Simulacra Naturae: Bio-Hybrid Generative Ecosystem

• Simulacra Naturae is a hybrid generative ecosystem where pre-recorded organoid neural signals co-create a dynamic, multisensory environment.
• It employs advanced signal processing and agent-based models to map 131-channel neural data onto real-time interactive simulations and material installations.
• The system redefines collective agency by merging biological rhythms, AI algorithms, and natural elements to foster ethical, multispecies co-creation.

Simulacra Naturae is a hybrid generative environment in which pre-recorded electrophysiological signals from human brain organoids act as the co-creative substrate for a real-time, multi-modal ecosystem composed of agent-based simulations, spatial audio, living plant matter, and computationally fabricated clay artifacts. The installation abandons the conventional paradigm of biosignal-controlled interfaces, instead positing organoid neural rhythms as agential “simulacra” that materially shape the emergence, topology, and behavior of a 54-square-meter artificial “forest” inhabited by algorithmic lifeforms. This system foregrounds a distributed collective agency among neurons, algorithms, material objects, and plants, with the goal of re-materializing abstract neurophysiology into tangible, multisensory phenomena and establishing visualization as a practice of care within a bio-hybrid, more-than-human ecology (Manoudaki et al., 3 Sep 2025).

1. Generative System Conceptualization

Simulacra Naturae positions electrophysiological spikes from 131-channel iPSC-derived human brain organoids as active “simulacra”—not direct control signals but dynamic, latent determinants that co-compose the environment alongside artificial and biological agents. Neural events do not function as mere triggers or drivers; rather, they act in parallel with agent-based behavioral models (stigmergic termites, Physarum-inspired slime mold, and flocking boids), acoustic systems, light, and living substrate, creating a shared field in which agency is thoroughly decentralized. The system’s scale and physical instantiation—an immersive 9 × 6 m sensory “forest” featuring AI-generated visuals, 16.2-channel audio, and embedded ceramics—advances the discourse on ecological attunement and collective authorship among heterogeneous actants. The central innovation is the real-time mapping pipeline spanning 131 neural streams and up to 60 million GPU agents, dynamically reconfiguring the material and computational substrate in correspondence with the temporal structure of organoid activity (Manoudaki et al., 3 Sep 2025).

2. Neural Signal Processing and Mapping

2.1 Data Acquisition and Preprocessing

Neural data are acquired from iPSC-derived brain organoids interfaced via high-density CMOS microelectrode arrays at 20 kHz sampling rates. Raw extracellular voltage traces are processed using Kilosort2 to segment and sort spikes, yielding 131 active neuron channels. Spike times $t_{i,k}$ for neuron $i$ are registered with 1 ms resolution.

Population firing rate, $\mathrm{FR}(t)$, is computed as

$$\mathrm{FR}(t) = \frac{1}{\Delta T}\sum_{i=1}^{131}\sum_{k} \mathbf{1}_{[t_{i,k}\in (t-\Delta T/2,\;t+\Delta T/2)]}$$

utilizing a sliding window of $\Delta T = 100$ ms. Dense clusters of spike events delineate burst event markers.

2.2 Mapping to Simulation Parameters

Neural signals are further mapped to agent behavior and system outputs via defined functions. For each neuron $i$, the spike train is

$s_i(t) = \sum_{k} \delta\bigl(t-t_{i,k}\bigr)$

and key translation rules are as follows:

• Agent Speed: $$v_i(t) = v_{\mathrm{base}} + \alpha_v (s_i * h_v)(t)$$, where $*$ indicates convolution with a causal exponential kernel $h_v$.
• Trail Deposit Rate: $i$0 for chemotactic/pheromonal models.
• Audio Synthesis: Acoustic texture densities for sustained tones and granular clouds are modulated by functions of $i$1, e.g., $i$2.

This mapping orchestrates bi-directional co-fluctuations across computational and material layers, such that periods of high neural bursting precipitate marked shifts in agent collective states, audio timbre, ceramic activations, and visual dynamics (Manoudaki et al., 3 Sep 2025).

3. Agent-Based Simulation Formalism

3.1 Stigmergic Termites

A distributed set of $i$3 agents is governed by state $i$4 and interacts via a global pheromone field $i$5 described by

$i$6

where agents sense local gradients to update heading and deposit neural-modulated spikes, resulting in stigmergic emergent trail networks.

3.2 Physarum-Inspired Slime Molds

Agents sense and deposit in a chemical field $i$7, update heading towards maximal local concentrations, and move with $i$8. Collective behavior forms labyrinthine networks reflecting neural modularity.

3.3 Flocking Boids

Classic Reynolds-style rules operate on positions and velocities $i$9 with agent-agent force weights (alignment, cohesion, separation) modulated by $\mathrm{FR}(t)$0. Maximum scale is $\mathrm{FR}(t)$150,000 boids in real time because of $\mathrm{FR}(t)$2 neighbor search complexity.

3.4 Performance

Simulation	Agents	Complexity	Runtime (ms/frame)
Termites	50 million	O(n)	11 ms
Physarum	60 million	O(n)	14 ms
Boids	50 thousand	O(n²)	9 ms

GPU compute shaders are employed for real-time execution, leveraging frame-accurate synchronization with all material outputs (Manoudaki et al., 3 Sep 2025).

4. System Architecture and Material Integration

A distributed orchestration stack underpins the ecosystem:

• Software: TouchDesigner (master clock, organoid parsing), Unity and Processing (a-life render), Max/MSP (audio, solenoids), OSC and MQTT for synchronization, Stable Diffusion 2.1 + LoRA for AI-based floor imagery.
• Hardware: Dual RTX 4090 workstations (10 Gbps LAN), 4K laser projection (floor, 3×4K wall), 27 solenoid-embedded ceramic vessels (Arduino/MQTT), RGB LED matrix fiber-optics, and a 16.2-channel audio array.
• Synchronization: All time-dependent modalities (visuals, audio, solenoid strikes, fiber-optics) are aligned through a shared millisecond master clock originating in TouchDesigner, with inter-process communication over OSC/MQTT and NDI texture streaming.

Material ecologies are referenced directly: Monstera, Alocasia, and related taxa are placed following a planar abstraction of 27 neuron clusters; vessel design is procedurally generated by a Python/Grasshopper differential growth algorithm, then coil-printed and installed with one-to-one mappings between neuro-clusters and solenoid actuation (Manoudaki et al., 3 Sep 2025).

5. Emergent Behavioral Dynamics

System output is characterized by continuous, emergent phenomena resulting from the high-dimensional interplay of neural, artificial, and material signals:

• At low firing rates, stigmergic agents display sparse, modular trail formation, and flocking agents aggregate in small flocks, paralleling organoid neural modules.
• During burst phases, the Physarum model exhibits dense, interconnected networks and audio transitions to high-density, sustained textures; solenoid strikes accelerate in synchrony with spike bursts.
• Case analyses over 10-minute burst intervals show $\mathrm{FR}(t)$3 increasing 3×, translating to a doubling of sustained-tone audio density, a halving of granular density, and major topological transitions in both digital substrates (expansion of trail hubs, flocking vortices) and material feedback.

These phenomena are not scripted but emerge from the negotiation of mapped neural dynamics and agent-based rule sets (Manoudaki et al., 3 Sep 2025).

6. Ethical, Ecological, and Experiential Dimensions

Simulacra Naturae articulates a theory and praxis of distributed responsibility and care:

• Neural data are not mined or surveilled but “let to speak for themselves” as co-creative actants. No live organoid manipulation occurs; all signals are pre-recorded—an explicit ethical boundary to avoid biological interference.
• Plants, clay vessels, sound, and light are configured as autonomous ecologies, foregrounding a multispecies, more-than-human agency.
• Participants traverse an immersive, dynamically shifting terrain designed to promote contemplative, synesthetic embodied experience, facilitating a distributed, non-anthropocentric understanding of cognition.

This approach foregrounds ethics, empathy, and ecological attunement within bio-hybrid creative systems, proposing a reconfiguration of visualization from interpretation toward care and co-evolution (Manoudaki et al., 3 Sep 2025).

7. Mathematical Summary and Key Equations

Several formal mappings are central:

• Population firing rate:

$\mathrm{FR}(t)$4

• Agent speed modulation:

$\mathrm{FR}(t)$5

• Pheromone update with neural-driven deposition:

$\mathrm{FR}(t)$6

These formalizations provide a template for future computational ecologies grounded in biosignals, enabling real-time, large-scale distributed emergence (Manoudaki et al., 3 Sep 2025).