The confounding effect of response amplitude on MVPA performance measures

Smith, Andrew T.; Kosillo, Polina; Williams, AE

doi:10.1016/j.neuroimage.2010.05.079

Cited by 39 publications

(26 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For contexts in which there are voxel-level interactions with experimental variables (e.g., some voxels in an ROI do not activate), as would likely occur in many ROIs, increasing the between-subject variability in mean activation may also increase voxel-level variability and the between-subject variability of the voxel-level variances (the τ matrix). Thus, in cases of voxel-by-condition interactions, group-level MVPA may be affected by between-subject variability in activation (see e.g., LaRocque et al 2013; Smith et al 2011; Tong et al 2012). Because of the infinite ways that voxel-by-condition interactions can manifest, it is not possible to give precise a priori predictions for how they impact MVPA in all contexts.…”

Section: Resultsmentioning

confidence: 99%

What do differences between multi-voxel and univariate analysis mean? How subject-, voxel-, and trial-level variance impact fMRI analysis

et al. 2014

View full text Add to dashboard Cite

Multi-voxel pattern analysis (MVPA) has led to major changes in how fMRI data are analyzed and interpreted. Many studies now report both MVPA results and results from standard univariate voxel-wise analysis, often with the goal of drawing different conclusions from each. Because MVPA results can be sensitive to latent multidimensional representations and processes whereas univariate voxel-wise analysis cannot, one conclusion that is often drawn when MVPA and univariate results differ is that the activation patterns underlying MVPA results contain a multidimensional code. In the current study, we conducted simulations to formally test this assumption. Our findings reveal that MVPA tests are sensitive to the magnitude of voxel-level variability in the effect of a condition within subjects, even when the same linear relationship is coded in all voxels. We also find that MVPA is insensitive to subject-level variability in mean activation across an ROI, which is the primary variance component of interest in many standard univariate tests. Together, these results illustrate that differences between MVPA and univariate tests do not afford conclusions about the nature or dimensionality of the neural code. Instead, targeted tests of the informational content and/or dimensionality of activation patterns are critical for drawing strong conclusions about the representational codes that are indicated by significant MVPA results.

show abstract

Section: Resultsmentioning

confidence: 99%

What do differences between multi-voxel and univariate analysis mean? How subject-, voxel-, and trial-level variance impact fMRI analysis

et al. 2014

View full text Add to dashboard Cite

show abstract

“…Because baseline differences in correlations across ROIs are hard to interpret (higher overall correlation in one region compared to another could be due to many factors, including the distance of a region from the coils, amount of vascularization, or region size; see Smith, Kosillo, & Williams, 2011), we used difference scores (within condition correlation - across condition correlation) as the dependent variable. We found a main effect of ROI (F(4,28)=3.03, p=0.03, partial η 2 =0.27), suggesting that the regions contain varying amounts of information about source modality in their neural pattern.…”

Section: Resultsmentioning

confidence: 99%

Thinking about seeing: Perceptual sources of knowledge are encoded in the theory of mind brain regions of sighted and blind adults

2014

View full text Add to dashboard Cite

Blind people's inferences about how other people see provide a window into fundamental questions about the human capacity to think about one another's thoughts. By working with blind individuals, we can ask both what kinds of representations people form about others’ minds, and how much these representations depend on the observer having had similar mental states themselves. Thinking about others’ mental states depends on a specific group of brain regions, including the right temporo-parietal junction (RTPJ). We investigated the representations of others’ mental states in these brain regions, using multivoxel pattern analyses (MVPA). We found that, first, in the RTPJ of sighted adults, the pattern of neural response distinguished the source of the mental state (did the protagonist see or hear something?) but not the valence (did the protagonist feel good or bad?). Second, these neural representations were preserved in congenitally blind adults. These results suggest that the temporo-parietal junction contains explicit, abstract representations of features of others’ mental states, including the perceptual source. The persistence of these representations in congenitally blind adults, who have no first-person experience with sight, provides evidence that these representations emerge even in the absence of first-person perceptual experiences.

show abstract

“…This would seem to indicate that a simple lack of features is not the cause of this difference in performance. Other factors that could lead to reduced performance even if equivalent (or greater) selectivity exists in MT+/V5 are the arrangement or distribution of direction selective columns leading to very small voxel-wise biases (Kamitani and Tong, 2006;Malonek et al, 1994), or differences in the amplitude of BOLD response in different parts of the cortex (Smith et al, 2010). Due to the wide range of possible factors involved in determining classification accuracy in a given area, there are many difficulties in comparing classification performance across cortical areas and inferring differences in selectivity from classification results.…”

Section: Discussionmentioning

confidence: 99%

Contribution of large scale biases in decoding of direction-of-motion from high-resolution fMRI data in human early visual cortex

et al. 2012

View full text Add to dashboard Cite

Previous studies have demonstrated that the perceived direction of motion of a visual stimulus can be decoded from the pattern of functional magnetic resonance imaging (fMRI) responses in occipital cortex using multivariate analysis methods (Kamitani and Tong, 2006). One possible mechanism for this is a difference in the sampling of direction selective cortical columns between voxels, implying that information at a level smaller than the voxel size might be accessible with fMRI. Alternatively, multivariate analysis methods might be driven by the organization of neurons into clusters or even orderly maps at a much larger scale. To assess the possible sources of the direction selectivity observed in fMRI data, we tested how classification accuracy varied across different visual areas and subsets of voxels for classification of motion-direction. To enable high spatial resolution functional MRI measurements (1.5mm isotropic voxels), data were collected at 7T. To test whether information about the direction of motion is represented at the scale of retinotopic maps, we looked at classification performance after combining data across different voxels within visual areas (V1-3 and MT+/V5) before training the multivariate classifier. A recent study has shown that orientation biases in V1 are both necessary and sufficient to explain classification of stimulus orientation (Freeman et al., 2011). Here, we combined voxels with similar visual field preference as determined in separate retinotopy measurements and observed that classification accuracy was preserved when averaging in this 'retinotopically restricted' way, compared to random averaging of voxels. This insensitivity to averaging of voxels (with similar visual angle preference) across substantial distances in cortical space suggests that there are large-scale biases at the level of retinotopic maps underlying our ability to classify direction of motion.

show abstract

The confounding effect of response amplitude on MVPA performance measures

Cited by 39 publications

References 27 publications

What do differences between multi-voxel and univariate analysis mean? How subject-, voxel-, and trial-level variance impact fMRI analysis

What do differences between multi-voxel and univariate analysis mean? How subject-, voxel-, and trial-level variance impact fMRI analysis

Thinking about seeing: Perceptual sources of knowledge are encoded in the theory of mind brain regions of sighted and blind adults

Contribution of large scale biases in decoding of direction-of-motion from high-resolution fMRI data in human early visual cortex

Contact Info

Product

Resources

About