| |
Abstract:
The dopaminergic cells of the substantia nigra pars compacta
(SNc) fire phasically in response to unpredicted, rewarding stimuli
and play a key role in reinforcement learning. Recently, attempts
have been made to explain and model the dopamine cell responses,
including adaptive critics and temporal differences,
direct/indirect pathway mechanisms, and adaptive timing models. The
recent work has left many relevant neural mechanisms untreated or
unexplained. A detailed survey of known anatomy and
electrophysiology of afferents to the SNc has led to an explanation
of dopamine signaling in terms of known biological mechanisms. Our
model describes seven cell types within the SNc, PPTN, lateral
hypothalamus, ventral striatum, and striosomes. Excitatory
predictions of reward and actual reward signals in the
pedunculo-pontine tegmental nucleus (PPTN) excite the SNc.
Habituation in the PPTN explains the phasic nature of dopamine
bursts. The lateral hypothalamus generates primary reward signals
to the PPTN, and the limbic cortical-accumbal-ventral
pallidal/SNr-PPTN path underlies CS-induced dopamine bursts.
Meanwhile, adaptively-timed inhibitory signals in the striosomes
learn to prevent dopamine bursts in response to predictable,
rewarding events. When expected rewards are not received,
striosomal inhibition is unopposed, resulting in a phasic drop in
dopamine cell activity. Real-time, computational model mechanisms
can simulate neurophysiological data about these cell types.
|
