Sub-second Dopamine and Serotonin Signaling in Human Striatum during Perceptual Decision-Making

Abstract

Recent animal research indicates that dopamine and serotonin, neuromodulators traditionally linked to appetitive and aversive processes, are also involved in sensory inference and decisions based on such inference. We tested this hypothesis in humans by monitoring sub-second striatal dopamine and serotonin signaling during a visual motion discrimination task that separates sensory uncertainty from decision difficulty in a factorial design. Caudate nucleus recordings (n = 4) revealed multi-scale encoding: in three participants, serotonin tracked sensory uncertainty, and, in one participant, both dopamine and serotonin tracked deviations from expected trial transitions within our factorial design. Putamen recordings (n = 1) supported a cognition-action separation between caudate nucleus and putamen-a striatal sub-division unique to primates-with both dopamine and serotonin tracking decision times. These first-of-their-kind observations in the human brain reveal a role for sub-second dopamine and serotonin signaling in non-reward-based aspects of cognition and action.

Keywords: action; cognition; decision-making; dopamine; fast scan cyclic voltammetry; human; neuromodulation; perception; serotonin; striatum.

Figure 1

Experimental Framework (A) Continuous direction random dot motion task with variable reference. Participants had to judge whether the net direction of dot motion (sampled from the range 1–360°) was counterclockwise (CCW) or clockwise (CW) to a reference direction that appeared after stimulus offset (maximum absolute angular distance from motion direction: 45°). On around a third of trials, participants were required to estimate their confidence in the perceptual decision on a discrete visual scale indicating probability correct (50% to 100% in steps of 10%). Participant 5 viewed the motion stimulus for 0.8 s on day 2 due to a different configuration of the display monitor in the operating room. (B) Factorial design. We varied the fraction of coherently moving dots (sensory uncertainty) and the absolute angular distance between the motion direction and the reference direction (decision difficulty) in a two-by-two design. (C) Workflow. On day 1, we calibrated stimulus parameters to achieve target levels of performance and trained participants on the task. On day 2, we measured dopamine and serotonin fluctuations while participants performed the task during neurosurgery.

Figure 2

Behavioral Performance Behavioral data shown separately for presurgery (top, green) and surgery (bottom, pink). (A) Choice accuracy. (B) Choice reaction time as measured from onset of reference direction. (C) Confidence estimates as elicited on around a third of trials. In (A)–(C), blue lines denote low coherence and red lines denote high coherence. Black lines indicate main effects (vertical: low versus high coherence; horizontal: low versus high distance). Shaded gray bands denote the ranges of behavioral data observed in a non-patient sample. Data are represented as group mean ± SEM. Symbols indicate participant number and disease state (PD: Parkinson’s disease. ET: essential tremor). See Table S1 for regression statistics and analysis split by session.

Figure 3

Caudate Nucleus: Serotonin Signaling Tracks Sensory Uncertainty (A) Serotonin time series from caudate nucleus grouped by level of coherence in participants 1–3. Marker indicates that time series for low coherence (blue) and high coherence (red) are statistically different (p < 0.05, independent-samples t test). Data are represented as mean ± SEM. See Figure S2A for dopamine time series and Figure S2B for an analysis of dopamine and serotonin responses to motion coherence as a function of the laterality of the motion direction with respect to the hemisphere in which the recording electrode is located. (B) Schematic of sliding-window regression approach used to quantify relationship between task variables and neuromodulatory responses. We deployed a multiple linear regression across all trials where we predicted neuromodulatory responses at each time point (sliding window indicated by orange band) using coherence (pink) and distance (cyan) as well as their interaction (green). We included choice accuracy and choice reaction time as nuisance variables (not shown). All predictors were Z scored. The estimated regression coefficients (β’s) quantifies the encoding of task variables in neuromodulatory responses. (C) Dopamine and serotonin encoding profiles from caudate nucleus in participants 1–3. Marker indicates that a coefficient is statistically different from zero (p < 0.05) as estimated by the regression approach described in (B). Group-level analysis was conducted by combining data across participants and including a random intercept for each participant. See Figure S2C for individual participants. In (A) and (C), time series were locked to the onset of the motion stimulus, spanning a period from 1 s before stimulus onset to 5 s after stimulus onset, Z scored separately for each trial, and smoothed using a running average (.5 s). Top right-hand corner indicates participant number and disease state (PD: Parkinson’s disease; ET: essential tremor). The fourth caudate nucleus participant was not included in this analysis because their neuromodulatory responses—as explained in the main text—were qualitatively different. DA: dopamine. 5-HT: serotonin.

Figure 4

Caudate Nucleus: Dopamine and Serotonin Signaling Tracks Experienced Trial Type Transitions (A) Illustration of experienced task statistics for participant 4. Horizontal bars show the probability of encountering each trial type on trial t after having encountered the trial type defined by high coherence and low distance on trial t-1. The trial type transition probabilities, P(type_t|type_t-1), were computed as the normalized counts of the number of times each trial type succeeded a given trial type. The expected value of P(type_t|type_t-1) under randomization is 0.25. The least likely trial type transition is indicated by green color in (A) and (B). (B) Correlation between neuromodulatory responses and trial type transition probabilities in participant 4. To compute the neuromodulatory responses, we first averaged time points across a window from 0 s to 1.5 s (the period during which a trial type is revealed) within each trial and then averaged across all trials for each trial type transition. Time series were locked to the onset of the motion stimulus, spanning a period from 1 s before stimulus onset to 5 s after stimulus onset, and Z scored separately for each trial. For comparison between dopamine and serotonin, data points are colored according to the associated trial type transition probability. Lines are best-fitting lines from a linear regression. Top right-hand corner indicates participant number and disease state (PD: Parkinson’s disease; ET: essential tremor). DA: dopamine. 5-HT: serotonin.

Figure 5

Putamen: Dopamine and Serotonin Signaling Tracks Choice Submission (A) Dopamine and serotonin time series from putamen locked to the onset of the motion stimulus and overlaid onto distribution over choice reaction time (pink histogram) in participant 5. See Figure S2D for participants 1–4. (B) Dopamine and serotonin time series from putamen locked to stimulus onset (left) or choice submission (right) and grouped by terciles over choice reaction time in participant 5. (C) Same as in (A) but separated by choice accuracy. (D) Same as in right-hand side of (B) but separated by choice accuracy. In (A)–(D), time series were Z scored separately for each trial (stimulus-locked: period spanned from 1 s before stimulus onset to 5 s after stimulus onset; choice-locked: period spanned from 4 s before choice submission to 4 s after choice submission) and smoothed using a running average (.5 s). Data are represented as mean ± SEM in (A) and (C) and as mean in (B) and (D). Marker indicates that a time point is statistically different from zero (p < 0.05, one-sample t test). Top right-hand corner indicates participant number and disease state (PD: Parkinson’s disease; ET: essential tremor). DA: dopamine. 5-HT: serotonin.