Since the beginning of information processing by electronic components, the nervous system has served as a metaphor for the organization of computational primitives. Brain-inspired computing today encompasses a class of approaches ranging from using novel nano-devices for computation to research into large-scale neuromorphic architectures, such as TrueNorth, SpiNNaker, BrainScaleS, Tianjic, and Loihi. While implementation details differ, spiking neural networks—sometimes referred to as the third generation of neural networks—are the common abstraction used to model computation with such systems. Here we describe the second generation of the BrainScaleS neuromorphic architecture, emphasizing applications enabled by this architecture. It combines a custom analog accelerator core supporting the accelerated physical emulation of bio-inspired spiking neural network primitives with a tightly coupled digital processor and a digital event-routing network.

Sensory events produced by ourselves are known to lead to lower neural and perceptual impact than sensory events from other environmental sources. This sensory attenuation is widely assumed to result from control processes that are specific to our own motor actions, potentially helping us to distinguish effects produced by ourselves and others. However, previous research cannot rule out that the putative self-attenuation in fact reflect actor-independent, general predictive mechanisms, which, in direct comparison, just highlight external events due to lower predictability of their onset and thus higher surprise. By measuring the auditory-evoked N1 component, we show that self-generation of sounds only lead to cortical attenuation when the onset of other-generated sounds is less predictable due to the absence of any predictive cues. The presence of a cue predicting the onset of auditory stimuli, in contrast, led to a reversal of the attenuation effect, with lower N1 amplitudes for other-generated sounds in contrast to self-generated sounds. Thus, contrary to prevalent assumptions sensory attenuation is not bound to self-generation per se. Rather, it appears to be the result of general mechanisms that does not reliably and selectively attenuate self-induced stimulation but is determined by a flexible processing of sensory input based on its predictability, contextual relevance and attentional salience.

Viewing facial expressions often evokes facial responses in the observer. These spontaneous facial reactions (SFRs) are believed to play an important role for social interactions. However, their developmental trajectory and the underlying neurocognitive mechanisms are still little understood. In the current study, 4- and 7-month old infants were presented with facial expressions of happiness, anger, and fear. Electromyography (EMG) was used to measure activation in muscles relevant for forming these expressions: zygomaticus major (smiling), corrugator supercilii (frowning), and frontalis (forehead raising). The results indicated no selective activation of the facial muscles for the expressions in 4-month-old infants. For 7-month-old infants, evidence for selective facial reactions was found especially for happy (leading to increased zygomaticus major activation) and fearful faces (leading to increased frontalis activation), while angry faces did not show a clear differential response. These results suggest that emotional SFRs may be the result of complex neurocognitive mechanisms which lead to partial mimicry but are also likely to be influenced by evaluative processes. Such mechanisms seem to undergo important developments at least until the second half of the first year of life.

No abstract