What a session saves

A session writes everything it produces into one folder, including a manifest that records the configuration used and the units of every column.

Sessions are saved under:

base_output_path / experiment_mode / [--day label] / time_<timestamp>_<mouseID> /

Examples:

maze_recordings/grammar/habituation/time_2026-05-23_10_00_00_mouse1/
maze_recordings/grammar/day_1/time_2026-05-23_14_30_00_mouse1/
maze_recordings/grammar/day_2/time_2026-05-24_10_00_00_mouse1/

Figures are generated automatically at the end of every session and saved inside the session folder alongside the CSVs. To regenerate them for sessions already collected:

# Single session
amaze-analyse-session "C:\path\to\session_folder"

# Multiple sessions at once
amaze-analyse-session "C:\path\to\session1" "C:\path\to\session2"

File	What it shows
fig1_arm_totals.png	Total time (min) and visit count per arm, coloured by stimulus type
fig2_ee_vs_sc.png	EE vs SC arms grouped by predictability tier - time and visits
fig3_block_evolution.png	Time per stimulus category across active blocks - checks whether preference shifts
fig4_visit_duration.png	Boxplot of individual visit durations per stimulus type
fig5_maze_time.png	Total time inside the maze per trial block
fig6_location_preference.png	Heatmap of time per arm per block with stimulus labels - distinguishes location bias from stimulus preference

After running all mice for a day (or across multiple days), generate summary figures with run_summary_analysis.py. Saved into the folder you pass.

# All mice on one day
amaze-summary --day "C:\...\maze_recordings\grammar\day_1"

# All mice across all days collected so far
amaze-summary --all "C:\...\maze_recordings\grammar"

Silent-baseline sessions are automatically excluded - only active test-day sessions contribute.

EE vs SC preference:

File	What it shows
summary_A_ee_sc_per_mouse.png	EE vs SC total time - one pair of bars per mouse, one panel per day
summary_B_preference_index.png	EE preference index (−1 to +1) per mouse; positive = EE preference
summary_C_group_summary.png	Group mean ± SEM time and visit count on EE vs SC arms per day
summary_D_cross_day_pi.png	PI trajectory per mouse + group mean ± SEM across days (multi-day only)

Predictive complexity (dominant / secondary / rare):

File	What it shows
summary_E_tier_breakdown_per_mouse.png	Stacked bars per mouse: EE bar and SC bar each split by tier (dark → light = dominant → rare)
summary_F_group_tier_breakdown.png	Group mean ± SEM for all 6 tier × environment combinations
summary_G_cross_day_tiers.png	Per-tier preference across days - group mean ± SEM for each complexity level, EE and SC panels (multi-day only)

---