def convolution(self, regressor, step_size=2):
"""
Computes the predicted convolved hemodynamic response function signals
for a given regressor.
Parameters
----------
regressor : str
Name of the regressor whose predicted neural time course and whose
hemodynamic response function will be convolved: select from "gain",
"loss", "dist2indiff"
step_size : float
Size of temporal steps (in seconds) at which to generate signals
Return
------
convolution : np.ndarray
Array containing the predicted hemodynamic response function values
for the given regressor
"""
time_course = self.time_course(regressor, step_size)
# Hemodynamic responses typically last 30 seconds
hr_func = hrf(np.arange(0, 30, step_size))
convolution = np.convolve(time_course, hr_func)[:len(time_course)]
return convolution
def test_hrf():
# Define the only two parameters of interest
times = np.arange(30)
hr = hrf(times)
# They should contain the same number of elements
assert_equal(len(times), len(hr))
# The hemodynamic response is 0 at onset
assert hr[0] == 0
# HRF initially increases, then decreases, then increases/stabilizes to 0
for i in range(0, 4): assert hr[i] < hr[i + 1]
for i in range(5, 10): assert hr[i] > hr[i + 1]
for i in range(15, 29): assert hr[i] < hr[i + 1]
