Stop-signal reaction time (SSRT) is one of the most popular metrics of behavioral disinhibition. It can be estimated with a number of different methods. The exact method which you should use depends on the exeperimental design. If one uses a number of pre-determined stop-signal delays (SSD), then the method of choice is the integration method.
To use this method, we need to first calculate accuracy for each predetermined stop-signal delay, and select one for which accuracy in stop trials equals 0.5 (50 %). In this picture, no specific SSD (on y-axis, expressed in ms) correpsonds to 0.5 accuracy.
To estimate SSD corresponding to 50% accuracy, we need to interpolate it.
First, let's import the necessary packages:
import pandas as pd
from scipy.interpolate import UnivariateSpline
from scipy import interpolate
from scipy.interpolate import interp1d
from scipy.interpolate import InterpolatedUnivariateSpline
Second, assuming we calculated accuracy for each individual (expressed as subjID, which is set as index), we need to transform the dataframe first.
df_ssrt = stop_error_rate_overall.reset_index()
df = df_ssrt.melt(id_vars=['subjID'])
df['variable'] = df['variable'].map({'100_stop_acc': 100, '200_stop_acc': 200, '300_stop_acc':300, '400_stop_acc':400,'500_stop_acc':500,'600_stop_acc':600})
Finally, define the smoothing function:
def SmootheSplines(x):
x1 = x['variable'].values
y1 = x['value'].values
s = InterpolatedUnivariateSpline(x1, y1)
return s
splinedf = df.groupby('UIN').apply(SmootheSplines)
Done! Now we only have to subtract median "go" time from the value in splinedf.
def evaluate_spline_distance(row):
spline_obj, splice_val = row
return spline_obj(splice_val)
df['results'] = df[[0, 'splice']].apply(evaluate_spline_distance, axis=1)
