boydmeredith_piet_data.zip (3.47 GB)
Rat FOF Recordings During Dynamic Clicks Task
Version 2 2021-09-29, 13:24
Version 1 2021-09-29, 13:22
dataset
posted on 2021-09-29, 13:24 authored by Tyler Boyd-MeredithTyler Boyd-MeredithFOF recordings from 5 rats performing a dynamic evidence accumulation task. These recordings yielded 738 units from these 5 rats. Data was collected using unilateral electrode arrays (n=4) and bilateral tetrodes (n=1). Each rat has an accompanying behavioral dataset containing an average of 108,126 trials performing the dynamic clicks task. These trials were used to fit a previously described accumulation model to their behavior (Piet et al, 2018). The dynamic clicks task presents the animal with two streams of randomly-timed auditory clicks on the left and right side of a central fixation port. These left and right click trains are generated with at a high rate (38Hz) on one side and a low rate (2Hz) on the other. The underlying click rates alternate a rate of 1Hz and the rat is rewarded for correctly identifying the side generating the higher click rate at the end of the trial. To solve this task, rats learn to discount older evidence so that the most recently presented clicks have the greatest effect on the decision. This induces changes of mind throughout each trial, which we can predict with our behavioral model and relate to neural activity.
This dataset, along with the accompanying code (https://github.com/Brody-Lab/dynamic_ephys) can be used to regenerate all the results and figures in the manuscript titled Stable choice coding in FOF across model predicted changes of mind (Boyd-Meredith and Piet et al., under review; preprint here: https://www.biorxiv.org/content/10.1101/2021.05.13.444020v2).
Behavioral data:
Each rat (H###) has behavioral data saved in a file titled data/H###.mat
This contains a matlab struct called data with fields like:
T - the duration of the trial
leftbups - the timing of the left clicks relative to stimulus onset in each trial
pokedR - boolean indicated whether the rat went to the rightward port
sessid - a unique identifier for each session
genSwitchTimes - the timing of the state switches relative to the stimulus onset on that trial
Model fits:
Each rat's accumulation model fits are saved in a file titled results/fit_analytical_H###.mat
This contains a matlab struct with fields like:
final - the best fit parameters to the accumulation model
param_names - the names associated with each parameter
auto_se - standard error derived from automatic differentiation in julia
Electrophysiology data:
Each cell has an associated file titled: data/phys/bin/phys_data_#####.mat
This file contains an associated ratname and sessid as well as
a struct array of length ntrials called array_data whose fields are things like
stim_start - the timing (in seconds) of the stimulus start relative to the beginning of the trial
spikes - the timing of the spikes on the trial
The easiest way to begin working with the data is to clone the accompanying git repository https://github.com/Brody-Lab/dynamic_ephys
