From Neurons to Computation: Biological Reservoir Computing for Pattern Recognition
Ludovico Iannello *, Luca Ciampi *, Gabriele Lagani *, Fabrizio Tonelli, Eleonora Crocco, Lucio Maria Calcagnile, Angelo Di Garbo, Federico Cremisi, Giuseppe Amato · 2025
Read From Neurons to Computation: Biological Reservoir Computing for Pattern Recognition by Ludovico Iannello *, Luca Ciampi *, Gabriele Lagani *, Fabrizio Tonelli, Eleonora Crocco, Lucio Maria Calcagnile, Angelo Di Garbo, Federico Cremisi, Giuseppe Amato (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-03510.txt.
Introduction
Ludovico Iannello and co-authors use living neurons as the core of a computing system, in a research paper from 2025. Their setup, called biological reservoir computing, grows a pool of cultured neurons on an electrode grid. Signals enter through some electrodes, and the rest record how the network answers.
The living network turns each input into a rich, high-dimensional pattern of firing. A simple linear classifier then reads those patterns to sort inputs, much as an echo state network does in software. The team tested the system on position codes, bars at varied angles, and a digit task. Results showed the cell network could carry out pattern recognition, a step toward bio-hybrid computing.
Catalog record
| CREATOR | Ludovico Iannello *, Luca Ciampi *, Gabriele Lagani *, Fabrizio Tonelli, Eleonora Crocco, Lucio Maria Calcagnile, Angelo Di Garbo, Federico Cremisi, Giuseppe Amato |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-03510
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
From Neurons to Computation: Biological Reservoir Computing for Pattern Recognition Ludovico Iannello *, Luca Ciampi *, Gabriele Lagani *, Fabrizio Tonelli, Eleonora Crocco, Lucio Maria Calcagnile, Angelo Di Garbo, Federico Cremisi, Giuseppe Amato Abstract In this paper, we introduce a paradigm for reservoir computing (RC) that leverages a pool of cultured biological neurons as the reservoir substrate, creating a biological reservoir computing (BRC). This system operates similarly to an echo state network (ESN), with the key distinction that the neural activity is generated by a network of cultured neurons, rather than being modeled by traditional artificial computational units. The neuronal activity is recorded using a multi-electrode array (MEA), which enables high-throughput recording of neural signals. In our approach, inputs are introduced into the network through a subset of the MEA electrodes, while the remaining electrodes capture the resulting neural activity. This generates a nonlinear mapping of the input data to a high-dimensional biological feature space, where distinguishing between data becomes more efficient and straightforward, allowing a simple linear classifier to perform pattern recognition tasks effectively. To evaluate the performance of our proposed system, we present an experimental study that includes various input patterns, such as positional codes, bars with different orientations, and a digit recognition task. The results demonstrate the feasibility of using biological neural networks to perform tasks traditionally handled by artificial neural networks, paving the way for further exploration of biologically-inspired computing systems, with potential applications in neuromorphic engineering and bio-hybrid computing. 1 1 footnotetext: Corresponding authors, they contributed equally to this work: [email protected] , [email protected] , , [email protected] 1 Introduction Reservoir Computing (RC) [ 40 ] is a powerful machine learning paradigm that leverages a highly nonlinear dynamical system with numerous degrees of freedom to project input data into a high-dimensional latent space. In this space, data samples are often more easily linearly separable, which facilitates efficient downstream tasks such as classification or regression. Several models have been proposed to implement the reservoir, among which the Liquid State Machine (LSM) [ 41 ] and the Echo State Network (ESN) [ 25 , 17 ] are the most prominent. The LSM employs a pool of nonlinear spiking neural units to generate a dynamic representation of input signals, while the ESN uses continuous-valued neurons and does not aim to replicate biologically realistic spiking behavior. These models have enabled successful applications in various domains, including time-series prediction [ 3 ] and speech recognition [ 57 ] . Figure 1 : Overview of our biological reservoir computing (BRC) approach. A multi-electrode array (MEA) serves as an interface to both stimulate and record activity from a cultured biological neural network. Each input is mapped to a specific subset of MEA electrodes, which deliver the corresponding stimulus to the network. The resulting spiking activity is recorded by a different set of electrodes and assembled into a vector that represents the input in a high-dimensional latent space. The complex dynamics of the BRC system ensure that this mapping is highly nonlinear. Finally, a linear classifier is trained to recognize the category of the original input based on its latent representation. In this paper, we introduce an extension to the RC paradigm: the use of a cultured network of real biological neurons as the reservoir substrate. This approach, which we refer to as the biological reservoir computing (BRC), employs a biological dynamical system to map input samples into a high-dimensional space of biological features. Unlike traditional RC systems, where the reservoir is typically composed of artificial neurons or units, our biological system leverages real neurons grown in vitro , providing a natural bridge between computational and biological systems. This paradigm has the potential to improve energy efficiency – a significant limitation of current deep learning technologies [ 2 , 26 ] – while also advancing our understanding of biological neuron behavior. Indeed, neuroscientific observations suggest that transient dynamics in the brain carry a significant portion of the input information, indicating that some form of RC may play a role in high-level biological cognition [ 14 ] . In more detail, we derive our neurons from stem cells using well-established neurobiological techniques [ 7 , 8 ] . These cultured networks are then interfaced with a high-density multi-electrode array (MEA) device [ 4 ] , which contains 4096 electrodes for both electrical stimulation and neural response recording (see also Fig. 1 for a visual overview of our approach). To evaluate the ability of our BRC system to generate discriminative feature representations across varying input complexities, we conduct an experimental study spanning tasks of several difficulties. Specifically, we feed features derived from biological culture responses into a single-layer perceptron and assess classification performance of the combined system (i.e., our BRC system followed by the perceptron). We begin with simple pointwise stimuli as basic input patterns, then progress to more complex geometric stimuli such as bars with varying orientations. Finally, we evaluate digit recognition to test the utility of the BRC system in real-world classification scenarios. Results are promising, indicating that biologically cultured networks can function as effective substrates for reservoir computing, producing high-dimensional features that improve pattern recognition accuracy. In summary, the contributions of this paper are as follows: • We introduce a reservoir computing paradigm that leverages biologically cultured networks as substrates for generating high-dimensional feature representations, which we term biological reservoir computing (BRC). Our approach interfaces with these cultured networks using a high-density multi-electrode array (MEA) device, enabling both electrical stimulation and the recording of neural responses. • We define and calibrate a stimulation protocol for interacting with neurons via the MEA device, ensuring robust and consistent neuronal responses across different stimulation parameters. • We present an experimental investigation to validate the proposed BRC system in several pattern recognition tasks, assessing the efficacy of the biological feature representations in distinguishing various categories of input patterns. The remainder of this manuscript is structured as follows: Sec. 2 provides an overview of related work in the field of reservoir computing and biological computing; Sec. 3 details the proposed methodology, including the neurobiological techniques and MEA interfacing; Sec. 4 discusses the experimental setup, results, and analysis; and finally, Sec. 5 presents our conclusions and outlines future directions. 2 Related Work A central theme in deep learning (DL) research is the biological plausibility of current methods, with several studies questioning whether existing architectures reflect the complexity and functionality of biological neural systems [ 43 , 20 , 46 , 55 ] . In response, there has been a growing interest in exploring biologically inspired alternatives to push the boundaries of machine learning and cognitive computation [ 35 , 34 ] . A wide range of models has been proposed to bridge this gap, either by adopting biologically realistic neural computations [ 12 , 16 , 38 , 54 ] or by refining synaptic dynamics and learning mechanisms [ 30 , 23 , 32 , 33 , 31 , 37 , 27 , 36 , 9 , 10 ] . Reservoir Computing (RC) [ 40 , 50 ] represents one of the bio-inspired paradigms that has gained considerable attention for its potential to model complex neural dynamics. Specifically, Echo State Networks (ESNs) [ 17 , 25 ] and Liquid State