Irregular Changes in Network Topology During Monotonic Learning in Health and Schizophrenia

686 

Abstract Submission 

Clifford Abel II¹, John Kopchick², Dhruval Bhatt³, Hady Saad³, Patricia Thomas², Dalal Khatib², Usha Rajan², Caroline Zajac-Benitez³, Luay Haddad², Alireza Amirsadri², Jeffrey Stanley², Vaibhav Diwadkar³

¹Wayne State University, Translational Neuroscience Program, Department of Psychiatry, Detroit, MI, ²Wayne State University, Department of Psychiatry, Detroit, MI, ³Wayne State University, Detroit, MI

Clifford Abel II, BS  
Wayne State University, Translational Neuroscience Program, Department of Psychiatry
Detroit, MI

John Kopchick  
Wayne State University, Department of Psychiatry
Detroit, MI

Dhruval Bhatt  
Wayne State University
Detroit, MI

Hady Saad  
Wayne State University
Detroit, MI

Patricia Thomas  
Wayne State University, Department of Psychiatry
Detroit, MI

Dalal Khatib  
Wayne State University, Department of Psychiatry
Detroit, MI

Usha Rajan  
Wayne State University, Department of Psychiatry
Detroit, MI

Caroline Zajac-Benitez  
Wayne State University
Detroit, MI

Luay Haddad  
Wayne State University, Department of Psychiatry
Detroit, MI

Alireza Amirsadri  
Wayne State University, Department of Psychiatry
Detroit, MI

Jeffrey Stanley  
Wayne State University, Department of Psychiatry
Detroit, MI

Vaibhav Diwadkar  
Wayne State University
Detroit, MI

Task-based neuroimaging has recapitulated relationships between behavioral proficiency and brain imaging measures in specific examples such as retrieval success or retention (Barnett et al., 2023) but the issue is more complicated in for performance dynamics over time. In associative learning tasks, retrieval proficiency increases (as associations are consolidated), and has been associated with changes in effective connectivity that are disordered in schizophrenia (Banyai et al., 2011). However, these relationships have not been examined under the simple principle of monotonicity prevalent in human psychology (Grice et al., 2023). Thus, while performance improves in a weakly monotonic way (i.e., each iteration is greater or equal to the previous one), do connectomic changes reflect such monotonicity? Here, we implement this question as follows: 1) fMRI data were collected while healthy controls (HC) and patients with schizophrenia (SCZ) learned object location associations (resulting in negatively accelerated learning)(Hasan et al., 2023); 2) Network profiles across task conditions were summarized using the graph theoretic measure Betweenness Centrality (BC) that estimates the "hubness" of a region (node); 3) Finally, BC dynamics were tested for weak monotonicity. We demonstrate that at best, connectomic changes have obscure relationships with behavioral proficiency over time. ...

Participants (n=88, SCZ=49, Ages:18-45) gave informed consent to participate in fMRI acquisition (Siemens Verio 3T) while learning associations between memoranda (objects and locations) over eight epochs. Each epoch contained four conditions: Encoding, Post-Encoding Rest, Retrieval, Post-Retrieval Rest. For each participant, in all conditions and epochs, the functional connectome across 246 functionally defined nodes (Fan et al., 2016) was estimated using zero-lag functional connectivity. From each undirected graph, each node's BC was estimated and rank ordered (BCRO).

Both HC and SCZ showed increasing task proficiency over the eight epochs (Fig. 1a), but a higher proportion of SCZ did so non-monotonically (Fig. 1b, p < 0.05). Linear regression estimation of the relationship between epoch and BCRO for each node in each group/condition identified nodes displaying significant effects (pFDR<.01) but none of these nodes showed weak-monotonicity (p < 0.05). We therefore considered the subject-level within condition nodal variation in BCRO across the epochs. Separately for HC and SCZ, agglomerative clustering using the Ward method was applied across subjects to give five clusters of regions displaying similar variation (Fig 2).

Our failure to recapitulate the performance dynamics (weakly monotonic) in the connectomic dynamics (devoid of monotonicity) is a successful dissociation of behavior and the underlying brain activation; consistent behavioral dynamics are generated from highly variable connectomic fluctuations. A proper investigation of these connectomic dynamics must therefore treat the nodal BCRO variability as part of learning. Exploratory analyses suggest more a more heterogeneous distribution of clusters throughout the HC brain than the SCZ brain (Fig. 2).

Banyai, M., Diwadkar, V., Erdi, P., 2011. Model-based dynamical analysis of functional disconnection in schizophrenia. NeuroImage 58(3), 870-877.

Barnett, Alexander J. et al. (2023) "Hippocampal-cortical interactions during event boundaries support retention of complex narrative events." Neuron, S0896-6273(23)00766-3

Fan, L. et al. (2016) The Human Brainnetome Atlas: A new brain atlas based on connectional architecture. Cerebral cortex, 26(8), 3508–3526

Grice, Matt et al. (2023) "The psychological scaffolding of arithmetic." Psychological review, 10.1037/rev0000431

Hasan, S. et al. (2023) Learning without contingencies: A loss of synergy between memory and reward circuits in schizophrenia. Schizophrenia research 258, 21-35

Psychiatric (eg. Depression, Anxiety, Schizophrenia) ¹

Learning and Memory Other ²

Activation (eg. BOLD task-fMRI)

Connectivity (eg. functional, effective, structural)

fMRI Connectivity and Network Modeling

Cognition

Data analysis

FUNCTIONAL MRI

Learning

Memory

MRI

Plasticity

Psychiatric

Psychiatric Disorders

Schizophrenia