Encoding and decoding in fMRI

Abstract

Over the past decade fMRI researchers have developed increasingly sensitive techniques for analyzing the information represented in BOLD activity. The most popular of these techniques is linear classification, a simple technique for decoding information about experimental stimuli or tasks from patterns of activity across an array of voxels. A more recent development is the voxel-based encoding model, which describes the information about the stimulus or task that is represented in the activity of single voxels. Encoding and decoding are complementary operations: encoding uses stimuli to predict activity while decoding uses activity to predict information about stimuli. However, in practice these two operations are often confused, and their respective strengths and weaknesses have not been made clear. Here we use the concept of a linearizing feature space to clarify the relationship between encoding and decoding. We show that encoding and decoding operations can both be used to investigate some of the most common questions about how information is represented in the brain. However, focusing on encoding models offers two important advantages over decoding. First, an encoding model can in principle provide a complete functional description of a region of interest, while a decoding model can provide only a partial description. Second, while it is straightforward to derive an optimal decoding model from an encoding model it is much more difficult to derive an encoding model from a decoding model. We propose a systematic modeling approach that begins by estimating an encoding model for every voxel in a scan and ends by using the estimated encoding models to perform decoding.

Commonality of neural representations of sentences across languages: Predicting brain activation during Portuguese sentence comprehension using an English-based model of brain function

“…In a previous investigation of English monolingual speakers, both directions yielded similar accuracies in the mapping between the activation patterns and the NPSFs of sentences (Wang, Cherkassky, & Just, submitted). Neural activation prediction models, such as the one used in the current study, have several scientific advantages, at least in the initial stage of model development (Naselaris, Kay, Nishimoto, & Gallant, 2011). The most salient advantage is that it can yield a functional characterization of specific brain regions that can be compared to regional characterizations in other studies.…”

A statistical framework for neuroimaging data analysis based on mutual information estimated via a gaussian copula

“…This improved multivariate performance allows estimation of quantities such as conditional mutual information (CMI) [Ince et al, 2012], directed information (DI; also called transfer entropy [TE]) [Ince et al, 2015; Massey, 1990; Schreiber, 2000] as well as measures quantifying pairwise interactions between variables [Chicharro, 2014; Panzeri et al, 2008]. We believe these higher order information theoretic quantities have the potential to provide transformative new interpretations of neuroimaging data, by providing a unified framework for analyses based on the information content of neural signals [Kriegeskorte and Bandettini, 2007; Naselaris et al, 2011; Schyns et al, 2009]. The methods we present enable study of the representation, processing, and communication in the brain of multiple features of the external world [Ince et al, 2015].…”

Fast brain decoding with random sampling and random projections

“…It has become a central tool in neuroimage [1]. In particular, linear estimators can highlight the brain maps that lead to the identification of cognitive labels [2] [3]. Yet, to date, decoding is still orders of magnitude slower than standard analysis.…”