This dataset contains raw electrophysiological and behavioral data from 3 experimental cohorts comprising 17 mice.

File Organization

The experimental data are organized in subdirectories corresponding to individual recording sessions. File names typically include the recording session identifier (e.g., jp486-18032023-0108) as the basename. The example session jp486-18032023-0108 represents an electrophysiological recording session with associated behavioral data.

Session-Level Files

{session}.desel

• Specifies the brain region targeted by each probe
• Each line corresponds to one shank (Neuronexus/Cambridge probes were used)

{session}.desen

• Contains environment codes for each recording trial
• Random foraging trials: "circ80"
• Rest periods: "rest"
• AutoPI task trials: "autopi"

{session}.environmentFamiliarity

• Indicates whether the environment was novel or familiar
• All environments in this dataset are familiar, denoted as "fam"

{session}.par

• Contains parameters describing the recording session
• Structure:

  number_channels_in_dat_file number_bits_per_data_point_in_dat_file
  sampling_rate unused_number
  number_shanks unused_number
  number_channels_shank_1 channel_0_e1 channel_1_e1 channel_2_e1 channel_3_e1
  number_channels_shank_2 channel_0_e2 channel_1_e2 channel_2_e2 channel_3_e2
  ...
  number_of_session_trials
  name_of_session_trial_01
  name_of_session_trial_02
  ...
  

• Sampling rate format: 1000000/frequency (value of 50 = 20000 Hz sampling rate)
• Time conversion: sample_no/sampling_rate = seconds

{session}.protocol

• Protocol name (autopi_task.py_ml)
• Duration in seconds (7200)
• Rosmaster computer identifier and IP address

{session}.px_per_cm

• Pixel-to-centimeter conversion factor for position data

{session}.sampling_rate_dat

• Sampling rate in Hz for electrophysiological data

{session}.setup

• Recording room identifier where experiment was conducted

{session}.stimulation

• Laser stimulation status for each trial
• All trials show "none" (no stimulation) in this dataset

{session}.log

• Text file documenting event names, parameters, and timestamps during task
• Format: event_name time_in_seconds parameter_value
• Example: door 1679144594.012599534 1.000000

arenaCoordinates and bridgeCoordinates

• Coordinates of the arena and bridge during the AutoPI task extracted via object detector using AutopiPy.

{session}.mouseLeverPosition.csv

• Position data for mouse and lever generated via DeepLabCut
• Columns:
  • leverX, leverY: Lever position coordinates
  • leverOri: Lever orientation per trial
  • leverXHeading, leverYHeading: Angles between lever and bridge
  • leverPressX, leverPressY: Lever position at press time
  • leverBoxPLX, leverBoxPLY: Lever box posterior left corner coordinates
  • leverBoxPRX, leverBoxPRY: Lever box posterior right corner coordinates
  • mouseX, mouseY: Mouse position coordinates
  • mouseOri: Mouse orientation
  • mouseXHeading, mouseYHeading: Mouse heading values (mean between ear positions)

leverPose

• Lever position data for each trial generated via DeepLabCut

leverBoxDistance.npy

• Mouse occupancy histogram data with distance from lever on x-axis

leverZoneMaxDistance

• Maximum distance from lever to be considered within the lever zone

{session}.pose.npy

• 2D numpy array containing animal position and orientation information
• Columns: (time, x, y, z, yaw, pitch, roll)
• Position in centimeters, orientation in Euler angles (radians)
• Missing data marked as np.NAN
• Yaw ranges from -180 to 180 degrees; 0 aligns with positive x-axis, π/2 upward, -π/2 downward

{session}.pose_kf.npy

• Kalman-filtered position and orientation data
• Filter applied to mouse x,y positions to reduce tracking noise
• 2D array implementing constant velocity model for state [x, y, dx, dy]ᵀ

navPathInstan.csv

• Instantaneous behavioral variables for each navPath timepoint
• Columns:
  • name: Trial identifier (session_name_trial_number_journey_number)
  • trialNo: Trial number
  • timeRos: Computer clock time
  • timeRes: Recording time
  • iTime: Time from trial start
  • iTimeProp: Proportional trial time (0-1)
  • distance: Distance traveled (cm)
  • distanceProp: Proportional distance within trial (0-1; NaN indicates insufficient tracking accuracy)
  • speed: Animal velocity
  • x, y: Animal position
  • targetDistance: Distance to lever
  • targetToAnimalX, targetToAnimalY: Lever position relative to animal
  • targetToAnimalAngle: Angle between animal and lever
  • cumSumDiffAngleAroundTarget: Cumulative angular change around target
• Note: NaN values indicate insufficient tracking accuracy for parameter calculation

navPathSummary.csv

• Comprehensive list of all NavPaths in the session
• NavPaths represent behavioral segments
• Columns:
  • name: Trial identifier including session, journey, and path type
  • trial: Session plus trial name
  • trialNo: Trial number
  • journey: Journey number
  • type: Path classification (all, search, homing, at lever)
  • light: Lighting condition (light/dark)
  • nLeverPresses: Lever presses per journey
  • startTimeRos, endTimeRos: Journey start/end in ROS time
  • startTimeRes, endTimeRes: Journey start/end in recording session time

trial_table_simple

• DataFrame containing session name, trial number, start/end times in samples (startTime, endTime), within-session times (startTimeWS, endTimeWS in ROS time), light conditions, and recording start/end times

sessionIntervals.npy

• Numpy array containing time offsets for each session trial, generated from the dat file

ses.pickle

• SpikeA session object generated with AutopiPy
• Enables NavPath selection (search, homing, at lever) for analysis
• The "name" column links to instantaneous data for electrophysiological analysis
• Contains time intervals for constraining spike train analysis to specific behavioral epochs

ifr_autopi.pkl

• Instantaneous firing rates for all neurons as a function of time within the session

{session}.shank_neuron.csv

• Contains neuron ID, neuron number, and recording layer. This file may not be included in every session.

shank_layer.csv

• DataFrame mapping shank numbers to recording layer. This file may not be included in every session.

{session}.grid_cell_parameters.pkl

• Grid cell spatial parameters including period, offset, peak rate, and orientation for both rigid and flexible models, along with model fit metrics. This file may not be included in every session. This file is only generated for sessions with sufficient numbers of co-recorded grid cells suitable for model reconstruction.

{session}.rnn_ifr_normalization.pkl

• Normalization parameters (means and standard deviations) for instantaneous firing rate data used in RNN analysis.  This file may not be included in every session. This file is only generated for sessions with sufficient numbers of co-recorded grid cells suitable for model reconstruction.

Spike Sorting Files (Kilosort and Phy)

The following files are generated using Kilosort and Phy. For detailed information, visit: https://phy.readthedocs.io/en/latest/terminology/

• params.py: Python script containing spike sorting and data processing configuration parameters
• amplitudes.npy: Amplitude values for each spike, saved by the spike sorting algorithm
• channel_map.npy: 1D integer array of size (n_channels,) containing channel mapping. Raw data has n_channels_dat columns (may differ from n_channels if dead channels exist)
• channel_positions.npy: Channel positions on the probe shank
• pc_features.npy: Principal component features utilized in spike sorting
• pc_feature_ind.npy: Principal component indices
• spike_templates.npy: Template assignments for each spike prior to manual clustering
• spike_times.npy: Spike detection timestamps
• spike_clusters.npy: 1D array with n_spikes elements, providing cluster ID for each spike
• cluster_group.tsv: Cluster information where each row represents a putative neuron or neuron group, including properties and classification
• cluster_info.tsv: Detailed cluster information from spike sorting
• cluster_KSLabel.tsv: Cluster label information
• cluster_ContamPct.tsv: Principal component contamination percentage per cluster (cluster_id\tContamPct)
• cluster_Amplitude.tsv: Cluster ID and amplitude values (cluster_id\tAmplitude)
• templates.npy: Waveform templates for specific channels

Note: Some subjects have identifiers beginning with "jp" instead of "mn". File structures remain consistent across all animals.

Data Directory

protocol_m_rerender.png

• Graphical abstract of the experiment protocol.

Results Directory

cells.csv

• DataFrame containing neuron classification information
• Columns:
  • cluId: Unique cell identifier
  • session: Recording session name
  • mouse: Animal identifier
  • refractory: Refractory period metric
  • refractoryClean: Boolean indicating clean refractory period
  • usable: Whether cell is usable for analysis
  • meanRate: Mean firing rate
  • [Additional columns as per original data]

clu_table.csv

• Probe mapping information
• Columns:
  • channel: Recording channel number
  • shank: Shank number
  • hemisphere: Brain hemisphere (left/right)
  • iso_distance: Isolation distance
  • session: Recording session
  • cluId: Cell identifier

behavior_180_EastReferenceQuadrant.csv

• Behavioral metrics with East as reference (0°)
• Contains parameters including search length, duration, speed, heading errors, and homing accuracy
• Columns:
  • sessionName: Session identifier
  • name: Trial name
  • valid: Whether trial is valid
  • trialNo: Trial number
  • startTime, endTime: Trial start and end times
  • startTimeWS, endTimeWS: Timestamps
  • duration: Trial duration
  • light: Light condition
  • arenaRadiusCm: Arena radius in centimeters
  • nLeverPresses: Number of lever presses
  • nJourneys: Number of journeys
  • [Various search and homing behavioral metrics]
  • leverX, leverY: Lever position coordinates
  • subject: Animal identifier
  • date: Session date

instan_behavior_180_EastReference.csv

• Instantaneous behavioral variables at each timepoint
• Columns:
  • name: Trial identifier
  • trialNo: Trial number
  • timeRos, timeRes: Timestamps
  • iTime: Instantaneous time
  • iTimeProp: Time proportion
  • distance: Distance measure
  • distanceProp: Distance proportion
  • speed: Instantaneous speed
  • x, y: Position coordinates

trialsTable

• Comprehensive trial information including timing, light conditions, search/homing metrics, and lever positions

allSessionDf_with_leverVector_and_last_cohort.csv

• Dataset with lever vector calculations and final cohort data with model predictions in toroidal coordinates from RNN model
• Columns:
  • index: Index
  • mouse: Animal identifier
  • session: Session name
  • condition: Experimental condition
  • trial: Trial identifier
  • trialNo: Trial number
  • xPose, yPose: Animal position coordinates
  • hdPose: Head direction pose
  • x, y: Processed position coordinates
  • px, py: Predicted x, predicted y
  • withinPathTime: Time within path
  • recTime: Recording time
  • speed: Animal speed
  • distance: Distance traveled
  • v0_cos, v0_sin, v1_cos, v1_sin: Predicted toroidal coordinates of the animal position
  • lv0_cos, lv0_sin, lv1_cos, lv1_sin: Animal position in toroidal coordinates
  • mvtDirError: Movement direction error
  • absMvtDirError: Absolute movement direction error
  • shortCondition: Abbreviated condition name
  • light: Light condition (light/dark)
  • path: Path type

myMaps.pickle

• Dictionary containing 2D firing rate maps for all cells across 34 different conditions (e.g., 'circ80_1', 'searchToLeverPath_light', etc.)

myMapStats.csv

• Statistical measures for spatial maps
• Columns:
  • peakRate: Peak firing rate
  • gridScore: Grid score metric
  • gridSpacing: Grid spacing (when applicable)
  • meanRate: Mean firing rate
  • info: Information content
  • sparsity: Sparsity measure
  • condition: Recording condition
  • session: Session name
  • mouse: Animal identifier
  • cellId: Cell identifier

mySpatialAutos.pickle

• Spatial autocorrelations for all conditions (34 keys matching myMaps.pickle)

mySpatialAutosLever.pickle

• Spatial autocorrelations for lever-centered conditions

mapCorDf.csv

• Correlation coefficients between firing rate maps across conditions
• Contains r-values for map comparisons (NaN indicates insufficient data)
• Columns:
  • condition: Experimental condition pair
  • r: Pearson correlation coefficient (NaN means no valid correlation)
  • session: Session name
  • mouse: Animal identifier
  • cellId: Cell identifier

mapCorDf_circ80_inter.csv

• Map correlations for inter-session
• Same column structure as mapCorDf.csv

assignment_gridScore.csv

• Grid scores for all cells across different trial conditions and shuffling iterations
• Columns:
  • session: Session name
  • cluId: Cluster/cell identifier
  • gridScore: Computed grid score value
  • trialCode: Trial condition code
  • shuffling: Whether data was shuffled (True/False)

assignment_HDScore.csv

• Head direction scores including peak rates for directional tuning analysis
• Columns:
  • session: Session name
  • cluId: Cluster/cell identifier
  • HDScore: Head direction score
  • peakRate: Peak firing rate
  • trialCode: Trial condition code
  • shuffling: Whether data was shuffled (True/False)

gridRotationDf.csv

• Grid rotation analysis comparing reference frames
• Includes session, condition, light state, rotation angle, and light name

occ_matched_all_scenario.csv

• Occupancy-matched grid scores across different matching scenarios
• Columns:
  • Unnamed: 0: Row index
  • cellId: Cell identifier
  • condition: Experimental condition
  • grid_score: Grid score
  • mouse: Animal identifier
  • match: Matching criterion

MatcheGridScoreWithOccHdDownsampling.csv

• Grid scores after occupancy and head direction downsampling procedures

myLeverMaps.pickle

• 2D firing rate maps for lever-anchored cells in lever-centered coordinates

myLeverMapsArena.pickle

• Lever-anchored cell maps in arena coordinates

myLeverMapsStats.pickle

• Statistical measures for lever-centered maps including angular information

myLeverHistos.pickle

• 1D firing rate histograms as function of distance from lever for light/dark conditions

leverMapCor.pickle

• Correlation matrix between lever-centered maps across conditions and rotations

myLeverMapsHDInfo_new_function.pickle

• Head direction information for lever maps using updated analysis function

myLeverMapsHDInfoShuffle100.pickle

• Shuffled head direction statistics (100 iterations) for lever maps

myLeverMaps_last_lever_sessions_80.pickle

• Lever maps from last day of recording where the lever is placed on the arena for a random foraging session (but no task is involved)

searchIFRDict.pickle

• Instantaneous firing rates during search paths for all neurons

homingIFRDict.pickle

• Instantaneous firing rates during homing paths for all neurons

leverIFRDict.pickle

• Instantaneous firing rates relative to lever position

gc_pairs.csv

• Grid cell pairs for correlation analysis
• Contains cell IDs, indices, session, and mouse information
• Columns:
  • cluId1, cluId2: Identifiers for paired cells
  • index1, index2: Indices for paired cells
  • session: Session name
  • mouse: Animal identifier

res_ifr.csv

• Instantaneous firing rate associations between cell pairs across multiple conditions

gc_crosscorrs.pickle

• Cross-correlations between grid cell pairs

gc_crosscorrs_10_0.5_shuffle200.pickle

• Shuffled cross-correlations

gc_stcfm_10_0.5.pickle

• Spike-triggered cross-correlation matrices

gc_pairs_stcfm.pickle

• Pair-wise spike-triggered correlations across different conditions

atlever_gc_stcfm_10_0.5.pickle

• Spike-triggered correlations at the lever

atlever_gc_pairs_stcfm.pickle

• Pair-wise spike-triggered correlations at the lever

atlever_gc_stcfm_rotations_internal.pickle

• Internal rotational analysis of spike-triggered correlations at the lever

atlever_gc_stcfm_rotations_internal_stats.pickle

• Stats for internal rotational analysis of spike-triggered correlations at the lever

atlever_gc_stcfm_rotations.pickle

• Spike-triggered correlations at the lever with rotational analysis

atlever_gc_stcfm_rotations_stats.pickle

• Stats for spike-triggered correlations at the lever with rotational analysis

gc_stcfm_rotations.pickle

• Spike-triggered cross-correlation matrices with rotational analysis

gc_stcfm_rotations_stats.pickle

• Stats for spike-triggered correlations with rotational analysis

dfBridgeVonMises.csv

• Bi-variate Von Mises distribution parameters for bridge anchoring strength
• Columns:
  • condition: Experimental condition
  • session: Session name
  • trialBin: Trial bin number
  • level_3: Hierarchical level
  • k0, k1: Concentration parameters
  • k: Combined concentration parameter
  • mu0, mu1: Mean direction parameters

dfLeverVonMises.csv

• Bi-variate Von Mises distribution parameters for lever anchoring strength (k0, k1, k, mu0, mu1)
• Same column structure as dfBridgeVonMises.csv

dfLeverAnchoring_leverlocked.csv

• Anchoring strength analysis by time interval
• Columns:
  • Unnamed: 0: Row index
  • session: Session name
  • interval: Time interval
  • level_2: Hierarchical level
  • Von Mises parameters (k0, k1, k, mu0, mu1)

reconstuctionDFAutoPI.csv

• Path reconstruction data for AutoPI task including movement vectors and predictions
• Contains similar structure as allSessionDf

reconstuctionDFOF.csv

• Path reconstruction for open field sessions
• Contains similar structure as allSessionDf

reconstruction_shuffling.csv

• Shuffled reconstruction analysis
• Columns:
  • Unnamed: 0: Row index
  • mvtDirError: Movement direction error
  • session: Session name
  • condition: Experimental condition
  • shuffleIter: Shuffle iteration number
  • mvl: Mean vector length

sessionList_pathReconstruction.csv

• Sessions used for path reconstruction by grid cell count
• Columns:
  • session: Session name
  • gridCells: Number of grid cells in the session

model_distance_with_bins.csv

• Distance predictions with spatial binning for model validation
• Columns:
  • predictedLeverCoordv0, predictedLeverCoordv1: Predicted lever coordinates
  • actualCoordv0, actualCoordv1: Actual coordinates
  • distance_pred_to_actual: Distance between predicted and actual positions
  • bin: Distance bin assignment

model_mvl_shuffled.csv

• Mean vector length analysis with shuffled controls

myControlMaps.pickle

• Control condition firing rate maps

myControlMapStats.csv

• Statistics for control condition maps
• Columns:
  • gridScore: Grid score
  • peakRate: Peak firing rate
  • info: Information content
  • sparsity: Sparsity measure
  • condition: Experimental condition
  • session: Session name
  • mouse: Animal identifier
  • cellId: Cell identifier

myControlSpatialAutos.pickle

• Spatial autocorrelations for control conditions

rf_lever_control_light.csv

• Reference frame control analysis in light conditions
• Columns:
  • cellId: Cell identifier
  • mvl: Mean vector length
  • session: Session name
  • mouse: Animal identifier

rf_lever_control_dark.csv

• Reference frame control analysis in dark conditions
• Same structure as rf_lever_control_light.csv

myDoughnutCorrelations.pickle

• Doughnut correlation analysis (rotate the firing map of grid cells and correlate it with itself)

myLeverCellStatsSplitByMedian_homingDir_all.pickle

• Lever cell statistics split by median homing direction

myLeverCellStatsSplitBySearchLengthMedian_faster.pickle

• Statistics split by search length median

myLeverCellStatsTrialMatrix_all.pickle

• Trial matrix statistics for lever cells

rotation_per_trial_after_correction.csv

• Trial-by-trial rotation values after correction procedures
• Columns:
  • Unnamed: 0: Row index
  • session: Session name
  • condition: Experimental condition
  • light: Light condition
  • trialNo: Trial number
  • rotation: Rotation angle
  • abs_rotation: Absolute rotation angle

rr_anchoring_strength_analysis_by_trial.csv

• Anchoring strength (k-values) across trial bins for different conditions
• Columns:
  • condition: Experimental condition
  • session: Session name
  • trialBin: Trial bin
  • Von Mises parameters (k0, k1, k, mu0, mu1)
  • shortCondition: Abbreviated condition
  • light: Light condition
  • trialBinName: Trial bin name
  • anchor: Anchor type

shuffle_rf3_lever_dataframe.csv

• Shuffled map similarity for random foraging with lever on the arena
• Columns:
  • Unnamed: 0: Row index
  • cellName: Cell identifier
  • correlation: Correlation value

shuffled_lever_grid_coordinates_anchoring_strength_analysis.pickle

• Shuffled model prediction coordinates for anchoring strength analysis

within_trial_orientation_subtraction.csv

• Within-trial orientation differences for trial-by-trial rotation analysis
• Columns:
  • session: Session name
  • trialNo: Trial number
  • difference: Orientation difference value
• cartesian_maps_cropped.png: Stack of grid cell firing rate maps in Cartesian coordinates
• toroidal_maps_cropped.png: Stack of grid cell firing rate maps in toroidal coordinates
• heading_error.png: Graphical abstract of heading error
• error_at_peri.png: Graphical abstract of error at periphery
• leverPred2.png: Explanation figure of how anchoring strength is derived and how predicted lever coordinates are calculated
• anchoring_exclude_jp486_top.png: Same figure as Figure 6, excluding the animal with the most grid cells recorded
• anchoring_exclude_jp486_bottom.png: Same figure as Figure 7-8, excluding the animal with the most grid cells recorded
• schematic_cartesian_to_toroidal_3.png: Schematic illustrating Cartesian to toroidal transformation
• jp451_lstm_firstRF.npy: Test session used for simulations. Contains first random foraging animal position data
• jp451_lstm_lastRF.npy: Test session used for simulations. Contains last random foraging animal position data

Analysis Directories

• grid_spacing_and_dispersion/: Contains grid spacing and dispersion simulation analyses
• hd_grid_simulation/: Head direction and grid cell simulation analyses
• simReanchor/: Simulation results for reference frame re-anchoring
• simReanchorRotation/: Re-anchoring simulations with different scenarios

Units of Measure and Data Conventions

Units:

• Distance measurements: centimeters (cm)
• Time/duration measurements: seconds (s)
• Position coordinates: centimeters (cm)
• Angles: radians (unless otherwise specified)
• Firing rates: Hz (spikes per second)

Empty Cells:
Empty cells or NaN values in CSV/TSV files indicate missing data due to insufficient tracking accuracy, failed computations, or data not applicable to specific conditions. This applies consistently across all data files in the dataset.

Data Access Instructions

The dataset is split into 20 files due to its large size. To combine and extract the data:

Linux/Unix Terminal Commands:

# Combine files
cat folder.tar.gz.part.* > folder.tar.gz

# Extract archive
tar -xzf folder.tar.gz -C /path/to/destination_directory

Code Availability

Analysis code is available in the following GitHub repositories:

• https://github.com/PJJ19/Peng_et.al_2025_noInt
• https://github.com/PJJ19/autopi_grid_cell_decoding

File Format Specifications

• .npy: NumPy binary format
• .pickle / .pkl: Python pickle format
• .csv: Comma-separated values
• .tsv: Tab-separated values
• .png: Portable Network Graphics images