Limits...
Cortical network architecture for context processing in primate brain.

Chao ZC, Nagasaka Y, Fujii N - Elife (2015)

Bottom Line: We extracted five context-related network structures including a bottom-up network during encoding and, seconds later, cue-dependent retrieval of the same network with the opposite top-down connectivity.These findings show that context is represented in the cortical network as distributed communication structures with dynamic information flows.This study provides a general methodology for recording and analyzing cortical network neuronal communication during cognition.

View Article: PubMed Central - PubMed

Affiliation: Laboratory for Adaptive Intelligence, RIKEN Brain Science Institute, Wako-shi, Japan.

ABSTRACT
Context is information linked to a situation that can guide behavior. In the brain, context is encoded by sensory processing and can later be retrieved from memory. How context is communicated within the cortical network in sensory and mnemonic forms is unknown due to the lack of methods for high-resolution, brain-wide neuronal recording and analysis. Here, we report the comprehensive architecture of a cortical network for context processing. Using hemisphere-wide, high-density electrocorticography, we measured large-scale neuronal activity from monkeys observing videos of agents interacting in situations with different contexts. We extracted five context-related network structures including a bottom-up network during encoding and, seconds later, cue-dependent retrieval of the same network with the opposite top-down connectivity. These findings show that context is represented in the cortical network as distributed communication structures with dynamic information flows. This study provides a general methodology for recording and analyzing cortical network neuronal communication during cognition.

No MeSH data available.


PARAFAC revealed five dominant structures in the 3D tensor.A core consistency diagnostic was used to evaluate how well the tensor can be represented by different numbers of structures. During deconvolving the tensor into different numbers of structures, core consistencies were measured by two methods: DTLD/GRAM and random values (see the ‘Materials and methods’). For the random values method, 100 core consistency values were measured and their means and standard deviations are shown. In both methods, a sharp decrease in consistency was found when the number of structures increased from 5 to 6, indicating that five structures yielded the optimal fit.DOI:http://dx.doi.org/10.7554/eLife.06121.016
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4584448&req=5

fig3s2: PARAFAC revealed five dominant structures in the 3D tensor.A core consistency diagnostic was used to evaluate how well the tensor can be represented by different numbers of structures. During deconvolving the tensor into different numbers of structures, core consistencies were measured by two methods: DTLD/GRAM and random values (see the ‘Materials and methods’). For the random values method, 100 core consistency values were measured and their means and standard deviations are shown. In both methods, a sharp decrease in consistency was found when the number of structures increased from 5 to 6, indicating that five structures yielded the optimal fit.DOI:http://dx.doi.org/10.7554/eLife.06121.016

Mentions: (A) Event-related causalities (ERCs) between cortical areas. Example ERCs for a connection (IC 8 to IC 14, the corresponding cortical areas shown on the top) in two scenarios (CmRf and CwRf in C+) from Subject 1 are shown. Each ERC represents the spectro-temporal dynamics of causality evoked by a scenario, calculated as the logarithmic ratio between the direct directed transfer function (dDTF) and corresponding baseline values (baseline: gray bar), and measured in decibel (dB). Black vertical lines represent task events explained in Figure 2. (B) ∆ERCs, or the significant differences in ERCs between the two trial types (CmRf − CwRf) (αFDR = 0.05, false discovery rate correction) are shown. The results were either 0 (no significant difference), +1 (significantly greater), or −1 (significantly weaker). (C) 3D tensor of ∆ERCs. The data for the entire study were organized in three dimensions: dynamics (top), function (middle), and anatomy (bottom). Top: ∆ERCs shown in B describe the dynamics of difference in causality of a connection between two trial types, presented as a vector in 3D space (illustrated as a bar, where each segment represents a ∆ERC value). Middle: For the same connection, ∆ERCs from other comparisons were pooled to describe the functional dynamics of the connection (illustrated as a plate). Bottom: Functional dynamics from all connections were pooled to summarize the functional network dynamics in a subject (illustrated as a block). The data from all subjects were further combined to assess common functional network dynamics across subjects. (D) Parallel factor analysis (PARAFAC) extracted five dominant structures from the 3D tensor with consistency (>80%, also see Figure 3—figure supplement 2). Each structure represented a unique pattern of network function, dynamics, and anatomy (e.g., Func. 1, Dyn. 1, and Anat. 1 for Structure 1).


Cortical network architecture for context processing in primate brain.

Chao ZC, Nagasaka Y, Fujii N - Elife (2015)

PARAFAC revealed five dominant structures in the 3D tensor.A core consistency diagnostic was used to evaluate how well the tensor can be represented by different numbers of structures. During deconvolving the tensor into different numbers of structures, core consistencies were measured by two methods: DTLD/GRAM and random values (see the ‘Materials and methods’). For the random values method, 100 core consistency values were measured and their means and standard deviations are shown. In both methods, a sharp decrease in consistency was found when the number of structures increased from 5 to 6, indicating that five structures yielded the optimal fit.DOI:http://dx.doi.org/10.7554/eLife.06121.016
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4584448&req=5

fig3s2: PARAFAC revealed five dominant structures in the 3D tensor.A core consistency diagnostic was used to evaluate how well the tensor can be represented by different numbers of structures. During deconvolving the tensor into different numbers of structures, core consistencies were measured by two methods: DTLD/GRAM and random values (see the ‘Materials and methods’). For the random values method, 100 core consistency values were measured and their means and standard deviations are shown. In both methods, a sharp decrease in consistency was found when the number of structures increased from 5 to 6, indicating that five structures yielded the optimal fit.DOI:http://dx.doi.org/10.7554/eLife.06121.016
Mentions: (A) Event-related causalities (ERCs) between cortical areas. Example ERCs for a connection (IC 8 to IC 14, the corresponding cortical areas shown on the top) in two scenarios (CmRf and CwRf in C+) from Subject 1 are shown. Each ERC represents the spectro-temporal dynamics of causality evoked by a scenario, calculated as the logarithmic ratio between the direct directed transfer function (dDTF) and corresponding baseline values (baseline: gray bar), and measured in decibel (dB). Black vertical lines represent task events explained in Figure 2. (B) ∆ERCs, or the significant differences in ERCs between the two trial types (CmRf − CwRf) (αFDR = 0.05, false discovery rate correction) are shown. The results were either 0 (no significant difference), +1 (significantly greater), or −1 (significantly weaker). (C) 3D tensor of ∆ERCs. The data for the entire study were organized in three dimensions: dynamics (top), function (middle), and anatomy (bottom). Top: ∆ERCs shown in B describe the dynamics of difference in causality of a connection between two trial types, presented as a vector in 3D space (illustrated as a bar, where each segment represents a ∆ERC value). Middle: For the same connection, ∆ERCs from other comparisons were pooled to describe the functional dynamics of the connection (illustrated as a plate). Bottom: Functional dynamics from all connections were pooled to summarize the functional network dynamics in a subject (illustrated as a block). The data from all subjects were further combined to assess common functional network dynamics across subjects. (D) Parallel factor analysis (PARAFAC) extracted five dominant structures from the 3D tensor with consistency (>80%, also see Figure 3—figure supplement 2). Each structure represented a unique pattern of network function, dynamics, and anatomy (e.g., Func. 1, Dyn. 1, and Anat. 1 for Structure 1).

Bottom Line: We extracted five context-related network structures including a bottom-up network during encoding and, seconds later, cue-dependent retrieval of the same network with the opposite top-down connectivity.These findings show that context is represented in the cortical network as distributed communication structures with dynamic information flows.This study provides a general methodology for recording and analyzing cortical network neuronal communication during cognition.

View Article: PubMed Central - PubMed

Affiliation: Laboratory for Adaptive Intelligence, RIKEN Brain Science Institute, Wako-shi, Japan.

ABSTRACT
Context is information linked to a situation that can guide behavior. In the brain, context is encoded by sensory processing and can later be retrieved from memory. How context is communicated within the cortical network in sensory and mnemonic forms is unknown due to the lack of methods for high-resolution, brain-wide neuronal recording and analysis. Here, we report the comprehensive architecture of a cortical network for context processing. Using hemisphere-wide, high-density electrocorticography, we measured large-scale neuronal activity from monkeys observing videos of agents interacting in situations with different contexts. We extracted five context-related network structures including a bottom-up network during encoding and, seconds later, cue-dependent retrieval of the same network with the opposite top-down connectivity. These findings show that context is represented in the cortical network as distributed communication structures with dynamic information flows. This study provides a general methodology for recording and analyzing cortical network neuronal communication during cognition.

No MeSH data available.