def make_left_contrast_higher_neuron(trial_features, number_of_time_bins):
number_of_trials = trial_features.shape[1]
earliest_response = trial_features[-1].min() * 100
# setting a reasonable range for peak
forward_time_shift_relative_to_response = np.random.uniform(
2, earliest_response - 5)
sigma = np.random.uniform(0, number_of_time_bins / 4)
if trial_features[3][0] == 1: # stimulus was higher left contrast
mu = trial_features[-1][
0] * 100 - forward_time_shift_relative_to_response
firings_on_trial = math_utility.my_gaussian(
np.arange(number_of_time_bins), mu, sigma)
else:
firings_on_trial = np.zeros(number_of_time_bins)
for trial in range(number_of_trials - 1):
if trial_features[3][trial +
1] == 1: # stimulus was higher left contrast
mu = trial_features[-1][
trial + 1] * 100 - forward_time_shift_relative_to_response
next_firings = math_utility.my_gaussian(
np.arange(number_of_time_bins), mu, sigma)
else:
next_firings = np.zeros(number_of_time_bins)
firings_on_trial = np.vstack((firings_on_trial, next_firings))
return firings_on_trial
def make_response_neuron(trial_features, number_of_time_bins):
number_of_trials = trial_features.shape[1]
earliest_response = trial_features[-1].min() * 100
# setting a reasonable range for peak
forward_time_shift_relative_to_response = np.random.uniform(
2, earliest_response - 5)
sigma = np.random.uniform(0, number_of_time_bins / 4)
mu = trial_features[-1][0] * 100 - forward_time_shift_relative_to_response
firings_on_trial = math_utility.my_gaussian(np.arange(number_of_time_bins),
mu, sigma)
for trial in range(number_of_trials - 1):
mu = trial_features[-1][
trial + 1] * 100 - forward_time_shift_relative_to_response
next_firings = math_utility.my_gaussian(np.arange(number_of_time_bins),
mu, sigma)
firings_on_trial = np.vstack((firings_on_trial, next_firings))
return firings_on_trial