Abstract
The role of serotonin in human behaviour is informed by approaches which allow in vivo modification of synaptic serotonin. However, characterising the effects of increased serotonin signalling in human models of behaviour is challenging given the limitations of available experimental probes, notably selective serotonin reuptake inhibitors. Here we use a now-accessible approach to directly increase synaptic serotonin in humans (a selective serotonin releasing agent) and examine its influence on domains of behaviour historically considered core functions of serotonin. Computational techniques, including reinforcement learning and drift diffusion modelling, explain participant behaviour at baseline and after week-long intervention. Reinforcement learning models reveal that increasing synaptic serotonin reduces sensitivity for outcomes in aversive contexts. Furthermore, increasing synaptic serotonin enhances behavioural inhibition, and shifts bias towards impulse control during exposure to aversive emotional probes. These effects are seen in the context of overall improvements in memory for neutral verbal information. Our findings highlight the direct effects of increasing synaptic serotonin on human behaviour, underlining its role in guiding decision-making within aversive and more neutral contexts, and offering implications for longstanding theories of central serotonin function.
Similar content being viewed by others
Introduction
Understanding the function of central serotonin (or 5-hydroxytryptamine, 5-HT) has been a focal goal of neuroscience research for nearly a century1, not least because of its central role in the effects of many psychiatric drugs, predominantly selective serotonin reuptake inhibitors (SSRIs) and street drugs (e.g., ±3,4-methylenedioxymethamphetamine [MDMA] and lysergic acid diethylamide)2,3. Serotonin is phylogenetically ancient, and its function translates across species to many lower- and higher-level behaviours; from feeding and sexual functioning to goal-directed, flexible cognition4,5,6,7. Amongst these, behavioural inhibition, memory, and aversive processing are historically considered the core, specialised functions of serotonin8,9,10,11. This is underpinned by converging preclinical and human work involving in vivo manipulation of synaptic 5-HT, predominantly with SSRIs or depletion of its amino acid precursor tryptophan (TRP)7,12, and observing behavioural change. In humans, however, marked differences in the direction of behavioural effects are observed across similar experimental approaches. For example, several studies report seemingly contradictory effects of SSRIs on tasks of aversive and reward processing (reinforcement learning); specifically, different reports show that SSRIs increase reward sensitivity13, increase loss sensitivity and decrease reward sensitivity14, and decrease sensitivity to both reinforcement valences15. Inconsistent behavioural effects of SSRIs are also observed across other domains, including behavioural inhibition and memory processing12,16,17,18,19,20,21,22; in some cases, these behavioural changes align with those seen after tryptophan depletion (e.g., reduced cognitive flexibility) despite the expectation that they would have opposing effects on net synaptic 5-HT16,23.
Determining a causal link between increased synaptic 5-HT and behaviour in humans via SSRIs is difficult due to the complex effects of SSRIs on 5-HT and colocalised neurotransmitter systems. For example, negative signalling feedback along the serotonergic pathway following autoreceptor activation early in treatment can limit cell firing, and therefore 5-HT release, in a regionally-specific manner24,25,26. Furthermore, deactivation of 5-HT transporters results in 5-HT uptake via dopamine transporters, leading to subsequent co-release of dopamine and 5-HT27. The effect of increased dopaminergic content and signalling is seen in acute and subchronic SSRI administration28,29,30,31,32, observable in striatal, prefrontal, and hippocampal structures implicated in reward processing, behavioural inhibition and memory functioning27,33,34,35.
Given the complex molecular and behavioural profile of SSRIs, alternative probes which increase synaptic 5-HT may help further clarify the role of 5-HT in human behaviour and cognition. One such alternative involves the use of a selective serotonin releasing agent (SSRA) (Fig. 1): unlike SSRIs which increase 5-HT levels indirectly through prolonging synaptic 5-HT, SSRAs stimulate direct exocytic release of 5-HT, without broad monoaminergic efflux (as seen in non-selective 5-HT releasers, such as MDMA)36,37. While SSRIs require ongoing neural firing for vesicular release of 5-HT into the synapse, the SSRA mechanism is not firing-dependent and thus not negated by dorsal raphe autoreceptor negative feedback which delays the therapeutic onset of action of SSRIs38,39,40.
Until recently, it has been challenging to characterise the effects of SSRAs in humans because of the lack of available licensed pharmacological probes. However, in 2020, low dose fenfluramine (up to 26 mg daily; racemic mixture) was licensed for the treatment of Dravet epilepsy41. Preclinical work suggests low dose fenfluramine directly and rapidly increases synaptic 5-HT without modifying extracellular dopamine concentration in regions involved in mood regulation such as the striatum and hippocampus, in contrast to SSRIs40,42,43,44,45,46,47,48,49,50,51,52,53. Fenfluramine led to substantially greater extracellular 5-HT levels than the SSRI, fluoxetine, when administered at similar doses54. Further preclinical work suggests acute administration of fenfluramine increases synaptic 5-HT by 182-200% vs basal state53,55, while subchronic administration (4-5 days) retains increases in net 5-HT without influencing 5-HT terminal structural integrity53,56. In humans, acute and subchronic administration of d-fenfluramine decreases serotonergic radioligand binding of [18F]altanserin in a dose-dependent manner, suggestive of increased synaptic serotonin release in the brain57,58. With its recent relicensing for epilepsy syndromes41, fenfluramine provides an opportunity to probe the neurobehavioural effects of SSRAs in humans to answer outstanding questions about the role of synaptic 5-HT in human behaviour.
Here, we use this now-accessible approach to directly increase synaptic 5-HT in humans, examining its influence on domains of behaviour historically considered core functions of serotonin: aversive processing, behavioural inhibition, and memory. In line with our hypothesis that fenfluramine would result in a pattern of behaviour opposite to that seen with tryptophan depletion12,16,17,20,59,60,61,62, we show that increasing synaptic serotonin reduces reinforcement sensitivity for aversive outcomes; in addition, our findings indicate that increasing synaptic serotonin enhances behavioural inhibition and shifts bias towards impulse control during aversive interference. Finally, increasing synaptic serotonin increased memory for neutral verbal information.
Results
Does increased synaptic serotonin change reinforcement sensitivity for reward and loss?
We investigated the effect of increased synaptic 5-HT (via fenfluramine) on reinforcement sensitivity for reward and loss outcomes during a probabilistic instrumental learning task described in Fig. 2A63,64. During this task, participants learned the probability of outcomes associated with symbols within pairs. Each pair represented a task condition: win trials (win money or no change) and loss trials (lose money or no change). Within a given task block, each trial type was associated with one pair of symbols and appeared 30 times per block (across three blocks). Optimal choices were made when selecting symbols which had a greater probability (70%) of leading to a favourable outcome (i.e., win in win trials and no change in loss trials). A computational reinforcement learning model was fit to participant choice behaviour to formalise the predicted change in optimal choices between allocation groups. Model parameters for each trial type were derived, providing a distinct explanation of learning and decision-making behaviour throughout the task: learning rate (\(\alpha\)), explaining the rate at which outcomes modify expectations, which was estimated separately for win and loss trials; and outcome sensitivity (\({{{{\rm{\rho }}}}}\)), explaining the effective magnitude of experienced outcomes. Further information about the model, including parameter recovery and simulation procedures, are detailed within the Supplementary Methods.
In line with our hypothesis, fenfluramine allocation reduced the number of optimal choices during loss trials (ANCOVA group x task condition: F[1,50] = 5.14, p = 0.03, ηp2 = 0.07 [95% CI 0.00, 0.24]; reward condition EMM = 0.68 ± 3.18, p = 0.83; loss condition EMM ± SE = −8.62 ± 3.18, p < 0.01, Cohen’s d = −0.75 [95% CI −1.30, −0.19]) (Fig. 2B-C). Consistent with this, learning models fit to the data revealed increased synaptic 5-HT reduced outcome sensitivity for loss trials (ANCOVA group x task condition: F[1,50] = 5.73, p = 0.02, ηp2 = 0.10 [0.00, 0.28]; reward condition EMM = 0.10 ± 0.43, p = 0.82; loss condition EMM = −0.90 ± 0.43, p = 0.04, d = −0.57 [−1.11, −0.03]) (Fig. 2D). In contrast, modelled learning rate for both conditions did not vary across groups in a statistically significant manner (ANCOVA group x task condition: F[1,50] = 1.22, p = 0.27; main effect of group: F[1,50] = 0.92, p = 0.34) (Supplementary Fig. 10). Similarly, fenfluramine allocation increased time to choice selection during loss conditions (ANCOVA group x task condition: F[1,50] = 5.52, p = 0.02, ηp2 = 0.11 [0.00, 0.29]; reward condition EMM = 13.9 ± 95.6, p = 0.89; loss condition EMM = 246.0 ± 95.6, p = 0.01, d = 0.71 [0.15, 1.26]) (Fig. 2F), which would also be consistent with a relative reduction in loss sensitivity in this group.
Overall, these findings demonstrate that net increases in synaptic 5-HT (via serotonin releasing agent fenfluramine) decrease reinforcement sensitivity to loss outcomes, opposite to the effect of 5-HT depletion (TRP) where loss sensitivity increases61,62. While alternative computational accounts for the observed behaviour could include increased value decay or choice stochasticity, there are no reports of 5-HT manipulation influencing these components of behaviour64.
Does elevated 5-HT modulate behavioural inhibition, choice impulsivity, and vulnerability to aversive emotional interference?
Next, we assessed the impact of increased synaptic 5-HT on response inhibition (an index of behavioural inhibition), choice impulsivity, and interference during the Affective Interference Go/No-Go task. In this task, participants respond (go) or withhold responses (no-go) according to rules which change over time (e.g., instructions: do not press the button if you see a blue/yellow image) while being exposed to emotional distractors (fearful or happy faces, or control images) (Fig. 3A). Fenfluramine allocation increased response inhibition, measured by mean percentage of accurately withheld responses to no-go trials (ANCOVA main effect of group: F[1,47] = 11.26, p < 0.01, ηp2 = 0.15 [0.00, 0.37]; all conditions EMM = 9.69 ± 2.63, p < 0.001, d = 0.60 [0.27, 0.93]) (Fig. 3B). Further, there was no statistically significant group effect for go trial accuracy (ANCOVA main effect of group: F[1,47] = 0.83, p = 0.37) (Supplementary Fig. 14B).
Signal detection theory analyses was undertaken to determine if group differences in response inhibition were driven by perceptual decision-making (Fig. 3C). Fenfluramine allocation resulted in more cautious decision-making throughout (log criterion \(c\); ANCOVA main effect of group: F[1,47] = 13.54, p < 0.001, ηp2 = 0.19 [0.02, 0.39]; all conditions EMM = 0.08 ± 0.02, p < 0.001, d = 0.39 [0.16, 0.62]) (Fig. 3D), but there was no statistically significant group effect on signal discriminability (see Supplementary Note 4).
There was a positive relationship between go trial response times and both response inhibition (r = 0.61, p < 0.001, two-tailed) and cautious decision-making (log decision criterion [\(c\)]; r = 0.77, p < 0.001, two-tailed), as shown in Supplementary Fig. 15. These findings represent an inherent speed-accuracy trade-off within the limited response window for the task (400 ms), where a favourable optimisation for task strategy is slowing response times to ensure greater response inhibition accuracy. As a result, faster response time for go trials may be considered an index of choice impulsivity. Indeed, it is argued that the tendency to react in a premature manner without adequate signal processing constitutes the fundamental aspects of impulsive behaviour65,66.
Increasing synaptic 5-HT (via fenfluramine) resulted in reductions in choice impulsivity, indicated by increased time to choice for go trials, across all task conditions (ANCOVA main effect of group: F[1,47] = 22.00, p < 0.001; ηp2 = 0.27 [0.07, 0.46]) (Fig. 4). Moreover, there was an interaction between group and task interference (happy, fearful or control distractors) on choice impulsivity (ANCOVA group x task condition: F[2,95] = 3.22, p < 0.05, ηp2 = 0.08 [0.00, 0.20]). Specifically, choice impulsivity was most reduced when aversive emotional distractors were present in the fenfluramine group (EMM = 21.3 ± 4.71, p < 0.0001, d = 1.28 [0.70, 1.86]) compared with both control (EMM = 14.6 ± 4.71, p < 0.01, d = 0.88 [0.31, 1.44]) and positive emotional distractors (EMM = 15.4 ± 4.71, p < 0.01, d = 0.93 [0.36, 1.50]). In a separate analysis of the fenfluramine group, there was a main effect for valence on choice impulsivity when isolating aversive vs control conditions (ANCOVA main effect: F[1,22] = 4.87, p = 0.04, ηp2 < 0.01 [0.00, 0.19]); however, there was no statistically significant main effect of valence in the placebo group (see Supplementary Table 8).
Computational drift diffusion modelling (Fig. 5A) was undertaken to investigate evidence accumulation patterns throughout the Affective Interference Go/No-Go task (for further model details, including recovery and simulation procedures, see Supplementary Methods). Fenfluramine allocation shifted initial choice bias (\(z \cdot a\)) toward impulse control (no-go, lower boundary) during aversive interference only (ANCOVA group x task condition: F[2,95] = 3.45, p = 0.03, ηp2 = 0.06 [0.00, 0.17]; control condition EMM = −0.01 ± 0.15, p = 0.96; positive interference EMM = −0.16 ± 0.15, p = 0.31; aversive interference EMM = −0.33 ± 0.15, p = 0.03, d = −0.60 [−1.17, −0.04]) (Fig. 5B). There was no statistically significant group effect across other model parameters, including boundary separation (a) and drift rate (v) (see Supplementary Note 4). As 75% of task trials fit to the DDM were go trials, and there was no statistically significant group effect on accuracy for these trials, group differences in model parameters may not occur when accuracy is similar despite differences in choice time67.
Taken together, these findings suggest that increasing synaptic 5-HT enhances behavioural inhibition across emotional and more neutral contexts, an effect which was driven by more cautious decision-making. Moreover, increased 5-HT levels appear to shift bias towards impulse control during aversive affective interference at the start of evidence accumulation, consequentially lowering choice impulsivity.
Assessing the influence of increased synaptic serotonin on memory processing
Finally, we assessed the influence of fenfluramine administration on memory function. During a task of complex verbal working memory processing (Verbal n-back; Fig. 6A), participants were required to recall if a target letter occurred within a pre-specified sequential pattern (i.e., 0-, 1-, 2-, or 3-back letters ago). There was no statistically significant group effect for total number of correctly recalled targets (ANCOVA group analysis: F[1,47] = 0.70, p = 0.41) (Fig. 6B). However, during the highest task difficulty (3-back), fenfluramine allocation resulted in faster recall for correct targets (ANCOVA group x task condition: F[3,143] = 3.66, p = 0.01, ηp2 = 0.05 [0.00, 0.13]; 0-back EMM: 35.9 ± 50.20, p = 0.48; 1-back EMM: −16.7 ± 50.20, p = 0.74; 2-back EMM: −51.8 ± 50.20, p = 0.30; 3-back EMM: −118.20 ± 50.20, p = 0.02, d = −0.67 [−1.23, −0.10]) (Fig. 6C).
During a task of long-term memory encoding and retrieval (Auditory Verbal Learning Task; Fig. 6D), participants were required to learn a list of 15 verbal items and correctly recall these items during learning (immediate recall) and after a short period (delayed recall). Fenfluramine allocation resulted in higher total accuracy during delayed recall, while there was no statistically significant group effect for learning trials (ANCOVA group x task condition: F[2,1474] = 6.23, p = 0.01, ηp2 < 0.01 [0.00, 0.02]; delayed recall EMM = 0.84 ± 0.35, p = 0.02, d = 0.34 [0.06, 0.61]; learning EMM = −0.07 ± 0.14, p = 0.63; distraction recall EMM = −0.90 ± 0.70, p = 0.20) (Fig. 6E). There was no statistically significant group effect for word repetitions or intrusions (Supplementary Fig. 19).
There was no statistically significant group effect for performance on tasks of visuo-spatial working memory (Oxford Memory Task) and implicit visual learning (Contextual Cueing Task) (see Supplementary Note 7−8, Supplementary Table 4 and Supplementary Figs. 22, 24).
Taken as a whole, these findings suggest increasing synaptic 5-HT enhances memory processing for verbal information.
Relationship of elevated 5-HT with self-report questionnaire measures, gender-related covariance, and cortisol concentration
There was no statistically significant group effect on self-report ratings of subjective cognition, side effects, motivation, and affect (Tables 1–2; Supplementary Table 3). At study completion, the placebo group (70%) were better than chance at correctly guessing their allocation compared with the fenfluramine group (50%), however this was not a statistically significant difference (χ² = 3.92, p = 0.27). Gender did not covary with the effects of fenfluramine administration on task behaviour reported in previous sections (Supplementary Table 5). There was no statistically significant difference between groups in salivary cortisol during the initial dosing period, suggesting a lack of acute modulation of hypothalamic-pituitary-adrenal (HPA) axis function from fenfluramine.
Discussion
The present findings demonstrate the direct effects of increased synaptic serotonin on human behaviour, underlining its role in guiding decision-making across aversive and more neutral contexts (i.e., where valence is not explicitly manipulated). Specifically, we observed reduced sensitivity for outcomes in aversive contexts; enhanced behavioural inhibition and increased bias favouring impulse control during aversive interference; and enhanced memory function for verbally encoded information. These findings offer broad implications for longstanding theories of how central 5-HT influences human behaviour and contributes to psychiatric aetiology.
Throughout instrumental learning, increased synaptic 5-HT (via fenfluramine) reduced sensitivity to aversive outcomes. This effect is opposite to that described following central depletion of serotonin with tryptophan-depletion, where enhanced negative prediction errors during probabilistic instrumental learning and bias toward aversive stimuli during Pavlovian conditioning have been observed8,60,62,68. Further, in a Pavlovian-to-instrumental transfer paradigm, independent depletion of 5-HT and dopamine respectively enhanced aversive and decreased rewarding Pavlovian-to-instrumental transfer69. As SSRAs and TRP result in opposite effects on net synaptic 5-HT, the opposite behavioural pattern observed here is consistent with a key role for serotonin in modulating loss sensitivity12,38.
There was no statistically significant difference in reward processing between the SSRA and placebo groups. Despite the shared purpose of increasing synaptic 5-HT, SSRI administration has been associated with decreased sensitivity for rewarding outcomes in some studies14,15,70. Reduced reward sensitivity has been attributed to unwanted SSRI treatment effects, notably emotional blunting and reduced efficacy in targeting anhedonia71. Importantly, in preclinical work, SSRI administration results in indirect modulation of dopaminergic signalling pathways involved in reward processing28,29,30,31. In similar work, however, the SSRA used here (low dose fenfluramine, racemic mixture) retains selectivity for 5-HT45,55,72,73 and is inactive at dopaminergic synapses56,74, while having a binding affinity for 5-HT transporters which is <0.5% of that typically seen in SSRIs such as citalopram75 (see the Supplementary Discussion for further details on the past uses of the experimental probe). Thus, these results highlight potentially specific effects of serotonin on loss processing, whereas contradictory effects of SSRIs previously reported may relate to effects beyond the serotonin system. It would be worthwhile to investigate if the effect of the SSRA on aversive processing could prove advantageous for the treatment of depression without exacerbating features of anhedonia.
In the preclinical literature, pharmacological (fenfluramine) and optogenetic stimulation of serotonergic neurons in the dorsal raphe nucleus (DRN) results in no change in reward learning in animal models; however, stimulation of non-serotonergic DRN neurons via amphetamine and optogenetics results in increases in reward choice preference76. Moreover, increased firing of amygdala 5-HT projecting neurons is observed during aversive but not reward prediction errors, an effect which appears to be modulated by a functionally discrete DRN to basal amygdala 5-HT pathway77,78. Further optogenetic labelling work suggests that increased DRN 5-HT firing promotes aspects of reward processing79,80,81, potentially facilitated by co-release of glutamate and subsequent activation of mesoaccumbens dopamine neurons82. Given the methodological disparity between this literature and the present work, drawing direct parallels is challenging, particularly as associations between neuronal firing patterns and synaptic serotonin are region-specific83,84,85,86,87.
Interpreting the ecological meaning of performance during reinforcement learning is challenging. A reduction in sensitivity to loss outcomes may be adaptive or detrimental depending on real-world context. Despite reduced optimal choices for loss trials in the SSRA group, there was no difference in total money earned across groups (see Supplementary Note 3). Moreover, reduced loss learning cannot be attributed to undesired effects such as impaired cognition; indeed, concurrent improvements in learning and memory tasks were observed in the SSRA group.
Increasing synaptic 5-HT (via fenfluramine) enhanced behavioural inhibition, an effect driven by more cautious decision-making. Impairment of 5-HT function decreasing behavioural inhibition is well-observed in animals, and to a lesser extent in humans9,17. However, the opposite approach of increasing synaptic 5-HT with SSRIs yields a comparably less clear picture cross-species. In humans, SSRI challenge results in improvement or no change in action cancellation ability (stop signal)16,88, while action restraint ability (go/no-go) remains unchanged or impaired17,18,19,89. Frontal functional activity increases during action restraint following SSRI challenge, however this is not linked to a corresponding change in ability18,89. Likewise, SSRIs yield no clear effect on behavioural inhibition in animals17,90. The seemingly irreconcilable effects of SSRIs on behavioural inhibition may be attributed to the vulnerability of the agent to experimental noise; notably, its acute-to-chronic mechanistic shift and off-target dopaminergic effects. Nevertheless, the present study demonstrates objective improvements in action restraint by increasing synaptic 5-HT. Given disorders of behavioural control and impulsivity (e.g., ADHD) are associated with 5-HT dysregulation90, exploring potential clinical applications of SSRAs within these populations may prove beneficial.
During behavioural inhibition, increased synaptic 5-HT resulted in a bias for impulse control during aversive interference, alongside a corresponding drop in choice impulsivity. This indicates an optimisation in task strategy during aversive interference, which is congruent with increased 5-HT reducing sensitivity to aversive outcomes. These findings align with the longstanding conceptualisation of 5-HT as an inhibitor which becomes active in aversive contexts91,92. Indeed, in individuals with depression and tryptophan-depleted healthy adults, choice impulsivity increases for explicit negative emotional targets in a go/no-go paradigm93,94,95,96. However, the effects of increased 5-HT on behavioural inhibition reported here were not experimentally confined to aversive contexts; notably, we observed a decrease in choice impulsivity during a control condition without affective interference. Potentially then, 5-HT performs an active role of limiting impulsive action across emotional and neutral contexts, but this is amplified in aversive contexts.
Increasing synaptic 5-HT also enhanced retrieval and speed of processing during memory tasks involving verbal information. Observable changes in verbal but not visuospatial learning is reliably observed following TRP depletion12. The effects of tryptophan depletion on complex verbal working memory are less clear, owing to insufficient studies in this area. SSRI challenge, however, leads to highly variable effects on memory function; while improvements have been observed, typically null findings are reported21,22,97. Unlike fenfluramine, the threshold of synaptic 5-HT required for observable change may not be achieved during the brief SSRI regimen of most studies ( ≤ 7 days), where the problem of autoreceptor supersensitivity persists39,98. Importantly, 5-HT receptor subtypes strongly associated with memory functioning (i.e., 5-HT3,4,6 receptors) have significantly lower binding affinities for endogenous 5-HT relative to other 5-HT receptors (e.g., 5-HT1A,B,D,E,F; 5-HT2A-C)99,100,101,102. Thus, crossing a putative 5-HT concentration threshold may be required to observe change in memory function, potentially explaining our findings.
As with most psychoactive substances, including SSRIs103,104,105,106,107, the SSRA fenfluramine may produce ancillary off-target pharmacological effects alongside its primary mechanism. While retaining neurotransmitter selectivity for the serotonergic system at low doses, fenfluramine has modest affinity for the 5-HTR subtypes and sigma-1 (σ1) receptors. In the case of 5-HTR, these effects appear to be specific to 5-HT2A/B/CR108, while there is inconclusive evidence of the involvement of other receptors, such as 5-HT4109; however, the binding affinity of fenfluramine for 5-HT2A/B/CR is at most <1% of that of competitive endogenous 5-HT100,108,110. Indeed, given the high concentration of 5-HT following exocytic release and the finite availability of 5-HT receptors38,55,111, the resulting ancillary effects of 5-HTR agonism/antagonism may be negligible. Moreover, fenfluramine appears to be both a positive allosteric modulator and antagonist of σ1R112,113; while endogenous neurosteroid sigma-1 agonists may be potentiated through positive allosteric modulation, leading to improvements in cognitive ability113,114,115, it is unclear to what extent this effect may be offset by the sigma-1 antagonist properties of fenfluramine in the healthy brain112,115,116,117.
Pharmacodynamic data on norfenfluramine, the neuroactive metabolite of fenfluramine, is limited; in one study, d-norfenfluramine administration in mice resulted in small increases (relative to 5-HT) of synaptic noradrenaline118. In two further studies, however, noradrenaline levels were unaltered by d-norfenfluramine and dl-norfenfluramine administration119,120. Moreover, dl-fenfluramine administration in humans produces plasma concentrations of dl-fenfluramine and dl-norfenfluramine at a 1:3 ratio, respectively, with only a fraction of that being d-norfenfluramine118,121. Nevertheless, contrasting the neurobehavioural profile of SSRA fenfluramine with S-enantiomers (selective to 5-HT) of SRAs such as 4-methyl-N-methylcathinone39, once clinically available, could provide further insight. Beyond this, while there is human evidence of in vivo serotonergic modulation by d-fenfluramine with [18F]altanserin PET57, further human PET/SPECT investigations of low dose fenfluramine may offer additional insights into its neurochemical profile. Finally, when interpreting the present findings in the context of past studies of TRP, worth noting is the disputed contribution of non-serotonergic modulation via the kynurenine pathway from TRP107,122.
Methodological differences across studies complicate the interpretation of findings, past and present, within a broader literature context. While earlier we mentioned key differences in response inhibition paradigms (i.e., Stop Signal vs Go/No-Go tasks), paradigms which capture reinforcement learning in humans also splinter in a manner which hinders direct comparison. Notably, reinforcement learning paradigms with or without components such as reversal or model-free/model-based learning involve different computational models and neural pathways64,92,123,124,125,126,127; consequentially, model-free learning implemented in the present work is challenging to compare with TRP/SSRI work involving model-based and reversal learning128,129,130. For example, previous work suggests 5-HT depletion (via pharmacological lesioning) modulates reinforcement sensitivity to misleading punishments and rewards during reversal learning131. It is important, therefore, to interpret the present findings within the context of its methodological fit to past literature.
In summary, we demonstrate direct effects of increased synaptic serotonin on human behaviour, underlining its role in guiding decision-making within aversive and neutral contexts. In aversive contexts, increased synaptic serotonin appears to reduce sensitivity for loss outcomes, and promotes a bias toward impulse control during behavioural inhibition. In neutral contexts, increased synaptic serotonin appears to enhance behavioural inhibition by promoting cautious decisions, as well as enhancing memory recall for verbal information.
Not only do the present findings offer implications for longstanding theories of central serotonin, but they also demonstrate the promise of the SSRA as an experimental probe, furthering the scope of fundamental work which aims to characterise the involvement of serotonin in human behaviour, and its contribution to psychiatric aetiology in clinical samples.
Given the prominence of impaired cognition and aversive/negative emotional biases as transdiagnostic targets within psychiatry (e.g., unipolar and bipolar depression; schizophrenia)21,71,132,133,134,135, investigating the therapeutic potential of the SSRA in clinical populations may be worthwhile. Such investigations may allow greater targeting of specific neurocognitive mechanisms across disorders in the absence of widespread, and often unwanted, effects including emotional blunting.
Methods
Participants and design
Fifty-six participants (28:28, SSRA:placebo; mean age = 20.2) were randomised to take part in the study. Recruitment occurred between June 2021 and June 2022. Potential participants were screened to exclude those who had recently used recreational drugs (3-month wash-out, except MDMA which had a wash out period of ≥ 1 year), who were pregnant, trying to become pregnant, or who were currently breastfeeding. All participants had a BMI between 18–30 and were fluent speakers of English. For full exclusion and inclusion criteria, please see Supplementary Methods. For full details of the recruitment process, see the study CONSORT flow diagram (Fig. 7). The final sample consisted of 53 young, non-clinical participants (mean age = 20.15; 32 female) which were allocated to administration of serotonin releasing agent fenfluramine (n = 26) or placebo (n = 27) for subchronic administration (8 ± 1 days). All participants in the final sample attended sessions before treatment and at follow-up. Participants were reimbursed 175 GBP upon completion of the study. Participants were requested to self-report on gender, and the potential covariance of gender with treatment effects was investigated post-hoc. The detailed results of these analyses are provided in the Supplementary Tables 5–7.
Eligible participants were randomised to administration of SSRA fenfluramine hydrochloride (15 mg b.i.d.; racemic mixture) or placebo for subchronic administration (8 ± 1 days), in a double-blind design. Both fenfluramine and the placebo were administered orally in a flavoured aqueous solution, with the placebo lacking an active pharmaceutical ingredient. Randomisation was performed by the Clinical Pharmacy Support Unit, Oxford Health NHS Foundation Trust (Oxfordshire, United Kingdom) using a stratified block randomisation algorithm, with stratification for gender and task stimulus version (for further details on task stimulus versions, see Supplementary Methods).
The study was approved by the University of Oxford Central University Research Ethics Committee (MSD-IDREC reference R69642/RE004) and pre-registered on the National Institute of Health Clinical Trials Database (NCT05026398). All primary outcomes within the pre-registered are reported within the current paper. Prior to study participation, participants provided informed consent. All study visits were conducted at the Department of Psychiatry, University of Oxford.
Procedure
Participants undertook two screening visits to assess study eligibility. In the first session, medical history and current medication use was assessed and the Structured Clinical Interview for DSM-V was conducted to screen for current or past psychiatric illness. In the second session, cardiovascular health (blood pressure; electrocardiography), renal and liver health (liver function, urea, and electrolyte blood tests) were assessed, and drug and pregnancy urine tests were performed. Eligible participants attended two study visits, baseline and post-intervention occurring 7, 8 or 9 days later. This study period was scheduled to avoid the premenstrual week for female participants. At baseline, participants completed a battery of cognitive and emotional computer tasks and questionnaires. Participants were then given their first dose of the SSRA or placebo, and monitored for three hours during which regular blood pressure and observational checks were made. Measurements of early changes in salivary cortisol concentration were taken to investigate potential modulation of HPA axis function that has been previously observed in studies of acute high dose fenfluramine136. Saliva samples were collected immediately before initial dose, one hour post-dose, and three hours post-dose. These samples were immunoassayed for cortisol levels over linear calibration curves (for further details, see Supplementary Methods). After the initial dose visit, participants were asked to independently take the fenfluramine or placebo daily, in addition to completing daily questionnaires (see Questionnaires measures section). At the post-intervention visit, participants completed the same task and questionnaire battery as at baseline and were then requested to estimate their allocation prior to debriefing.
Questionnaire measures
At each study visit, participants completed self-report questionnaires measuring affect, mood, anxiety, subjective cognitive functioning, and side-effects; the Spielberger State-Trait Anxiety Inventory [STAI-T], Beck Depression Inventory II [BDI], Positive and Negative Affect Schedule [PANAS], Visual Analogue Scale [VAS], Perceived Deficit Questionnaire – Depression [PDQ-D], and side effects profile questionnaire. Participants completed the VAS and side effects questionnaire once per day from the baseline visit until the follow-up visit.
Cognitive and Emotional Task Battery
Participants undertook an extensive cognitive and emotional task battery at both the initial dose visit (baseline) and follow-up visit. Participants undertook the following tasks in order: 1) Auditory Verbal Learning Task (AVLT) (Fig. 6D) – a measure of declarative memory encoding and retrieval where accuracy of recall was the measured outcome. Behavioural data for the task consisted of N = 51 individuals (mean age = 20.15; 31 female). 2) Affective Interference Go/No-Go Task (Fig. 3A) – a measure of behavioural inhibition under affective interference (positive [happy faces], aversive/negative [fearful faces], and neutral distractors) where accuracy of inhibited response to no-go trials (response inhibition), accuracy, and response time for go trials (an index of impulsivity137 were the non-model outcome measures. The block design of the task allows for analysis of set-shifting effects (executive shifting for task condition rule changes) on accuracy and response time. Participant task data was fit to a computational drift diffusion model (see Supplementary Methods for further details) which provided the following model parameters: boundary separation, initial choice bias, non-decision time, drift rate, and drift criterion. Behavioural and computational data for the task consisted of N = 50 individuals (mean age = 20.22; 32 female). 3) Verbal n-back task (Fig. 6A) – a measure of complex verbal working memory where accuracy and response time to target letters (i.e., matching a letter which appeared n-back [0, 1, 2, or 3] trials ago) were the outcome measures. Behavioural data for the n-back consisted of N = 50 individuals (mean age = 20.1; 31 female). 4) Probabilistic instrumental learning task ([Fig. 2A]; adapted from63) – a measure of reward and loss sensitivity during instrumental learning, which produced non-model outcome measures which were fit to computational reinforcement learning models. Non-model outcomes were optimal choice outcome (i.e., selecting the stimulus with a higher probability of a favourable outcome under each task condition: wins during win trials and no change during loss trials) and response time. The main computational model parameters were outcome sensitivity and learning rate, where learning rate was estimated separately for both win and loss trials (see Supplementary Methods for further details). Behavioural and computational data for the task consisted of N = 53 individuals (mean age = 20.15; 32 female). 5) Oxford Memory Task – a measure of visuospatial working memory which included localisation speed and stimulus selection accuracy outcomes. Behavioural data for the task consisted of N = 51 individuals (mean age = 20.16; 31 female). 6) Contextual cueing task – a measure of implicit learning and visual search ability where the outcome measure was accuracy and response times under novel/implicit cueing conditions. Behavioural data for the task consisted of N = 53 individuals (mean age = 20.15; 32 female). Full details of tasks included in this battery are included in the Supplementary Methods.
Statistical analysis
Data pre-processing and statistical analyses were carried out using R Software (version 4.3.1), and computational modelling was undertaken using MATLAB (R2022a), Docker (version 4.22.0), and Python (version 3.8.8); required software dependencies and packages are listed within the Supplementary Methods. Homogeneity in demographic variables across allocation groups was assessed using chi-squared independence tests (categorical, binary variables) and two-tailed independent t-tests (continuous, discrete variables). The effect of the SSRA on outcomes across the task battery and questionnaire ratings was analysed using between-groups (SSRA vs. placebo) mixed effects ANCOVA models (two-tailed) on post-intervention data, with baseline visit performance serving as a regressor and participant as a random effect for nested multilevel data structures (e.g., multiple task conditions). The approach of using baseline score as a regressor in this manner was selected as it allows isolation of the group effect from potential sources of bias (e.g., learning and practice effects)138,139. In comparison with other baseline-adjustment techniques, such as change score between before and after intervention, baseline-adjusted ANCOVA yields greater statistical efficiency and precision (irrespective of baseline balance/imbalance)140,141. Planned comparisons were carried out on outcome measures collected at follow-up using two-tailed estimated marginal means tests, where estimates are reported alongside standard mean error. Family-wise error for estimated marginal means tests was adjusted using the Bonferroni-Holm procedure. Models were assessed through histogram examination of standardised residuals to determine normality of distribution (non-normal computational values were log-transformed), and homogeneity of regression slopes and covariate independence were tested by modelling covariate interactions across factors (covariate × group × task condition). Two-tailed effect sizes metrics are reported for significant ANCOVA (partial eta squared, ηp2) and EMM (Cohen’s d, d) models alongside corresponding 95% confidence intervals (for effect size calculations, see Supplementary Methods). Post-hoc correlational analyses on the Affective Interference Go/No-Go task were performed using two-tailed Pearson’s product-moment correlation analysis. Baseline-adjusted ANCOVA models were generated post-hoc to investigate the potential covariance of gender with group effects across study outcomes, which are reported in Supplementary Tables 5–7. Group allocation guesses were analysed using chi-squared independence tests. In addition to ANCOVA analysis of questionnaire data at follow-up, daily questionnaire data (VAS and side effects profile) was joined longitudinally with initial dose and follow-up visit data and analysed using linear mixed effects models with restricted maximum likelihood estimation where time served as a regressor. Salivary cortisol was analysed across three timepoints (before dose, 1- and 3-hr post-dose) using linear mixed effects modelling with time as a regressor. Baseline task battery, cortisol and self-report questionnaire data are included in Supplementary Figs. 5, 7–9, 11–13, 16, 18, 20–23, 25. All inferential analyses were carried out at the 0.05 alpha level. A priori sample size calculation (Power [1 - β error probability]: 80%) determined that a sample of N = 52 (26 per group) was required to undertake two-tailed between-groups analysis.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Raw and modelled datasets generated for this study have been deposited on Zenodo142 and Github: https://github.com/mjcolwell/SSRA_human_behaviour_data_and_code.
Code availability
The code used to undertake data preprocessing, computational modelling, and inferential analyses in this study has been deposited on Zenodo142 and GitHub: https://github.com/mjcolwell/SSRA_human_behaviour_data_and_code.
References
Whitaker-Azmitia, P. M. The Discovery of Serotonin and its Role in Neuroscience. Neuropsychopharmacology 21, 2–8 (1999).
Cowen, P. J. & Browning, M. What has serotonin to do with depression? World Psychiatry 14, 158 (2015).
Olivier, B. Serotonin: A never-ending story. Eur. J. Pharm. 753, 2–18 (2015).
Voigt, J.-P. & Fink, H. Serotonin controlling feeding and satiety. Behav. Brain Res. 277, 14–31 (2015).
Olivier, B., Van Oorschot, R. & Waldinger, M. D. Serotonin, serotonergic receptors, selective serotonin reuptake inhibitors and sexual behaviour. Int Clin. Psychopharmacol. 13, S9–S14 (1998).
van der Plasse, G. et al. Medial prefrontal serotonin in the rat is involved in goal-directed behaviour when affect guides decision making. Psychopharmacol. (Berl.) 195, 435–449 (2007).
Roberts, C., Sahakian, B. J. & Robbins, T. W. Psychological mechanisms and functions of 5-HT and SSRIs in potential therapeutic change: lessons from the serotonergic modulation of action selection, learning, affect, and social cognition. Neurosci. Biobehav Rev. 119, 138–167 (2020).
Crockett, M. J., Clark, L., Apergis-Schoute, A. M., Morein-Zamir, S. & Robbins, T. W. Serotonin Modulates the Effects of Pavlovian Aversive Predictions on Response Vigor. Neuropsychopharmacology 37, 2244–2252 (2012).
Soubrié, P. Reconciling the role of central serotonin neurons in human and animal behavior. Behav. Brain Sci. 9, 319–335 (1986).
Deakin, J. F. W. Roles of serotonergic systems in escape, avoidance and other behaviours. Theory Psychopharmacol. 2, 149–193 (1983).
McEntee, W. J. & Crook, T. H. Serotonin, memory, and the aging brain. Psychopharmacol. (Berl.) 103, 143–149 (1991).
Cowen, P. & Sherwood, A. C. The role of serotonin in cognitive function: evidence from recent studies and implications for understanding depression. J. Psychopharmacol. 27, 575–583 (2013).
Scholl, J. et al. Beyond negative valence: 2-week administration of a serotonergic antidepressant enhances both reward and effort learning signals. PLoS Biol. 15, e2000756 (2017).
Michely, J., Eldar, E., Erdman, A., Martin, I. M. & Dolan, R. J. Serotonin modulates asymmetric learning from reward and punishment in healthy human volunteers. Commun. Biol. 5, 812 (2022).
Langley, C. et al. Chronic escitalopram in healthy volunteers has specific effects on reinforcement sensitivity: a double-blind, placebo-controlled semi-randomised study. Neuropsychopharmacology 48, 664–670 (2023).
Skandali, N. et al. Dissociable effects of acute SSRI (escitalopram) on executive, learning and emotional functions in healthy humans. Neuropsychopharmacology 43, 2645–2651 (2018).
Eagle, D. M., Bari, A. & Robbins, T. W. The neuropsychopharmacology of action inhibition: cross-species translation of the stop-signal and go/no-go tasks. Psychopharmacol. (Berl.) 199, 439–456 (2008).
Macoveanu, J. et al. Serotonin 2 A Receptors, Citalopram and Tryptophan-Depletion: a Multimodal Imaging Study of their Interactions During Response Inhibition. Neuropsychopharmacology 38, 996–1005 (2013).
Guitart-Masip, M. et al. Differential, but not opponent, effects of l-DOPA and citalopram on action learning with reward and punishment. Psychopharmacol. (Berl.) 231, 955–966 (2014).
Mendelsohn, D., Riedel, W. J. & Sambeth, A. Effects of acute tryptophan depletion on memory, attention and executive functions: A systematic review. Neurosci. Biobehav Rev. 33, 926–952 (2009).
Colwell, M. J. et al. Pharmacological targeting of cognitive impairment in depression: recent developments and challenges in human clinical research. Transl. Psychiatry 12, 484 (2022).
Cassel, J.-C. CHAPTER 3.9 - Experimental Studies on the Role(s) of Serotonin in Learning and Memory Functions. in Handbook of Behavioral Neuroscience (eds Müller, C. P. & Jacobs, B. L.) vol. 21 429–447 (Elsevier, 2010).
Evers, E. A. T., Van der Veen, F. M., Fekkes, D. & Jolles, J. Serotonin and cognitive flexibility: neuroimaging studies into the effect of acute tryptophan depletion in healthy volunteers. Curr. Med Chem. 14, 2989–2995 (2007).
Barton, C. L. & Hutson, P. H. Inhibition of hippocampal 5‐HT synthesis by fluoxetine and paroxetine: Evidence for the involvement of both 5‐HT1A and 5‐HT1B/D autoreceptors. Synapse 31, 13–19 (1999).
El Mansari, M., Sánchez, C., Chouvet, G., Renaud, B. & Haddjeri, N. Effects of Acute and Long-Term Administration of Escitalopram and Citalopram on Serotonin Neurotransmission: an In Vivo Electrophysiological Study in Rat Brain. Neuropsychopharmacology 30, 1269–1277 (2005).
de Groote, L., Olivier, B. & Westenberg, H. G. Extracellular serotonin in the prefrontal cortex is limited through terminal 5-HT1B autoreceptors: a microdialysis study in knockout mice. Psychopharmacol. (Berl.) 162, 419–424 (2002).
Zhou, F.-M. et al. Corelease of Dopamine and Serotonin from Striatal Dopamine Terminals. Neuron 46, 65–74 (2005).
Sekine, Y., Suzuki, K., Ramachandran, P. V., Blackburn, T. P. & Ashby, J. R. C. R. Acute and repeated administration of fluoxetine, citalopram, and paroxetine significantly alters the activity of midbrain dopamine neurons in rats: An in vivo electrophysiological study. Synapse 61, 72–77 (2007).
Yoshino, T., Nisijima, K., Katoh, S., Yui, K. & Nakamura, M. Tandospirone potentiates the fluoxetine-induced increases in extracellular dopamine via 5-HT1A receptors in the rat medial frontal cortex. Neurochem. Int. 40, 355–360 (2002).
Shuto, T. et al. Obligatory roles of dopamine D1 receptors in the dentate gyrus in antidepressant actions of a selective serotonin reuptake inhibitor, fluoxetine. Mol. Psychiatry 25, 1229–1244 (2020).
Dremencov, E., el Mansari, M. & Blier, P. Effects of sustained serotonin reuptake inhibition on the firing of dopamine neurons in the rat ventral tegmental area. J. Psychiatry Neurosci. 34, 223–229 (2009).
Montgomery, A. M. J., Rose, I. C. & Herberg, L. J. 5-HT1A agonists and dopamine: the effects of 8-OH-DPAT and buspirone on brain-stimulation reward. J. Neural Transm. / Gen. Sect. JNT 83, 139–148 (1991).
Pozzi, L., Invernizzi, R., Garavaglia, C. & Samanin, R. Fluoxetine increases extracellular dopamine in the prefrontal cortex by a mechanism not dependent on serotonin: a comparison with citalopram. J. Neurochem. 73, 1051–1057 (1999).
Kobayashi, K., Haneda, E., Higuchi, M., Suhara, T. & Suzuki, H. Chronic Fluoxetine Selectively Upregulates Dopamine D1-Like Receptors in the Hippocampus. Neuropsychopharmacology 37, 1500–1508 (2012).
Kitaichi, Y. et al. Sertraline increases extracellular levels not only of serotonin, but also of dopamine in the nucleus accumbens and striatum of rats. Eur. J. Pharm. 647, 90–96 (2010).
Müller, F. et al. MDMA-induced changes in within-network connectivity contradict the specificity of these alterations for the effects of serotonergic hallucinogens. Neuropsychopharmacology 46, 545–553 (2021).
Mustafa, N. S. et al. MDMA and the Brain: A Short Review on the Role of Neurotransmitters in Neurotoxicity. Basic Clin. Neurosci. 11, 381 (2020).
Rothman, R. B. & Baumann, M. H. Serotonin releasing agents: neurochemical, therapeutic and adverse effects. Pharm. Biochem Behav. 71, 825–836 (2002).
Mayer, F. P. et al. Serotonin-releasing agents with reduced off-target effects. Mol. Psychiatry https://doi.org/10.1038/s41380-022-01843-w (2022).
Udo de Haes, J. I., Harada, N., Elsinga, P. H., Maguire, R. P. & Tsukada, H. Effect of fenfluramine‐induced increases in serotonin release on [18 F] MPPF binding: A continuous infusion PET study in conscious monkeys. Synapse 59, 18–26 (2006).
Knupp, K. G. et al. Efficacy and Safety of Fenfluramine for the Treatment of Seizures Associated With Lennox-Gastaut Syndrome: A Randomized Clinical Trial. JAMA Neurology 79, 554–564 (2022).
Finnema, S. J., Varrone, A., Hwang, T.-J., Halldin, C. & Farde, L. Confirmation of fenfluramine effect on 5-HT1B receptor binding of [11 C] AZ10419369 using an equilibrium approach. J. Cereb. Blood Flow. Metab. 32, 685–695 (2012).
Duval, F. et al. Serotonergic and noradrenergic function in depression: clinical correlates. Dialogues Clin. Neurosci. 2, 299–308 (2022).
Marona-Lewicka, D. & Nichols, D. E. Behavioral effects of the highly selective serotonin releasing agent 5-methoxy-6-methyl-2-aminoindan. Eur. J. Pharm. 258, 1–13 (1994).
Fattaccini, C. M., Gozlan, H. & Hamon, M. Differential effects of d-fenfluramine and p-chloroamphetamine on H7512-induced depletion of 5-hydroxytryptamine and dopamine in the rat brain. Neuropharmacology 30, 15–23 (1991).
Bonanno, G., Fassio, A., Severi, P., Ruelle, A. & Raiteri, M. Fenfluramine Releases Serotonin from Human Brain Nerve Endings by a Dual Mechanism. J. Neurochem 63, 1163–1166 (1994).
Baumann, M. H., Ayestas, M. A., Dersch, C. M. & Rothman, R. B. 1-(m-Chlorophenyl)piperazine (mCPP) Dissociates In Vivo Serotonin Release from Long-Term Serotonin Depletion in Rat Brain. Neuropsychopharmacology 24, 492–501 (2001).
Leonardi, E. T. K. & Azmitia, E. C. MDMA (Ecstasy) Inhibition of MAO Type A and Type B: Comparisons with Fenfluramine and Fluoxetine (Prozac). Neuropsychopharmacology 10, 231–238 (1994).
Campbell, S. & MacQueen, G. The role of the hippocampus in the pathophysiology of major depression. J. Psychiatry Neurosci. 29, 417–426 (2004).
Lanzenberger, R. et al. Prediction of SSRI treatment response in major depression based on serotonin transporter interplay between median raphe nucleus and projection areas. Neuroimage 63, 874–881 (2012).
Jørgensen, L. M. et al. Cerebral 5-HT release correlates with [11 C] Cimbi36 PET measures of 5-HT2A receptor occupancy in the pig brain. J. Cereb. Blood Flow. Metab. 37, 425–434 (2017).
Hume, S. et al. Effect of 5‐HT on binding of [11 C] WAY 100635 to 5‐HT1A receptors in rat brain, assessed using in vivo microdialysis and PET after fenfluramine. Synapse 41, 150–159 (2001).
Balcioglu, A. & Wurtman, R. J. Effects of fenfluramine and phentermine (fen–phen) on dopamine and serotonin release in rat striatum: in vivo microdialysis study in conscious animals. Brain Res 813, 67–72 (1998).
Sabol, K. E., Richards, J. B. & Seiden, L. S. Fluoxetine attenuates the DL-fenfluramine-induced increase in extracellular serotonin as measured by in vivo dialysis. Brain Res 585, 421–424 (1992).
Baumann, M. H. et al. Effects of phentermine and fenfluramine on extracellular dopamine and serotonin in rat nucleus accumbens: therapeutic implications. Synapse 36, 102–113 (2000).
Zaczek, R. et al. Effects of repeated fenfluramine administration on indices of monoamine function in rat brain: pharmacokinetic, dose response, regional specificity and time course data. J. Pharmacol. Exp. Therapeutics 253, 104–112 (1990).
Quednow, B. B. et al. Assessment of serotonin release capacity in the human brain using dexfenfluramine challenge and [18 F]altanserin positron emission tomography. Neuroimage 59, 3922–3932 (2012).
Ikoma, Y. et al. Measurement of changes in endogenous serotonin level by positron emission tomography with [18 F] altanserin. Ann. Nucl. Med 35, 955–965 (2021).
Moore, P. et al. Clinical and Physiological Consequences of Rapid Tryptophan Depletion. Neuropsychopharmacology 23, 601–622 (2000).
Geurts, D. E. M., Huys, Q. J. M., den Ouden, H. E. M. & Cools, R. Serotonin and Aversive Pavlovian Control of Instrumental Behavior in Humans. J. Neurosci. 33, 18932 (2013).
Robinson, O. J., Cools, R. & Sahakian, B. J. Tryptophan depletion disinhibits punishment but not reward prediction: implications for resilience. Psychopharmacol. (Berl.) 219, 599–605 (2012).
Cools, R., Robinson, O. J. & Sahakian, B. Acute Tryptophan Depletion in Healthy Volunteers Enhances Punishment Prediction but Does not Affect Reward Prediction. Neuropsychopharmacology 33, 2291–2299 (2008).
Pessiglione, M., Seymour, B., Flandin, G., Dolan, R. J. & Frith, C. D. Dopamine-dependent prediction errors underpin reward-seeking behaviour in humans. Nature 442, 1042–1045 (2006).
Halahakoon, D. C. et al. Pramipexole Enhances Reward Learning by Preserving Value Estimates. Biol. Psychiatry https://doi.org/10.1016/j.biopsych.2023.05.023 (2023).
Dalley, J. W., Everitt, B. J. & Robbins, T. W. Impulsivity, compulsivity, and top-down cognitive control. Neuron 69, 680–694 (2011).
Campanella, S. et al. Short-term impact of tDCS over the right inferior frontal cortex on impulsive responses in a go/no-go task. Clin. EEG Neurosci. 49, 398–406 (2018).
Wiecki, T. V., Sofer, I. & Frank, M. J. HDDM: Hierarchical Bayesian estimation of the drift-diffusion model in Python. Front Neuroinform. 14, 55610 (2013).
Hindi Attar, C., Finckh, B. & Büchel, C. The influence of serotonin on fear learning. PLOS ONE 7, e42397 (2012).
Hebart, M. N. & Gläscher, J. Serotonin and dopamine differentially affect appetitive and aversive general Pavlovian-to-instrumental transfer. Psychopharmacol. (Berl.) 232, 437–451 (2015).
McCabe, C., Mishor, Z., Cowen, P. J. & Harmer, C. J. Diminished Neural Processing of Aversive and Rewarding Stimuli During Selective Serotonin Reuptake Inhibitor Treatment. Biol. Psychiatry 67, 439–445 (2010).
Vrieze, E. et al. Reduced reward learning predicts outcome in major depressive disorder. Biol. Psychiatry 73, 639–645 (2013).
Giambalvo, C. T. & Price, L. H. Effects of fenfluramine and antidepressants on protein kinase C activity in rat cortical synaptoneurosomes. Synapse 50, 212–222 (2003).
Raiteri, M., Bonanno, G. & Vallebuona, F. In vitro and in vivo effects of d-fenfluramine: no apparent relation between 5-hydroxytryptamine release and hypophagia. J. Pharmacol. Exp. Therapeutics 273, 643 (1995).
Kannengiesser, M.-H., Hunt, P. F. & Raynaud, J.-P. Comparative actionof fenfluramine on the uptake and release of serotonin and dopamine. Eur. J. Pharm. 35, 35–43 (1976).
Owens, M. J., Knight, D. L. & Nemeroff, C. B. Second-generation SSRIs: human monoamine transporter binding profile of escitalopram and R-fluoxetine. Biol. Psychiatry 50, 345–350 (2001).
McDevitt, R. A. et al. Serotonergic versus Nonserotonergic Dorsal Raphe Projection Neurons: Differential Participation in Reward Circuitry. Cell Rep. 8, 1857–1869 (2014).
McHugh, S. B. et al. Aversive prediction error signals in the amygdala. J. Neurosci. 34, 9024–9033 (2014).
Sengupta, A. & Holmes, A. A Discrete Dorsal Raphe to Basal Amygdala 5-HT Circuit Calibrates Aversive Memory. Neuron 103, 489–505.e7 (2019).
Cohen, J. Y., Amoroso, M. W. & Uchida, N. Serotonergic neurons signal reward and punishment on multiple timescales. Elife 4, e06346 (2015).
Zhong, W., Li, Y., Feng, Q. & Luo, M. Learning and stress shape the reward response patterns of serotonin neurons. J. Neurosci. 37, 8863–8875 (2017).
Paquelet, G. E. et al. Single-cell activity and network properties of dorsal raphe nucleus serotonin neurons during emotionally salient behaviors. Neuron 110, 2664–2679 (2022).
Wang, H.-L. et al. Dorsal raphe dual serotonin-glutamate neurons drive reward by establishing excitatory synapses on VTA mesoaccumbens dopamine neurons. Cell Rep. 26, 1128–1142 (2019).
Athilingam, J. C., Ben-Shalom, R., Keeshen, C. M., Sohal, V. S. & Bender, K. J. Serotonin enhances excitability and gamma frequency temporal integration in mouse prefrontal fast-spiking interneurons. Elife 6, e31991 (2017).
Pompeiano, M., Palacios, J. M. & Mengod, G. Distribution and cellular localization of mRNA coding for 5-HT1A receptor in the rat brain: correlation with receptor binding. J. Neurosci. 12, 440–453 (1992).
De Almeida, J. & Mengod, G. Quantitative analysis of glutamatergic and GABAergic neurons expressing 5‐HT2A receptors in human and monkey prefrontal cortex. J. Neurochem 103, 475–486 (2007).
Hentall, I. D., Kurle, P. J. & White, T. R. Correlations between serotonin level and single-cell firing in the rat’s nucleus raphe magnus. Neuroscience 95, 1081–1088 (1999).
Scuvée-Moreau, J. & Dresse, A. Influence of fenfluramine and norfenfluramine stereoisomers on the firing rate of central monoaminergic neurons in the rat. Eur. J. Pharm. 179, 211–215 (1990).
Chamberlain, S. R. et al. Neurochemical modulation of response inhibition and probabilistic learning in humans. Science (1979) 311, 861–863 (2006).
Del-Ben, C. M. et al. The effect of citalopram pretreatment on neuronal responses to neuropsychological tasks in normal volunteers: an FMRI study. Neuropsychopharmacology 30, 1724–1734 (2005).
Winstanley, C. A. The utility of rat models of impulsivity in developing pharmacotherapies for impulse control disorders. Br. J. Pharm. 164, 1301–1321 (2011).
Dayan, P. & Huys, Q. J. M. Serotonin in affective control. Annu Rev. Neurosci. 32, 95–126 (2009).
Robinson, O. J. & Roiser, J. P. The role of serotonin in aversive inhibition: behavioural, cognitive and neural perspectives. Psychopathol. Rev. 3, 29–40 (2016).
Roiser, J. P. et al. The Effect of Acute Tryptophan Depletion on the Neural Correlates of Emotional Processing in Healthy Volunteers. Neuropsychopharmacology 33, 1992–2006 (2008).
Murphy, F., Smith, K., Cowen, P., Robbins, T. & Sahakian, B. The effects of tryptophan depletion on cognitive and affective processing in healthy volunteers. Psychopharmacol. (Berl.) 163, 42–53 (2002).
Erickson, K. et al. Mood-congruent bias in affective go/no-go performance of unmedicated patients with major depressive disorder. Am. J. Psychiatry 162, 2171–2173 (2005).
Crockett, M. J., Clark, L. & Robbins, T. W. Reconciling the role of serotonin in behavioral inhibition and aversion: acute tryptophan depletion abolishes punishment-induced inhibition in humans. J. Neurosci. 29, 11993–11999 (2009).
Harmer, C. J., Bhagwagar, Z., Cowen, P. J. & Goodwin, G. M. Acute administration of citalopram facilitates memory consolidation in healthy volunteers. Psychopharmacol. (Berl.) 163, 106–110 (2002).
Cowen, P. J. Psychopharmacology of 5-HT1A receptors. Nucl. Med Biol. 27, 437–439 (2000).
Hartig, P. R. Molecular biology of 5-HT receptors. Trends Pharm. Sci. 10, 64–69 (1989).
Paterson, L. M., Tyacke, R. J., Nutt, D. J. & Knudsen, G. M. Measuring endogenous 5-HT release by emission tomography: promises and pitfalls. J. Cereb. Blood Flow. Metab. 30, 1682–1706 (2010).
Murphy, S. E., Wright, L. C., Browning, M., Cowen, P. J. & Harmer, C. J. A role for 5-HT4 receptors in human learning and memory. Psychol. Med 50, 2722–2730 (2020).
Smith, J. et al. Vortioxetine reduces BOLD signal during performance of the N-back working memory task: a randomised neuroimaging trial in remitted depressed patients and healthy controls. Mol. Psychiatry 23, 1127–1133 (2018).
Dionisie, V., Filip, G. A., Manea, M. C., Manea, M. & Riga, S. The anti-inflammatory role of SSRI and SNRI in the treatment of depression: a review of human and rodent research studies. Inflammopharmacology 29, 75–90 (2021).
Cruceanu, C., Lopez, J. P., Tsai, W.-T. & Turecki, G. Dysregulation of the glutamatergic receptors after antidepressant treatment in human neural progenitor cells. Mol. Psychiatry 22, 1228–1229 (2017).
Hashimoto, K. Sigma-1 receptors and selective serotonin reuptake inhibitors: clinical implications of their relationship. Cent. Nerv. Syst. Agents Medicinal Chem. (Former. Curr. Medicinal Chem.-Cent. Nerv. Syst. Agents) 9, 197–204 (2009).
Damsa, C. et al. Dopamine-dependent’ side effects of selective serotonin reuptake inhibitors: a clinical review. J. Clin. Psychiatry 65, 1064–1068 (2004).
van Donkelaar, E. L. et al. Mechanism of acute tryptophan depletion: is it only serotonin? Mol. Psychiatry 16, 695–713 (2011).
Fitzgerald, L. W. et al. Possible role of valvular serotonin 5-HT2B receptors in the cardiopathy associated with fenfluramine. Mol. Pharm. 57, 75–81 (2000).
Sourbron, J. & Lagae, L. Serotonin receptors in epilepsy: Novel treatment targets? Epilepsia Open 7, 231 (2022).
di Giovanni, G., di Matteo, V. & Esposito, E. Serotonin-dopamine interaction: experimental evidence and therapeutic relevance. Prog. Brain Res. 172, 122–123 (2008).
Gerhardt, C. C. & van Heerikhuizen, H. Functional characteristics of heterologously expressed 5-HT receptors. Eur. J. Pharm. 334, 1–23 (1997).
Sourbron, J. et al. Serotonergic modulation as effective treatment for Dravet syndrome in a zebrafish mutant model. ACS Chem. Neurosci. 7, 588–598 (2016).
Martin, P. et al. Fenfluramine acts as a positive modulator of sigma-1 receptors. Epilepsy Behav. 105, 106989 (2020).
Hindmarch, I. & Hashimoto, K. Cognition and depression: the effects of fluvoxamine, a sigma‐1 receptor agonist, reconsidered. Hum. Psychopharmacol: Clin. Exp. 25, 193–200 (2010).
Hashimoto, K. Activation of sigma-1 receptor chaperone in the treatment of neuropsychiatric diseases and its clinical implication. J. Pharm. Sci. 127, 6–9 (2015).
Hashimoto, K., Fujita, Y. & Iyo, M. Phencyclidine-Induced Cognitive Deficits in Mice are Improved by Subsequent Subchronic Administration of Fluvoxamine: Role of Sigma-1 Receptors. Neuropsychopharmacology 32, 514–521 (2007).
Guo, L. et al. SKF83959 Attenuates Memory Impairment and Depressive-like Behavior during the Latent Period of Epilepsy via Allosteric Activation of the Sigma-1 Receptor. ACS Chem. Neurosci. 13, 3198–3209 (2022).
Rothman, R. B., Clark, R. D., Partilla, J. S. & Baumann, M. H. (+)-Fenfluramine and its major metabolite,(+)-norfenfluramine, are potent substrates for norepinephrine transporters. J. Pharmacol. Exp. Ther. 305, 1191–1199 (2003).
Invernizzi, R., Fracasso, C., Caccia, S., Garattini, S. & Samanin, R. Effects of intracerebroventricular administration of d-fenfluramine and d-norfenfluramine, as a single injection or 2-HR infusion, on serotonin in brain: Relationship to concentrations of drugs in brain. Neuropharmacology 30, 119–123 (1991).
Scheurink, A. J. W., Leuvenink, H. & Steffens, A. B. Metabolic and hormonal responses to hypothalamic administration of norfenfluramine in rats. Physiol. Behav. 53, 889–898 (1993).
Caccia, S., Conforti, I., Duchier, J. & Garattini, S. Pharmacokinetics of fenfluramine and norfenfluramine in volunteers given d- and dl-fenfluramine for 15 days. Eur. J. Clin. Pharm. 29, 221–224 (1985).
Crockett, M. J. et al. Converging evidence for central 5-HT effects in acute tryptophan depletion. Mol. Psychiatry 17, 121–123 (2012).
Skvortsova, V. et al. A Causal Role for the Pedunculopontine Nucleus in Human Instrumental Learning. Curr. Biol. 31, 943–954.e5 (2021).
Fellows, L. K. & Farah, M. J. Ventromedial frontal cortex mediates affective shifting in humans: evidence from a reversal learning paradigm. Brain 126, 1830–1837 (2003).
Izquierdo, A., Brigman, J. L., Radke, A. K., Rudebeck, P. H. & Holmes, A. The neural basis of reversal learning: an updated perspective. Neuroscience 345, 12–26 (2017).
Browning, M. & Lan, D. What can reinforcement learning models of dopamine and serotonin tell us about the action of antidepressants? Comput. Psychiatry 6, 166 (2022).
Gläscher, J., Daw, N., Dayan, P. & O’Doherty, J. P. States versus rewards: dissociable neural prediction error signals underlying model-based and model-free reinforcement learning. Neuron 66, 585–595 (2010).
Kanen, J. W. et al. Serotonin depletion impairs both Pavlovian and instrumental reversal learning in healthy humans. Mol. Psychiatry 26, 7200–7210 (2021).
Worbe, Y. et al. Valence-dependent influence of serotonin depletion on model-based choice strategy. Mol. Psychiatry 21, 624–629 (2016).
Luo, Q. et al. Comparable roles for serotonin in rats and humans for computations underlying flexible decision-making. Neuropsychopharmacology 49, 600–608 (2024).
Rygula, R. et al. Role of Central Serotonin in Anticipation of Rewarding and Punishing Outcomes: Effects of Selective Amygdala or Orbitofrontal 5-HT Depletion. Cereb. Cortex 25, 3064–3076 (2015).
Bilderbeck, A. C. et al. Associations between mood instability and emotional processing in a large cohort of bipolar patients. Psychol. Med 46, 3151–3160 (2016).
Quarmley, M. et al. Reduced safety processing during aversive social conditioning in psychosis and clinical risk. Neuropsychopharmacology 44, 2247–2253 (2019).
McCutcheon, R. A., Keefe, R. S. E. & McGuire, P. K. Cognitive impairment in schizophrenia: aetiology, pathophysiology, and treatment. Mol. Psychiatry 28, 1902–1918 (2023).
Robinson, L. J. & Nicol Ferrier, I. Evolution of cognitive impairment in bipolar disorder: a systematic review of cross‐sectional evidence. Bipolar Disord. 8, 103–116 (2006).
Schürmeyer, T. H., Brademann, G. & von zur Mühlen, A. Effect of fenfluramine on episodic ACTH and cortisol secretion. Clin. Endocrinol. (Oxf.) 45, 39–45 (1996).
Chowdhury, N. S., Livesey, E. J., Blaszczynski, A. & Harris, J. A. Pathological Gambling and Motor Impulsivity: A Systematic Review with Meta-Analysis. J. Gambl. Stud. 33, 1213–1239 (2017).
Van Breukelen, G. J. P. ANCOVA versus change from baseline had more power in randomized studies and more bias in nonrandomized studies. J. Clin. Epidemiol. 59, 920–925 (2006).
Wright, D. B. & Osborne, J. E. Dissociation, cognitive failures, and working memory. Am. J. Psychol. 118, 103–114 (2005).
Egbewale, B. E., Lewis, M. & Sim, J. Bias, precision and statistical power of analysis of covariance in the analysis of randomized trials with baseline imbalance: a simulation study. BMC Med. Res. Methodol. 14, 1–12 (2014).
Clifton, L. & Clifton, D. A. The correlation between baseline score and post-intervention score, and its implications for statistical analysis. Trials 20, 1–6 (2019).
Colwell, M. J. SSRA and Human Behaviour Study - Associated Raw Data and Scripts. Zenodo https://doi.org/10.5281/zenodo.12599247 (2024).
Colwell, M. J., Murphy, S. & Harmer, C. J. Emotional Go/No-Go Task (Oxford). https://doi.org/10.5281/ZENODO.6207865 (2022).
Conley, M. I. et al. The racially diverse affective expression (RADIATE) face stimulus set. Psychiatry Res. 270, 1059–1067 (2018).
Tottenham, N. et al. The NimStim set of facial expressions: Judgments from untrained research participants. Psychiatry Res. 168, 242–249 (2009).
Halgren, E., Raij, T., Marinkovic, K., Jousmäki, V. & Hari, R. Cognitive response profile of the human fusiform face area as determined by MEG. Cereb. Cortex 10, 69–81 (2000).
Ratcliff, R., Huang-Pollock, C. & McKoon, G. Modeling individual differences in the go/no-go task with a diffusion model. Decision 5, 42 (2018).
de Gee, J. W. et al. Pupil-linked phasic arousal predicts a reduction of choice bias across species and decision domains. Elife 9, e54014 (2020).
Acknowledgements
This study was funded by a grant from Zogenix International Ltd. (awarded to C.H.), prior to merge with UCB Pharma, and supported by the NIHR Oxford Health Biomedical Research Centre and the NIHR Oxford Cognitive Health Clinical Research Facility. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health. We thank Dr Sandra Tamm, Dr Angharad de Cates, Dr Sara Costi, and Dr Alexander Smith for their assistance with medical screening and diagnostics procedures. We thank Prof Valerie Voon for suggestions for data analysis. We thank Dr Margarita Chibalina for their assistance with biological sample handling and processing. We thank Dr Jan Willem de Gee for publishing openly available computational modelling scripts. We thank Tara Pusinelli for assistance with data collection and entry. We thank Sorcha Hamilton for help with verifying code reproducibility.
Author information
Authors and Affiliations
Contributions
M.C., S.M., C.H. and P.C. designed the study. M.C., H.T., H.I.C. and C.E.W. undertook data collection. M.C., H.T., H.I.C. and M.B. produced preprocessing task scripts. M.C. and C.E.W. undertook biological specimen processing. F.S. and M.B. provided computational modelling support. M.C. undertook all computational modelling and inferential analyses. H.T. and F.S. contributed equally to this work as second authors. M.C., S.M., C.H., P.C., and M.B. undertook data interpretation. M.C. drafted the article and produced illustrations. All authors contributed to revisions and approval of the final draft.
Corresponding authors
Ethics declarations
Competing interests
C.H. has received consultancy fees from P1vital Ltd., Janssen Pharmaceuticals, UCB Pharma, Sage Therapeutics, Pfizer, Zogenix, Compass Pathways, and Lundbeck, and is a director of the company TnC Psychiatry and Neuroscience. S.M. has received consultancy fees from Zogenix, Sumitomo Dainippon Pharma, P1vital Ltd., UCB Pharma and Janssen Pharmaceuticals. C.H. and S.M. hold grant income from Zogenix, UCB Pharma, Syndesi and Janssen Pharmaceuticals. C.H. and P.C. hold grant income from a collaborative research project with Pfizer. M.B. has received travel expenses from Lundbeck for attending conferences and has acted as a consultant for J&J, Novartis, Boehringher and CHDR. He previously owned shares in P1vital Ltd. The other authors report no conflicts of interest.
Peer review
Peer review information
Nature Communications thanks the anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Colwell, M.J., Tagomori, H., Shang, F. et al. Direct serotonin release in humans shapes aversive learning and inhibition. Nat Commun 15, 6617 (2024). https://doi.org/10.1038/s41467-024-50394-x
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-024-50394-x
- Springer Nature Limited