Emergent reliability in sensory cortical coding and inter-area communication

Ebrahimi, Sadegh; Lecoq, Jérôme; Rumyantsev, Oleg; Tasci, Tugce; Zhang, Yanping; Irimia, Cristina; Li, Jane; Ganguli, Surya; Schnitzer, Mark J.

doi:10.1038/s41586-022-04724-y

Cited by 49 publications

(43 citation statements)

References 65 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent electrophysiology studies have found that neural representations of natural stimuli in mouse visual cortex are not stable, but rather change, or drift, at timescales ranging from minutes to weeks [1][2][3][4] . The discovery of this phenomenon, termed 'representational drift', was made possible by technology enabling large-scale chronic recordings of populations of neurons in rodent cortex over extended periods of time.…”

Section: Main Textmentioning

confidence: 99%

Representations in human primary visual cortex drift over time

Roth

Merriam

2022

Preprint

View full text Add to dashboard Cite

Primary sensory regions are believed to instantiate stable neural representations, yet a number of recent rodent studies suggest instead that representations drift over time. We analyzed a massive fMRI dataset using an image-computable encoding model and found systematic changes in model fits that exhibited cumulative drift over many months. Convergent analyses pinpoint changes in neural responsivity as the source of the drift, while population-level representational dissimilarities between visual stimuli were unchanged, suggesting that downstream cortical areas may read-out a stable representation, even as representations within V1 drift.

show abstract

Section: Main Textmentioning

confidence: 99%

Representations in human primary visual cortex drift over time

Roth

Merriam

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Therefore, demixing covariability classes may be a crucial step when considering large-scale multi-region recordings [IBL et al, 2022, Wagner et al, 2019, Ebrahimi et al, 2022, Ahrens et al, 2012.…”

Section: Discussionmentioning

confidence: 99%

“…Our results support this hypothesis, and further show that these different classes can be demixed by sliceTCA. Therefore, demixing covariability classes may be a crucial step when considering large-scale multi-region recordings [IBL et al, 2022, Wagner et al, 2019, Ebrahimi et al, 2022, Ahrens et al, 2012].…”

Section: Discussionmentioning

confidence: 99%

Disentangling Mixed Classes of Covariability in Large-Scale Neural Data

Pellegrino

Stein

Cayco-Gajic

2023

Preprint

View full text Add to dashboard Cite

Recent work has argued that large-scale neural recordings are often well described by low-dimensional ‘latent’ dynamics identified using dimensionality reduction. However, the view that task-relevant variability is shared across neurons misses other types of structure underlying behavior, including stereotyped neural sequences or slowly evolving latent spaces. To address this, we introduce a new framework that simultaneously accounts for variability that is shared across neurons, trials, or time. To identify and demix these covariability classes, we develop a new unsupervised dimensionality reduction method for neural data tensors called sliceTCA. In three example datasets, including motor cortical dynamics during a classic reaching task and recent multi-region recordings from the International Brain Laboratory, we show that sliceTCA can capture more task-relevant structure in neural data using fewer components than traditional methods. Overall, our theoretical framework extends the classic view of low-dimensional population activity by incorporating additional classes of latent variables capturing higher-dimensional structure.

show abstract

“…For example, an unexpectedly large fraction of the variance in the activity of neurons in “sensory” areas such as the primary visual cortex can be explained as a function of carefully monitored movement variables [84, 85]. As another example, Ebrahimi et al [86] recorded neuronal activity throughout spatially distributed regions in the mouse cortex during visual decision making. In their data, a considerable portion of neurons in anatomically defined “ motor ” or “ somatosensory ” regions coded for stimulus identity, even as early as stimulus onset time.…”

Section: Discussionmentioning

confidence: 99%

Multimodal measures of spontaneous brain activity reveals both common and divergent patterns of cortical functional organization

Vafaii

Mandino

Desrosiers-Grégoire

et al. 2022

Preprint

View full text Add to dashboard Cite

Understanding the organization of large-scale brain networks remains a central problem in neuroscience. Work in both humans and rodents shows that the brain can be decomposed in terms of several networks (e.g., "default network"). Whereas the bulk of what we know is based on the blood-oxygenation level-dependent (BOLD) signal, the relationship between BOLD and neuronal activity is complex, which poses several challenges to interpreting networks revealed by this technique. To resolve these challenges, here we employed wide-field Ca2+imaging simultaneously recorded with fMRI-BOLD in a highly-sampled group of GCaMP6f-expressing mice. This allowed us to determine the relationship between networks discovered by BOLD and Ca2+signals both at the group level and in individual animals. Network partition used a mixed-membership stochastic blockmodel algorithm, which allows networks to be overlapping, such that a brain region may be assigned to multiple networks with varying strengths. Here, we tested the hypothesis that functional networks of the mouse brain are organized in an overlapping manner for both BOLD and Ca2+data. Our findings demonstrate that BOLD large-scale networks can be detected via Ca2+signals. In addition: (1) Overlapping networks were reliably estimated at the group level via random-effects statistical analysis. (2) Functional organization was generally similar for BOLD and Ca2+; e.g., with seven networks, cosine similarity was very high in five instances (values between 0.82-0.90; max possible value of 1), moderate in one case (0.73) and different in one case (0.26). (3) We discovered overlapping network organization in both data modalities, which was quantified in multiple ways; e.g., the proportion of regions that belonged to multiple networks was 75\% for BOLD and 60\% for Ca2+. (4) The large-scale functional organization of the mouse cortex as determined by Ca2+signals is considerably more similar to that with BOLD when both signal modalities are considered in the low temporal frequency range (0.01 to 0.5 Hz). (5) Although many similarities were observed between data modalities, finer characterization uncovered key differences related to functional hubs believed to play important roles in the integration and/or segregation of signals. In particular, the spatial distribution of membership diversity (i.e., the extent to which a region affiliates with multiple networks), differed for the two types of signals. In conclusion, Ca2+mesoscale signals revealed that the mouse cortex is functionally organized in terms of large-scale networks in a manner that reflects many of the properties observed using BOLD. Despite many similarities, important differences were also uncovered, suggesting that mesoscale Ca2+has a strong potential to uncover additional properties of the large-scale organization of the mouse brain.

show abstract

Emergent reliability in sensory cortical coding and inter-area communication

Cited by 49 publications

References 65 publications

Representations in human primary visual cortex drift over time

Representations in human primary visual cortex drift over time

Disentangling Mixed Classes of Covariability in Large-Scale Neural Data

Multimodal measures of spontaneous brain activity reveals both common and divergent patterns of cortical functional organization

Contact Info

Product

Resources

About