Synchronicity: The Role of Midbrain Dopamine in Whole-Brain Coordination

Abstract

Midbrain dopamine seems to play an outsized role in motivated behavior and learning. Widely associated with mediating reward-related behavior, decision making, and learning, dopamine continues to generate controversies in the field. While many studies and theories focus on what dopamine cells encode, the question of how the midbrain derives the information it encodes is poorly understood and comparatively less addressed. Recent anatomical studies suggest greater diversity and complexity of afferent inputs than previously appreciated, requiring rethinking of prior models. Here, we elaborate a hypothesis that construes midbrain dopamine as implementing a Bayesian selector in which individual dopamine cells sample afferent activity across distributed brain substrates, comprising evidence to be evaluated on the extent to which stimuli in the on-going sensorimotor stream organizes distributed, parallel processing, reflecting implicit value. To effectively generate a temporally resolved phasic signal, a population of dopamine cells must exhibit synchronous activity. We argue that synchronous activity across a population of dopamine cells signals consensus across distributed afferent substrates, invigorating responding to recognized opportunities and facilitating further learning. In framing our hypothesis, we shift from the question of how value is computed to the broader question of how the brain achieves coordination across distributed, parallel processing. We posit the midbrain is part of an "axis of agency" in which the prefrontal cortex (PFC), basal ganglia (BGS), and midbrain form an axis mediating control, coordination, and consensus, respectively.

Keywords: coherence; dopamine; phasic dopamine; striatum; synchronous dopamine activity.

Figure 1.

Random distribution of afferents to dopamine cells as sampling the input space. A, Projections from each afferent region construed as a vector randomly distributed to individual dopamine cells. B, Afferent projections to midbrain comprise a total vectorized input space from which individual dopamine cells represent random samples of that space.

Figure 2.

Schematic of “axis of agency.” Primary processing substrates (distributed, parallel processing) represented abstractly in gray as either top-down or bottom-up substrates with reciprocal connections to the prefrontal cortex (PFC) and basal ganglia system (BGS), not detailed here. The three conceptual nodes in the axis of agency are indicated in boxes with their hypothetical role in mediating coordinated activity across distributed substrates noted in bold below. Only the inputs/outputs of dopamine, the focus of this perspective, are colored, with green and orange representing excitatory and inhibitory inputs and blue dopamine outputs. The role of the BGS and midbrain dopamine is elaborated below.

Figure 3.

Two axes model of dopamine integrative consensus signaling. A, Overall conceptual rendering of proposed model where midbrain dopamine integrates two primary axes of input, (1) a (dis)inhibitory axis arising from the BGS (ventral pallidum, striosomes, accumbens) and (2) a largely excitatory axis arising from distributed afferents across the brain, reflecting both top-down (e.g., cortical inputs, amygdala, hippocampus, BNST) and bottom-up information processing (e.g., collicular, multiple brainstem afferents). B, A more anatomic rendering incorporating cortical and subcortical loops through the BGS. For detailed cataloging of dopamine inputs, see Watabe-Uchida et al. (2012), Lerner et al. (2015), and Beier et al. (2015). Basal ganglia system (BGS), bed nucleus of stria terminalis (BNST), pedunculopontine tegmental nucleus, (PPTg), laterodorsal tegmental nucleus (LTDg).

Figure 4.

Illustration of midbrain dopamine as a Bayesian selector. Random distribution of vectorized inputs from an afferent input space, as illustrated in Figure 1, where each dopamine cell samples on-going activity, mapped onto a Bayesian construal. The excitatory axis (green) is assigned as the prior (the advocate) and the (dis)inhibitory axis (orange) is assigned as the likelihood (the skeptic). The posterior (blue) arises from the integration of these two axes (i.e., Fig. 3) at both the level of individual dopamine cells (firing rate) and at a population level, where synchrony determines the degree to which increases in firing rates in individual cells sum to produce a population-based phasic signal, which we construe as a consensus index, both consensus across dopamine cells as Bayesian units and consensus across the sampled input space, reflecting widespread afferent activity in response to current stimuli.

Figure 5.

Cascading learning. Learning occurs progressively interleaved at three levels: (1) in primary models of individual afferent substrates, both top-down and bottom-up (blue), (2) in a secondary, integrative selection model in the basal ganglia system (BGS, orange), and (3) in the midbrain dopamine system itself (green). Arrows indicate how learning at different levels influences learning at other levels.