Biology subjects

Soch, J.

Publications and source records attributed to Soch, J..

At least 19 recordsLinked to original sources

bioRxiv2025

Multivariate Bayesian Inversion for Classification and Regression

We propose the statistical modelling approach to supervised learning (i.e. predicting labels from features) as an alternative to algorithmic machine learning (ML). The approach is demonstrated by employing a multivariate general linear model (MGLM) describing the effects of labels on features, possibly accounting for covariates of no interest, in combination with prior distributions on the model parameters. ML "training" is translated into estimating the MGLM parameters via Bayesian inference and ML "testing" or application is translated into Bayesian model comparison - a reciprocal relationship we refer to as multivariate Bayesian inversion (MBI). We devise MBI algorithms for the standard cases of supervised learning, discrete classification and continuous regression, derive novel classification rules and regression predictions, and use practical examples (simulated and real data) to illustrate benefits of the statistical modelling approach: interpretability, incorporation of prior knowledge, probabilistic predictions. We close by discussing further advantages, disadvantages and the future potential of MBI.

neuroscience↗

bioRxiv2025

High-field fMRI at 7 Tesla reveals topographic responses tuned to number in the developing human brain

In the adult brain, hemodynamic responses to visually presented numerosities reveal receptive field-like tuning curves in topographically organized maps across association cortices. It is currently unknown whether such tuned topographic responses to numerosities can also be detected in the developing brain. Here we conducted a 7 Tesla fMRI experiment in which we presented a large set of visual dot displays to 11-12-year-old children and adults. We found that developing hemodynamic responses indeed revealed logarithmic Gaussian tuning to quantitative information. Tuning models explained comparable amounts of variance in children and adults. In most subjects, six bilateral cortical maps consistently exhibited topographic responses to numerosities. The present study goes beyond previous work by uncovering a population code for quantity detection in individual developing human brains. Our work lays a foundation for a model-based neuroimaging approach to individual cognitive differences in the context of developmental dyscalculia and mathematical giftedness.

neuroscience↗

bioRxiv2024

Neural dynamics of visual working memory representation during sensory distraction

Recent studies have provided evidence for the concurrent encoding of sensory percepts and visual working memory contents (VWM) across visual areas; however, it has remained unclear how these two types of representations are concurrently present. Here, we reanalyzed an open-access fMRI dataset where participants memorized a sensory stimulus while simultaneously being presented with sensory distractors. First, we found that the VWM code in several visual regions did not fully generalize between different time points, suggesting a dynamic code. A more detailed analysis revealed that this was due to shifts in coding spaces across time. Second, we collapsed neural signals across time to assess the degree of interference between VWM contents and sensory distractors, specifically by testing the alignment of their encoding spaces. We find that VWM and feature-matching sensory distractors are encoded in coding spaces that do not fully overlap, but the separation decreases when distractors negatively impact behavioral performance in recalling the target. Together, these results indicate a role of dynamic coding and temporally stable coding spaces in helping multiplex perception and VWM within visual areas.

neuroscience↗

bioRxiv2023

Searchlight-based trial-wise fMRI decoding in the presence of trial-by-trial correlations

In multivariate pattern analysis (MVPA) for functional magnetic resonance imaging (fMRI) signals, trial-wise response amplitudes are sometimes estimated using a general linear model (GLM) with one onset regressor for each trial. When using rapid event-related designs with trials closely spaced in time, those estimates can be highly correlated due to the temporally smoothed shape of the hemodynamic response function. In previous work (Soch, J., Allefeld, C., & Haynes, J.-D. (2020). Inverse transformed encoding models - a solution to the problem of correlated trial-by-trial parameter estimates in fMRI decoding. NeuroImage, 209, 116449, 1-19. https://doi.org/10.1016/j.neuroimage.2019.116449), we have proposed inverse transformed encoding modelling (ITEM), a principled approach for trial-wise decoding from fMRI signals in the presence of trial-by-trial correlations. Here, we (i) perform simulation studies addressing its performance for multivariate signals and (ii) present searchlight-based ITEM analysis - which allows to predict a variable of interest from the vicinity of each voxel in the brain. We empirically validate the approach by confirming a priori plausible hypotheses about the well-understood visual system.

neuroscience↗

bioRxiv2023

Age-related differences in fMRI subsequent memory effects are directly linked to local grey matter volume differences

Episodic memory performance declines with increasing age, and older adults typically show reduced activation of inferior temporo-parietal cortices in functional magnetic resonance imaging (fMRI) studies of episodic memory formation. Given the age-related cortical volume loss, it is conceivable that age-related reduction of memory-related fMRI activity may be partially attributable to reduced grey matter volume (GMV). We performed a voxel-wise multimodal neuroimaging analysis of fMRI correlates of successful memory encoding, using regional GMV as covariate. In a large cohort of healthy adults (106 young, 111 older), older adults showed reduced GMV across the entire neocortex and reduced encoding-related activation of inferior temporal and parieto-occipital cortices compared to young adults. Importantly, these reduced fMRI activations during successful encoding in older adults could in part be attributed to lower regional GMV. Our results highlight the importance of controlling for structural MRI differences in fMRI studies in older adults but also demonstrate that age-related differences in memory-related fMRI activity cannot be attributed to structural variability alone. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=117 SRC="FIGDIR/small/529668v2_ufig1.gif" ALT="Figure 1"> View larger version (37K): org.highwire.dtl.DTLVardef@b43e79org.highwire.dtl.DTLVardef@1717c58org.highwire.dtl.DTLVardef@19aa13borg.highwire.dtl.DTLVardef@cf2f77_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2023

Working memory and imagery in early visual cortex

It has been suggested that visual images are memorized across brief periods of time by vividly imagining them as if they still were there. In line with this, the contents of both working memory and visual imagery are known to be encoded already in early visual cortex. If these signals in early visual areas were indeed to reflect a combined imagery and memory code, one would predict them to be weaker for individuals with reduced visual imagery vividness. Here, we systematically investigated this question in two groups of participants. Strong and weak imagers were asked to remember images across brief delay periods. We were able to reliably reconstruct the memorized stimuli from early visual cortex during the delay. Importantly, in contrast to the prediction, the quality of reconstruction was equally accurate for both strong and weak imagers. The decodable information also closely reflected behavioral precision in both groups, suggesting it could contribute to behavioral performance, even in the extreme case of completely aphantasic individuals. Our data thus suggest that working memory signals in early visual cortex can be present even in the (near) absence of phenomenal imagery.

neuroscience↗

bioRxiv2023

Encoding of continuous perceptual choices in human early visual cortex

Research on the neural mechanisms of perceptual decision-making has typically focused on simple categorical choices, say between two alternative motion directions. Studies on such discrete alternatives have often suggested that choices are encoded either in a motor-based or in an abstract, categorical format in regions beyond sensory cortex. However, many sensory features are graded rather than discrete, raising the question how choices are encoded when they span the full sensory continuum. Here we assessed this using motion stimuli that could vary anywhere between 0{degrees} and 360{degrees}. We employed a combination of neuroimaging and encoding models based on Gaussian Process Regression to assess how either stimuli or choices were encoded in brain responses. We found that single-voxel tuning patterns could be used to reconstruct the trial-by-trial physical direction of motion as well as the participants continuous choices. Importantly, these continuous choice signals were primarily observed in early visual areas. The tuning properties in this region generalized between choice encoding and stimulus encoding, even for reports that reflected pure guessing. We found only little information related to the decision outcome in regions beyond visual cortex, such as parietal cortex, possibly because our task did not involve differential motor preparation. This could suggest that decisions for continuous stimuli take can place already in sensory brain regions, potentially using similar mechanisms to the sensory recruitment in visual working memory.

neuroscience↗

bioRxiv2022

Inhibitory temporo-parietal effective connectivity is associated with explicit memory performance in older adults

Successful explicit memory encoding is associated with inferior temporal activations and medial parietal deactivations, which are attenuated in aging. Here we used Dynamic Causal Modeling (DCM) of functional magnetic resonance imaging data to elucidate effective connectivity patterns between hippocampus, parahippocampal place area (PPA) and precuneus during encoding of novel visual scenes. In 117 young adults, DCM revealed pronounced activating input from the PPA to the hippocampus and inhibitory connectivity from the PPA to the precuneus during novelty processing, with both being enhanced during successful encoding. This pattern could be replicated in two cohorts (N = 141 and 148) of young and older adults. In both cohorts, older adults selectively exhibited attenuated inhibitory PPA-precuneus connectivity, which correlated negatively with memory performance. Our results provide insight into network dynamics underlying explicit memory encoding and suggest that age-related differences in memory-related network activity are, at least partly, attributable to altered temporo-parietal neocortical connectivity.

neuroscience↗

bioRxiv2022

Openness to experience is associated with neural and performance measures of memory in older adults

Age-related decline in episodic memory performance is a well-replicated finding across numerous studies. Recent studies focusing on aging and individual differences found that the Big Five personality trait Openness to Experience is associated with better episodic memory performance in older adults, but the associated neural mechanisms are largely unclear. Here we investigated the relationship between Openness and memory network function in a sample of 352 participants (143 older adults, 50-80 years; 209 young adults, 18-35 years). Participants underwent functional magnetic resonance imaging (fMRI) during a visual memory encoding task. Functional memory brain-network integrity was assessed using the SAME scores (similarity of activations during memory encoding), which reflect the similarity of a participants memory network activity compared to prototypical fMRI activity patterns of young adults. Openness was assessed using the NEO Five Factor Inventory (NEO-FFI). Older vs. young adults showed lower memory performance and higher deviation of fMRI activity patterns (i.e., lower SAME scores). Specifically in older adults, high Openness was associated with better memory performance, and mediation analysis showed that this relationship was partially mediated by higher SAME scores. Our results suggest that trait Openness may constitute a protective factor in cognitive aging by better preservation of the brains memory network.

neuroscience↗

bioRxiv2022

Indirect decoding of behavior from brain signals via reconstruction of experimental conditions

The data acquired during a functional magnetic resonance imaging (fMRI) experiment usually comprise experimental conditions, brain signals and behavioral responses. This reflects the underlying causal flow where the experimental conditions evoke brain responses that in turn result in behavior. In multivariate analyses, a common approach is to focus on the second step of that chain and to decode behavioral responses from brain signals. However, a different approach would be to first reconstruct the experimental conditions from brain signals and in turn use these to predict behavior. While this indirect approach would go against the causal chain of events, it might work better under certain circumstances, especially when the experimental conditions evoke much stronger measurable brain signals than the overt motor behavior. Here we tested this question directly by assessing the various mappings between conditions, brain signals and behavior in an open dataset. We found that the path of first decoding experimental conditions works surprisingly well, even though in our example data set, it is still outperformed by directly decoding the behavior from brain responses.

neuroscience↗

bioRxiv2022

Structural and functional MRI data differentially predict chronological age and behavioral memory performance

Human cognitive abilities decline with increasing chronological age, with decreased explicit memory performance being most strongly affected. However, some older adults show "successful aging", that is, relatively preserved cognitive ability in old age. One explanation for this could be higher brain structural integrity in these individuals. Alternatively, the brain might recruit existing resources more efficiently or employ compensatory cognitive strategies. Here, we approached this question by testing multiple candidate variables from structural and functional neuroimaging for their ability to predict chronological age and memory performance, respectively. Prediction was performed using support vector machine (SVM) classification and regression across and within two samples of young (N = 106) and older (N = 153) adults. The candidate variables were (i) behavioral response frequencies in an episodic memory test, (ii) recently described fMRI scores reflecting preservation of functional memory networks, (iii) whole-brain fMRI contrasts for novelty processing and subsequent memory, (iv) resting-state fMRI maps quantifying voxel-wise signal fluctuation and (v) gray matter volume estimated from structural MR images. While age group could be reliably decoded from all variables, chronological age within young and older subjects was best predicted from gray matter volume. In contrast, memory performance was best predicted from task-based fMRI contrasts and particularly single-value fMRI scores, whereas gray matter volume has no predictive power with respect to memory performance in healthy adults. Our results suggest that superior memory performance in healthy older adults is better explained by efficient recruitment of memory networks rather than by preserved brain structure.

neuroscience↗

bioRxiv2022

Summary statistics of memory-related fMRI activity reflect dissociable neuropsychological and anatomical signatures of neurocognitive aging

Memory-related functional magnetic resonance imaging (fMRI) activations show age-related differences across multiple brain regions that can be captured in summary statistics like single-value scores. Recently, we described two single-value scores reflecting deviations from prototypical whole-brain fMRI activity of young adults during novelty processing and successful encoding. Here, we investigate the brain-behavior associations of these scores with age-related neurocognitive changes in 153 healthy older adults. All scores were associated with episodic recall performance. The memory network scores, but not the novelty network scores, additionally correlated with medial temporal gray matter and a composite measure comprising pro-active inhibition, episodic memory, tonic alertness, flexibility, and working memory. Our results thus suggest that novelty-network-based fMRI scores show high brain-behavior associations with episodic memory and that encoding-network-based fMRI scores additionally capture individual differences in global cognitive function. More generally, our results suggest that single-value scores of memory-related fMRI provide a comprehensive measure of individual differences in network dysfunction that may contribute to age-related cognitive decline.

neuroscience↗

bioRxiv2021

A comprehensive score reflecting memory-related fMRI activations and deactivations as potential biomarker for neurocognitive aging

Older adults and particularly those at risk for developing dementia typically show a decline in episodic memory performance, which has been associated with altered memory network activity detectable via functional magnetic resonance imaging (fMRI). To quantify the degree of these alterations, a score has been developed as a putative imaging biomarker for successful aging in memory for older adults (Functional Activity Deviations during Encoding, FADE; Duzel et al., 2011). Here, we introduce and validate a more comprehensive version of the FADE score, termed FADE-SAME (Similarity of Activations during Memory Encoding), which differs from the original FADE score by considering not only activations but also deactivations in fMRI contrasts of stimulus novelty and successful encoding, and by taking into account the variance of young adults activations. We computed both scores for novelty and subsequent memory contrasts in a cohort of 217 healthy adults, including 106 young and 111 older participants, as well as a replication cohort of 117 young subjects. We further tested the stability and generalizability of both scores by controlling for different MR scanners and gender, as well as by using different data sets of young adults as reference samples. Both scores showed robust age-group-related differences for the subsequent memory contrast, and the FADE-SAME score additionally exhibited age-group-related differences for the novelty contrast. Furthermore, both scores correlate with behavioral measures of cognitive aging, namely memory performance. Taken together, our results suggest that single-value scores of memory-related fMRI responses may constitute promising biomarkers for quantifying neurocognitive aging.

neuroscience↗

bioRxiv2020

Distributional Transformation improves Decoding Accuracy when Predicting Chronological Age from Structural MRI

When predicting a certain subject-level variable (e.g. age in years) from measured biological data (e.g. structural MRI scans), the decoding algorithm does not always preserve the distribution of the variable to predict. In such a situation, distributional transformation (DT), i.e. mapping the predicted values to the variables distribution in the training data, might improve decoding accuracy. Here, we tested the potential of DT within the 2019 Predictive Analytics Competition (PAC) which aimed at predicting chronological age of adult human subjects from structural MRI data. In a low-dimensional setting, i.e. with less features than observations, we applied multiple linear regression, support vector regression and deep neural networks for out-of-sample prediction of subject age. We found that (i) when the number of features is low, no method outperforms linear regression; and (ii) except when using deep regression, distributional transformation increases decoding performance, reducing the mean absolute error (MAE) by about half a year. We conclude that DT can be advantageous when predicting variables that are non-controlled, but have an underlying distribution in healthy or diseased populations.

neuroscience↗

bioRxiv2020

Suppress me if you can: neurofeedback of the readiness potential

Voluntary movements are usually preceded by a slow, negative-going brain signal over motor areas, the so-called readiness potential (RP). To date, the exact nature and causal role of the RP in movement preparation have remained heavily debated. One important open question is whether people can exert conscious control over their RP, for example by learning to suppress it. If people were able to initiate spontaneous movements without eliciting an RP, this would challenge the idea that the RP is a necessary stage of the causal chain leading up to a voluntary movement. We tested the ability of participants to control the magnitude of their RP in a neurofeedback experiment. Participants performed self-initiated movements and after every movement they were provided with immediate feedback about the magnitude of their RP. They were asked to find a mental strategy to perform voluntary movements such that the RPs were as small as possible. We found no evidence that participants were able to to willfully modulate or suppress their RPs while still eliciting voluntary movements. This suggests that the RP might be an involuntary component of voluntary action over which people cannot exert conscious control.

neuroscience↗

bioRxiv2020

Bayesian model selection favors parametric over categorical models of the fMRI subsequent memory effect in young and elderly adults

Subsequent memory paradigms allow to identify neural correlates of successful encoding by separating brain responses as a function of memory performance during later retrieval. In functional magnetic resonance imaging (fMRI), the paradigm typically elicits activations of medial temporal lobe, prefrontal and parietal cortical structures in young, healthy participants. This categorical approach is, however, limited by insufficient memory performance in older and particularly memory-impaired individuals. A parametric modulation of encoding-related activations with memory confidence could overcome this limitation. Here, we applied cross-validated Bayesian model selection (cvBMS) for first-level fMRI models to a visual subsequent memory paradigm in young (18-35 years) and older (51-80 years) adults. Nested cvBMS revealed that parametric models, especially with non-linear transformations of memory confidence ratings, outperformed categorical models in explaining the fMRI signal variance during encoding. We thereby provide a framework for improving the modeling of encoding-related activations and for applying subsequent memory paradigms to memory-impaired individuals.

neuroscience↗

bioRxiv2019

Inverse Transformed Encoding Models - a solution to the problem of correlated trial-by-trial parameter estimates in fMRI decoding

Techniques of multivariate pattern analysis (MVPA) can be used to decode the discrete experimental condition or a continuous modulator variable from measured brain activity during a particular trial. In functional magnetic resonance imaging (fMRI), trial-wise response amplitudes are sometimes estimated from the measured signal using a general linear model (GLM) with one onset regressor for each trial. When using rapid event-related designs with trials closely spaced in time, those estimates are highly variable and serially correlated due to the temporally extended shape of the hemodynamic response function (HRF). Here, we describe inverse transformed encoding modelling (ITEM), a principled approach of accounting for those serial correlations and decoding from the resulting estimates, at low computational cost and with no loss in statistical power. We use simulated data to show that ITEM outperforms the current standard approach in terms of decoding accuracy and analyze empirical data to demonstrate that ITEM is capable of visual reconstruction from fMRI signals.

neuroscience↗

bioRxiv2017

MACS - a new SPM toolbox for model assessment, comparison and selection

BackgroundIn cognitive neuroscience, functional magnetic resonance imaging (fMRI) data are widely analyzed using general linear models (GLMs). However, model quality of GLMs for fMRI is rarely assessed, in part due to the lack of formal measures for statistical model inference.\n\nNew MethodWe introduce a new SPM toolbox for model assessment, comparison and selection (MACS) of GLMs applied to fMRI data. MACS includes classical, information-theoretic and Bayesian methods of model assessment previously applied to GLMs for fMRI as well as recent methodological developments of model selection and model averaging in fMRI data analysis.\n\nResultsThe toolbox - which is freely available from GitHub - directly builds on the Statistical Parametric Mapping (SPM) software package and is easy-to-use, general-purpose, modular, readable and extendable. We validate the toolbox by reproducing model selection and model averaging results from earlier publications.\n\nComparison with Existing MethodsA previous toolbox for model diagnosis in fMRI has been discontinued and other approaches to model comparison between GLMs have not been translated into reusable computational resources in the past.\n\nConclusionsIncreased attention on model quality will lead to lower false-positive rates in cognitive neuroscience and increased application of the MACS toolbox will increase the reproducibility of GLM analyses and is likely to increase the replicability of fMRI studies.

neuroscience↗