DA Source-Level Gain Control
The tonic/phasic circuit (Grace et al. 2007)
DA neurons in the VTA display three activity states: inactive (hyperpolarized), tonic firing (slow, irregular, 2-10 Hz pacemaker-driven), and phasic burst firing (fast, afferent-driven). Approximately 50% of VTA DA neurons are held in a hyperpolarized, inactive state by tonic GABAergic inhibition from the ventral pallidum (VP).
The hippocampal ventral subiculum (vSub) regulates DA population activity through a disinhibition pathway: vSub excites NAcc neurons -> NAcc inhibits VP -> VP releases VTA DA neurons from inhibition -> more DA neurons fire tonically. The pedunculopontine tegmental nucleus (PPTg) drives phasic burst firing, but only in neurons that are already tonically active.
Critical property: the hippocampal pathway and the burst-firing pathway are functionally independent but interact multiplicatively. The vSub pathway controls how many DA neurons are available for phasic response. The PPTg pathway controls whether those neurons burst in response to reward-relevant signals. Total DA output depends on both population size and firing rate. This is a gain control mechanism operating at the population level.
Functional compartmentalization
Phasic DA release produces high-amplitude, transient, spatially restricted signals within the synaptic cleft (estimated millimolar range), curtailed by DAT. Acts selectively on postsynaptic D1 receptors and potentiates limbic inputs to NAcc.
Tonic DA is the low-concentration extrasynaptic pool (nanomolar range), regulated by the overall population of active DA neurons. Acts on presynaptic D2 receptors and attenuates PFC inputs to NAcc.
Information flow shift
The DA system shifts the balance of information flow in the NAcc away from PFC control (via tonic D2-mediated suppression of PFC inputs) and toward limbic/hippocampal control (via phasic D1-mediated potentiation of vSub inputs). Increasing DA transmission produces a double shift: PFC is degraded locally (catecholamine-PFC mechanism) AND its influence on downstream processing is simultaneously attenuated relative to limbic inputs. The reflexive-over-reflective transition is reinforced at the circuit level.
The MODAL model (Inglis, Valentin, & Ashby 2022)
Formalizes the Grace circuit as a biologically detailed spiking-neuron computational model (Modulation of Dopamine for Adaptive Learning). Key features:
- NAcc MSNs have distributed resting membrane potentials, creating smooth monotonic relationship between modulating variable value and active VTA DA population size
- Model is agnostic about which environmental variable drives the modulating signal. Candidates: uncertainty, volatility, estimation uncertainty, state-feedback contingency
- Novel predictions: (1) decreasing the modulating variable should reduce slope of extracellular DA as a function of RPE, (2) decreasing should decrease tonic extracellular DA levels
Connection to controllability
State-feedback contingency is one of MODAL's proposed modulating variables. Low contingency (uncontrollability) -> reduced modulating variable -> smaller active DA population -> reduced DA gain -> lower tonic DA -> flattened learning rate. This provides a mechanistic pathway from perceived uncontrollability to DA disruption that degrades PFC function.
The Maier/Seligman controllability circuit (vmPFC -> DRN) and the Grace/MODAL gain control circuit (vSub -> NAcc -> VP -> VTA) may operate in parallel -- one gating the serotonergic stress response, the other modulating dopaminergic learning gain -- both driven by the same or related controllability signals.
Source verification
Grace, Floresco, Goto, & Lodge 2007 (Trends Neurosci. 30(5):220-227) -- verified. Inglis, Valentin, & Ashby 2022 (Computational Brain & Behavior) -- verified. Grace 2010 (Neurotoxicity Research 18:367-376) -- cited but not independently verified.