Spatial Computing in Modular Spiking Neural Networks with a Robotic Embodiment

Abstract

One of the challenges in modern neuroscience is creating a brain-on-a-chip. Such a semiartificial device based on neural networks grown in vitro should interact with the environment when embodied in a robot. A crucial point in this endeavor is developing a neural network architecture capable of associative learning. This work proposes a mathematical model of a midscale modular spiking neural network (SNN) to study learning mechanisms within the brain-on-a-chip context. We show that besides spike-timing-dependent plasticity (STDP), synaptic and neuronal competitions are critical factors for successful learning. Moreover, the shortest pathway rule can implement the synaptic competition responsible for processing conditional stimuli coming from the environment. This solution is ready for testing in neuronal cultures. The neuronal competition can be implemented by lateral inhibition actuating over the SNN modulus responsible for unconditional responses. Empirical testing of this approach is challenging and requires the development of a technique for growing cultures with a given ratio of excitatory and inhibitory neurons. We test the modular SNN embedded in a mobile robot and show that it can establish the association between touch (unconditional) and ultrasonic (conditional) sensors. Then, the robot can avoid obstacles without hitting them, relying on ultrasonic sensors only.

STDP-Driven Rewiring in Spiking Neural Networks under Stimulus-Induced and Spontaneous Activity

“…Then, the central neuron is stimulated, which induces the propagation of spikes from the center to the periphery. The activity vector field illustrates these dynamic events in a single static picture in Figure 3 B. Consequently, this repetitive spike propagation causes STDP-driven weight rearrangement (the details can be found in [ 39 , 40 , 49 ]), which is visualized by the weight vector field ( Figure 3 C). In turn, structural plasticity replaces unused centripetal connections with multiple centrifugal ones involved in long-term activity, as illustrated in Figure 3 D. Thus, this stereotypical stimulation leads to network rearrangements at different levels: spikes → synaptic weights → anatomical connectome, visualized in the static pictures of Figure 3 by the vector fields.…”

“…The activity vector field illustrates these dynamic events in a single static picture in Figure 3B. Consequently, this repetitive spike propagation causes STDP-driven weight rearrangement (the details can be found in [39,40,49]), which is visualized by the weight vector field (Figure 3C). In turn, structural plasticity replaces unused centripetal connections with multiple centrifugal ones involved in long-term activity, as illustrated in Figure 3D.…”

“…The approach described previously [ 39 , 40 ] was employed to simulate the dynamics of the SNN. Specifically, the dynamics of a neuron were described by Izhikevich’s model [ 41 ].…”

The technology, opportunities, and challenges of Synthetic Biological Intelligence

“…( Barlow, 2012 ; Barto et al, 2013 ; Friston et al, 2012 , Friston et al, 2018 , Friston et al, 2009 ; Kangassalo et al, 2020 ; Madhav and Cowan, 2020 ; Schwartenbeck et al, 2019 , Schwartenbeck et al, 2015 ; Schwartz, 2016 ; Sinapayen et al, 2017 )). One of the limitations of this area is that many theories about how internal states such as intelligence, cognition, sentience, consciousness et cetera may arise and the implications of this are exceedingly difficult to empirically test and interrogate in vivo ( Friston, 2023 ; Goddard et al, 2023 ; Lobov et al, 2023 ; Pereira et al, 2023 ). Therefore, while enormous conceptual advancements have been made in this area that can potentially facilitate basic SBI, the ability to test these theories requires SBI techniques to co-develop more controlled research methods ( Fig.…”