Author name cluster

Jörn Diedrichsen

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

16 papers

1 author row

YNIMG Journal 2022 Journal Article

Predicting brain activation maps for arbitrary tasks with cognitive encoding models

Jonathon Walters
Maedbh King
Patrick G. Bissett
Richard B. Ivry
Jörn Diedrichsen
Russell A. Poldrack

A deep understanding of the neural architecture of mental function should enable the accurate prediction of a specific pattern of brain activity for any psychological task, based only on the cognitive functions known to be engaged by that task. Encoding models (EMs), which predict neural responses from known features (e.g., stimulus properties), have succeeded in circumscribed domains (e.g., visual neuroscience), but implementing domain-general EMs that predict brain-wide activity for arbitrary tasks has been limited mainly by availability of datasets that 1) sufficiently span a large space of psychological functions, and 2) are sufficiently annotated with such functions to allow robust EM specification. We examine the use of EMs based on a formal specification of psychological function, to predict cortical activation patterns across a broad range of tasks. We utilized the Multi-Domain Task Battery, a dataset in which 24 subjects completed 32 ten-minute fMRI scans, switching tasks every 35 s and engaging in 44 total conditions of diverse psychological manipulations. Conditions were annotated by a group of experts using the Cognitive Atlas ontology to identify putatively engaged functions, and region-wise cognitive EMs (CEMs) were fit, for individual subjects, on neocortical responses. We found that CEMs predicted cortical activation maps of held-out tasks with high accuracy, outperforming a permutation-based null model while approaching the noise ceiling of the data, without being driven solely by either cognitive or perceptual-motor features. Hierarchical clustering on the similarity structure of CEM generalization errors revealed relationships amongst psychological functions. Spatial distributions of feature importances systematically overlapped with large-scale resting-state functional networks (RSNs), supporting the hypothesis of functional specialization within RSNs while grounding their function in an interpretable data-driven manner. Our implementation and validation of CEMs provides a proof of principle for the utility of formal ontologies in cognitive neuroscience and motivates the use of CEMs in the further testing of cognitive theories.

YNIMG Journal 2021 Journal Article

Sharing voxelwise neuroimaging results from rhesus monkeys and other species with Neurovault

Andrew S. Fox
Daniel Holley
Peter Christiaan Klink
Spencer A. Arbuckle
Carol A. Barnes
Jörn Diedrichsen
Sze Chai Kwok
Colin Kyle

Animal neuroimaging studies can provide unique insights into brain structure and function, and can be leveraged to bridge the gap between animal and human neuroscience. In part, this power comes from the ability to combine mechanistic interventions with brain-wide neuroimaging. Due to their phylogenetic proximity to humans, nonhuman primate neuroimaging holds particular promise. Because nonhuman primate neuroimaging studies are often underpowered, there is a great need to share data amongst translational researchers. Data sharing efforts have been limited, however, by the lack of standardized tools and repositories through which nonhuman neuroimaging data can easily be archived and accessed. Here, we provide an extension of the Neurovault framework to enable sharing of statistical maps and related voxelwise neuroimaging data from other species and template-spaces. Neurovault, which was previously limited to human neuroimaging data, now allows researchers to easily upload and share nonhuman primate neuroimaging results. This promises to facilitate open, integrative, cross-species science while affording researchers the increased statistical power provided by data aggregation. In addition, the Neurovault code-base now enables the addition of other species and template-spaces. Together, these advances promise to bring neuroimaging data sharing to research in other species, for supplemental data, location-based atlases, and data that would otherwise be relegated to a "file-drawer". As increasing numbers of researchers share their nonhuman neuroimaging data on Neurovault, this resource will enable novel, large-scale, cross-species comparisons that were previously impossible.

YNIMG Journal 2019 Journal Article

A representative template of the neonatal cerebellum

Carlos R. Hernandez-Castillo
Catherine Limperopoulos
Jörn Diedrichsen

The cerebellum plays an important role in human brain development. To improve the spatial specificity of the analysis of human cerebellar magnetic resonance imaging (MRI) data, we present a new template of the neonatal human cerebellum and brainstem based on the anatomy of 20 full-term healthy neonates. The template is spatially unbiased, which means that the location of each structure is not biased by the anatomy of the individuals used to create the template. In comparison to current whole-brain templates, it allows for an improved voxel-by-voxel normalization for MRI analysis. To align the cerebellum to the template, it needs to be isolated from the surrounding tissue, a process for which an automated algorithm has been developed. Our methodology outperforms normalization to a whole-brain neonatal template, using either linear or nonlinear transformations. Our algorithm reduces the spatial variability of the infratentorial area, while simultaneously increasing the overlap of the cerebellum. The template and the related software are freely available as part of SUIT v3. 3 SPM toolbox.

YNIMG Journal 2019 Journal Article

Stability of representational geometry across a wide range of fMRI activity levels

Spencer A. Arbuckle
Atsushi Yokoi
J. Andrew Pruszynski
Jörn Diedrichsen

Fine-grained activity patterns, as measured with functional magnetic resonance imaging (fMRI), are thought to reflect underlying neural representations. Multivariate analysis techniques, such as representational similarity analysis (RSA), can be used to test models of brain representation by quantifying the representational geometry (the collection of pair-wise dissimilarities between activity patterns). One important caveat, however, is that non-linearities in the coupling between neural activity and the fMRI signal may lead to significant distortions in the representational geometry estimated from fMRI activity patterns. Here we tested the stability of representational dissimilarity measures in primary sensory-motor (S1 and M1) and early visual regions (V1/V2) across a large range of activation levels. Participants were visually cued with different letters to perform single finger presses with one of the 5 fingers at a rate of 0. 3–2. 6 Hz. For each stimulation frequency, we quantified the difference between the 5 activity patterns in M1, S1, and V1/V2. We found that the representational geometry remained relatively stable, even though the average activity increased over a large dynamic range. These results indicate that the representational geometry of fMRI activity patterns can be reliably assessed, largely independent of the average activity in the region. This has important methodological implications for RSA and other multivariate analysis approaches that use the representational geometry to make inferences about brain representations.

YNIMG Journal 2019 Journal Article

Variational representational similarity analysis

Karl J. Friston
Jörn Diedrichsen
Emma Holmes
Peter Zeidman

This technical note describes a variational or Bayesian implementation of representational similarity analysis (RSA) and pattern component modelling (PCM). It considers RSA and PCM as Bayesian model comparison procedures that assess the evidence for stimulus or condition-specific patterns of responses distributed over voxels or channels. On this view, one can use standard variational inference procedures to quantify the contributions of particular patterns to the data, by evaluating second-order parameters or hyperparameters. Crucially, this allows one to use parametric empirical Bayes (PEB) to infer which patterns are consistent among subjects. At the between-subject level, one can then assess the evidence for different (combinations of) hypotheses about condition-specific effects using Bayesian model comparison. Alternatively, one can select a single hypothesis that best explains the pattern of responses using Bayesian model selection. This note rehearses the technical aspects of within and between-subject RSA using a worked example, as implemented in the Statistical Parametric Mapping (SPM) software. En route, we highlight the connection between univariate and multivariate analyses of neuroimaging data and the sorts of analyses that are possible using component modelling and representational similarity analysis.

YNIMG Journal 2018 Journal Article

Concentric radiofrequency arrays to increase the statistical power of resting-state maps in monkeys

Kyle M. Gilbert
David J. Schaeffer
Peter Zeman
Jörn Diedrichsen
Stefan Everling
Julio C. Martinez-Trujillo
J. Andrew Pruszynski
Ravi S. Menon

The close homology of monkeys and humans has increased the prevalence of non-human-primate models in functional MRI studies of brain connectivity. To improve upon the attainable resolution in functional MRI studies, a commensurate increase in the sensitivity of the radiofrequency receiver coil is required to avoid a reduction in the statistical power of the analysis. Most receive coils are comprised of multiple loops distributed equidistantly over a surface to produce spatially independent sensitivity profiles. A larger number of smaller elements will in turn provide a higher signal-to-noise ratio (SNR) over the same field of view. As the loops become physically smaller, noise originating from the sample is reduced relative to noise originating from the coil. In this coil-noise-dominated regime, coil elements can have overlapping sensitivity profiles, yet still possess only mildly correlated noise. In this manuscript, we demonstrate that inductively decoupled, concentric coil arrays can improve temporal SNR when operating in the coil-noise-dominated regime—in contrast to what is expected for the more ubiquitous sample-noise-dominated array. A small, thin, 7-channel flexible coil is developed and operated in conjunction with an existing whole-head monkey coil. The mean and maximum noise correlation between the two arrays was 5% and 23%, respectively. When the flex coil was placed over the sensorimotor cortex, the temporal SNR improved by up to 2. 3-fold in the peripheral cortex and up to 1. 3-fold at a 2- to 3-cm depth within the brain. When the flex coil was placed over the frontal eye fields, resting-state maps showed substantially elevated sensitivity to correlations in the prefrontal cortex (54%), supplementary eye fields (39%), and anterior cingulate cortex (41%). The concentric-coil topology provided a pragmatic and robust means to significantly improve local temporal SNR and the statistical power of functional connectivity maps.

YNIMG Journal 2018 Journal Article

Pattern component modeling: A flexible approach for understanding the representational structure of brain activity patterns

Jörn Diedrichsen
Atsushi Yokoi
Spencer A. Arbuckle

Representational models specify how complex patterns of neural activity relate to visual stimuli, motor actions, or abstract thoughts. Here we review pattern component modeling (PCM), a practical Bayesian approach for evaluating such models. Similar to encoding models, PCM evaluates the ability of models to predict novel brain activity patterns. In contrast to encoding models, however, the activity of individual voxels across conditions (activity profiles) are not directly fitted. Rather, PCM integrates over all possible activity profiles and computes the marginal likelihood of the data under the activity profile distribution specified by the representational model. By using an analytical expression for the marginal likelihood, PCM allows the fitting of flexible representational models, in which the relative strength and form of the encoded feature spaces can be estimated from the data. We present here a number of different ways in which such flexible representational models can be specified, and how models of different complexity can be compared. We then provide a number of practical examples from our recent work in motor control, ranging from fixed models to more complex non-linear models of brain representations. The code for the fitting and cross-validation of representational models is provided in an open-source software toolbox.

YNICL Journal 2018 Journal Article

Unique degeneration signatures in the cerebellar cortex for spinocerebellar ataxias 2, 3, and 7

Carlos R. Hernandez-Castillo
Maedbh King
Jörn Diedrichsen
Juan Fernandez-Ruiz

Spinocerebellar ataxias (SCAs) are a heterogeneous group of neurodegenerative diseases that selectively affect vulnerable neuronal populations in the cerebellum and other subcortical regions. While previous studies have reported subtype differences in the absolute amount of degeneration in specific regions of interest, they failed to account for two important factors. First, they did not control for overall differences in the severity of the degeneration pattern, and second, they did not fully characterize the spatial pattern of degeneration for each SCA subtype. Here, we provide a systematic characterization of the spatial degeneration patterns for three polyQ SCAs (55 patients, either SCA2, SCA3, or SCA7) while controlling for the severity of the degeneration pattern. After this correction, the cerebellar degeneration pattern can successfully classify between the three different SCA subtypes with high cross-validated accuracy. Specifically, degeneration in SCA3 disproportionally affects motor regions of the cerebellar cortex, which explains the relatively severe motor symptoms observed in this subtype. Our results demonstrate that each of the three studied SCA subtypes has a unique cerebellar degeneration signature, hinting at differences in the disease process. Clinically, these differentiable patterns of cerebellar degeneration can be used to reliably discern subtypes, even at relatively early stages of the disease.

YNIMG Journal 2016 Journal Article

Reliability of dissimilarity measures for multi-voxel pattern analysis

Alexander Walther
Hamed Nili
Naveed Ejaz
Arjen Alink
Nikolaus Kriegeskorte
Jörn Diedrichsen

Representational similarity analysis of activation patterns has become an increasingly important tool for studying brain representations. The dissimilarity between two patterns is commonly quantified by the correlation distance or the accuracy of a linear classifier. However, there are many different ways to measure pattern dissimilarity and little is known about their relative reliability. Here, we compare the reliability of three classes of dissimilarity measure: classification accuracy, Euclidean/Mahalanobis distance, and Pearson correlation distance. Using simulations and four real functional magnetic resonance imaging (fMRI) datasets, we demonstrate that continuous dissimilarity measures are substantially more reliable than the classification accuracy. The difference in reliability can be explained by two characteristics of classifiers: discretization and susceptibility of the discriminant function to shifts of the pattern ensemble between imaging runs. Reliability can be further improved through multivariate noise normalization for all measures. Finally, unlike conventional distance measures, crossvalidated distances provide unbiased estimates of pattern dissimilarity on a ratio scale, thus providing an interpretable zero point. Overall, our results indicate that the crossvalidated Mahalanobis distance is preferable to both the classification accuracy and the correlation distance for characterizing representational geometries.

YNIMG Journal 2013 Journal Article

A multivariate method to determine the dimensionality of neural representation from population activity

Jörn Diedrichsen
Tobias Wiestler
Naveed Ejaz

How do populations of neurons represent a variable of interest? The notion of feature spaces is a useful concept to approach this question: According to this model, the activation patterns across a neuronal population are composed of different pattern components. The strength of each of these components varies with one latent feature, which together are the dimensions along which the population represents the variable. Here we propose a new method to determine the number of feature dimensions that best describes the activation patterns. The method is based on Gaussian linear classifiers that use only the first d most important pattern dimensions. Using cross-validation, we can identify the classifier that best matches the dimensionality of the neuronal representation. We test this method on two datasets of motor cortical activation patterns measured with functional magnetic resonance imaging (fMRI), during (i) simultaneous presses of all fingers of a hand at different force levels and (ii) presses of different individual fingers at a single force level. As expected, the new method shows that the representation of force is low-dimensional; the neural activation for different force levels is scaled versions of each other. In comparison, individual finger presses are represented in a full, four-dimensional feature space. The approach can be used to determine an important characteristic of neuronal population codes without knowing the form of the underlying features. It therefore provides a novel tool in the building of quantitative models of neuronal population activity as measured with fMRI or other approaches.

YNIMG Journal 2011 Journal Article

A comparison of volume-based and surface-based multi-voxel pattern analysis

Nikolaas N. Oosterhof
Tobias Wiestler
Paul E. Downing
Jörn Diedrichsen

For functional magnetic resonance imaging (fMRI), multi-voxel pattern analysis (MVPA) has been shown to be a sensitive method to detect areas that encode certain stimulus dimensions. By moving a searchlight through the volume of the brain, one can continuously map the information content about the experimental conditions of interest to the brain. Traditionally, the searchlight is defined as a volume sphere that does not take into account the anatomy of the cortical surface. Here we present a method that uses a cortical surface reconstruction to guide voxel selection for information mapping. This approach differs in two important aspects from a volume-based searchlight definition. First, it uses only voxels that are classified as grey matter based on an anatomical scan. Second, it uses a surface-based geodesic distance metric to define neighbourhoods of voxels, and does not select voxels across a sulcus. We study here the influence of these two factors onto classification accuracy and onto the spatial specificity of the resulting information map. In our example data set, participants pressed one of four fingers while undergoing fMRI. We used MVPA to identify regions in which local fMRI patterns can successfully discriminate which finger was moved. We show that surface-based information mapping is a more sensitive measure of local information content, and provides better spatial selectivity. This makes surface-based information mapping a useful technique for a data-driven analysis of information representation in the cerebral cortex.

YNIMG Journal 2011 Journal Article

Comparing the similarity and spatial structure of neural representations: A pattern-component model

Jörn Diedrichsen
Gerard R. Ridgway
Karl J. Friston
Tobias Wiestler

In recent years there has been growing interest in multivariate analyses of neuroimaging data, which can be used to detect distributed patterns of activity that encode an experimental factor of interest. In this setting, it has become common practice to study the correlations between patterns to make inferences about the way a brain region represents stimuli or tasks (known as representational similarity analysis). Although it would be of great interest to compare these correlations from different regions, direct comparisons are currently not possible. This is because sample correlations are strongly influenced by voxel-selection, fMRI noise, and nonspecific activation patterns, all of which can differ widely between regions. Here, we present a multivariate modeling framework in which the measured patterns are decomposed into their constituent parts. The model is based on a standard linear mixed model, in which pattern components are considered to be randomly distributed over voxels. The model allows one to estimate the true correlations of the underlying neuronal pattern components, thereby enabling comparisons between different regions or individuals. The pattern estimates also allow us to make inferences about the spatial structure of different response components. Thus, the new model provides a theoretical and analytical framework to study the structure of distributed neural representations.

YNIMG Journal 2009 Journal Article

A probabilistic MR atlas of the human cerebellum

Jörn Diedrichsen
Joshua H. Balsters
Jonathan Flavell
Emma Cussans
Narender Ramnani

The functional organization of the cerebellum is reflected in large part by the unique afferent and efferent connectivity of the individual cerebellar lobules. This functional diversity on a relatively small spatial scale makes accurate localization methods for human functional imaging and anatomical patient-based research indispensable. Here we present a probabilistic atlas of the cerebellar lobules in the anatomical space defined by the MNI152 template. We separately masked the lobules on T1-weighted MRI scans (1 mm isotropic resolution) of 20 healthy young participants (10 male, 10 female, average age 23. 7 yrs). These cerebella were then aligned to the standard or non-linear version of the whole-brain MNI152 template using a number of commonly used normalization algorithms, or to a previously published cerebellum-only template (Diedrichsen, J. , 2006. A spatially unbiased atlas template of the human cerebellum. NeuroImage 33, 127–138.). The resulting average overlap was higher for the cerebellum-only template than for any of the whole-brain normalization methods. The probabilistic maps allow for the valid assignment of functional activations to specific cerebellar lobules, while providing a quantitative measure of the uncertainty of such assignments. Furthermore, maximum probability maps derived from these atlases can be used to define regions of interest (ROIs) in functional neuroimaging and neuroanatomical research. The atlas, made freely available online, is compatible with a number of widely used analysis packages.

YNIMG Journal 2008 Journal Article

Neural substrates of visuomotor learning based on improved feedback control and prediction

Scott T. Grafton
Paul Schmitt
John Van Horn
Jörn Diedrichsen

Motor skills emerge from learning feedforward commands as well as improvements in feedback control. These two components of learning were investigated in a compensatory visuomotor tracking task on a trial-by-trial basis. Between-trial learning was characterized with a state–space model to provide smoothed estimates of feedforward and feedback learning, separable from random fluctuations in motor performance and error. The resultant parameters were correlated with brain activity using magnetic resonance imaging. Learning related to the generation of a feedforward command correlated with activity in dorsal premotor cortex, inferior parietal lobule, supplementary motor area and cingulate motor area, supporting a role of these areas in retrieving and executing a predictive motor command. Modulation of feedback control was associated with activity in bilateral posterior superior parietal lobule as well as right ventral premotor cortex. Performance error correlated with activity in a widespread cortical and subcortical network including bilateral parietal, premotor and rostral anterior cingulate cortex as well as the cerebellar cortex. Finally, trial-by-trial changes of kinematics, as measured by mean absolute hand acceleration, correlated with activity in motor cortex and anterior cerebellum. The results demonstrate that incremental, learning-dependent changes can be modeled on a trial-by-trial basis and neural substrates for feedforward control of novel motor programs are localized to secondary motor areas.

YNIMG Journal 2006 Journal Article

A spatially unbiased atlas template of the human cerebellum

Jörn Diedrichsen

This article presents a new high-resolution atlas template of the human, cerebellum and brainstem, based on the anatomy of 20 young healthy individuals. The atlas is spatially unbiased, i. e. , the location of each structure is equal to the expected location of that structure across individuals in MNI space, a result that is cross-validated with an independent sample of 16 individuals. At the same time, the new template preserves the anatomical detail of cerebellar structures through a nonlinear atlas generation algorithm. In comparison to current whole-brain templates, it allows for an improved voxel-by-voxel normalization for functional MRI and lesion analysis. Alignment to the template requires that the cerebellum and brainstem are isolated from the surrounding tissue, a process for which an automated algorithm has been developed. Compared to normalization to the MNI whole-brain template, the new method strongly improves the alignment of individual fissures, reducing their spatial spread by 60%, and improves the overlap of the deep cerebellar nuclei. Applied to functional MRI data, the new normalization technique leads to a 5–15% increase in peak t values and in the activated volume in the cerebellar cortex for movement vs. rest contrasts. This indicates that the new template significantly improves the overlap of functionally equivalent cerebellar regions across individuals. The template and software are freely available as an SPM-toolbox, which also allows users to relate the new template to the annotated volumetric (Schmahmann, J. D. , Doyon, J. , Toga, A. , Petrides, M. , Evans, A. (2000). MRI atlas of the human cerebellum. San Diego: Academic Press) and surface-based (Van Essen, D. C. (2002a) Surface-based atlases of cerebellar cortex in the human, macaque, and mouse. Ann. N. Y. Acad. Sci. 978: 468–479.) atlas of one individual, the “colin27”-brain.

YNIMG Journal 2005 Journal Article

Detecting and adjusting for artifacts in fMRI time series data

Jörn Diedrichsen
Reza Shadmehr

We present a new method to detect and adjust for noise and artifacts in functional MRI time series data. We note that the assumption of stationary variance, which is central to the theoretical treatment of fMRI time series data, is often violated in practice. Sporadic events such as eye, mouth, or arm movements can increase noise in a spatially global pattern throughout an image, leading to a non-stationary noise process. We derive a restricted maximum likelihood (ReML) algorithm that estimates the variance of the noise for each image in the time series. These variance parameters are then used to obtain a weighted least squares estimate of the regression parameters of a linear model. We apply this approach to a typical fMRI experiment with a block design and show that the noise estimates strongly vary across different images and that our method detects and appropriately weights images that are affected by artifacts. Furthermore, we show that the noise process has a global spatial distribution and that the variance increase is multiplicative rather than additive. The new algorithm results in significantly increased sensitivity in the ability to detect regions of activation. The new method may be particularly useful for studies that involve special populations (e. g. , children or elderly) where sporadic, artifact-generating events are more likely.