Dataset Information

What do across-subject analyses really tell us about neural coding?

ABSTRACT: A key challenge in human neuroscience is to gain information about patterns of neural activity using indirect measures. Multivariate pattern analysis methods testing for generalization of information across subjects have been used to support inferences regarding neural coding. One critical assumption of an important class of such methods is that anatomical normalization is suited to align spatially-structured neural patterns across individual brains. We asked whether anatomical normalization is suited for this purpose. If not, what sources of information are such across-subject cross-validated analyses likely to reveal? To investigate these questions, we implemented two-layered feedforward randomly-connected networks. A key feature of these simulations was a gain-field with a spatial structure shared across networks. To investigate whether total-signal imbalances across conditions-e.g. differences in overall activity-affect the observed pattern of results, we manipulated the energy-profile of images conforming to a pre-specified correlation structure. To investigate whether the level of granularity of the data also influences results, we manipulated the density of connections between network layers. Simulations showed that anatomical normalization is unsuited to align neural representations. Pattern similarity-relationships were explained by the observed total-signal imbalances across conditions. Further, we observed that deceptively complex representational structures emerge from arbitrary analysis choices, such as whether the data are mean-subtracted during preprocessing. These simulations also led to testable predictions regarding the distribution of low-level features in images used in recent fMRI studies that relied on leave-one-subject-out pattern analyses. Image analyses broadly confirmed these predictions. Finally, hyperalignment emerged as a principled alternative to test across-subject generalization of spatially-structured information. We illustrate cases in which hyperalignment proved successful, as well as cases in which it only partially recovered the latent correlation structure in the pattern of responses. Our results highlight the need for robust, high-resolution measurements from individual subjects. We also offer a way forward for across-subject analyses. We suggest ways to inform hyperalignment results with estimates of the strength of the signal associated with each condition. Such information can usefully constrain ensuing inferences regarding latent representational structures as well as population tuning dimensions.

SUBMITTER: Ramirez FM

PROVIDER: S-EPMC8596303 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundRecent work using large datasets (>500 records per subject) has demonstrated seemingly high levels of step-to-step variation in peak plantar pressure within human individuals during walking. One intuitive consequence of this variation is that smaller sample sizes (e.g., 10 steps per subject) may be quantitatively and qualitatively inaccurate and fail to capture the variance in plantar pressure of individuals seen in larger data sets. However, this remains quantitatively unexplored reflecting a lack of detailed investigation of intra-subject sample size effects in plantar pressure analysis.MethodsHere we explore the sensitivity of various plantar pressure metrics to intra-subject sample size (number of steps per subject) using a random subsampling analysis. We randomly and incrementally subsample large data sets (>500 steps per subject) to compare variability in three metric types at sample sizes of 5-400 records: (1) overall whole-record mean and maximum pressure; (2) single-pixel values from five locations across the foot; and (3) the sum of pixel-level variability (measured by mean square error, MSE) from the whole plantar surface.ResultsOur results indicate that the central tendency of whole-record mean and maximum pressure within and across subjects show only minor sensitivity to sample size >200 steps. However, <200 steps, and particularly <50 steps, the range of overall mean and maximum pressure values yielded by our subsampling analysis increased considerably resulting in potential qualitative error in analyses of pressure changes with speed within-subjects and in comparisons of relative pressure magnitudes across subjects at a given speed. Our analysis revealed considerable variability in the absolute and relative response of the single pixel centroids of five regions to random subsampling. As the number of steps analysed decreased, the absolute value ranges were highest in the areas of highest pressure (medial forefoot and hallux), while the largest relative changes were seen in areas of lower pressure (the midfoot). Our pixel-level measure of variability by MSE across the whole-foot was highly sensitive to our manipulation of sample size, such that the range in MSE was exponentially larger in smaller subsamples. Random subsampling showed that the range in pixel-level MSE only came within 5% of the overall sample size in subsamples of >400 steps. The range in pixel-level MSE at low subsamples (<50) was 25-75% higher than that of the full datasets of >500 pressure records per subject. Overall, therefore, we demonstrate a high probability that the very small sample sizes (n < 20 records), which are routinely used in human and animal studies, capture a relatively low proportion of variance evident in larger plantar pressure data set, and thus may not accurately reflect the true population mean.

Project description:Studies investigating neural information processing often implicitly ask both, which processing strategy out of several alternatives is used and how this strategy is implemented in neural dynamics. A prime example are studies on predictive coding. These often ask whether confirmed predictions about inputs or predictions errors between internal predictions and inputs are passed on in a hierarchical neural system-while at the same time looking for the neural correlates of coding for errors and predictions. If we do not know exactly what a neural system predicts at any given moment, this results in a circular analysis-as has been criticized correctly. To circumvent such circular analysis, we propose to express information processing strategies (such as predictive coding) by local information-theoretic quantities, such that they can be estimated directly from neural data. We demonstrate our approach by investigating two opposing accounts of predictive coding-like processing strategies, where we quantify the building blocks of predictive coding, namely predictability of inputs and transfer of information, by local active information storage and local transfer entropy. We define testable hypotheses on the relationship of both quantities, allowing us to identify which of the assumed strategies was used. We demonstrate our approach on spiking data collected from the retinogeniculate synapse of the cat (N = 16). Applying our local information dynamics framework, we are able to show that the synapse codes for predictable rather than surprising input. To support our findings, we estimate quantities applied in the partial information decomposition framework, which allow to differentiate whether the transferred information is primarily bottom-up sensory input or information transferred conditionally on the current state of the synapse. Supporting our local information-theoretic results, we find that the synapse preferentially transfers bottom-up information.

Dataset Information

What do across-subject analyses really tell us about neural coding?

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets