Towards Lateral Inhibition and Collective Perception in Unorganised Non-Neural Systems

Could simple organisms such as slime mould approximate LI without recourse to neural tissue? We describe a model whereby LI can emerge without explicit inhibitory wiring, using only bulk transport effects. We use a multi-agent model of slime mould to reproduce the char- acteristic edge contrast amplification effects of LI using excitation via attractant based stimuli. We also explore a counterpart behaviour, Lateral Activation (where stimulated regions are inhibited and lateral regions are excited), using simulated exposure to light irradiation. In both cases restoration of baseline activity occurs when the stimuli are removed. In addition to the enhancement of local edge contrast the long-term change in population density distribution corresponds to a collective response to the global brightness of 2D image stimuli, including the scalloped inten- sity profile of the Chevreul staircase and the perceived difference of two identically bright patches in the Simultaneous Brightness Contrast (SBC) effect. This simple modelapproximatesLIcontrastenhancementphenomenaandglobalbrightnessper- ception in collective unorganised systems without fixed neural architectures. This may encourage further research into unorganised analogues of neural processes in simple organisms and suggests novel mechanisms to generate collective perception of contrast and brightness in distributed computing and robotic devices.

💡 Research Summary

The paper investigates whether a non‑neural organism—specifically the slime mould Physarum polycephalum—can exhibit phenomena analogous to neural lateral inhibition (LI), a mechanism that enhances contrast by exciting stimulated neurons while suppressing their immediate neighbours. To explore this, the authors employ a multi‑agent particle model that captures the essential physical processes of the plasmodium: local sensing, movement, and deposition of a chemoattractant that diffuses across a two‑dimensional lattice. Each particle possesses three forward‑biased sensors; based on the strongest sensed chemoattractant gradient it rotates (parameterized by sensor angle SA and rotation angle RA) and attempts to move one pixel forward, depositing chemoattractant on successful steps. The collective behaviour of thousands of such particles reproduces the bulk flow of protoplasm observed in real slime mould.

Two types of stimuli are introduced. An “excitatory” stimulus is modelled by locally increasing the chemoattractant concentration, thereby drawing particles toward the region. An “inhibitory” stimulus mimics light avoidance by reducing sensor sensitivity and chemoattractant deposition in illuminated zones. In the first set of experiments a horizontal tube‑shaped arena (300 × 100 pixels) is populated with 8 000 particles. When a central third of the arena receives an attractant pulse, particles accumulate inside the stimulus region, raising local density, while density drops in the surrounding zones. This creates a sharp contrast at the stimulus borders, directly analogous to neural LI where the excited region is amplified and neighbours are suppressed. Upon removal of the stimulus, the particle distribution gradually returns to a uniform baseline, mirroring the reversal of neural activity after the removal of a visual stimulus.

The second experiment replaces the attractant with a simulated light field. Particles at the illuminated border preferentially move away, producing a “lateral activation” effect: the illuminated region becomes depleted of particles (inhibition) while the adjacent dark regions become enriched (excitation). This inverse response demonstrates that the same simple rules can generate both LI‑like and opposite behaviours depending on the sign of the stimulus.

A third series of simulations addresses global perception of brightness. The authors project greyscale images (the Chevreul staircase and simultaneous brightness contrast patterns) onto the lattice by scaling chemoattractant levels according to pixel intensity. Over many iterations, particles drift toward brighter areas, forming a density landscape that reproduces the classic illusory gradients of the Chevreul staircase and the perceived brightness difference between two identically illuminated patches when surrounded by different backgrounds. Thus, a collective, long‑term redistribution of material encodes a global estimate of image brightness, analogous to higher‑order visual processing in the brain.

The analysis highlights that contrast enhancement and global brightness perception emerge without any explicit inhibitory wiring; they arise from local coupling (sensor offset, sensor angle) and diffusion of a shared chemical field. The model therefore provides a minimal, physics‑based substrate for LI‑like computation, suggesting that distributed robotic swarms or material‑based computers could achieve edge enhancement and ambient light estimation without dedicated inhibitory circuits. The work bridges biological observation (Physarum’s ability to solve mazes, form efficient transport networks, and avoid light) with computational theory, offering a new paradigm for “collective perception” in unorganised systems.

In conclusion, the study demonstrates that simple bulk transport and diffusion mechanisms are sufficient to reproduce key aspects of neural lateral inhibition and global brightness perception. This insight opens avenues for bio‑inspired distributed algorithms, novel robotic control strategies, and deeper understanding of how primitive organisms may perform sophisticated sensory processing without a nervous system.