Skip to content

Latest commit

 

History

History
622 lines (534 loc) · 44.6 KB

File metadata and controls

622 lines (534 loc) · 44.6 KB

Standardized Interfaces

  • Schema version: 1
  • Registry fingerprint: cf7df9f21d3a0527bd676a63b010ee22a72edd16e86aa52610a2cb564cd45ce8
  • Proposal interfaces: visual_cortex, sensory_cortex, hunger_center, sleep_center, alert_center, perception_center, homeostasis_center, threat_center, alert_v1, alert_v2, alert_v3, hunger_v1, hunger_v2, hunger_v3, sensory_v1, sensory_v2, sensory_v3, sleep_v1, sleep_v2, sleep_v3, visual_v1, visual_v2, visual_v3
  • Context interfaces: action_center_context, motor_cortex_context

Compatibility

  • The current policy is exact_match_required for save/load.
  • Changes to the name, observation_key, signal order, outputs, or version require explicit incompatibility handling.
  • There is no automatic migration for older checkpoints.

Perception And Active Sensing

Breaking Change: Action Space

  • The locomotion action space expanded from 5 to 9 actions.
  • Existing indices are unchanged: MOVE_UP=0, MOVE_DOWN=1, MOVE_LEFT=2, MOVE_RIGHT=3, STAY=4.
  • New active-sensing actions are ORIENT_UP=5, ORIENT_DOWN=6, ORIENT_LEFT=7, and ORIENT_RIGHT=8.

Foveal And Peripheral Vision

  • Visual targets are classified relative to the spider heading as foveal, peripheral, or outside.
  • fov_half_angle defaults to 45.0 degrees and defines the foveal cone.
  • The peripheral_half_angle defaults to 70.0 degrees and defines the outer visual cone.
  • A default peripheral_certainty_penalty of 0.35 reduces peripheral visual certainty.
  • Targets beyond peripheral_half_angle are outside the visual field (visible=0).
  • Heading zones are not exposed as observation fields; agents receive the resulting visible, certainty, occlusion, and direction signals.
  • Smell and olfactory gradients remain omnidirectional and are not gated by heading zones.

Scan Recency

  • foveal_scan_age reports how recently the current heading was actively scanned.
  • max_scan_age defaults to 10.0 ticks for normalizing scan age into [0, 1].
  • A value of 0.0 means the current foveal heading was scanned this tick; 1.0 means stale or never scanned.
  • PerceptTrace.heading_dx and PerceptTrace.heading_dy record the body or gaze heading active when a food, shelter, or predator trace was refreshed.
  • Trace headings surface as food_trace_heading_*, shelter_trace_heading_*, and predator_trace_heading_* signals.
  • Scan recency is self-knowledge derived from the spider's own heading changes, not hidden world-state access.

ORIENT Actions

  • ORIENT_* actions update heading_dx and heading_dy from ORIENT_HEADINGS; they are not movement translations in ACTION_DELTAS.
  • After the heading update, the world refreshes the current tick's perception buffer immediately, before the normal end-of-step tick commit.
  • The refresh itself does not advance the tick or run hunger, fatigue, predator, memory, or perceptual-delay stages.
  • ORIENT_* actions still consume the step: after the refresh, the normal commit advances the tick and applies hunger, fatigue, predator, and perceptual-buffer progression.
  • ORIENT_* actions do not cause displacement and do not trigger motor slip.

Percept Traces

  • Trace signals carry target direction, decayed strength, and the heading active when the trace was refreshed.
  • *_trace_heading_dx and *_trace_heading_dy are scan context for the trace; they are not target directions.

Perceptual Delay

  • perceptual_delay_ticks defaults to 1.0 and returns observations from the previous tick when buffered.
  • perceptual_delay_noise defaults to 0.5 and scales delay-noise effects.
  • NoiseConfig.delay.certainty_decay_per_tick defaults to 0.0 and decays certainty as observations age.
  • NoiseConfig.delay.direction_jitter_per_tick defaults to 0.0 and adds age-proportional jitter to delayed direction signals.
  • Delay noise uses a dedicated RNG stream to preserve reproducibility.
  • Trace signals such as food_trace_strength reflect the delayed percept state, not the current world state.

Perception Categories

  • Direct sight means visual certainty is above the debug threshold and the current scan is fresh.
  • Trace means a decayed percept trace is active after direct sight has gone stale or disappeared.
  • Delayed means perceptual delay is active and the current payload was not refreshed by an active scan.
  • Uncertain means no direct, trace, or delayed classification applies.
  • These categories are debug and audit labels from _perception_category(); they are not observation signals.

Explicit Memory Signals

  • food_memory_dx/food_memory_dy/food_memory_age, shelter_memory_*, predator_memory_*, and escape_memory_* are derived from perception-grounded memory slots.
  • The environment pipeline maintains mechanical timing only: refresh cycles age slots, enforce TTLs, and clear expired targets.
  • Data sources are limited to local visual perception, contact events, and movement history; the pipeline cannot inject information the spider has not perceived.
  • spider_cortex_sim.memory.MEMORY_LEAKAGE_AUDIT is the authoritative classification source for explicit memory ownership and leakage risk.

Interface Variants

  • Interface variants are diagnostic sufficiency probes for local tasks and organism-level ablations; they are not production save/load interfaces.
  • Use the variant ladder to ask how much signal a module needs before escalating to broader architectural diagnosis.

visual_cortex

  • Progression: visual_v1 -> visual_v2 -> visual_v3 -> visual_cortex
Level Interface Signal Count Added At This Level
v1 visual_v1 5 food_visible, food_dx, food_dy, heading_dx, heading_dy
v2 visual_v2 11 shelter_visible, shelter_dx, shelter_dy, predator_visible, predator_dx, predator_dy
v3 visual_v3 26 foveal_scan_age, food_trace_strength, food_trace_heading_dx, food_trace_heading_dy, shelter_trace_strength, shelter_trace_heading_dx, shelter_trace_heading_dy, predator_trace_strength, predator_trace_heading_dx, predator_trace_heading_dy, predator_motion_salience, visual_predator_threat, olfactory_predator_threat, day, night
v4 visual_cortex 32 food_certainty, food_occluded, shelter_certainty, shelter_occluded, predator_certainty, predator_occluded (canonical interface)

sensory_cortex

  • Progression: sensory_v1 -> sensory_v2 -> sensory_v3 -> sensory_cortex
Level Interface Signal Count Added At This Level
v1 sensory_v1 5 recent_pain, recent_contact, predator_smell_strength, predator_smell_dx, predator_smell_dy
v2 sensory_v2 8 food_smell_strength, food_smell_dx, food_smell_dy
v3 sensory_v3 11 health, hunger, fatigue
v4 sensory_cortex 12 light (canonical interface)

hunger_center

  • Progression: hunger_v1 -> hunger_v2 -> hunger_v3 -> hunger_center
Level Interface Signal Count Added At This Level
v1 hunger_v1 5 hunger, on_food, food_visible, food_dx, food_dy
v2 hunger_v2 8 food_smell_strength, food_smell_dx, food_smell_dy
v3 hunger_v3 13 food_trace_dx, food_trace_dy, food_trace_strength, food_trace_heading_dx, food_trace_heading_dy
v4 hunger_center 18 food_certainty, food_occluded, food_memory_dx, food_memory_dy, food_memory_age (canonical interface)

sleep_center

  • Progression: sleep_v1 -> sleep_v2 -> sleep_v3 -> sleep_center
Level Interface Signal Count Added At This Level
v1 sleep_v1 3 fatigue, on_shelter, night
v2 sleep_v2 4 shelter_role_level
v3 sleep_v3 9 shelter_trace_dx, shelter_trace_dy, shelter_trace_strength, shelter_trace_heading_dx, shelter_trace_heading_dy
v4 sleep_center 18 hunger, health, recent_pain, sleep_phase_level, rest_streak_norm, sleep_debt, shelter_memory_dx, shelter_memory_dy, shelter_memory_age (canonical interface)

alert_center

  • Progression: alert_v1 -> alert_v2 -> alert_v3 -> alert_center
Level Interface Signal Count Added At This Level
v1 alert_v1 3 predator_visible, predator_dx, predator_dy
v2 alert_v2 5 predator_certainty, predator_occluded
v3 alert_v3 9 predator_smell_strength, recent_pain, recent_contact, on_shelter
v4 alert_center 27 predator_motion_salience, visual_predator_threat, olfactory_predator_threat, dominant_predator_none, dominant_predator_visual, dominant_predator_olfactory, night, predator_trace_dx, predator_trace_dy, predator_trace_strength, predator_trace_heading_dx, predator_trace_heading_dy, predator_memory_dx, predator_memory_dy, predator_memory_age, escape_memory_dx, escape_memory_dy, escape_memory_age (canonical interface)

A5 Candidate Interfaces

  • A5 candidate modules are gated by the prerequisite checklist, module admission criteria, and operating rules in architectural_ladder.md.
  • These candidate entries are protocol-defined proposal surfaces for A5_future_bio_refined; they are not part of interface_registry(), do not affect save/load compatibility, and do not change the current registry fingerprint.
  • Placeholder signals documented here are allowed at the protocol level even when the runtime does not yet expose them. They exist so each future module proposal has a concrete interface hypothesis before implementation starts.
  • homeostasis_center remains the coarse A2 control surface. Any A5 homeostatic refinement is an alternative experiment path, not an automatic extension of the current proposer stack.

olfactory_center

  • Status: proposal_only
  • Candidate input count: 9
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Connection: feeds action_center with valence role "foraging" or "threat" based on the dominant smell-driven signal. If the current runtime arbitration vocabulary is kept unchanged, "foraging" would map to the existing hunger lane.
  • Experimental hypothesis: a dedicated smell-specialized proposer will improve odor-led food approach and odor-led threat avoidance in mixed-scent conditions, especially when direct vision is weak or absent.
  • Ablation test: drop_olfactory_center must selectively reduce performance on smell-dominant tasks such as olfactory_visual_dissociation without producing the same failure pattern as drop_sensory_cortex or drop_alert_center.
  • Coupling risk: HIGH because the candidate overlaps both the current smell integrator (sensory_cortex) and smell-aware threat handling (alert_center).
# Signal Source Range Description
1 food_smell_strength sensory_cortex 0.0..1.0 Strength of the local food odor field.
2 food_smell_dx sensory_cortex -1.0..1.0 Horizontal food odor gradient.
3 food_smell_dy sensory_cortex -1.0..1.0 Vertical food odor gradient.
4 predator_smell_strength sensory_cortex + alert_center 0.0..1.0 Strength of the local predator odor field.
5 predator_smell_dx sensory_cortex -1.0..1.0 Horizontal predator odor gradient.
6 predator_smell_dy sensory_cortex -1.0..1.0 Vertical predator odor gradient.
7 olfactory_predator_threat alert_center 0.0..1.0 Aggregated threat estimate from olfactory-oriented predators.
8 dominant_predator_olfactory alert_center 0.0..1.0 Indicator that olfactory hunters are the dominant current predator type.
9 smell_ambiguity derived from sensory_cortex + alert_center 0.0..1.0 Conflict score for mixed food and predator gradients or unstable odor localization.

nociceptive_center

  • Status: proposal_only
  • Candidate input count: 8
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Connection: feeds action_center with valence role "threat" as a body-state emergency proposer.
  • Experimental hypothesis: a pain-specialized proposer will shorten escape latency after damage or contact and reduce repeated contact events after the first hit, especially when the predator is no longer directly visible.
  • Ablation test: drop_nociceptive_center must degrade pain_escape_latency or post-contact survival more than smell-led detection tasks, while remaining distinguishable from the broader drop_alert_center failure pattern.
  • Coupling risk: MEDIUM because the candidate overlaps current threat signals but proposes a narrower post-contact recovery role than alert_center.
# Signal Source Range Description
1 recent_pain sensory_cortex + alert_center 0.0..1.0 Immediate recent pain caused by a predator hit.
2 recent_contact sensory_cortex + alert_center 0.0..1.0 Immediate recent physical contact with the predator.
3 health sensory_cortex + alert_center 0.0..1.0 Current body health available for damage-sensitive retreat control.
4 damage_rate derived from health delta 0.0..1.0 Short-horizon rate of health loss after contact.
5 pain_memory_dx derived from alert_center memory traces -1.0..1.0 Horizontal direction of the last harmful contact or pain source.
6 pain_memory_dy derived from alert_center memory traces -1.0..1.0 Vertical direction of the last harmful contact or pain source.
7 pain_memory_age derived from alert_center memory traces 0.0..1.0 Age of the last harmful contact memory.
8 contact_intensity derived from recent_contact + recent_pain 0.0..1.0 Combined severity estimate for recent bodily impact.

arousal_center

  • Status: proposal_only
  • Candidate input count: 10
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Connection: feeds action_center with valence role "sleep" when arousal is low and "exploration" when arousal is high, acting as a state-transition proposer rather than a direct threat or feeding specialist.
  • Experimental hypothesis: a dedicated arousal proposer will improve sleep-wake transitions, scan initiation, and exploratory reactivation under changing circadian and threat pressure without taking over the core sheltering job of sleep_center.
  • Ablation test: drop_arousal_center must selectively hurt arousal_state_transition, timely ORIENT_* use, or post-rest reactivation without collapsing into the same deficit profile as drop_sleep_center.
  • Coupling risk: HIGH because the candidate overlaps sleep_center, contextual arbitration, and activity-state features that may already be expressible without another proposer.
# Signal Source Range Description
1 fatigue sensory_cortex + sleep_center + action_context 0.0..1.0 Immediate fatigue pressure acting against wakeful exploration.
2 sleep_debt sleep_center + action_context 0.0..1.0 Accumulated need for restorative sleep.
3 night sleep_center + action_context 0.0..1.0 Nighttime state indicator.
4 day_night_transition derived from day/night history 0.0..1.0 Recent circadian transition marker for waking or winding down.
5 arousal_level derived from sleep_center + action_context 0.0..1.0 Latent wakefulness estimate combining rest and vigilance state.
6 threat_induced_arousal derived from sensory_cortex + action_context 0.0..1.0 Transient activation caused by distant threat evidence.
7 activity_level derived from action_context.last_move_* history 0.0..1.0 Short-horizon locomotor activity or quiescence estimate.
8 metabolic_state derived from sensory_cortex body-state signals 0.0..1.0 Compact scalar summarizing available internal energy and recovery pressure.
9 circadian_phase derived from day/night cycle phase 0.0..1.0 Normalized position within the circadian cycle.
10 rest_urgency derived from sleep_center 0.0..1.0 Immediate pressure to stay in shelter or resume rest.

homeostasis_center A5 Refinement Pathway

  • Current status: homeostasis_center is the coarse A2_three_center homeostatic control with 28 canonical signals and remains the rollback target when finer splits regress.
  • Refinement admission rule: consider an A5 homeostatic refinement only if the current A4_current_full_modular split still shows reproducible interference that the existing hunger_center and sleep_center separation does not resolve.
  • Refinement topology rule: treat metabolic_regulation_center and thermoregulation_center as alternative sixth-module experiments. Do not admit both at once without new evidence, so the first A5 comparison stays modular_full versus a hypothetical 6-module proposer stack.

metabolic_regulation_center

  • Status: proposal_only, blocked unless A4 homeostatic splits prove insufficient
  • Candidate input count: 14
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Connection: would feed action_center with valence role "hunger" as a narrower metabolic-drive proposer.
# Signal Source Range Description
1 hunger homeostasis_center subset 0.0..1.0 Primary metabolic deficit signal.
2 health homeostasis_center subset 0.0..1.0 Body viability constraint on foraging persistence.
3 on_food homeostasis_center subset 0.0..1.0 Immediate feeding context.
4 food_visible homeostasis_center subset 0.0..1.0 Direct food visibility gate.
5 food_certainty homeostasis_center subset 0.0..1.0 Confidence in the visible food cue.
6 food_smell_strength homeostasis_center subset 0.0..1.0 Food odor strength.
7 food_smell_dx homeostasis_center subset -1.0..1.0 Horizontal food odor gradient.
8 food_smell_dy homeostasis_center subset -1.0..1.0 Vertical food odor gradient.
9 food_trace_dx homeostasis_center subset -1.0..1.0 Horizontal direction of the short food trace.
10 food_trace_dy homeostasis_center subset -1.0..1.0 Vertical direction of the short food trace.
11 food_trace_strength homeostasis_center subset 0.0..1.0 Strength of the short food trace.
12 food_memory_dx homeostasis_center subset -1.0..1.0 Horizontal direction of remembered food.
13 food_memory_dy homeostasis_center subset -1.0..1.0 Vertical direction of remembered food.
14 food_memory_age homeostasis_center subset 0.0..1.0 Age of remembered food.

thermoregulation_center

  • Status: proposal_only, blocked unless A4 homeostatic splits prove insufficient
  • Candidate input count: 14
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Connection: would feed action_center with valence role "sleep" as a narrower shelter and recovery proposer.
# Signal Source Range Description
1 fatigue homeostasis_center subset 0.0..1.0 Immediate fatigue pressure.
2 on_shelter homeostasis_center subset 0.0..1.0 Current shelter occupancy.
3 day homeostasis_center subset 0.0..1.0 Daytime exposure proxy.
4 night homeostasis_center subset 0.0..1.0 Nighttime sheltering proxy.
5 sleep_phase_level homeostasis_center subset 0.0..1.0 Current sleep depth or rest phase.
6 rest_streak_norm homeostasis_center subset 0.0..1.0 Recent rest continuity.
7 sleep_debt homeostasis_center subset 0.0..1.0 Accumulated unmet rest demand.
8 shelter_role_level homeostasis_center subset 0.0..1.0 Current depth inside the shelter.
9 shelter_trace_dx homeostasis_center subset -1.0..1.0 Horizontal direction of the short shelter trace.
10 shelter_trace_dy homeostasis_center subset -1.0..1.0 Vertical direction of the short shelter trace.
11 shelter_trace_strength homeostasis_center subset 0.0..1.0 Strength of the short shelter trace.
12 shelter_memory_dx homeostasis_center subset -1.0..1.0 Horizontal direction of remembered shelter.
13 shelter_memory_dy homeostasis_center subset -1.0..1.0 Vertical direction of remembered shelter.
14 shelter_memory_age homeostasis_center subset 0.0..1.0 Age of remembered shelter.
  • Refinement experimental hypothesis: only pursue an A5 homeostatic refinement when the current split still shows a reproducible interference pattern, and the candidate sixth-module configuration improves that pattern with Cohen's d >= 0.3 while preserving inherited gates.
  • Ablation design: compare modular_full against a hypothetical 6-module A5 configuration that introduces exactly one homeostatic refinement module at a time, then require the added module's removal or rollback comparison to isolate the intended interference reduction under the same Cohen's d >= 0.3 admission threshold.
  • Coupling risk: HIGH because the current A4 architecture already decomposes coarse homeostasis into hunger_center and sleep_center, so additional splitting is likely to duplicate existing behavior unless the interference pattern is cleanly separable.

visual_cortex

  • Observation key: visual
  • Role: proposal
  • Version: 7
  • Description: Visual proposer driven by food, shelter, and predator cues inside the local visual field.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 food_visible -1.0 1.0 1 when food is detected with sufficient confidence in the visual field.
2 food_certainty 0.0 1.0 Visual confidence in the perceived food target.
3 food_occluded 0.0 1.0 1 when the nearest food is within range but occluded.
4 food_dx -1.0 1.0 Relative horizontal offset of the detected food.
5 food_dy -1.0 1.0 Relative vertical offset of the detected food.
6 shelter_visible -1.0 1.0 1 when the shelter is visible with sufficient confidence.
7 shelter_certainty 0.0 1.0 Visual confidence in the perceived shelter.
8 shelter_occluded 0.0 1.0 1 when the shelter is within visual range but occluded.
9 shelter_dx -1.0 1.0 Relative horizontal offset of the detected shelter.
10 shelter_dy -1.0 1.0 Relative vertical offset of the detected shelter.
11 predator_visible -1.0 1.0 1 when the predator is visible with sufficient confidence.
12 predator_certainty 0.0 1.0 Visual confidence in the perceived predator.
13 predator_occluded 0.0 1.0 1 when the predator is within visual range but occluded.
14 predator_dx -1.0 1.0 Relative horizontal offset of the detected predator.
15 predator_dy -1.0 1.0 Relative vertical offset of the detected predator.
16 heading_dx -1.0 1.0 Current horizontal body or gaze orientation.
17 heading_dy -1.0 1.0 Current vertical body or gaze orientation.
18 foveal_scan_age 0.0 1.0 Normalized ticks since the current foveal heading was actively scanned; 0 is fresh and 1 is stale or never scanned.
19 food_trace_strength 0.0 1.0 Decayed strength of the short food trace.
20 food_trace_heading_dx -1.0 1.0 Horizontal heading active when the short food trace was refreshed.
21 food_trace_heading_dy -1.0 1.0 Vertical heading active when the short food trace was refreshed.
22 shelter_trace_strength 0.0 1.0 Decayed strength of the short shelter trace.
23 shelter_trace_heading_dx -1.0 1.0 Horizontal heading active when the short shelter trace was refreshed.
24 shelter_trace_heading_dy -1.0 1.0 Vertical heading active when the short shelter trace was refreshed.
25 predator_trace_strength 0.0 1.0 Decayed strength of the short predator trace.
26 predator_trace_heading_dx -1.0 1.0 Horizontal heading active when the short predator trace was refreshed.
27 predator_trace_heading_dy -1.0 1.0 Vertical heading active when the short predator trace was refreshed.
28 predator_motion_salience 0.0 1.0 Explicit motion salience of the predator.
29 visual_predator_threat 0.0 1.0 Aggregated threat from visually oriented predators.
30 olfactory_predator_threat 0.0 1.0 Aggregated threat from olfactory predators.
31 day -1.0 1.0 1 during daytime.
32 night -1.0 1.0 1 during nighttime.

sensory_cortex

  • Observation key: sensory
  • Role: proposal
  • Version: 2
  • Description: Sensory proposer that integrates pain, body state, smell, and light.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 recent_pain 0.0 1.0 Recent pain caused by a predator attack.
2 recent_contact 0.0 1.0 Recent physical contact with the predator.
3 health 0.0 1.0 Body health.
4 hunger 0.0 1.0 Body hunger.
5 fatigue 0.0 1.0 Body fatigue.
6 food_smell_strength 0.0 1.0 Intensity of the food scent.
7 food_smell_dx -1.0 1.0 Horizontal gradient of the food scent.
8 food_smell_dy -1.0 1.0 Vertical gradient of the food scent.
9 predator_smell_strength 0.0 1.0 Intensity of the predator scent.
10 predator_smell_dx -1.0 1.0 Horizontal gradient of the predator scent.
11 predator_smell_dy -1.0 1.0 Vertical gradient of the predator scent.
12 light 0.0 1.0 Simple light level.

hunger_center

  • Observation key: hunger
  • Role: proposal
  • Version: 4
  • Description: Homeostatic proposer aimed at foraging and returning to the last perceived food.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 hunger 0.0 1.0 Internal hunger state.
2 on_food 0.0 1.0 1 when the spider is standing on food.
3 food_visible 0.0 1.0 1 when food is visible.
4 food_certainty 0.0 1.0 Visual confidence in the perceived food.
5 food_occluded 0.0 1.0 1 when nearby food exists but is occluded.
6 food_dx -1.0 1.0 Horizontal direction of the detected food.
7 food_dy -1.0 1.0 Vertical direction of the detected food.
8 food_smell_strength 0.0 1.0 Intensity of the food scent.
9 food_smell_dx -1.0 1.0 Horizontal gradient of the food scent.
10 food_smell_dy -1.0 1.0 Vertical gradient of the food scent.
11 food_trace_dx -1.0 1.0 Horizontal direction of the short food trace.
12 food_trace_dy -1.0 1.0 Vertical direction of the short food trace.
13 food_trace_strength 0.0 1.0 Decayed strength of the short food trace.
14 food_trace_heading_dx -1.0 1.0 Horizontal heading active when the short food trace was refreshed.
15 food_trace_heading_dy -1.0 1.0 Vertical heading active when the short food trace was refreshed.
16 food_memory_dx -1.0 1.0 Horizontal direction of the last seen food.
17 food_memory_dy -1.0 1.0 Vertical direction of the last seen food.
18 food_memory_age 0.0 1.0 Normalized age of the food memory.

sleep_center

  • Observation key: sleep
  • Role: proposal
  • Version: 5
  • Description: Homeostatic proposer aimed at returning to shelter, resting, and seeking safe depth.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 fatigue 0.0 1.0 Internal fatigue state.
2 hunger 0.0 1.0 Internal hunger state.
3 on_shelter 0.0 1.0 1 when the spider is on the shelter.
4 night 0.0 1.0 1 during nighttime.
5 health 0.0 1.0 Body health.
6 recent_pain 0.0 1.0 Recent pain.
7 sleep_phase_level 0.0 1.0 Current sleep phase level.
8 rest_streak_norm 0.0 1.0 Recent rest continuity.
9 sleep_debt 0.0 1.0 Accumulated sleep debt.
10 shelter_role_level 0.0 1.0 Current depth inside the shelter.
11 shelter_trace_dx -1.0 1.0 Horizontal direction of the short shelter trace.
12 shelter_trace_dy -1.0 1.0 Vertical direction of the short shelter trace.
13 shelter_trace_strength 0.0 1.0 Decayed strength of the short shelter trace.
14 shelter_trace_heading_dx -1.0 1.0 Horizontal heading active when the short shelter trace was refreshed.
15 shelter_trace_heading_dy -1.0 1.0 Vertical heading active when the short shelter trace was refreshed.
16 shelter_memory_dx -1.0 1.0 Horizontal direction of the nearest visible shelter cell.
17 shelter_memory_dy -1.0 1.0 Vertical direction of the nearest visible shelter cell.
18 shelter_memory_age 0.0 1.0 Normalized age of the nearest visible shelter cell memory.

alert_center

  • Observation key: alert
  • Role: proposal
  • Version: 8
  • Description: Defensive proposer aimed at threat response, escape, and shelter prioritization under risk.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 predator_visible 0.0 1.0 1 when the predator is visible.
2 predator_certainty 0.0 1.0 Visual confidence in the perceived predator.
3 predator_occluded 0.0 1.0 1 when the predator is within range but occluded.
4 predator_dx -1.0 1.0 Horizontal direction of the detected predator.
5 predator_dy -1.0 1.0 Vertical direction of the detected predator.
6 predator_smell_strength 0.0 1.0 Intensity of the predator scent.
7 predator_motion_salience 0.0 1.0 Explicit motion salience of the predator.
8 visual_predator_threat 0.0 1.0 Aggregated threat from visually oriented predators.
9 olfactory_predator_threat 0.0 1.0 Aggregated threat from olfactory predators.
10 dominant_predator_none 0.0 1.0 1 when no predator type is currently dominant.
11 dominant_predator_visual 0.0 1.0 1 when visual predators are the dominant threat type.
12 dominant_predator_olfactory 0.0 1.0 1 when olfactory predators are the dominant threat type.
13 recent_pain 0.0 1.0 Recent pain.
14 recent_contact 0.0 1.0 Recent physical contact.
15 on_shelter 0.0 1.0 1 when the spider is on the shelter.
16 night 0.0 1.0 1 during nighttime.
17 predator_trace_dx -1.0 1.0 Horizontal direction of the short predator trace.
18 predator_trace_dy -1.0 1.0 Vertical direction of the short predator trace.
19 predator_trace_strength 0.0 1.0 Decayed strength of the short predator trace.
20 predator_trace_heading_dx -1.0 1.0 Horizontal heading active when the short predator trace was refreshed.
21 predator_trace_heading_dy -1.0 1.0 Vertical heading active when the short predator trace was refreshed.
22 predator_memory_dx -1.0 1.0 Horizontal direction of the predator position perceived during visual detection or inferred from contact event.
23 predator_memory_dy -1.0 1.0 Vertical direction of the predator position perceived during visual detection or inferred from contact event.
24 predator_memory_age 0.0 1.0 Normalized age of the predator position perceived during visual detection or inferred from contact event.
25 escape_memory_dx -1.0 1.0 Horizontal direction of the recent escape target derived from movement history without walkability assumptions.
26 escape_memory_dy -1.0 1.0 Vertical direction of the recent escape target derived from movement history without walkability assumptions.
27 escape_memory_age 0.0 1.0 Normalized age of the escape memory derived from movement history without walkability assumptions.

perception_center

  • Observation key: perception
  • Role: proposal
  • Version: 1
  • Description: Coarse perception proposer that fuses visual, olfactory, trace, and perceptual-memory cues.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 food_visible 0.0 1.0 1 when food is visible.
2 food_certainty 0.0 1.0 Visual confidence in the perceived food.
3 food_dx -1.0 1.0 Horizontal direction of the detected food.
4 food_dy -1.0 1.0 Vertical direction of the detected food.
5 shelter_visible 0.0 1.0 1 when shelter is visible.
6 shelter_certainty 0.0 1.0 Visual confidence in the perceived shelter.
7 shelter_dx -1.0 1.0 Horizontal direction of the detected shelter.
8 shelter_dy -1.0 1.0 Vertical direction of the detected shelter.
9 predator_visible 0.0 1.0 1 when a predator is visible.
10 predator_certainty 0.0 1.0 Visual confidence in the perceived predator.
11 predator_dx -1.0 1.0 Horizontal direction of the detected predator.
12 predator_dy -1.0 1.0 Vertical direction of the detected predator.
13 heading_dx -1.0 1.0 Current horizontal body or gaze orientation.
14 heading_dy -1.0 1.0 Current vertical body or gaze orientation.
15 foveal_scan_age 0.0 1.0 Normalized ticks since the current foveal heading was actively scanned.
16 food_smell_strength 0.0 1.0 Intensity of the food scent.
17 food_smell_dx -1.0 1.0 Horizontal gradient of the food scent.
18 food_smell_dy -1.0 1.0 Vertical gradient of the food scent.
19 predator_smell_strength 0.0 1.0 Intensity of the predator scent.
20 predator_smell_dx -1.0 1.0 Horizontal gradient of the predator scent.
21 predator_smell_dy -1.0 1.0 Vertical gradient of the predator scent.
22 light 0.0 1.0 Simple light level.
23 day 0.0 1.0 1 during daytime.
24 night 0.0 1.0 1 during nighttime.
25 food_trace_strength 0.0 1.0 Decayed strength of the short food trace.
26 food_trace_heading_dx -1.0 1.0 Horizontal heading active when the short food trace was refreshed.
27 food_trace_heading_dy -1.0 1.0 Vertical heading active when the short food trace was refreshed.
28 shelter_trace_strength 0.0 1.0 Decayed strength of the short shelter trace.
29 shelter_trace_heading_dx -1.0 1.0 Horizontal heading active when the short shelter trace was refreshed.
30 shelter_trace_heading_dy -1.0 1.0 Vertical heading active when the short shelter trace was refreshed.
31 predator_trace_strength 0.0 1.0 Decayed strength of the short predator trace.
32 predator_trace_heading_dx -1.0 1.0 Horizontal heading active when the short predator trace was refreshed.
33 predator_trace_heading_dy -1.0 1.0 Vertical heading active when the short predator trace was refreshed.
34 food_memory_dx -1.0 1.0 Horizontal direction of the last seen food.
35 food_memory_dy -1.0 1.0 Vertical direction of the last seen food.
36 food_memory_age 0.0 1.0 Normalized age of the food memory.
37 shelter_memory_dx -1.0 1.0 Horizontal direction of the remembered shelter.
38 shelter_memory_dy -1.0 1.0 Vertical direction of the remembered shelter.
39 shelter_memory_age 0.0 1.0 Normalized age of the shelter memory.
40 predator_memory_dx -1.0 1.0 Horizontal direction of the remembered predator.
41 predator_memory_dy -1.0 1.0 Vertical direction of the remembered predator.
42 predator_memory_age 0.0 1.0 Normalized age of the predator memory.

homeostasis_center

  • Observation key: homeostasis
  • Role: proposal
  • Version: 1
  • Description: Coarse homeostatic proposer that blends hunger, fatigue, sheltering, and recovery drives.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 hunger 0.0 1.0 Internal hunger state.
2 fatigue 0.0 1.0 Internal fatigue state.
3 health 0.0 1.0 Body health.
4 on_food 0.0 1.0 1 when the spider is standing on food.
5 on_shelter 0.0 1.0 1 when the spider is on the shelter.
6 day 0.0 1.0 1 during daytime.
7 night 0.0 1.0 1 during nighttime.
8 sleep_phase_level 0.0 1.0 Current sleep phase level.
9 rest_streak_norm 0.0 1.0 Recent rest continuity.
10 sleep_debt 0.0 1.0 Accumulated sleep debt.
11 shelter_role_level 0.0 1.0 Current depth inside the shelter.
12 food_visible 0.0 1.0 1 when food is visible.
13 food_certainty 0.0 1.0 Visual confidence in the perceived food.
14 food_smell_strength 0.0 1.0 Intensity of the food scent.
15 food_smell_dx -1.0 1.0 Horizontal gradient of the food scent.
16 food_smell_dy -1.0 1.0 Vertical gradient of the food scent.
17 food_trace_dx -1.0 1.0 Horizontal direction of the short food trace.
18 food_trace_dy -1.0 1.0 Vertical direction of the short food trace.
19 food_trace_strength 0.0 1.0 Decayed strength of the short food trace.
20 food_memory_dx -1.0 1.0 Horizontal direction of the last seen food.
21 food_memory_dy -1.0 1.0 Vertical direction of the last seen food.
22 food_memory_age 0.0 1.0 Normalized age of the food memory.
23 shelter_trace_dx -1.0 1.0 Horizontal direction of the short shelter trace.
24 shelter_trace_dy -1.0 1.0 Vertical direction of the short shelter trace.
25 shelter_trace_strength 0.0 1.0 Decayed strength of the short shelter trace.
26 shelter_memory_dx -1.0 1.0 Horizontal direction of the remembered shelter.
27 shelter_memory_dy -1.0 1.0 Vertical direction of the remembered shelter.
28 shelter_memory_age 0.0 1.0 Normalized age of the shelter memory.

threat_center

  • Observation key: threat
  • Role: proposal
  • Version: 1
  • Description: Coarse threat proposer that blends threat sensing, pain/contact, and escape memory.
  • Outputs: MOVE_UP, MOVE_DOWN, MOVE_LEFT, MOVE_RIGHT, STAY, ORIENT_UP, ORIENT_DOWN, ORIENT_LEFT, ORIENT_RIGHT
  • Save policy: exact_match_required
# Signal Min Max Description
1 predator_visible 0.0 1.0 1 when a predator is visible.
2 predator_certainty 0.0 1.0 Visual confidence in the perceived predator.
3 predator_dx -1.0 1.0 Horizontal direction of the detected predator.
4 predator_dy -1.0 1.0 Vertical direction of the detected predator.
5 predator_smell_strength 0.0 1.0 Intensity of the predator scent.
6 predator_smell_dx -1.0 1.0 Horizontal gradient of the predator scent.
7 predator_smell_dy -1.0 1.0 Vertical gradient of the predator scent.
8 predator_motion_salience 0.0 1.0 Explicit motion salience of the predator.
9 visual_predator_threat 0.0 1.0 Aggregated threat from visually oriented predators.
10 olfactory_predator_threat 0.0 1.0 Aggregated threat from olfactory predators.
11 dominant_predator_none 0.0 1.0 1 when no predator type is currently dominant.
12 dominant_predator_visual 0.0 1.0 1 when visual predators are the dominant threat type.
13 dominant_predator_olfactory 0.0 1.0 1 when olfactory predators are the dominant threat type.
14 recent_pain 0.0 1.0 Recent pain.
15 recent_contact 0.0 1.0 Recent physical contact.
16 health 0.0 1.0 Body health.
17 on_shelter 0.0 1.0 1 when the spider is on the shelter.
18 night 0.0 1.0 1 during nighttime.
19 predator_trace_dx -1.0 1.0 Horizontal direction of the short predator trace.
20 predator_trace_dy -1.0 1.0 Vertical direction of the short predator trace.
21 predator_trace_strength 0.0 1.0 Decayed strength of the short predator trace.
22 predator_memory_dx -1.0 1.0 Horizontal direction of the remembered predator.
23 predator_memory_dy -1.0 1.0 Vertical direction of the remembered predator.
24 predator_memory_age 0.0 1.0 Normalized age of the predator memory.
25 escape_memory_dx -1.0 1.0 Horizontal direction of the remembered escape target.
26 escape_memory_dy -1.0 1.0 Vertical direction of the remembered escape target.
27 escape_memory_age 0.0 1.0 Normalized age of the escape memory.

action_center_context

  • Observation key: action_context
  • Role: context
  • Version: 3
  • Description: Raw context used by the action_center to arbitrate competing locomotion proposals.
  • Outputs: ``
  • Save policy: exact_match_required
# Signal Min Max Description
1 hunger 0.0 1.0 Body hunger.
2 fatigue 0.0 1.0 Body fatigue.
3 health 0.0 1.0 Body health.
4 recent_pain 0.0 1.0 Recent pain.
5 recent_contact 0.0 1.0 Recent contact with the predator.
6 on_food 0.0 1.0 1 when standing on food.
7 on_shelter 0.0 1.0 1 when standing on the shelter.
8 predator_visible 0.0 1.0 1 when the predator is visible.
9 predator_certainty 0.0 1.0 Visual confidence in the predator.
10 day 0.0 1.0 1 during daytime.
11 night 0.0 1.0 1 during nighttime.
12 last_move_dx -1.0 1.0 Last executed horizontal displacement.
13 last_move_dy -1.0 1.0 Last executed vertical displacement.
14 sleep_debt 0.0 1.0 Accumulated sleep debt.
15 shelter_role_level 0.0 1.0 Current depth inside the shelter.

motor_cortex_context

  • Observation key: motor_context
  • Role: context
  • Version: 5
  • Description: Raw context used by the motor_cortex to execute locomotion intent within local embodiment constraints.
  • Outputs: ``
  • Save policy: exact_match_required
# Signal Min Max Description
1 on_food 0.0 1.0 1 when standing on food.
2 on_shelter 0.0 1.0 1 when standing on the shelter.
3 predator_visible 0.0 1.0 1 when the predator is visible.
4 predator_certainty 0.0 1.0 Visual confidence in the predator.
5 day 0.0 1.0 1 during daytime.
6 night 0.0 1.0 1 during nighttime.
7 last_move_dx -1.0 1.0 Last executed horizontal displacement.
8 last_move_dy -1.0 1.0 Last executed vertical displacement.
9 shelter_role_level 0.0 1.0 Current depth inside the shelter.
10 heading_dx -1.0 1.0 Current horizontal body orientation.
11 heading_dy -1.0 1.0 Current vertical body orientation.
12 terrain_difficulty 0.0 1.0 Local movement cost from the terrain under the spider.
13 fatigue 0.0 1.0 Body fatigue affecting execution reliability.
14 momentum 0.0 1.0 Bounded execution momentum state; motor-only body state, not decision-priority context.