Received: May 4, 2016; Revised: July 26, 2016; Accepted: August 13, 2016
© The Author 2016. Published by Oxford University Press on behalf of CINP.
International Journal of Neuropsychopharmacology (2017) 20(1): 58–66
doi:10.1093/injp/pyw075
Advance Access Publication: September 16, 2016
Regular Research Article
58
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.
org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is
properly cited.
Peripheral Serotonin 1B Receptor Transcription
Predicts the Effect of Acute Tryptophan Depletion on
Risky Decision-Making
PaulFaulkner, PhD; FedericoMancinelli, MRes; Patricia L.Lockwood, PhD;
MarMatarin, PhD; Raymond J.Dolan, PhD, FRS; Nick W.Wood, MRCP, PhD;
PeterDayan, PhD; Jonathan P.Roiser, PhD
Institute of Cognitive Neuroscience, University College London, London, United Kingdom (Drs Faulkner and Roiser);
Psychiatry and Biobehavioral Sciences, Semel Institute, University of California, Los Angeles, California (Dr Faulkner);
Gatsby Computational Neuroscience Unit (Mr Mancinelli and Dr Dayan), and CoMPLEX Centre for
Mathematics, Physics and Engineering in the Life Sciences and Experimental Biology (Mr Mancinelli),
University College London, London, United Kingdom; Experimental Psychology, University of Oxford, Oxford,
United Kingdom (Dr Lockwood); Clinical and Experimental Epilepsy, Institute of Neurology (Dr Matarin), and
Wellcome Trust Centre for Neuroimaging (Dr Dolan), University College London, London, United Kingdom;
Molecular Neuroscience, UCL Institute of Neurology, Queen Square, London, United Kingdom (Dr Wood).
Correspondence: Paul Faulkner, PhD, Semel Institute, 760 Westwood Boulevard, University of California, Los Angeles, CA 90025 (pfaulkner@mednet.ucla.edu).
Abstract
Background: The effects of acute tryptophan depletion on human decision-making suggest that serotonin modulates the processing
of rewards and punishments. However, few studies have assessed which of the many types of serotonin receptors are responsible.
Methods: Using a within-subject, double-blind, sham-controlled design in 26 subjects, we examined whether individual
differences in serotonin system gene transcription, measured in peripheral blood, predicted the effect of acute tryptophan
depletion on decision-making. Participants performed a task in which they chose between successive pairs of xed, lower-
stakes (control) and variable, higher-stakes (experimental) gambles, each involving wins or losses. In 21 participants, mRNA
from 9 serotonin system genes was measured in whole blood prior to acute tryptophan depletion: 5-HT1B, 5-HT1F, 5-HT2A,
5-HT2B, 5-HT3A, 5-HT3E, 5-HT7 (serotonin receptors), 5-HTT (the serotonin transporter), and tryptophan hydroxylase1.
Results: Acute tryptophan depletion did not signicantly inuence participants’ sensitivity to probability, wins, or losses,
although there was a trend for a lower tendency to choose experimental gambles overall following depletion. Signicant
positive correlations, which survived correction for multiple comparisons, were detected between baseline 5-HT1B mRNA
levels and acute tryptophan depletion-induced increases in both the overall tendency to choose the experimental gamble
and sensitivity to wins. No signicant relationship was observed with any other peripheral serotonin system markers.
Computational analyses of decision-making data provided results consistent with these ndings.
Conclusions: These results suggest that the 5-HT1B receptor may modulate the effects of acute tryptophan depletion on risky
decision-making. Peripheral levels of serotonin markers may predict response to treatments that act upon the serotonin
system, such as selective serotonin reuptake inhibitors.
Keywords: serotonin, decision-making, acute tryptophan depletion, 5-HT1B, risk
Downloaded from https://academic.oup.com/ijnp/article/20/1/58/2629257 by guest on 20 August 2022
Faulkner et al. | 59
Introduction
The serotonin (5-HT) system has long been implicated in deci-
sion-making (Soubrie etal., 1986; Deakin and Graeff, 1991), yet
its precise role in reward and punishment processing remains
unclear (Dayan and Huys, 2009; Boureau and Dayan, 2011).
Several studies have attempted to characterize this relation-
ship by manipulating 5-HT levels via the dietary technique of
acute tryptophan depletion (ATD; see Faulkner and Deakin,
2014 for a review). The results of such studies are mixed, but
a common nding is that ATD can inuence the process-
ing of punishments on reaction time tasks (e.g., Cools etal.,
2008; Crockett etal., 2012; Robinson etal., 2012). For example,
Crockett etal. (2009) showed that ATD abolished punishment-
induced behavioral inhibition on a go/no-go task; and Geurts
etal. (2013) extended these results, reporting that ATD led to
decreased punishment-induced instrumental inhibition while
reversing the inhibitory effect of aversive (but not appetitive)
Pavlovian stimuli on instrumental responding (replicated by
Hebart and Glascher, 2014). Studies of the effects of ATD on
reward processing are also mixed. While Cools et al. (2005)
found that ATD attenuated reinforcement-related speeding on
a cued-reinforcement reaction time task (partially conrmed
by Roiser etal., 2006), den Ouden etal. (2014) reported that ATD
decreased behavioral inhibition to a similar extent for both
rewarding and punishing stimuli on a modied cued-reinforce-
ment reaction time task, though only in a high-feedback prob-
ability condition.
Studies that have examined the effect of ATD on probabilis-
tic choice tasks have also produced mixed results. Rogers etal.
(2003) administered a gambling task that assessed participants’
choices as a function of (1) probability, (2) the magnitude of
expected gains, and (3) the magnitude of expected losses. The
authors found that ATD altered participants’ decision-making
by attenuating their discrimination between large and small
wins, while the processing of losses and probability was unaf-
fected. Seymour etal. (2012) observed a result weakly consist-
ent with this using a probabilistic 4-armed bandit task: ATD
altered the exchange rate by which rewards and punishments
were compared, with this alteration being driven by a decrease
in the subjective value of rewards. However, Rogers etal. (1999b)
observed an ATD-induced reduction in high probability choices
on the Cambridge Gamble Task, while Talbot etal. (2006) found
the opposite result. The discordant results between these latter
2 studies suggest a need to examine the potential role of indi-
vidual differences in 5-HT function in moderating the effects
ofATD.
Very few studies have attempted to delineate the roles of
specic receptors in the effects of 5-HT on decision-making
(Homberg, 2012). In humans, Macoveanu etal. (2013) reported
that administration of the 5-HT
2A
antagonist ketanserin
increased risk aversion, and Faulkner et al. (2014) showed a
positive correlation in humans between hippocampal 5-HT
1A
receptor availability and sensitivity to probability on the gam-
bling task used in Rogers etal. (2003). These studies raise the
possibility that specic 5-HT receptor subtypes may moderate
the effect of ATD on decision-making. However, this is difcult
to test directly in humans due to the high cost and invasive
nature of molecular imaging techniques such as positron emis-
sion tomography.
Studies have shown that dopamine (Illani et al., 2001) and
serotonin (Ye etal., 2014) receptor mRNA levels in the blood cor-
respond to levels in the brain. This was exploited by Ersche etal.
(2011), who examined peripheral levels of mRNA pertaining to
the dopamine receptor 3 (D
3
) and D
4
genes in order to under-
stand variability in response to dopamine agonist treatment.
Variation in peripheral levels of D
3
mRNA explained over one-
quarter of the variation in improvements on a spatial working
memory task due to administration of the D
2/3
receptor agonist
pramipexole. Based on this nding, we reasoned that measur-
ing individual differences in levels of 5-HT receptor mRNA in
peripheral blood might help explain variation in the effects
ofATD.
The current study aimed to test the effects of ATD on
decision-making and to examine whether individual differ-
ences in peripheral 5-HT system gene mRNA levels moder-
ate such effects. Whole-blood mRNA levels were measured
for 9 5-HT-related genes: 5-HT
1B
, 5-HT
1F
, 5-HT
2A
, 5-HT
2B
, 5-HT
3A
,
5-HT
3E
, 5-HT
7
(serotonin receptors), 5-HTT (the serotonin
transporter), and tryptophan hydroxylase 1 (TPH1). Both
standard and computational analyses of decision-making
behavior were utilized. Based on previous results examining
decision-making under ATD, we predicted that participants
would display decreased ability to utilize information per-
taining to rewards following ATD; we also predicted that such
changes would be related to peripheral 5-HT system mRNA
levels.
Materials and Methods
Participants
Thirty participants were recruited via online advertisements. All
participants gave written informed consent and the University
College London (UCL) Research Ethics committee approved the
study. Exclusion criteria (assessed by the Mini International
Neuropsychiatric Inventory: Sheehan etal., 1998) included past/
present major depressive disorder, bipolar disorder, psychosis,
anxiety disorders, psychotropic medication usage, substance/
alcohol dependence or recent (<6months) abuse, or any neuro-
logical disorder. Participants were medically healthy and none
reported taking regular medication for any illness. Both men
Signicance Statement
In this manuscript, we show that the effects of depleting central serotonin levels, via acute tryptophan depletion, on risky
decision-making are moderated by peripheral levels of serotonin (5-HT)1B mRNA. The current results extend those of a
much-cited study (Rogers et al., 2003) - which show acute tryptophan depletion to reduce sensitivity to wins on a risky
decision-making task - by revealing that the ability of this treatment to affect such reward processing is moderated by
peripheral 5-HT1B transcription. The results of this study highlight the importance of considering individual differences
when examining responses to this treatment, which is utilized by many research groups. Further, they suggest that
examining peripheral levels of serotonin markers may help to predict response to treatments that act upon this
neurotransmitter system, such as selective serotonin reuptake inhibitors.
Downloaded from https://academic.oup.com/ijnp/article/20/1/58/2629257 by guest on 20 August 2022
60 | International Journal of Neuropsychopharmacology, 2017
and women were admitted to the study. However, (female) par-
ticipants’ menstrual cycle stage was not recorded.
Following exclusions (detailed in the supplementary Online
Materials), data from 22 participants were analyzed to assess
the biochemical effect of the ATD procedure, data from 26
participants were analyzed to assess ATD effects on decision-
making, and data from 21 participants were analyzed to assess
relationships between peripheral 5-HT system mRNA levels and
ATD-induced effects on decision-making.
Procedure
Testing took place at the Wellcome Trust Centre for
Neuroimaging, UCL. Participants completed 2 testing sessions
(ATD and sham) at least 1 week apart and fasted for 8 hours
prior to the start of each testing session. At the start of each
study day (T0), blood samples were obtained to determine
baseline plasma amino acid levels and 5-HT system periph-
eral mRNA levels. Participants then completed the State Trait
Anxiety Inventory (STAI; Speilberger et al., 1987) and ingested an
amino acid drink either selectively lacking (TRP-) or containing
(sham depletion, TRP+) tryptophan, in a within-subjects, double-
blind, sham-controlled design. The order of drink administra-
tion was counterbalanced across participants. Participants then
rested (and continued to fast) at the testing site for the next 5
hours. Participants then completed the STAI, and blood samples
were obtained once more at the end of this 5-hour period (T5)
to determine plasma amino acid levels post-treatment before
the gambling task was performed (as part of a larger cogni-
tive battery; data for other tasks will be reported elsewhere).
Participants were then provided with a tryptophan-rich meal
before leaving the laboratory.
The concentrations of each amino acid in the drink were
based upon those reported in Young etal. (1985) and are reported
in the supplementary Online Materials, together with the meth-
ods used for plasma amino acid analysis.
GamblingTask
Participants completed 80 trials, during each of which they had
to make a choice between 2 gambles. Each gamble was repre-
sented as a bar, the height of which indicated the probability
of winning or losing a number of points. The magnitudes of
potential gains and losses were displayed in green at the top
and in red at the bottom of each bar, respectively (supplemen-
tary Figure1). One of the gambles, called the control gamble, was
the same on every trial. It involved a lower-stakes 50% chance
of winning or losing 10 points. The other, experimental, gamble
involved either a 75% or 25% chance of winning 80 or 20 points
and the inverse chance of losing 80 or 20 points. There were
thus 8 trial types in total, which are described in supplementary
Table1. Two additional trial types, “gains only” and “losses only,”
were also administered, but data from these were not included
in the analyses.
There were 3 main basic (i.e., noncomputational) outcomes
variables from this task, namely the proportion of experimental
gambles chosen over the control gamble as a function of: (1) the
probability of winning (high vs low probability of winning); (2)
the magnitude of potential win (high vs low magnitude); and (3)
the magnitude of potential loss (high vs low magnitude) (sup-
plementary Figure1), which we term “sensitivity” to probability,
win, and loss after Rogers etal. (2003). Each of these 3 measures
was calculated by taking the difference between the proportion
of experimental gamble choices when each of these factors was
high compared with when it was low. For clarity in presenting
the results, the negative of loss sensitivity is depicted.
Questionnaire
The STAI is a 40-item, self-rating anxiety measure; the rst 20
items identify state anxiety, and the second 20 identify trait
anxiety. Participants must give a score of 1 (“do not agree at all”),
2 (“agree somewhat”), 3 (“agree moderately”), or 4 (“very much
agree”). Scores are then summed to give their state and trait
anxiety scores.
mRNA Extraction and Analysis
Blood samples (4 mL) were drawn into PAXgene tubes (Qiagen)
to maximize the stability of the mRNA and avoid potential deg-
radation. Levels of plasma mRNA were examined for 5-HT
1A
,
5-HT
1B
, 5-HT
1D
, 5-HT
1E
, 5-HT
1F
, 5-HT
2A
, 5-HT
2B
, 5-HT
2C
, 5-HT
3A
,
5-HT
3B
, 5-HT
3C
, 5-HT
3D
, 5-HT
3E
, 5-HT
4
, 5-HT
5A
, 5-HT
6
, 5-HT
7
, 5-HTT,
TPH1, and TPH2. However, mRNA levels were detectable for all
subjects for only 5-HT
1B
, 5-HT
1F
, 5-HT
2A
, 5-HT
2B
, 5-HT
3A
, 5-HT
3E
,
5-HT
7
, 5-HTT, and TPH1, and as such only these transcripts were
included in the correlational analyses. Details of sample pro-
cessing and mRNA measurement are provided in the SOM.
Statistical Analysis
Data were analyzed using IBM SPSS Statistics (http://www-01.
ibm.com/software/analytics/spss/products/statistics). An arc-
sine (square root) transform was applied to probability data
prior to analysis. The order of treatment administration (i.e.,
TRP- in week 1 or week 2)was added as a between-subjects fac-
tor for all analyses.
Plasma amino acid data were analyzed using a 2 x 2 ANOVA,
with treatment constituting 2 levels (ATD and sham) and time
constituting 2 levels (T0 and T5). Decision-making data were
analyzed using a 2 x 2 x 2 x 2 ANOVA, with treatment constituting
2 levels, and each of probability, win, and loss constituting 2 lev-
els (high and low); based on our a priori hypotheses, we focused
on the effects of probability, win, and loss and their interac-
tions with treatment. Mood data from the STAI were analyzed
using a 2 x 2 ANOVA, with treatment constituting 2 levels and
time constituting 2 levels (T0 and T5). Signicant interactions
were interrogated by constructing the simple main or interac-
tion effects. A signicance threshold of alpha = 0.05 (2-tailed)
was adopted for all analyses, while .05 < P < .10 was considered
a trend towards signicance. Relationships between peripheral
5-HT system mRNA levels and ATD-induced changes in deci-
sion-making behavior were analyzed using Pearson’s correla-
tion coefcients. Bonferroni correction (BC; for 9 transcripts*4
decision-making measures = 36 tests) was applied to the corre-
lational analyses.
Finally, exploratory analyses were conducted to examine
whether the effects of treatment, and correlations between
treatment-induced changes in task performance and seroto-
nin gene transcription, were moderated by gender. To examine
the former, a 2 x 2 x 2 x 2 ANOVA was performed, with treatment
(ATD and sham) and each of probability, win, and loss (high and
low) as within-subjects factors and gender added as a between-
subjects factor. The effect of gender on relationships between
peripheral 5-HT system mRNA levels and ATD-induced changes
in decision-making behavior was analyzed using general linear
models, with the treatment-induced change in sensitivity to
probability, win sensitivity, loss sensitivity, and the proportion of
Downloaded from https://academic.oup.com/ijnp/article/20/1/58/2629257 by guest on 20 August 2022
Faulkner et al. | 61
experimental gambles acting as dependent variables in separate
ANOVAs, peripheral 5-HT system mRNA levels added as contin-
uous covariates, and gender added as a between-subjects factor.
The interaction between gender and mRNA levels was included
to test for modulatory effects.
Power Analysis
For the effects of ATD upon decision-making, with data from 26
participants in a within-subjects design, this study had approxi-
mately 95% power to detect an effect size of d~0.75 (large, Cohen
et al., 1988), comparable with that observed in the study of
Rogers etal. (2003) (using a between-subjects design). For cor-
relations between peripheral 5-HT system mRNA levels and
the effect of ATD on decision-making, with data from 21 par-
ticipants, this study had 80% power to detect a correlation of
r~0.55, comparable with the correlation between peripheral D
3
mRNA levels and pramipexole-induced improvement in work-
ing memory observed by Ersche etal. (2011).
Computational Analysis
To explore behavior on the decision-making task in more detail,
we performed a hierarchical Bayesian mixed-effects analysis
of participants’ data (Daw, 2009), tting a set of 5 increasingly
complex, parametrized logistic regression models (supplemen-
tary Table2). The models included different components involv-
ing information about probability, reward, expected values, and
other factors.
All statistical analyses were based on permutation tests with
empirical null distributions created by randomizing TRP status
across testing sessions. Posterior distributions over the parame-
ters for each model for each participant and condition were esti-
mated using a Hamiltonian Monte Carlo sampler, implemented
in PyStan (http://mc-stan.org/pystan.html). We additionally
generated synthetic data from our winning model to assess the
delity with which it could recapitulate the statistical character-
istics of the original behavior. Further details are provided in the
supplementary Online Material.
Results
Plasma Amino Acid Concentrations
Total tryptophan concentration (nmol/mL) in the plasma was
lower overall on the TRP- day than the TRP+ day (signicant
main effect of treatment: F(1,20) = 160.712, P < .001) and was
lower overall at T5 than T0 (signicant main effect of time:
F(1,20) = 9.189, P = .007). As expected, there was a signicant
treatment-by-time interaction (F(1,20) = 133.174, P < .001): plasma
tryptophan concentration increased 126.5% from T0 to T5 on
the TRP+ day (mean T0 = 56.16, SD = 12.50, mean T5 = 127.24,
SD = 40.42, t(21) = 7.994, P < .001) and decreased 73.3% from T0 to
T5 on the TRP- day (mean T0 = 60.39, SD = 13.97, mean T5 = 16.12,
SD = 5.66, t(21) = 19.239, P < .001). There was no difference in mean
tryptophan concentrations between TRP+ and TRP- days at T0
(t[21] = 1.343, P = .194). There was no effect of drink order on tryp-
tophan levels (F(1,20) = .211, P = .651), and no signicant interac-
tion withorder.
The tryptophan:large neutral amino acid (LNAA) ratio was
also lower on the TRP- day than the TRP+ day (signicant main
effect of treatment: F(1,20) = 102.475, P < .001) and was lower at T5
than T0 (signicant main effect of time: F(1,20) = 29.972, P < .001).
Again, there was an expected treatment-by-time interaction
(F(1,20) = 73.105, P < .001): the plasma ratio increased 17.2% from T0
to T5 following TRP+, which showed a trend towards signicance
(mean T0 = 0.156, SD = 0.046, mean T5 = 0.189, SD = .015, t(21) = 1.931,
P = .067), and signicantly decreased 85.1% from T0 to T5 follow-
ing TRP- (mean T0 = 0.166, SD = 0.054, mean T5 = 0.025, SD = 0.012,
t(21) = 13.860, P < .001). Again, there was no difference between the
days on the plasma ratio measure at T0 (t[21] = 0.937, P = .359), no
effect of drink order on the ratio measure (F(1,20) = .541, P = .470),
and no signicant interaction with order.
Decision-Making Task: ProportionateChoice
Choice data are displayed in Table 1A. Participants chose the
experimental gamble signicantly more often when the prob-
ability of winning was high (F(1,24) = 223.812, P < .001), when the
amount that could be won was high (F(1,24) = 34.798, P < .001),
and when the amount that could be lost was low (F(1,24) = 34.364,
P < .001).
Following TRP- participants displayed a trend towards choos-
ing the experimental gamble less often compared with TRP+
(F(1,24) = 3.532, P = .072). Examining each sensitivity separately,
there was no effect of treatment upon sensitivity to probability
(F(1,24) = 0.668, P = .422), wins (F(1,24) = 0.210, P = .651), or losses
(F(1,24) = 0.143, P = .708).
There was no effect of drink order on participants’ overall
choices (F(1,24) = 0.007, P = .936). There was a signicant drink
order-by-treatment-by-probability interaction (F(1,24) = 6.445,
P = .018), indicating a practice effect by which sensitivity to prob-
ability was higher on the second testing session. No interactions
between order, treatment, and either wins or losses approached
signicance (both P > .169).
DeliberationTimes
Deliberation time data are displayed in Table 1B. Participants
were signicantly quicker to respond when the probability of
winning was high (F(1,24) = 7.437, P = .012) and when the magni-
tude of potential losses was low (F(1,24) = 6.887, P = .015), but the
effect of the magnitude of potential wins on reaction time was
nonsignicant (F(1,24) = .017, P = .898).
Table1. Proportion (SD) of Choices of the Experimental Gamble (A), and Mean (SD) Deliberation Times (ms; B) as a Function of the Probability
of Winning, the Magnitude of Potential Wins, and the Magnitude of Potential Losses under ATD (TRP-) and Sham (TRP+)
Probability of Winning Magnitude of Wins Magnitude of Losses
A High Low Large Small Large Small
TRP- 0.77 (.18) 0.18 (.19) 0.55 (.12) 0.39 (.12) 0.38 (.16) 0.56 (.10)
TRP+ 0.84 (.15) 0.17 (.14) 0.58 (.10) 0.44 (.12) 0.42 (.14) 0.60 (.09)
B High Low Large Small Large Small
TRP- 1766 (1015) 2013 (1055) 1902 (1078) 1878 (1008) 1935 (1066) 1845 (1018)
TRP+ 1769 (993) 1983 (930) 1882 (1006) 1871 (930) 1970 (993) 1793 (935)
Downloaded from https://academic.oup.com/ijnp/article/20/1/58/2629257 by guest on 20 August 2022
62 | International Journal of Neuropsychopharmacology, 2017
There was no main effect of treatment on participants’ delib-
eration times (F(1,24) = 1.102, P = .304). Treatment did not signi-
cantly inuence participants’ deliberation times as a function
of probability, potential wins, or potential losses (all interac-
tions: P > .321). There was a signicant drink order-by-treatment
interaction on participants’ deliberation times (F(1,24) = 14.625,
P = .042), indicating a practice effect by which responses were
faster on the second testing session. There was also a signicant
drink order-by-treatment-by-loss interaction (F(1,24) = 5.446,
P = .028). The increase in deliberation times on high potential
loss trials was attenuated on the second testing session, again
indicating a practice effect. No interactions between order, treat-
ment, and either probability or wins approached signicance
(both P > .292).
Effect of Acute Tryptophan Depletion on Anxiety
There was no signicant main effect of treatment upon par-
ticipants’ state anxiety scores (state STAI: F(1,24) = .121,
P = .732). There was also no main effect of time on state STAI
(F(1,24) = 2.211, P = .153). However, there was a signicant
treatment*time interaction (F(1,24) = 6.651, P = .019), with post-
hoc paired t tests revealing that while scores did not differ
signicantly between T0 and T5 on the TRP+ day (mean (SD)
T0 TRP+ = 10.71 (9.13), T5 TRP+ = 10.33 (8.32)), participants’ state
STAI scores increased signicantly from T0 to T5 on the TRP-
day (T0 TRP- = 9.86 (5.94), T5 TRP- = 12.00 (6.72), t(25) = 2.918,
P = .009).
Associations between Peripheral Serotonin mRNA
Expression and ATD Effects on Decision-Making
There was a signicant positive correlation between baseline
peripheral 5-HT
1B
mRNA levels and the ATD-induced increase
(TRP- minus TRP+) in the overall proportion of experimental
gambles chosen (r = .658, nominal P = .001, Bonferroni-corrected
P
BC
= .041), with participants with higher baseline 5-HT
1B
tran-
scription levels displaying the greatest increase (Figure 1A).
There was also a signicant positive correlation between base-
line 5-HT
1B
transcription levels and the ATD-induced increase
in win sensitivity (TRP- minus TRP+: r = .654, nominal P = .001,
P
BC
= .045), with participants with higher baseline 5-HT
1B
tran-
scription levels displaying the greatest increase (Figure 1B).
There was no other nominally signicant correlation between
peripheral 5-HT system mRNA levels and ATD-induced changes
in decision-making.
Associations between Peripheral Serotonin mRNA
Expression and ATD Effects on Anxiety
There was no signicant correlation between peripheral tran-
scription levels of serotonin receptor mRNA and the ATD-
induced increase (TRP- T5 minus TRP- T0 scores) in state STAI
scores (all P > .182).
Effect of Sex on Responses toATD
We conducted exploratory analyses examining the effect of sex
on treatment-induced changes in task performance. Results of
the repeated-measures ANOVA revealed no signicant interac-
tions between gender and treatment (F(1,24) = 1.848, P = .187),
between gender, treatment, and sensitivity to probability
(F(1,24) = .373, P = .547), between gender, treatment, and sensitiv-
ity to wins (F(1,24) = .083, P = .776), or between gender, treatment,
and sensitivity to losses (F(1,24) = .028, P = .868).
We conducted exploratory analyses examining the effect
of gender on the relationship between treatment-induced dif-
ferences in both win sensitivity and the overall proportion of
experimental gambles chosen with baseline serotonin recep-
tor transcription levels. Regarding win sensitivity, there was an
expected signicant main effect of baseline 5HT
1B
transcription
levels (F(1,17) = 11.697, P = .003) but no signicant main effect
of gender (F(1,17) = 1.334, P = .264) and no signicant interac-
tion between baseline 5HT
1B
transcription levels and gender
(F(1,17) = .201, P = .660). Regarding the overall proportion of exper-
imental gambles chosen, there was an expected signicant
main effect of baseline 5HT
1B
transcription levels (F(1,17) = 8.296,
P = .010), but no signicant main effect of gender (F(1,17) = 2.504,
P = .132) and no signicant interaction between baseline 5HT
1B
transcription levels and gender (F(1,17) = .2.575, P = .127). All mod-
els examining the effect of gender on the relationship between
transcription levels of the remaining serotonin genes revealed
no main effects of transcription levels or gender and no signi-
cant interactions between transcription levels and gender.
Computational Analysis
We considered a family of models involving different poten-
tial inuences of probability, reward, and expected value and
included a trial-type independent choice bias parameter (some-
times called a lapse rate) to account for rare, unlikely decisions.
We rst identied the most parsimonious model to param-
eterise performance on the decision-making task, independent
of depletion condition. Widely Applicable Information Criterion
Figure1. Relationships between baseline peripheral serotonin (5-HT)
1B
mRNA levels and the treatment-induced increase (acute tryptophan depletion [ATD] minus
sham) in the overall proportion of experimental gambles chosen (A) and the ATD-induced increase in sensitivity to wins (B).
Downloaded from https://academic.oup.com/ijnp/article/20/1/58/2629257 by guest on 20 August 2022