Reliability of dissimilarity measures for multi-voxel pattern analysis

Decoding Dynamic Brain Patterns from Evoked Responses: A Tutorial on Multivariate Pattern Analysis Applied to Time Series Neuroimaging Data

Multivariate pattern analysis (MVPA) or brain decoding methods have become standard practice in analyzing fMRI data. Although decoding methods have been extensively applied in brain–computer interfaces, these methods have only recently been applied to time series neuroimaging data such as MEG and EEG to address experimental questions in cognitive neuroscience. In a tutorial style review, we describe a broad set of options to inform future time series decoding studies from a cognitive neuroscience perspective. Using example MEG data, we illustrate the effects that different options in the decoding analysis pipeline can have on experimental results where the aim is to “decode” different perceptual stimuli or cognitive states over time from dynamic brain activation patterns. We show that decisions made at both preprocessing (e.g., dimensionality reduction, subsampling, trial averaging) and decoding (e.g., classifier selection, cross-validation design) stages of the analysis can significantly affect the results. In addition to standard decoding, we describe extensions to MVPA for time-varying neuroimaging data including representational similarity analysis, temporal generalization, and the interpretation of classifier weight maps. Finally, we outline important caveats in the design and interpretation of time series decoding experiments.

Representational models: A common framework for understanding encoding, pattern-component, and representational-similarity analysis

Representational models specify how activity patterns in populations of neurons (or, more generally, in multivariate brain-activity measurements) relate to sensory stimuli, motor responses, or cognitive processes. In an experimental context, representational models can be defined as hypotheses about the distribution of activity profiles across experimental conditions. Currently, three different methods are being used to test such hypotheses: encoding analysis, pattern component modeling (PCM), and representational similarity analysis (RSA). Here we develop a common mathematical framework for understanding the relationship of these three methods, which share one core commonality: all three evaluate the second moment of the distribution of activity profiles, which determines the representational geometry, and thus how well any feature can be decoded from population activity. Using simulated data for three different experimental designs, we compare the power of the methods to adjudicate between competing representational models. PCM implements a likelihood-ratio test and therefore provides the most powerful test if its assumptions hold. However, the other two approaches-when conducted appropriately-can perform similarly. In encoding analysis, the linear model needs to be appropriately regularized, which effectively imposes a prior on the activity profiles. With such a prior, an encoding model specifies a well-defined distribution of activity profiles. In RSA, the unequal variances and statistical dependencies of the dissimilarity estimates need to be taken into account to reach near-optimal power in inference. The three methods render different aspects of the information explicit (e.g. single-response tuning in encoding analysis and population-response representational dissimilarity in RSA) and have specific advantages in terms of computational demands, ease of use, and extensibility. The three methods are properly construed as complementary components of a single data-analytical toolkit for understanding neural representations on the basis of multivariate brain-activity data.

Transformed Neural Pattern Reinstatement during Episodic Memory Retrieval

Contemporary models of episodic memory posit that remembering involves the reenactment of encoding processes. Although encoding-retrieval similarity has been consistently reported and linked to memory success, the nature of neural pattern reinstatement is poorly understood. Using high-resolution fMRI on human subjects, our results obtained clear evidence for item-specific pattern reinstatement in the frontoparietal cortex, even when the encoding-retrieval pairs shared no perceptual similarity. No item-specific pattern reinstatement was found in the ventral visual cortex. Importantly, the brain regions and voxels carrying item-specific representation differed significantly between encoding and retrieval, and the item specificity for encoding-retrieval similarity was smaller than that for encoding or retrieval, suggesting different nature of representations between encoding and retrieval. Moreover, cross-region representational similarity analysis suggests that the encoded representation in the ventral visual cortex was reinstated in the frontoparietal cortex during retrieval. Together, these results suggest that, in addition to reinstatement of the originally encoded pattern in the brain regions that perform encoding processes, retrieval may also involve the reinstatement of a transformed representation of the encoded information. These results emphasize the constructive nature of memory retrieval that helps to serve important adaptive functions.SIGNIFICANCE STATEMENT Episodic memory enables humans to vividly reexperience past events, yet how this is achieved at the neural level is barely understood. A long-standing hypothesis posits that memory retrieval involves the faithful reinstatement of encoding-related activity. We tested this hypothesis by comparing the neural representations during encoding and retrieval. We found strong pattern reinstatement in the frontoparietal cortex, but not in the ventral visual cortex, that represents visual details. Critically, even within the same brain regions, the nature of representation during retrieval was qualitatively different from that during encoding. These results suggest that memory retrieval is not a faithful replay of past event but rather involves additional constructive processes to serve adaptive functions.

Avoiding illusory effects in representational similarity analysis: What (not) to do with the diagonal

Representational similarity analysis (RSA) is an important part of the methodological toolkit in neuroimaging research. The focus of the approach is the construction of representational dissimilarity matrices (RDMs), which provide a single format for making comparisons between different neural data types, computational models, and behavior. We highlight two issues for the construction and comparison of RDMs. First, the diagonal values of RDMs, which should reflect within condition reliability of neural patterns, are typically not estimated in RSA. However, without such an estimate, one lacks a measure of the reliability of an RDM as a whole. Thus, when carrying out RSA, one should calculate the diagonal values of RDMs and not take them for granted. Second, although diagonal values of a correlation matrix can be used to estimate the reliability of neural patterns, these values must nonetheless be excluded when comparing RDMs. Via a simple simulation we show that inclusion of these values can generate convincing looking, but entirely illusory, correlations between independent and entirely unrelated data sets. Both of these points are further illustrated by a critical discussion of Coggan et al. (2016), who investigated the extent to which category-selectivity in the ventral temporal cortex can be accounted for by low-level image properties of visual object stimuli. We observe that their results may depend on the improper inclusion of diagonal values in their analysis.

Measuring the representational space of music with fMRI: a case study with Sting

Functional brain imaging has revealed much about the neuroanatomical substrates of higher cognition, including music, language, learning, and memory. The technique lends itself to studying of groups of individuals. In contrast, the nature of expert performance is typically studied through the examination of exceptional individuals using behavioral case studies and retrospective biography. Here, we combined fMRI and the study of an individual who is a world-class expert musician and composer in order to better understand the neural underpinnings of his music perception and cognition, in particular, his mental representations for music. We used state of the art multivoxel pattern analysis (MVPA) and representational dissimilarity analysis (RDA) in a fixed set of brain regions to test three exploratory hypotheses with the musician Sting: (1) Composing would recruit neutral structures that are both unique and distinguishable from other creative acts, such as composing prose or visual art; (2) listening and imagining music would recruit similar neural regions, indicating that musical memory shares anatomical substrates with music listening; (3) the MVPA and RDA results would help us to map the representational space for music, revealing which musical pieces and genres are perceived to be similar in the musician's mental models for music. Our hypotheses were confirmed. The act of composing, and even of imagining elements of the composed piece separately, such as melody and rhythm, activated a similar cluster of brain regions, and were distinct from prose and visual art. Listened and imagined music showed high similarity, and in addition, notable similarity/dissimilarity patterns emerged among the various pieces used as stimuli: Muzak and Top 100/Pop songs were far from all other musical styles in Mahalanobis distance (Euclidean representational space), whereas jazz, R&B, tango and rock were comparatively close. Closer inspection revealed principaled explanations for the similarity clusters found, based on key, tempo, motif, and orchestration.

Strong Evidence for Pattern Separation in Human Dentate Gyrus

The hippocampus is proposed to be critical in distinguishing between similar experiences by performing pattern separation computations that create orthogonalized representations for related episodes. Previous neuroimaging studies have provided indirect evidence that the dentate gyrus (DG) and CA3 hippocampal subregions support pattern separation by inferring the nature of underlying representations from the observation of novelty signals. Here, we use ultra-high-resolution fMRI at 7 T and multivariate pattern analysis to provide compelling evidence that the DG subregion specifically sustains representations of similar scenes that are less overlapping than in other hippocampal (e.g., CA3) and medial temporal lobe regions (e.g., entorhinal cortex). Further, we provide evidence that novelty signals within the DG are stimulus specific rather than generic in nature. Our study, in providing a mechanistic link between novelty signals and the underlying representations, constitutes the first demonstration that the human DG performs pattern separation.

SIGNIFICANCE STATEMENT

A fundamental property of an episodic memory system is the ability to minimize interference between similar episodes. The dentate gyrus (DG) subregion of the hippocampus is widely viewed to realize this function through a computation referred to as pattern separation, which creates distinct nonoverlapping neural codes for individual events. Here, we leveraged 7 T fMRI to test the hypothesis that this region supports pattern separation. Our results demonstrate that the DG supports representations of similar scenes that are less overlapping than those in neighboring subregions. The current study therefore is the first to offer compelling evidence that the human DG supports pattern separation by obtaining critical empirical data at the representational level: the level where this computation is defined.