Recorded ecological growth. All 48 genotype–habitat worlds and four GAM-proposed confirmations completed 32 model seconds of material-funded construction. Compare inherited branching programs, scrub through actual committed structures, and inspect conditional GAM predictions in the new growth atlas ↗. These ecological runs used neutral body controls, without a CNS.
Current V5 private synapses. The full CNS, articulated bodies and private memory ran together for 128 physical ticks. Actual sensory history produced small efficacy changes on all 4,184 selected KC→MBON11 connections in each resident. Whole-life replay was exact, and the synaptic inspector did not alter state. This establishes the coupled mechanism, not learned walking or associative preference. Joined receipt ↗
Earlier V2 CNS response. With exactly matched initial worlds, a physical film screen changed 140,200 of 149,782 non-afferent neuron rows at some point during 30 model seconds, above a 10⁻⁶ difference threshold. Actions and trajectories also diverged. This is a graded synthetic model with an initialized controller, not measured fly firing or film understanding. Experiment and controls ↗
Earlier native physical integration. Eight articulated bodies shared 21 material regions, 29 routes and six outlets for 1,200 physical ticks. All six designated outlets activated finite packets; a complete packet exit conserved its contents and retired its physical body. Growth committed 12 times, and the coupled state restored exactly. This run used analyst-supplied actions. Executed receipt ↗
Reciprocal-wave recording. The embedded 360-frame episode is an authenticated research continuation from a coherent tick boundary, not a fresh birth. It exposes eight residents' recorded sensory and physical streams plus exact typed event receipts without exposing private memory content.
Two completed reciprocal waves. All 160 assigned research lives completed 2,048 ticks across ten environments: 327,680 resident transitions. The principal wave took 986 supervised seconds; the selected challenge wave took 969.03 seconds. They recorded five and sixteen mouth-contact ticks respectively. The cohorts differ, so this is not a measured improvement in feeding.
Connected evidence. The native Weave graph now contains 9,753 nodes and 28,802 edges, linking the earlier campaign, fitted mechanisms, explicit research branches and 9,081 recorded world events. Original failed histories remain preserved.
Archive and fits. After both waves, the population search holds 74 entries in 21 archive cells, informed by 160 completed life records. Fitted GAM models contribute bounded action scores and private corrections. Their predictive accuracy does not by itself establish a behavioral benefit. The inherited research controller is downloadable.
What remains open. This recording establishes that the joined mechanisms ran and records their physical consequences. A longer private continuation later paused after a material packet escaped the finite terrain; its checkpoint and diagnosis are preserved. It does not by itself demonstrate learned feeding, cross-feeding, sustained regulation, social competence, reproductive success or evolutionary improvement.
Earlier recordings. The preceding trained-organ episode, v4 founder region, and courtyard life remain pinned records of their own mechanisms and times.