A neural network approach to the acoustic startle reflex and prepulse inhibition.

Prepulse inhibition (PPI) is the normal suppression of the startle reflex when an intense stimulus is preceded by a weak non-startling prestimulus. PPI is widely used as a model for sensorimotor gating processes and has been shown to be impaired in various neuropsychiatric disorders, including schizophrenia. We have reproduced startle-like behavior and basic PPI modifications with a neural network. The network design was constrained by the attempt (1) to use as few connections as possible and (2) to relate neuroanatomical structures to the simulated network. Performance of the network was evaluated by the behavior of the simulated motor neurons in response to prepulse and pulse stimuli presented with various lead intervals and prepulse intensities. A delayed inhibitory pathway via the pedunculopontine nucleus (PPTg) to the caudal pontine reticular nucleus was found to be a necessary but insufficient requirement to reproduce basic PPI output patterns. Additional requirements included (a) a low threshold at or below the caudal pontine reticular formation, (b) signal amplification in the inhibitory pathway and (c) prolongation of activity in the inhibitory pathway. On the grounds of the most appropriate output patterns of the simulations, we propose a mechanism of sustained activation in the PPTg due to recursive connections. Relations between stimuli, behavior (motor output) and the underlying architecture are discussed. Potentially, this modeling technique can be extended to investigate the impact of drugs and higher brain regions on PPI.