Adapting to time: Why nature may have evolved a diverse set of neurons

Brains have evolved diverse neurons with varying morphologies and dynamics that impact temporal information processing. In contrast, most neural network models use homogeneous units that vary only in spatial parameters (weights and biases). To explore the importance of temporal parameters, we trained spiking neural networks on tasks with varying temporal complexity, holding different parameter subsets constant. We found that adapting conduction delays is crucial for solving all test conditions under tight resource constraints. Remarkably, these tasks can be solved using only temporal parameters (delays and time constants) with constant weights. In more complex spatio-temporal tasks, an adaptable bursting parameter was essential. Overall, allowing adaptation of both temporal and spatial parameters enhances network robustness to noise, a vital feature for biological brains and neuromorphic computing systems. Our findings suggest that rich and adaptable dynamics may be the key for solving temporally structured tasks efficiently in evolving organisms, which would help explain the diverse physiological properties of biological neurons.Author summary: The impressive successes of artificial neural networks (ANNs) in solving a range of challenging artificial intelligence tasks have led many researchers to explore the similarities between ANNs and brains. One obvious difference is that ANNs ignore many salient features of biology, such as the fact that neurons spike and that the timing of spikes plays a role in computation and learning. Here we explore the importance of adapting temporal parameters in spiking neural networks in an evolutionary context. We observed that adapting weights by themselves (as typical in ANNs) was not sufficient to solve a range of tasks in our small networks, and that adapting temporal parameters was needed (such as adapting the time constants and delays of units). Indeed, we showed that adapting weights is not even needed to achieve solutions to many problems. Our findings may provide some insights into why evolution produced a wide range of neuron types that vary in terms of their processing of time and suggest that adaptive time parameters will play an important role in developing models of the human brain. In addition, we showed that adapting temporal parameters makes networks more robust to noise, a feature that can prove beneficial for neuromorphic system design.