently been associated with the highest incidence of sports-related concussions in the country. This raises concerns about youth exposure to repetitive head impacts during critical periods of cognitive development. From 8 to 12 years of age, children rapidly improve their reading abilities, coinciding with a multitude of changes in white matter (WM). Previous research shows that a decrease in corpus callosum volume and a decrease in fractional anisotropy (FA) in the left arcuate fasciculus (AF) are associated with below-average reading skills in children. Do we observe blunted development of these WM tracts due to football- related head impacts and are these tracts associated with differences in reading ability in football players compared to controls? 

Methods: We scanned boys between 8 and 12 years of age (n=26) using fMRI before their initial season of tackle football to establish a baseline. We then scanned them post- season and compared their neurodevelopment to an age- and motion-matched control cohort (n=15) scanned within a similar timeframe. Participants completed anatomical T1 and DTI scans, as well as the KTEA Reading Comprehension subtest outside of the scanner. FA and volume data was obtained by running DTI scans through FreeSurfer’s TRACULA longitudinal pipeline. 

Results: We hypothesize that there will be a significant increase in the volume and FA of the left arcuate fasciculus in the control group, as well as increased volume in the corpus callosum, showing typical development between the first and second timepoints. However, we believe there will be no significant increases in any of these tract measures in football players between pre- and post-season. Control participants should show a significant increase in KTEA scores whereas football players should show a slight but insignificant increase. 

Discussion: Our hypothesized results would indicate that after a single season of tackle football, children show blunted development in tracts associated with reading development. These developmental differences coincide with behavioral differences in reading ability. Ongoing longitudinal investigations will explore whether football players’ white matter FA and volume and reading comprehension catch up to control participants with increasing time post-season. We will also explore the dose-response relationship of head impacts on reading comprehension.

Keywords: Youth Tackle Football, Corpus Callosum, Arcuate Fasciculate, Fractional Anisotropy, Volume 

Unfolding the spatiotemporal neural mechanisms of 3D perception in the human brain: an fMRI-EEG fusion study

Zitong Lu and Julie Golomb

Department of Psychology, The Ohio State University

Introduction: Although visual input is initially recorded in 2D on our retinas, we live in a 3D world, and our visual systems must integrate 2D representations with various depth cues to achieve 3D perception. However, it remains unclear exactly how and when during visual processing the brain integrates 2D and depth information into 3D representations. 

Methods: In this study, we collected fMRI and EEG data from participants viewing 3D stimuli with red-green anaglyph glasses. Participants first completed a behavioral session involving two depth judgment tasks and a 3D cube adjustment task to quantify different depth distances. They then participated in two EEG sessions and two fMRI sessions in which they viewed stimuli presented across 64 possible 3D locations. With a large number of trials (>7,000 trials per participant) and by integrating EEG, fMRI, and computational methods via representational similarity analysis, we are able to address previously unexplored questions. 

Results: In our preliminary data, we first tracked the spatiotemporal dynamics of different types of spatial representations, asking in which brain areas and at which points in time do representations emerge for spatial information about an object’s position in various coordinates: horizontal, vertical, radial, polar angle, depth, etc. Second, we ask whether there is evidence of additional integrated representations of 2D and/or 3D space beyond those individual components. In exploratory analyses we also investigated the representational format of spatial encoding and revealed a preference for a 2D polar coordinate system and 3D cylindrical coordinate system. 

Discussion: These findings provide valuable insights into the neural mechanisms underlying our ability to perceive the world in 3D.

Keywords: 3D perception, EEG, fMRI

Analysis of Temporal and Spatial Reliability of fMRI Signals in Emotion Regulation Task

Lingwen Ren, Nikki A. Puccetti, Athena L. Biggs, Stephanie Gorka, Luan Phan, Jay C. Fournier

The Ohio State University, Columbus, OH

Introduction: Recent research (Fröhner et al., 2019; Hajcak et al., 2017) raises serious concerns regarding the reliability of task-based functional magnetic resonance imaging (fMRI) metrics. Establishing the reliability of fMRI-based measurements is crucial for their clinical use, particularly in identifying meaningful biomarkers, tailoring personalized treatments, and guiding future research directions. Therefore, we evaluated three forms of reliability of a widely used emotion regulation fMRI task: split-half, test-retest, and spatial reliability.

Methods: Our sample consisted of forty adult participants with no history of psychiatric or neurologic illness that underwent two fMRI scans 12 weeks apart. During both scans, participants completed an emotion regulation task instructing them to view neutral or distressing images from the International Affective Picture System (IAPS). In the (1) “Look Neutral” and the (2) “Look Negative” (i.e., Maintain) conditions, participants naturally viewed neutral or negative images, respectively, without attempting to change their emotional response. In the (3) “Reappraise Negative” participants were asked to change their thoughts about the image to alter their emotional response. Between each image, a fixation cross was shown. We extracted BOLD signal from a group of ROIs (10 ROIs) from a meta-analysis (Morawetz et al., 2020) of emotion regulation brain activation. From these ROIs, we then calculated split-half reliability at time 1, test-retest reliability (intraclass correlation between time 1 and 2), Euclidean distance between peak voxels (from time 1 and 2), and cluster overlap (dice coefficients). 

Results: We found that split half reliability was highest for modelling Conditions alone, with the average falling in the “fair” range (average r=0.55, range= [-0.25, 0.82]). Both Condition-Fixation contrasts (average r =0.38, range= [-0.44, 0.74]) and the Condition-Look contrasts (average r =0.24, range= [-1.216, 0.616]) fell into the “poor” range (F (2,78) =13.66, p&lt;0.001). Test-retest reliabilities showed a different pattern, such that the ICC for contrasts between active conditions – fixation were the highest (F(2,78)=20.51, p&lt;0.001); average ICC = 0.40, range =[-0.26, 0.74]), followed by conditions alone (average ICC = 0.24, range =[-0.33, 0.54]), and active condition – look contrasts (average ICC = 0.07, range =[-0.31, 0.38]). We found higher spatial reliability when modelling Conditions alone, such that distances between peak activations during time 1 and 2 were shorter than for Contrasts (b = 2.01, SE = .37, t(704) = 5.45, p &lt; 0.001). 

Discussion: Reliability of the BOLD signal from task Conditions alone showed higher internal consistency across condition/contrast comparisons within a given scan whereas contrasts between active conditions – fixation had greatest test-retest reliability. Euclidean distance results show that conditions alone hold higher spatial consistency values at the group level from time 1 to time 2. These results suggest that future research using this task should be careful to select the most appropriate and reliable measures (conditions vs contrasts) based on their question of interest. Additionally, task-based fMRI researchers should consistently evaluate the psychometric properties of their data.

Keywords: Emotion Regulation, Internal Consistency, Spatial reliability

Investigating Visual and Auditory Language Reponses in the Ventral Temporal Cortex of Pre-Reading Children: A Multivariate Analysis

Lauren Rydel, Kelly Hiersche, and Zeynep Saygin

Department of Psychology, The Ohio State University

Introduction: The visual word form area (VWFA) is a region within ventral temporal cortex (VTC) that shows strong univariate preferences for visual words and letter strings but only after an individual learns how to read. Previous work suggests that distributed multivariate representations may form a scaffold for later-emerging univariate category-selectivity within the VTC for faces, bodies, and scenes: young children who show no hint of univariate selectivity to these categories show mature multivariate patterns in these regions. Does the VWFA also emerge from initially multivariate representations in pre- readers? And given that the VWFA is a ‘bridge’ between language and vision, does it show linguistic or visual multivariate representations? In this study, we first investigate whether pre-reading children exhibit a multivariate response to written words, line fragments of words/letters, faces, objects, auditory language, or other auditory stimuli across the VTC and within the VWFA. Next, we examine if the voxels that decode written words can also distinguish between auditory language and control conditions. 

Methods: Children (N=74), 35 pre-readers ages 2-6 and 39 readers ages 4-13, completed two runs of a high-level visual localizer task, with blocked conditions of words, scrambled words, objects, and faces, and 2 runs of an auditory language task with blocked conditions of linguistically meaningful sentences, matched nonword sentences, and texturized sound. Multi-voxel pattern analysis (MVPA) conducted on each subject identified voxels within the VTC with higher within-condition correlations than between conditions. Correlation values were Fisher Z-transformed before averaging across participants. 

Results: Preliminary results reveal that similar to readers, pre-readers show multivariate selectivity for words, as well as for auditory sentences within the VTC. 

Discussion: Further analyses will examine whether the voxels that decode words also distinguish between auditory language and control conditions, and univariate response selectivity in these voxels, as well as differences between hemispheres, in hopes of better understanding the emergence of new domains of knowledge in the human brain.

Keywords: ventral temporal cortex, visual word form area, multi-voxel pattern analysis, multivariate response, pre-readers, auditory language

Fetal structural and functional MRI scanning: experiences, challenges, and future development

Xinnan Wang¹, Kelly Hiersche¹, Anna Quatrale¹, Elizabeth Sponseller², Rachael F. Holt³, Zeynep M. Saygin¹

¹Department of Psychology, The Ohio State University; ²Center for Cognitive and Behavioral Brain Imaging, Ohio State University; ³Department of Speech and Hearing Science, Ohio State University

Introduction: Unraveling the initial stage of brain development is an important part of understanding the human brain. Structural MRI reveals crucial information on fetal neuroanatomy, charting development of gray and white matter micro- and macrostructure, while functional MRI allows us to examine both early underpinnings of mental function (task-based fMRI) as well as the role of spontaneous intrinsic activity in developing functional networks (functional connectivity). These fetal markers can potentially predict later cognition and unpack the role of prenatal and postnatal experience in shaping this cognition in both typical and atypical development. However, due to fetal motion, the quality of fetal scanning data is often suboptimal, leading to substantial data loss. It also complicates data acquisition and processing. These undesirable consequences underscore the importance of minimizing both the occurrence and impact of motion in fetal fMRI studies. This abstract outlines our current methods for fetal data collection and our preliminary work in processing structural, task-based, and resting state fetal MRI data. 

Methods: In the current study, T2-weighted and fMRI data were collected using a coil positioned on the mother's abdomen according to where the fetus’s head is situated. Unlike adult T2-weighted scans, fetal imaging involves high- resolution scanning in one dimension at a time, resulting in sequential acquisitions. Resting-state fMRI scans last 21 minutes, although this duration varies depending on the mother's comfort and can be shortened to as little as 15 minutes if needed. Task-based fMRI scans last 5 minutes per run and consist of auditory cues to the mother to either read a book or hum a melody out loud (which the fetus will presumably hear) for 4 blocks, with interspersed rest blocks. Mothers are instructed to lie on their side, entering the bore feet first. To date, we have completed scans on nine fetuses at 31 to 35 weeks of gestational age. 

Results: Planned data analyses include registration of T2 scans to fetal templates using fetal brain extraction tool Fetal- BET and measuring sulcal depth, cortical thickness, and white matter volumes. We will build functional connectomes using seed-based functional connectivity analysis. Task-based fMRI will be analyzed with contrasts of read &gt; hum as well as comparisons to rest to identify auditory cortices and early precursors to the language network. Fetal motion will be measured using fetal preprocessing pipeline and independent component analyses. We also plan to characterize the relationship between fetal motion, gestational age, scan duration, and other parameters like smoothing kernels for fMRI data, which we hope will contribute to the development of optimized models for future fetal MRI procedures. 

Discussion: Ultimately, refining preprocessing steps will enable more reliable insight into fetal brain development, and our analyses will start to reveal early precursors of the developing mind

Keywords: fetal fMRI, fetal structural MRI, ICA

Decoding identity from representations of traits, attitudes, and moral character

Dan Zhu, Dylan Wagner

Department of Psychology, The Ohio State University

Introduction: When we first meet someone, what kinds of information do we rely on or prioritize when forming impressions? Previous research suggests that social information such as personality trait, moral character, and attitude/opinion is important for person perception and interpersonal attraction (Byrne, 1961; Goodwin, 2015; Hassabis et al., 2014). In the present study, we examined whether different social identities could be reliably decoded in the medial prefrontal cortex (mPFC) based on personality, morality, and attitude information using multivariate pattern classification analysis. 

Methods: Twenty-three right-handed adults between the ages of 18 to 35 participated in this study. Subjects watched an episode of The Circle and completed a trait evaluation task (Broom et al., 2021) during fMRI scanning with 8 characters previously introduced in the show. Across six runs, subjects used a 5- button box to rate each target on 18 personality traits, 18 moral characters, and their attitudes to 18 daily activities on a 5-point scale. 

Results: The within-subject and between-subject classification accuracies were both significantly above chance (chance = 0.125) when using information across three categories (within acc = 0.21, between acc = 0.18). When we fed information in only one category into the classifier, different identities could still be reliably decoded (acc = 0.15 – 0.17). Furthermore, cross-categorical classification accuracies were significantly above chance level in the mPFC for 4/6 combinations (personality-morality, morality-personality, attitude-personality, attitude-morality), indicating that the cross-categorical social information is represented by a similar neural code. 

Discussion: Our study suggests that the brain incorporates personality, morality, and attitude information to form impressions of novel others. Although identities could be decoded with information in one category, the accuracy when using information across all three categories was higher. In conclusion, this work extends our understanding on person perception and suggests the cooperation of different types of social information in person perception.

Keywords: identity decoding, impression formation, social perception, medial prefrontal cortex