Research

My current research aims to develop more precise brain models of learning using machine learning and MRI.

Specifically, I am reseaching neural plasticity induced by motor skill learning using a visuomotor task that can assess accuracy, adaptation, and speed in participant responses to stimuli while in an MRI scanner over many sessions (dense sampling). As part of my research, I am also drawn to questions related to how to make more precise brain models, especially in relation to the parcellation question of how to functionally divide the brain into regions. Another way to put this is how to reduce the high dimensionality of MRI scans (millions of voxels or hundreds of thousands of vertices) into a lower dimension of hundreds of parcels or regions.

Previously, I've worked on a variety of research topics related to effort-based decision making, pain, and embodied neuroscience (e.g., proprioception, mind-body practices). More about my past research experience can be found in the About section. Journal publications from past research are listed below.

Related data, software, and code can be found in the Data+Code section.


Publications

Last updated: 2024-10-04; see my Google Scholar for most up-to-date.
* denotes co-first author; denotes trainee

Preprints

Do you ever get tired of being wrong? The unique impact of feedback on subjective experiences of effort-based decision-making.
[Preprint] [Code]
Anna Xu, Romy Frömer, Wanja Wolff, and Amitai Shenhav
Abstract

To achieve a goal, people have to keep track of how much effort they are putting in (effort monitoring) and how well they are performing (performance monitoring), which can be informed by endogenous signals, or exogenous signals providing explicit feedback about whether they have met their goal. Interventions to improve performance often focus on adjusting feedback to direct the individual on how to better invest their efforts, but is it possible that this feedback itself plays a role in shaping the experience of how effortful the task feels? Here, we examine this question directly by assessing the relationship between effort monitoring and performance monitoring. Participants (N = 68) performed a task in which their goal was to squeeze a handgrip to within a target force level (not lower or higher) for a minimum duration. On most trials, they were given no feedback as to whether they met their goal, and were largely unable to detect how they had performed. On a subset of trials, however, we provided participants with (false) feedback indicating that they had either succeeded or failed at meeting their goal (positive vs. negative feedback blocks, respectively). Sporadically, participants rated their experience of effort exertion, fatigue, and confidence in having met the target grip force on that trial. Despite being non-veridical to their actual performance, we found that the type of feedback participants received influenced their experience of effort. When receiving negative (vs. positive) feedback, participants fatigued faster and adjusted their grip strength more for higher target force levels. We also found that confidence gradually increased with increasing positive feedback and decreased with increasing negative feedback, again despite feedback being uniformly uninformative. These results suggest differential influences of feedback on experiences related to effort and further shed light on the relationship between experiences related to performance monitoring and effort monitoring.


hyve, a compositional visualisation engine for brain imaging data.
[Preprint] [Code]
Rastko Ćirić, Anna Xu, Russell Alan Poldrack
Abstract

Visualisations facilitate the interpretation of geometrically structured data and results. However, heterogeneous geometries—such as volumes, surfaces, and networks—have traditionally mandated different software approaches. We introduce hyve, a Python library that uses a compositional functional framework to enable parametric implementation of custom visualisations for different brain geometries. Under this framework, users compose a reusable visualisation protocol from geometric primitives for representing data geometries, input primitives for common data formats and research objectives, and output primitives for producing interactive displays or configurable snapshots. hyve also writes documentation for user-constructed protocols, automates serial production of multiple visualisations, and includes an API for semantically organising an editable multi-panel figure. Through the seamless composition of input, output, and geometric primitives, hyve supports creating visualisations for a range of neuroimaging research objectives.


2023

Structural imaging studies of patients with chronic pain: an anatomic likelihood estimate meta-analysis. PAIN.
[Link] [Code]
Alina T. Henn, Bart Larsen, Lennart Frahm, Anna Xu, Azeez Adebimpe, J. Cobb Scott, Sophia Linguiti, Vaishnavi Sharma , Allan I. Basbaum, Gregory Corder, Robert H. Dworkin, Robert R. Edwards, Clifford J. Woolf, Ute Habel, Simon B. Eickhoff, Claudia R. Eickhoff, Lisa Wagels, Theodore D. Satterthwaite
Abstract

Neuroimaging is a powerful tool to investigate potential associations between chronic pain and brain structure. However, the proliferation of studies across diverse chronic pain syndromes and heterogeneous results challenges data integration and interpretation. We conducted a preregistered anatomic likelihood estimate meta-analysis on structural magnetic imaging studies comparing patients with chronic pain and healthy controls. Specifically, we investigated a broad range of measures of brain structure as well as specific alterations in gray matter and cortical thickness. A total of 7,849 abstracts of experiments published between January 1, 1990 and April 26, 2021 were identified from eight databases and evaluated by two independent reviewers. Overall, 103 experiments with a total of 5,075 participants met the pre-registered inclusion criteria. After correction for multiple comparisons using the gold standard family-wise error correction (p<. 05), no significant differences associated with chronic pain were found. However, exploratory analyses using threshold-free cluster enhancement revealed several spatially distributed clusters showing structural alterations in chronic pain. The majority of clusters coincided with regions implicated in nociceptive processing including the amygdala, thalamus, hippocampus, insula, anterior cingulate cortex and inferior frontal gyrus. Taken together, these results suggest that chronic pain is associated with subtle, spatially distributed alterations of brain structure.


2022

Decision value signals in the ventromedial prefrontal cortex and motivational and hedonic symptoms across mood and psychotic disorders. Neuroimage: Clinical.
[Link] [Preprint]
Min K. Souther, Daniel H. Wolf, Rebecca Kazinka, Sangil Lee, Kosha Ruparel, Mark A. Elliott, Anna Xu, Matthew Cieslak, Greer Prettyman, Theodore D. Satterthwaite, Joseph W. Kable
Abstract

Deficits in motivation and pleasure are common across many psychiatric disorders, and manifest as symptoms of amotivation and anhedonia, which are prominent features of both mood and psychotic disorders. Here we provide evidence for an association between neural value signals and symptoms of amotivation and anhedonia across adults with major depression, bipolar disorder, schizophrenia, or no psychiatric diagnosis. We found that value signals in the ventromedial prefrontal cortex (vmPFC) during intertemporal decision-making were dampened in individuals with greater motivational and hedonic deficits, after accounting for primary diagnosis. This relationship remained significant while controlling for diagnosis-specific symptoms of mood and psychosis, such as depression as well as positive and negative symptoms. Our results demonstrate that dysfunction in the vmPFC during value-based decision-making is specifically linked to motivational and hedonic impairments. These findings provide a quantitative neural target for the potential development of novel treatments for amotivation and anhedonia.


2021

Brain responses to noxious stimuli in patients with chronic pain: A systematic review and meta-analysis. JAMA Network Open.
[Link] [Code]
Anna Xu, Bart Larsen, Alina Henn, Erica B. Baller, J. Cobb Scott, Vaishnavi Sharma , Azeez Adebimpe, Allan I. Basbaum, Gregory Corder, Robert H. Dworkin, Robert R. Edwards, Clifford J. Woolf, Simon B. Eickhoff, Claudia R. Eickhoff, Theodore D. Satterthwaite
Abstract

Importance Functional neuroimaging is a valuable tool for understanding how patients with chronic pain respond to painful stimuli. However, past studies have reported heterogenous results, highlighting opportunities for a quantitative meta-analysis to integrate existing data and delineate consistent associations across studies.

Objective To identify differential brain responses to noxious stimuli in patients with chronic pain using functional magnetic resonance imaging (fMRI) while adhering to current best practices for neuroimaging meta-analyses.

Data Sources All fMRI experiments published from January 1, 1990, to May 28, 2019, were identified in a literature search of PubMed/MEDLINE, EMBASE, Web of Science, Cochrane Library, PsycINFO, and SCOPUS.

Study Selection Experiments comparing brain responses to noxious stimuli in fMRI between patients and controls were selected if they reported whole-brain results, included at least 10 patients and 10 healthy control participants, and used adequate statistical thresholding (voxel-height P < .001 or cluster-corrected P < .05). Two independent reviewers evaluated titles and abstracts returned by the search. In total, 3682 abstracts were screened, and 1129 full-text articles were evaluated.

Data Extraction and Synthesis Thirty-seven experiments from 29 articles met inclusion criteria for meta-analysis. Coordinates reporting significant activation differences between patients with chronic pain and healthy controls were extracted. These data were meta-analyzed using activation likelihood estimation. Data were analyzed from December 2019 to February 2020.

Main Outcomes and Measures A whole-brain meta-analysis evaluated whether reported differences in brain activation in response to noxious stimuli between patients and healthy controls were spatially convergent. Follow-up analyses examined the directionality of any differences. Finally, an exploratory (nonpreregistered) region-of-interest analysis examined differences within the pain network.

Results The 37 experiments from 29 unique articles included a total of 511 patients and 433 controls (944 participants). Whole-brain meta-analyses did not reveal significant differences between patients and controls in brain responses to noxious stimuli at the preregistered statistical threshold. However, exploratory analyses restricted to the pain network revealed aberrant activity in patients.

Conclusions and Relevance In this systematic review and meta-analysis, preregistered, whole-brain analyses did not reveal aberrant fMRI activity in patients with chronic pain. Exploratory analyses suggested that subtle, spatially diffuse differences may exist within the pain network. Future work on chronic pain biomarkers may benefit from focus on this core set of pain-responsive areas.


2020

Convergent neural representations of experimentally-induced acute pain in healthy volunteers: A large-scale fMRI meta-analysis. Neuroscience & Biobehavioral Reviews.
[Link] [Code] [Preprint]
Anna Xu, Bart Larsen, Erica B. Baller, J. Cobb Scott, Vaishnavi Sharma , Azeez Adebimpe, Allan I. Basbaum, Robert H. Dworkin, Robert R. Edwards, Clifford J. Woolf, Simon B. Eickhoff, Claudia R. Eickhoff, Theodore D. Satterthwaite
Abstract

Characterizing a reliable, pain-related neural signature is critical for translational applications. Many prior fMRI studies have examined acute nociceptive pain-related brain activation in healthy participants. However, synthesizing these data to identify convergent patterns of activation can be challenging due to the heterogeneity of experimental designs and samples. To address this challenge, we conducted a comprehensive meta-analysis of fMRI studies of stimulus-induced pain in healthy participants. Following pre-registration, two independent reviewers evaluated 4,927 abstracts returned from a search of 8 databases, with 222 fMRI experiments meeting inclusion criteria. We analyzed these experiments using Activation Likelihood Estimation with rigorous type I error control (voxel height p < 0.001, cluster p < 0.05 FWE-corrected) and found a convergent, largely bilateral pattern of pain-related activation in the secondary somatosensory cortex, insula, midcingulate cortex, and thalamus. Notably, these regions were consistently recruited regardless of stimulation technique, location of induction, and participant sex. These findings suggest a highly-conserved core set of pain-related brain areas, encouraging applications as a biomarker for novel therapeutics targeting acute nociceptive pain.


Distinct insular functional connectivity changes related to mood and fatigue improvements in major depressive disorder following tai chi training: A pilot study. Frontiers in Integrative Neuroscience.
[Link]
Anna Xu, Chloe S. Zimmerman*, Sara W. Lazar, Yan Ma, Catherine E. Kerr, Albert Yeung
Abstract

Objective: Tai chi (TC), a contemplative practice combining slow movements and deep breathing, has been shown to be clinically effective in alleviating depressive symptoms. Feelings of fatigue or low vitality often accompany major depressive disorder (MDD) though they are commonly overlooked and not well understood neurologically. By using resting state functional connectivity (rs-FC) using the insula as the seed, this study examines the relationship between mood and vitality symptoms in MDD and how they are impacted by TC training.

Methods: Patients (N = 16) with MDD participated in a 10-week TC intervention. Self-report scores of vitality (using the SF-36 scale) and depressed mood (using the Beck Depression Inventory) as well as rs-fMRI were collected pre- and post-intervention. A seed-to-voxel approach was used to test whether changes in insular rs-FC were related to therapeutic improvement in MDD-related symptoms resulting from TC practice.

Results: We found decreased self-reported depressed mood and increased vitality following the TC intervention. Furthermore, decreases in depressed mood were associated with increased rs-FC between the right anterior insula (AIC) and superior temporal gyrus and caudate (cluster-corrected p < 0.05). Increased vitality was associated with increased rs-FC between the right posterior insula (PIC) and regions associated with sensorimotor processes (cluster-corrected p < 0.05).

Conclusion: These results provide support for differential changes in insula connectivity as neural correlates of symptom improvement in MDD.


2019

Epidemiology of chronic pain in Ukraine: findings from the World Mental Health Survey. PLoS ONE.
[Link]
Anna Xu, Elizabeth Hilton, Riley Arkema, Nathan L. Tintle, Luralyn M. Helming
Abstract

Chronic pain can pose a serious challenge in everyday life for many individuals globally, especially in developing countries, but studies explicitly exploring risk factors of chronic pain beyond demographic characteristics using survey data have been scarce. To address this problem, this study analyzed World Health Organization data on chronic pain in Ukraine to explore demographic, psychological, and treatment perception-related risk factors to chronic pain. We replicated previous reports of older age, female sex, married status, inadequate financial resources, and comorbidity of other physical conditions as significant demographic risk factors for chronic pain diagnosis but not necessarily for severe pain. We also found evidence for psychological risk factors and treatment perceptions as significant predictors for chronic pain diagnosis and its severity. These results provide a first step in examining beyond demographic risk factors for chronic pain diagnosis and severity and, instead, assessing potential psychological risk factors.


2018

Comparing embodiment experiences in expert meditators and non-meditators using the rubber hand illusion. Consciousness and Cognition.
[Link]
Anna Xu, Brendan H. Cullen, Cooper Penner, Chloe S. Zimmerman, Catherine E. Kerr, Laura Schmalzl
Abstract

One assessment of embodiment is the rubber hand illusion (RHI), a visuo-tactile illusion in which individuals attribute a sense of ownership to a rubber hand and disownership to their real hand. Interestingly, interoception seems to influence RHI susceptibility. In this study, we administered the RHI and the Multidimensional Assessment of Interoceptive Awareness (MAIA) to examine embodiment experiences and interoceptive awareness in experienced meditators (n = 15) and non-meditators (n = 15). We found that meditators reported less intensity in rubber hand ownership, but there was no significant difference between groups with respect to disownership of their real hand or drift in finger proprioception. Moreover, we found, from our MAIA results, that disownership experiences were associated with a feeling of trusting one's body in non-meditators and with the ability to maintain attention to unpleasant bodily sensations in meditators. These results suggest a unique relationship between interoceptive awareness and embodiment related to meditation.