def get_frinfo(paod, trial_briefs, completedOnly=True):
"""
:param paod: from class paoding
:return: first/second stage choices, stage2 state, reward probability, reward state
"""
num_Trials = len(paod.index_records)
trials = {
'chosen': [],
'reward': [],
'trial_length': [],
'choice_num': [],
'choices': []
}
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
if brief_info['completed'] != True:
continue
# only keep the trials in which the decision is made
trials['chosen'].append(brief_info['chosen'])
trials['reward'].append(brief_info['reward'])
trials['choices'].append(brief_info['choices'])
trials['choice_num'].append(brief_info['choice_num'])
trials['trial_length'].append(brief_info['trial_length'])
trials = pd.DataFrame(trials)
return trials
def get_rvlrinfo(paod, trial_briefs):
"""
:param paod: from class paoding
:return: coherence, sure trials, choice, reward, evidence difference for each trial
"""
# input_setting = tsDawbrief_config()
# choice_position = input_setting['choice']
# reward_position = input_setting['reward']
###
num_Trials = len(paod.index_records)
# num_Trials = 4900
trials = {
'choice': [],
'chosen': [],
'reward': [],
'block': [],
'completed': []
}
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
if brief_info['completed']:
trials['choice'].append(brief_info['choice'])
trials['chosen'].append(brief_info['chosen'])
trials['reward'].append(brief_info['reward'])
if not brief_info['completed']:
trials['choice'].append(np.nan)
trials['chosen'].append(np.nan)
trials['reward'].append(np.nan)
trials['completed'].append(brief_info['completed'])
trials['block'].append(brief_info['block'])
trials = pd.DataFrame(trials)
return trials
def get_multinfo(paod, trial_briefs):
"""
:param paod: from class paoding
:return: coherence, sure trials, choice, reward, evidence difference for each trial
"""
# input_setting = tsDawbrief_config()
# choice_position = input_setting['choice']
# reward_position = input_setting['reward']
###
num_Trials = len(paod.index_records)
trials = {
'choice': [],
'chosen': [],
'reward': [],
'modality': [],
'direction': [],
'estimates': []
}
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
if brief_info['completed'] != True:
continue
# only keep the trials in which the decision is made
trials['choice'].append(brief_info['choice'])
trials['chosen'].append(brief_info['chosen'])
trials['reward'].append(brief_info['reward'])
trials['modality'].append(brief_info['modality'])
trials['direction'].append(brief_info['directions'])
# trials['completed'].append(brief_info['completed'])
trials['estimates'].append(brief_info['estimates'])
trials = pd.DataFrame(trials)
return trials
def get_hidden_resp_all(paod, trial_briefs, gates=False):
"""
:param resp_hidden_group: from class paoding
:return: response of neurons in the hidden layer
"""
all_resp = []
num_Trials = len(paod.index_records)
if gates:
ig_all, rg_all, ng_all = [], [], []
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
# only keep the trials in which the decision is made
if not brief_info['completed']:
continue
_, _, _, nd, rg, ig, ng = paod.get_neuron_behavior_pair(i,
gates=True)
all_resp.append(nd)
rg_all.append(rg)
ig_all.append(ig)
ng_all.append(ng)
all_resp = np.array(all_resp)
rg_all = np.array(rg_all)
ig_all = np.array(ig_all)
ng_all = np.array(ng_all)
return all_resp, rg_all, ig_all, ng_all
else:
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
fd, gd, rd, nd = paod.get_neuron_behavior_pair(i)
if not brief_info['completed']:
continue
nd = nd.squeeze()
all_resp.append(nd)
all_resp = np.array(all_resp)
return all_resp
def get_output_resp_fixratio(paod, trial_briefs, gates=False):
"""
:param resp_hidden_group: from class paoding
:return: response of neurons in the hidden layer
"""
all_resp = []
num_Trials = len(paod.index_records)
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
fd, gd, rd, nd = paod.get_neuron_behavior_pair(i)
if brief_info['completed'] != True:
continue
gd = rd.squeeze()
all_resp.append(gd)
all_resp = np.array(all_resp)
return all_resp
def get_resp_ts(paod, trial_briefs, gates=False):
"""
:param resp_hidden_group: from class paoding
:return: response of neurons in the hidden layer
"""
resp_hidden = []
resp_output = []
num_Trials = len(paod.index_records)
if gates:
pass
else:
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
if brief_info['completed'] is not True:
continue
fd, gd, rd, nd = paod.get_neuron_behavior_pair(i)
resp_hidden.append(nd.squeeze())
resp_output.append(rd.squeeze())
# print([resp_output[i].shape[0] for i in range(len(resp_output))])
return np.array(resp_hidden), np.array(resp_output)
def get_bhvinfo(paod, trial_briefs, choices_trials=True):
"""
:param paod: from class paoding
:return: choice: choose left or right; finish choice or not; when do it make a choice
shapes: how many shapes are used for choice; temporal weight ; summed weight; shape on time; shape on
trials: whole trial length
"""
###
num_Trials = len(paod.index_records)
choice = {
'status': [],
'time': [],
'left': [],
'correct': [],
'correct_logLR': []
}
shapes = {
'rt': [],
'tempweight': [],
'sumweight': [],
'ontime': [],
'order': [],
'on': []
} #
trial = {'length': [], 'num': [], 'tarOn_time': []} #
reward = {'pred': [], 'time': []}
shape_weight_pair = {
-0.9: 1,
-.7: 2,
-.5: 3,
-.3: 4,
-.1: 5,
.0: 0,
.1: 6,
.3: 7,
.5: 8,
.7: 9,
.9: 10
}
shapeRT_max = 30
shape_dur = 5
a = 0
for i in range(num_Trials):
brief_info = base642obj3(trial_briefs['trial_brief_base64'][i])
# only keep the trials in which the decision is made
if choices_trials:
if not brief_info['completed']:
a += 1
continue
else:
if brief_info['completed']:
a += 1
continue
n_shape_curr = brief_info['rt']
tmpweight_tmp = np.zeros(shapeRT_max, )
tmpweight_tmp[:n_shape_curr] = brief_info['tmp_weight_sequence']
tmpweight_tmp = tmpweight_tmp.round(2)
shapes['rt'].append(n_shape_curr)
shapes['tempweight'].append(tmpweight_tmp)
shapes['sumweight'].append(np.sum(tmpweight_tmp))
shapes['ontime'].append(
np.arange(3, shape_dur * n_shape_curr, shape_dur, dtype=int))
shapeorder_tmp = []
for ii in range(shapeRT_max):
shapeorder_tmp.append(shape_weight_pair[tmpweight_tmp[ii]])
shapes['order'].append(np.array(shapeorder_tmp))
shapes['on'].append(np.arange(n_shape_curr) + 1)
##
choice['status'].append(brief_info['chosen'])
choice['time'].append(brief_info['choice_time'][0])
choice['left'].append(3 -
2 * brief_info['choice']) # 1 if left else -1
choice['correct_logLR'].append(True if (
np.sum(brief_info['tmp_weight_sequence']) >= 0) == (
brief_info['choice'] == 1) else False)
choice['correct'].append(True if brief_info['trialtype'] == (
2 - brief_info['choice']) else False)
##
trial['length'].append(brief_info['choice_time'][0] + 6)
trial['num'].append(i)
trial['tarOn_time'].append(2)
##
reward['pred'].append(brief_info['reward'])
reward['time'].append(brief_info['choice_time'][-1])
trial = pd.DataFrame(trial)
choice = pd.DataFrame(choice)
shapes = pd.DataFrame(shapes)
reward = pd.DataFrame(reward)
finish_rate = 0 if num_Trials == 0 else 1 - a / num_Trials
return trial, choice, shapes, reward, finish_rate