From Neurons to Computation: Biological Reservoir Computing for Pattern Recognition

2025-05-06

AI Summary

Conventional reservoir computing (RC) relies on artificial or simulated neural networks, limiting biological plausibility and hardware efficiency. Method: This work introduces Biologically Realized Reservoir Computing (BRC), wherein cultured rat cortical neurons grown on multielectrode arrays (MEAs) serve as a living physical reservoir. Input signals are delivered via pulsed electrical stimulation, and spatiotemporal neural responses are recorded in real time. Leveraging echo state network principles, inputs are projected into a high-dimensional biological feature space, and classification is performed using linear discriminant analysis. Contribution/Results: BRC represents the first experimental demonstration of an unmodified, living neuronal network directly functioning as a physically instantiated, empirically verifiable reservoir—bypassing silicon-based or simplified computational models. On position encoding, orientation bar, and MNIST digit recognition tasks, BRC achieves classification accuracy comparable to classical artificial ESNs, demonstrating the practical feasibility of biological neural networks for pattern recognition.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Bio-inspired Learning
• Computer Vision: Medical and Biological Imaging

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

In this paper, we introduce a novel 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.

Problem

• Leveraging biological neurons for reservoir computing in pattern recognition
• Using cultured neurons to replace artificial computational units in RC
• Evaluating BRC performance on diverse input patterns and tasks

Innovation

• Uses cultured biological neurons as reservoir substrate
• Employs multi-electrode array for neural signal recording
• Maps input data to high-dimensional biological feature space