Examine transitions between neighbors of different presentations of repeated items.
The repetition-neighbor CRP analysis examines transitions between items studied near different presentations of the same repeated item. For a repeated item at positions \(i\) and \(j\), the analysis tracks transitions from \(j\)’s neighbors to \(i\)’s neighborhood (j2i), from \(i\)’s neighbors to \(j\)’s neighborhood (i2j), or both.
These cross-occurrence neighbor transitions test whether recalling items near one presentation of a repeated item facilitates transitions to items near the other presentation.
import os
import matplotlib.pyplot as plt
from jaxcmr.analyses.repneighborcrp import (
plot_rep_neighbor_crp,
subject_rep_neighbor_crp,
test_rep_neighbor_crp_vs_control,
)
from jaxcmr.helpers import find_project_root, generate_trial_mask, load_data, save_figure
from jaxcmr.repetition import make_control_datasetdata_path = "data/LohnasKahana2014.h5"
figure_dir = "results/figures"
figure_str = ""
ylim = [0, .30]
mixed_trial_query = "data['list_type'] > 3"
control_trial_query = "data['list_type'] == 1"
control_shuffles = 10
min_lag = 4
max_lag = 3
use_lag2 = Trueproject_root = find_project_root()
figure_dir = os.path.join(project_root, figure_dir)
data_path = os.path.join(project_root, data_path)
data = load_data(data_path)
trial_mask = generate_trial_mask(data, mixed_trial_query)
control_dataset = make_control_dataset(data, mixed_trial_query, control_trial_query, control_shuffles)
control_mask = generate_trial_mask(control_dataset, mixed_trial_query)
datasets = [data, control_dataset]
masks = [trial_mask, control_mask]for direction in ["j2i", "i2j", "both"]:
plot_rep_neighbor_crp(
datasets=datasets,
trial_masks=masks,
max_lag=max_lag,
min_lag=min_lag,
direction=direction,
use_lag2=use_lag2,
contrast_name="Source",
labels=["Mixed", "Control"],
)
plt.title(f"Direction: {direction}")
if ylim is not None:
for ax in plt.gcf().axes:
ax.set_ylim(ylim)
save_figure(figure_dir, figure_str, suffix=direction)
for direction in ["j2i", "i2j", "both"]:
observed_crp = subject_rep_neighbor_crp(
data, trial_mask, direction, use_lag2, min_lag, max_lag
)
control_crp = subject_rep_neighbor_crp(
control_dataset, control_mask, direction, use_lag2, min_lag, max_lag
)
result = test_rep_neighbor_crp_vs_control(observed_crp, control_crp, max_lag, direction)
print(f"\n{'='*70}")
print(f"Statistical Test: {direction}")
print(f"{'='*70}")
if direction == "j2i":
print("Tests whether j+1/j+2 → i-neighbors transitions are elevated vs control.")
print("Significant positive effects support study-phase retrieval at P2.")
elif direction == "i2j":
print("Tests whether i+1/i+2 → j-neighbors transitions are elevated vs control.")
print("(Control condition for asymmetric predictions.)")
else:
print("Tests both directions combined.")
print()
print(result)
======================================================================
Statistical Test: j2i
======================================================================
Tests whether j+1/j+2 → i-neighbors transitions are elevated vs control.
Significant positive effects support study-phase retrieval at P2.
Direction: j2i
Lag | t-stat t p-val | W-stat W p-val | Mean Diff
------+----------------------+----------------------+------------
-3 | -0.398 0.6929 | 245.0 0.2585 | -0.0029
-2 | -0.005 0.9956 | 285.0 0.6330 | -0.0000
-1 | 2.082 0.0450 | 224.0 0.1397 | 0.0173
0 | 0.392 0.6975 | 308.0 0.9163 | 0.0049
1 | 0.813 0.4217 | 263.0 0.7545 | 0.0076
2 | 2.198 0.0348 | 189.0 0.0390 | 0.0216
3 | 1.016 0.3170 | 300.0 0.8146 | 0.0115
======================================================================
Statistical Test: i2j
======================================================================
Tests whether i+1/i+2 → j-neighbors transitions are elevated vs control.
(Control condition for asymmetric predictions.)
Direction: i2j
Lag | t-stat t p-val | W-stat W p-val | Mean Diff
------+----------------------+----------------------+------------
-3 | 1.190 0.2421 | 247.0 0.2726 | 0.0165
-2 | 0.874 0.3881 | 269.0 0.4609 | 0.0090
-1 | 1.222 0.2301 | 269.0 0.4609 | 0.0130
0 | 3.765 0.0006 | 118.0 0.0008 | 0.0594
1 | 1.952 0.0593 | 217.0 0.1110 | 0.0188
2 | 0.417 0.6793 | 312.0 0.9608 | 0.0038
3 | -1.322 0.1951 | 197.0 0.0533 | -0.0103
======================================================================
Statistical Test: both
======================================================================
Tests both directions combined.
Direction: both
Lag | t-stat t p-val | W-stat W p-val | Mean Diff
------+----------------------+----------------------+------------
-3 | 1.420 0.1646 | 231.0 0.1736 | 0.0107
-2 | 0.744 0.4617 | 282.0 0.5988 | 0.0053
-1 | 2.207 0.0342 | 201.0 0.0625 | 0.0138
0 | 2.951 0.0057 | 150.0 0.0060 | 0.0301
1 | 1.602 0.1185 | 230.0 0.1684 | 0.0105
2 | 2.017 0.0516 | 209.0 0.0840 | 0.0120
3 | -0.373 0.7115 | 259.0 0.3678 | -0.0027Plots show transition probabilities between neighbors of different presentations. Key patterns:
data_path — path to an HDF5 file containing a RecallDataset.figure_dir — directory for saving figures.figure_str — base filename for the saved figure. Leave empty to display without saving.ylim — y-axis limits, or None for automatic scaling.mixed_trial_query — query selecting trials with repeated items.control_trial_query — query selecting trials for the control.control_shuffles — number of shuffle iterations for building the control dataset.min_lag — minimum absolute lag included in the analysis.max_lag — maximum absolute lag displayed.use_lag2 — whether to include lag-2 neighbors in the analysis.