y_psb = ['us'],
order_fixed = True,
iti = iti,
n_rep = n_rep_train)
extra_us_ctl_stage = expr.stage(x_pn = [[]],
x_bg = ['ctx'],
y_psb = ['us'],
order_fixed = True,
iti = 0,
n_rep = (iti + 1)*n_rep_train)
##### DEFINE SCHEDULES #####
# no conditioning (as a very basic control)
no_cond = expr.schedule(resp_type = 'exct', stages = {'test': test_stage})
# basic conditioning
cond = expr.schedule(resp_type = 'exct', stages = {'cond': training_stage, 'test': test_stage})
# blocking
blocking = expr.schedule(resp_type = 'exct', stages = {'one_cue': one_cue_stage, 'two_cue': two_cue_stage, 'test': test_cs2_stage})
# two cue conditioning (e.g. overshadowing)
two_cue = expr.schedule(resp_type = 'exct', stages = {'two_cue': two_cue_stage, 'test': test_cs2_stage})
# pre-exposure (for latent inhibition)
pre_exp = expr.schedule(resp_type = 'exct', stages = {'pre_exposure': pre_exp_stage, 'cond': training_stage, 'test': test_stage})
# pre-exposure (ABC)
pre_exp_abc = expr.schedule(resp_type = 'exct', stages = {'pre_exposure': pre_exp_ctx_a_stage, 'cond': training_ctx_b_stage, 'test': test_ctx_c_stage})
import pandas as pd
from statsrat import expr
'''
Simplified category learning tasks (smaller versions of the tasks like those in 'cat' -> 'kitten').
'''
# learned predictiveness
design = expr.schedule(resp_type = 'choice',
stages = {'relevance': expr.stage(x_pn = [['a', 'x'], ['a', 'y'], ['b', 'x'], ['b', 'y']],
y = [['cat1'], ['cat1'], ['cat2'], ['cat2']],
n_rep = 10),
'transfer': expr.stage(x_pn = [['a', 'x'], ['b', 'y']],
y = [['cat3'], ['cat4']],
n_rep = 10),
'test': expr.stage(x_pn = [['a', 'y'], ['b', 'x']],
y_psb = ['cat3', 'cat4'],
lrn = False,
n_rep = 1)})
rel_irl_oat = expr.oat(schedule_pos = ['design'],
behav_score_pos = expr.behav_score(stage = 'test',
trial_pos = ['a.y -> nothing', 'b.x -> nothing'],
trial_neg = ['a.y -> nothing', 'b.x -> nothing'],
resp_pos = ['cat3', 'cat4'],
resp_neg = ['cat4', 'cat3']))
learned_predictiveness = expr.experiment(schedules = {'design': design},
oats = {'rel_irl': rel_irl_oat})
del design; del rel_irl_oat
iti=iti,
n_rep=n_rep_test)
test_different = expr.stage(x_pn=[['cs1']],
x_bg=['ctx_b2'],
y_psb=['us'],
order_fixed=True,
iti=iti,
n_rep=n_rep_test)
# group "same"
same = expr.schedule(resp_type='exct',
stages={
'cond': training,
'ex_cs1': extinction_cs1,
'ex_cs2': extinction_cs2,
'ex_cs2': extinction_cs2,
'ex_cs1': extinction_cs1,
'test': test_same
})
# group "different"
different = expr.schedule(resp_type='exct',
stages={
'cond': training,
'ex_cs1': extinction_cs1,
'ex_cs2': extinction_cs2,
'ex_cs2': extinction_cs2,
'ex_cs1': extinction_cs1,
'test': test_different
})
design = expr.schedule(resp_type = 'choice',
stages = {
'tutorial_0a': expr.stage(
x_pn = [['alpha'], ['beta']],
y = [['cati'], ['catii']],
y_psb = ['cati', 'catii'],
n_rep = 6),
'tutorial_0b': expr.stage(
x_pn = [['alpha'], ['beta']],
y = [['cati'], ['catii']],
y_psb = ['cati', 'catii'],
n_rep = 6),
'tutorial_0c': expr.stage(
x_pn = [['alpha', 'theta'], ['beta', 'theta'], ['alpha', 'phi'], ['beta', 'phi']],
y = 2*[['cati'], ['catii']],
y_psb = ['cati', 'catii'],
n_rep = 6),
'relevance': expr.stage(
x_pn = [['b1', 't1'], ['b1', 't2'], ['b2', 't1'], ['b2', 't2'], ['t3', 'b3'], ['t3', 'b4'], ['t4', 'b3'], ['t4', 'b4']],
y = 2*[['cat1'], ['cat1'], ['cat2'], ['cat2']],
y_psb = ['cat1', 'cat2'],
n_rep = 12),
'transfer': expr.stage(
x_pn = [['t5', 'b5'], ['t6', 'b6'], ['t1', 'b1'], ['t2', 'b2'], ['t3', 'b3'], ['t4', 'b4']],
y = 3*[['cat3'], ['cat4']],
y_psb = ['cat3', 'cat4'],
n_rep = 8),
'test': expr.stage(
x_pn = [['t5', 'b6'], ['t6', 'b5'], ['t1', 'b2'], ['t2', 'b1'], ['t3', 'b4'], ['t4', 'b3']],
y_psb = ['cat3', 'cat4'],
lrn = False,
n_rep = 2)},
x_dims = {'fruits': ['alpha', 'beta', 'theta', 'phi'], 'benign_faces': ['b1', 'b2', 'b3', 'b4', 'b5', 'b6'], 'angry_faces': ['t1', 't2', 't3', 't4', 't5', 't6']})
'''
Inverse base rate effect (IBRE) and related experiments.
'''
##### BASIC INVERSE BASE RATE EFFECT DESIGN #####
# Kruschke (1996), Experiment 1 (inverse base rate effect)
# I'm missing some of the test trial types.
design = expr.schedule(resp_type='choice',
stages={
'training':
expr.stage(freq=[3, 1, 3, 1],
x_pn=[['i1', 'pc1'], ['i1', 'pr1'],
['i2', 'pc2'], ['i2', 'pr2']],
y=[['c1'], ['r1'], ['c2'], ['r2']],
y_psb=['c1', 'r1', 'c2', 'r2'],
n_rep=15),
'test':
expr.stage(x_pn=[['pc1'], ['pr1'], ['pc2'], ['pr2'],
['pc1', 'pr1'], ['pc2', 'pr2']],
y_psb=['c1', 'r1', 'c2', 'r2'],
lrn=False,
n_rep=2)
})
pc_pr = expr.oat(schedule_pos=['design'],
behav_score_pos=expr.behav_score(
stage='test',
trial_pos=['pc1.pr1 -> nothing', 'pc2.pr2 -> nothing'],
trial_neg=['pc1.pr1 -> nothing', 'pc2.pr2 -> nothing'],
resp_pos=['r1', 'r2'],
resp_neg=['c1', 'c2']))
extra_us_ctl_stage_t2 = expr.stage(x_pn=[[]],
x_bg=['ctx', 'time'],
x_value=pd.Series({
'cs': 1.0,
'ctx': 1.0,
'time': 2.0
}),
y_psb=['us'],
order_fixed=True,
iti=0,
n_rep=(iti + 1) * n_rep_train)
##### DEFINE SCHEDULES #####
# no conditioning (as a very basic control)
no_cond = expr.schedule(resp_type='exct', stages={'test': test_stage_t0})
# basic conditioning
cond = expr.schedule(resp_type='exct',
stages={
'cond': training_stage_t0,
'test': test_stage_t1
})
# blocking
blocking = expr.schedule(resp_type='exct',
stages={
'one_cue': one_cue_stage_t0,
'two_cue': two_cue_stage_t1,
'test': test_cs2_stage_t2
})