Multivoxel Pattern of Blood Oxygen Level Dependent Activity can be sensitive to stimulus specific fine scale responses

Abstract

<jats:label>1.</jats:label><jats:title>Abstract</jats:title><jats:p>At ultra-high field, fMRI voxels can span the sub-millimeter range, allowing the recording of blood oxygenation level dependent (BOLD) responses at the level of fundamental units of neural computation, such as cortical columns and layers. This sub-millimeter resolution, however, is only nominal in nature as a number of factors limit the spatial acuity of functional voxels. Multivoxel Pattern Analysis (MVPA) may provide a means to detect information at finer spatial scales that may otherwise not be visible at the single voxel level due to limitations in sensitivity and specificity. Here, we evaluate the spatial scale of stimuli specific BOLD responses in multivoxel patterns exploited by linear Support Vector Machine, Linear Discriminant Analysis and Naïve Bayesian classifiers across cortical depths in V1. To this end, we artificially misaligned the testing relative to the training portion of the data in increasing spatial steps, then investigated the breakdown of the classifiers’ performances. A one voxel shift led to a significant decrease in decoding accuracy (p&lt;.05) across all cortical depths, indicating that stimulus specific responses in a multivoxel pattern of BOLD activity exploited by multivariate decoders can be as precise as the nominal resolution of single voxels (here .8 mm isotropic). Our results further indicate that large draining vessels, prominently residing in proximity of the pial surface, do not, in this case, hinder the ability of MVPA to exploit fine scale patterns of BOLD signals. We argue that tailored analytical approaches can help overcoming limitations in high-resolution fMRI and permit studying the mesoscale organization of the human brain with higher sensitivities.</jats:p>

Recurrent Processing Drives Perceptual Plasticity

“…These corrections resulted in similar BOLD magnitude and multi-voxel pattern classification accuracy before training across layers (Figure 3E), suggesting that our approach for correcting vasculature-related effects controlled substantially for the superficial bias. In particular, consistent with previous studies [34] showing reduced superficial bias for MVPA classification measures, we did not observe any significant differences in MVPA accuracy between trained and untrained orientations before training (e.g., orientation 3 location 3 layer interaction: F(2,24) = 0.891, p = 0.423; main effect of layer: F(2,24) = 0.287, p = 0.753). Thus, it is unlikely that our MVPA results after vasculature correction were significantly confounded by the superficial bias.…”

Recurrent processing drives experience-dependent plasticity for perceptual decisions

“…These corrections resulted in similar BOLD magnitude and multi-voxel pattern classification accuracy before training across layers ( Figure 3E), suggesting that our approach for correcting vasculature-related effects controlled substantially for the superficial bias. In particular, consistent with previous studies 28 showing reduced superficial bias for MVPA classification measures, we did not observe any significant differences in MVPA accuracy between trained and untrained orientations before training (e.g. orientation × location × layer interaction: F(2,24) = 0.891, p = 0.423; main effect of layer: F(2,24) = 0.287, p = 0.753).…”

Statistical Power or More Precise Insights into Neuro-Temporal Dynamics? Assessing the Benefits of Rapid Temporal Sampling in fMRI

“…(Dowdle et al, 2021)) and multivariate (e.g. (Kriegeskorte and Bandettini, 2007; Vizioli et al, 2020a)) parametric tests using subjects or runs as independent observations are in fact routinely used. Nonparametric statistical approaches, such as bootstrap confidence interval that make fewer assumptions about the data distribution, thus avoiding many of these concerns, are also a valid alternative.…”

“…To infer statistically significant differences across time courses, tMVPA builds on a sliding window approach (extensively tested on real and synthetic data elsewhere – Vizioli et al, 2018) that allows for the precise identification of the temporal window of an effect and whether this encompasses only a few time points or is sustained over a larger time window. Multivariate analyses methods have been shown to have increased sensitivity for the analyses of spatial maps (Kriegeskorte and Bandettini, 2007; Vizioli et al, 2020a) in fMRI. There is evidence that this is also the case for the temporal domain, with prior work finding earlier identification of statistically significant task differences on both real (Ramon et al, 2015) and synthetically simulated data (Vizioli et al, 2018).…”