custom script code in matlab r2016b  (MathWorks Inc)

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Place field formation from visual scenes enabled estimating the animal’s position. ( A ) Schematic of the simulated animal. ( B ) Examples of the animal’s perspective images showing local and distal cues. ( A ) and ( B ) were adapted from Google DeepMind Lab open source software ( https://www.deepmind.com/open-source/deepmind-lab ) . ( C ) Two typical example trajectories (S1 and S2) during random exploration of a 2D square arena. ( D ) Place field centers for S1 and S2. ( E , F ) Example place fields for S1 and S2 at different locations with different field sizes. ( G ) Example of absolute decoding error from the place cell population along the navigation. Plots indicated that there was a tendency for location error to decrease along the simulation for both trajectories. Blue: average decoding error in 100-time bins across the simulated time; error bars represent s.e.m. All panels except for ( A ) and ( B ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All panels except for ( A ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ).

Techniques: Software

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Place field emergence stabilized after the animal's early exploration. ( A ) Examples of entropy distributions across the simulated time. Insets represent the same data between 0 and 500 simulated time (a.u.). ( B ) Entropy statistics against the increase in the number of place fields during navigation. H is entropy, S is spatial entropy, and Z is information density. ( C ) Changes in information I, entropy H , number of place fields n , and complexity ratio R in the population of place cells across time. ( D ) Representational complexity stabilized after early exploration. The blue dashed line represents the initial time when a stable number of place fields was established. All panels were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All panels except for ( A ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ).

Techniques:

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Grid patterns were maintained across the animal’s trajectory but depended on the trajectory features. ( A , B ) Spikes across trajectories, spike density and rate map plots from dorsal (right side) to ventral (left side) grid network configurations. Note that for each column, the plots indicate the same neuron (5 different neurons in total separately for ( A ) and ( B ) groups of plots). Regular triangular tessellation can be observed for both trajectories. Yellow represents maximum activity in the spike density and rate map plots, and red indicates spike clusters of the same neurons as shown in the spike density plots along the trajectory. Over each rate map, the gridness score (G) and the squared gridness (GS) are shown. ( C ) Variance across modules over place estimates through time for both trajectories. ( D ) The Euclidean distance between estimates for each module from dorsal (module 1) to ventral (module 5) modules. The inset numbers represent mean ± sem for the associated data for all the panels. All panels were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All panels except for ( A ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ).

Techniques: Activity Assay

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Place field input to grid cells enabled error reduction for path integration. ( A ) Schematics of the place-grid model (further detailed in Suppl. Fig. ). Scenes obtained from Google DeepMind Lab open source software ( https://www.deepmind.com/open-source/deepmind-lab) . ( B ) Example of the “activity packet” (bump) for the grid cell network considering the dorsal to ventral modules individually. The squared dynamical space is represented by the 20 × 20 grid cells. ( C ) Example of the estimated trajectories made by the grid cell modules during an S1 movement across the arena for 1000-time steps at the top row with place field’s input. To evidence that the error accumulation deviates the estimated position from the actual trajectory, the bottom row represents the prediction of each module without place fields’ input for a short trajectory only (100-time steps; red trace) from the starting position (see Supp. Fig. for the whole trajectory). Plots indicate a better prediction of the actual animal’s trajectory when place field inputs are given to grid cells. ( D ) The same trajectories as previously shown for the ventral module were magnified for comparison purposes. The green dot represents the starting position, the magenta dot indicates the predicted trajectory ending, and the pink one is the actual ending position. Asterix represents the places where the activities of grid cells across modules could enable place fields. ( E ) An example of place field centers that emerged during the short trajectory is depicted in panned ( D ). ( F , I ) The Euclidean distance error measure compared the estimated and the actual trajectory for dorsal (module 1) to ventral (module 5) modules. Plots indicate that the distance was higher in ( I ) when place field information to the grid cell network was absent. ( G , J ) The measure of the variance across estimates showed a similar observation. ( H , K ) The estimated error was lower when place field input was provided ( H ) compared to the absence of input ( K ). The inset numbers represent mean ± sem for the associated data for all the panels. All panels except for ( A ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All panels except for ( A ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ).

Techniques: Software, Activity Assay, Comparison

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: A lower variance and estimated error across grid modules were observed with closer place fields to the animal trajectory. ( A ) A brief trajectory example and the estimated error from the grid cell network when a small (PF = 20) or high (PF = 200) number of place field centers were spread randomly across the arena. Blue lines represent actual trajectories, and red ones indicate trajectory estimates. ( B , C ) An example of Euclidean distance error measure shows that ( B ) when place field input was provided and place fields were set to PF = 20, the error was higher than ( C ) when the number of place fields was set to PF = 200. Module 1 refers to dorsal and module 5 to ventral. ( D , E ) The measure of variance in fewer place fields ( D ) vs a larger number of place fields ( E ), indicated a higher variance for the former case. ( F , G ) Example of the estimated error when ( F ) PF = 20 vs ( G ) when PF = 200 across modules. The inset equations represent mean ± sem for the associated data for all the panels. All panels were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All panels except for ( A ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ).

Techniques:

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Place field formation from visual scenes enabled estimating the animal’s position. ( A ) Schematic of the simulated animal. ( B ) Examples of the animal’s perspective images showing local and distal cues. ( A ) and ( B ) were adapted from Google DeepMind Lab open source software ( https://www.deepmind.com/open-source/deepmind-lab ) . ( C ) Two typical example trajectories (S1 and S2) during random exploration of a 2D square arena. ( D ) Place field centers for S1 and S2. ( E , F ) Example place fields for S1 and S2 at different locations with different field sizes. ( G ) Example of absolute decoding error from the place cell population along the navigation. Plots indicated that there was a tendency for location error to decrease along the simulation for both trajectories. Blue: average decoding error in 100-time bins across the simulated time; error bars represent s.e.m. All panels except for ( A ) and ( B ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All data analyses and scripting reported in this manuscript were made through custom code using Matlab R2016b.

Techniques: Software

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Place field emergence stabilized after the animal's early exploration. ( A ) Examples of entropy distributions across the simulated time. Insets represent the same data between 0 and 500 simulated time (a.u.). ( B ) Entropy statistics against the increase in the number of place fields during navigation. H is entropy, S is spatial entropy, and Z is information density. ( C ) Changes in information I, entropy H , number of place fields n , and complexity ratio R in the population of place cells across time. ( D ) Representational complexity stabilized after early exploration. The blue dashed line represents the initial time when a stable number of place fields was established. All panels were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All data analyses and scripting reported in this manuscript were made through custom code using Matlab R2016b.

Techniques:

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Grid patterns were maintained across the animal’s trajectory but depended on the trajectory features. ( A , B ) Spikes across trajectories, spike density and rate map plots from dorsal (right side) to ventral (left side) grid network configurations. Note that for each column, the plots indicate the same neuron (5 different neurons in total separately for ( A ) and ( B ) groups of plots). Regular triangular tessellation can be observed for both trajectories. Yellow represents maximum activity in the spike density and rate map plots, and red indicates spike clusters of the same neurons as shown in the spike density plots along the trajectory. Over each rate map, the gridness score (G) and the squared gridness (GS) are shown. ( C ) Variance across modules over place estimates through time for both trajectories. ( D ) The Euclidean distance between estimates for each module from dorsal (module 1) to ventral (module 5) modules. The inset numbers represent mean ± sem for the associated data for all the panels. All panels were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All data analyses and scripting reported in this manuscript were made through custom code using Matlab R2016b.

Techniques: Activity Assay

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: Place field input to grid cells enabled error reduction for path integration. ( A ) Schematics of the place-grid model (further detailed in Suppl. Fig. ). Scenes obtained from Google DeepMind Lab open source software ( https://www.deepmind.com/open-source/deepmind-lab) . ( B ) Example of the “activity packet” (bump) for the grid cell network considering the dorsal to ventral modules individually. The squared dynamical space is represented by the 20 × 20 grid cells. ( C ) Example of the estimated trajectories made by the grid cell modules during an S1 movement across the arena for 1000-time steps at the top row with place field’s input. To evidence that the error accumulation deviates the estimated position from the actual trajectory, the bottom row represents the prediction of each module without place fields’ input for a short trajectory only (100-time steps; red trace) from the starting position (see Supp. Fig. for the whole trajectory). Plots indicate a better prediction of the actual animal’s trajectory when place field inputs are given to grid cells. ( D ) The same trajectories as previously shown for the ventral module were magnified for comparison purposes. The green dot represents the starting position, the magenta dot indicates the predicted trajectory ending, and the pink one is the actual ending position. Asterix represents the places where the activities of grid cells across modules could enable place fields. ( E ) An example of place field centers that emerged during the short trajectory is depicted in panned ( D ). ( F , I ) The Euclidean distance error measure compared the estimated and the actual trajectory for dorsal (module 1) to ventral (module 5) modules. Plots indicate that the distance was higher in ( I ) when place field information to the grid cell network was absent. ( G , J ) The measure of the variance across estimates showed a similar observation. ( H , K ) The estimated error was lower when place field input was provided ( H ) compared to the absence of input ( K ). The inset numbers represent mean ± sem for the associated data for all the panels. All panels except for ( A ) were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All data analyses and scripting reported in this manuscript were made through custom code using Matlab R2016b.

Techniques: Software, Activity Assay, Comparison

Journal: Scientific Reports

Article Title: Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model

doi: 10.1038/s41598-022-25863-2

Figure Lengend Snippet: A lower variance and estimated error across grid modules were observed with closer place fields to the animal trajectory. ( A ) A brief trajectory example and the estimated error from the grid cell network when a small (PF = 20) or high (PF = 200) number of place field centers were spread randomly across the arena. Blue lines represent actual trajectories, and red ones indicate trajectory estimates. ( B , C ) An example of Euclidean distance error measure shows that ( B ) when place field input was provided and place fields were set to PF = 20, the error was higher than ( C ) when the number of place fields was set to PF = 200. Module 1 refers to dorsal and module 5 to ventral. ( D , E ) The measure of variance in fewer place fields ( D ) vs a larger number of place fields ( E ), indicated a higher variance for the former case. ( F , G ) Example of the estimated error when ( F ) PF = 20 vs ( G ) when PF = 200 across modules. The inset equations represent mean ± sem for the associated data for all the panels. All panels were made using custom code in Matlab R2016b ( https://www.mathworks.com/ ). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License ( https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: All data analyses and scripting reported in this manuscript were made through custom code using Matlab R2016b.

Techniques: