Spike-frequency adaptation provides a long short-term memory to networks of spiking neurons

D. Salaj, A. Subramoney, C. Kraišnikovic, G. Bellec, R. Legenstein, and W. Maass

Abstract:

Brains are able to integrate memory from the recent past into their current computations, seemingly without effort. This ability is critical for cognitive tasks such as speech understanding or working with sequences of symbols according to dynamically changing rules. But it has remained unknown how networks of spiking neurons in the brain can achieve that. We show that the presence of neurons with spike frequency adaptation makes a significant difference: Their inclusion in a network moves its performance for such computing tasks from a very low level close to the level of human performance. While artificial neural networks with special long short-term memory (LSTM) units had already reached such high performance levels, they lack biological plausibility. We find that neurons with spike-frequency adaptation, which occur especially frequently in higher cortical areas of the human brain, provide to brains a functional equivalent to LSTM units.



Reference: D. Salaj, A. Subramoney, C. Kraišnikovic, G. Bellec, R. Legenstein, and W. Maass. Spike-frequency adaptation provides a long short-term memory to networks of spiking neurons. bioRxiv, 2020.