The role of the noradrenaline and acetylcholine systems in uncertainty estimation and temporal attention

There are striking commonalities in the anatomical organization and low-level
actions of different neuromodulator systems, as well as direct interactions
between these systems (Briand et al., 2007). An important goal for cognitive
neuroscientists is to delineate the specific contributions of particular
neuromodulator systems to cognitive control and decision making. The primary
objective of the proposed research is to take a step toward that goal by
examining and comparing the involvement of the noradrenaline (NA) system and
acetylcholine (ACh) system in representing various forms of uncertainty
associated with a given behavioral context.

There are strong reciprocal interactions between the NA and ACh systems (Briand
et al., 2007), and they apparently contribute in a similar way to alerting and
arousal functions. In line with this view, a recent computational model of Yu
and Dayan (2005; Dayan & Yu, 2006) proposes that NA and ACh have specific but
complementary roles in coding uncertainty. According to this model, ACh is
involved in signaling expected uncertainty, which arises from known
unreliability of predictive relationships within a familiar environment. In
contrast, NA is proposed to be involved in reporting unexpected uncertainty,
induced by gross changes in the environment that produce sensory observations
that strongly violate top-down expectations. Yu and Dayan (2005) review a
number of pharmacological studies that have manipulated either NA or ACh levels
and have examined only one type of uncertainty. We intend to replicate some of
these key observations underlying the model, but in the same participants in a
task that manipulates both expected and unexpected uncertainty.

Recent research has demonstrated the critical involvement of the NA system in
the generation of the P300 (or P3) component of the event-related potential
(Nieuwenhuis et al., 2005), a well-known neural correlate of estimated phasic
uncertainty signals (Nieuwenhuis, in press). There is also substantial evidence
for an influence of the ACh system on P3 generation (Wang et al., 1997; Hammond
et al., 1987). However, the relative contributions of these two neuromodulator
systems to the generation of the P3 are not well understood. The goal of this
experiment is to test a hypothesis proposed by Ranganath and Rainer (2003).
Their hypothesis is that the two neuromodulator systems contribute to separate
subcomponents of the P3: the P3a (or novelty-P3) and the P3b. The P3b has a
parietocentral scalp distribution and is mainly sensitive to infrequent
task-relevant stimuli. The P3a has a prominent frontocentral scalp distribution
and is mainly sensitive to novel and highly deviant or salient task-irrelevant
stimuli. Most P3s are a mixture of these two subcomponents (Spencer et al.,
2001). Ranganath and Rainer discuss one study that found that administration of
a NA antagonist in monkeys selectively abolished the posterior P3b, but left
the frontal P3a unaffected. By contrast, in another study, administration of an
ACh antagonist in humans attenuated the P3a but not the P3b. These findings
suggest that the NA system may primarily affect the P3b, whereas the ACh system
may primarily affect the P3a. Indeed, the topography of cortical generators of
the P3a and P3b seems consistent with the primary projection areas of the basal
forebrain (i.e., ACh: frontal and anterior medial) and locus coeruleus (i.e.,
NA: parietal). However, note the apparent contradiction between this hypothesis
and the model by Yu and Dayan (2005) that we discussed above: novel and other
unexpected stimuli (ACh: P3a) seem a source of unexpected uncertainty, whereas
infrequent task-relevant stimuli (NA: P3b) seem a source of expected
uncertainty! Our intended experiment allows us to contrast these two hypotheses
by pharmacologically manipulating NA and ACh activity in participants.

Another important objective of the proposed research is to examine the role of
the NA and ACh systems in temporal attention. If a visual target stimulus is
immediately preceded by a salient yet task-irrelevant auditory stimulus, a
participant*s reaction time to the target is significantly decreased
(Bernstein, Clark, & Edelstein, 1969a; 1969b). Several researchers have
proposed that this accessory stimulus effect may be related to the NA system
(Hackley, 2009; Jepma et al., 2009; Stafford & Jacobs, 1990). According to this
view, the locus coeruleus (or LC), the major source of noradrenergic
innervation of the forebrain (Berridge & Waterhouse, 2003) is activated by the
saliency of the accessory stimulus, which in turn boosts activity in the motor
cortex through phasic NA release, thus speeding up the response. We will test
this hypothesis by examining the effects of pharmacologically decreasing NA
activity on the accessory stimulus effect, and comparing this with the
accessory stimulus effects associated with decreased ACh (drug control) and
placebo.

Another major paradigm for studying temporal attention is the attentional blink
task, in which participants are instructed to identify two targets (e.g.
numbers) in a rapidly presented series of distractors (e.g., letters). If these
two targets are presented in rapid succession, participants are often unable to
identify the second target (i.e. they exhibit an *attentional blink*).
Theoretical work has suggested that the attentional blink reflects the temporal
dynamics of LC activity and consequent effects on NA activity (Nieuwenhuis et
al., 2005b). We will test this account by examining the effects of a NA
manipulation on attentional blink performance. This experiment will replicate
earlier work by Nieuwenhuis et al. (2007), but with greater statistical power,
a within-subject design, and other design improvements. Again, an ACh
manipulation will serve as drug control.

There are large individual differences in the pharmacodynamics and cognitive
effects of the drugs we intend to administer, which complicates an
interpretation of the results. To capture these individual differences, it is
necessary to collect information about gene polymorphisms that affect NA and
ACh activity (Abbott, 2003). Personality questionnaires are another source of
critical information about individual differences that can help explain the
variance in drug effects on our experimental measures (e.g. Cools et al.,
2005).

Primary Objective: to ascertain whether administered clonidine and scopolamine
affect performance (as measured by modulation of reaction times, learning
and/or error rates) in a range of cognitive tasks designed to examine
uncertainty estimation and temporal attention, as well as specific components
of the EEG signal that will be used as neural correlates of specific cognitive
constructs like uncertainty estimation.

Secondary Objective(s): to explore possible individual differences in the
behavioural and EEG indices of uncertainty estimation and temporal attention by
means of genetic analyses and personality questionnaires. Several personality
trait questionnaires will be used as psychological corroborations of the
genetic analyses of polymorphisms. We will use a state questionnaire (the
Visual Analogue Scales) to gain insight in participants* levels of sedation,
but also in their current mood and how it is affected by clonidine and
scopolamine. This allows us to chart the psychological effects of both drugs.

The proposed study uses a double-blind, pseudo-randomized, placebo-controlled
double-dummy cross-over design. Due to different pharmacokinetic profiles of
clonidine and scopolamine, the two drugs will be administered at different
times relative to testing. Participants* EEG is measured in three sessions
(each one week apart), once following administration of 150µg of clonidine
verum (and scopolamine placebo), once following 1.2 mg of scopolamine verum
(and clonidine placebo, and once following clonidine and scopolamine placebo,
in a counterbalanced order. We will test 24 participants, four with each
possible order of treatments.

While the EEG is measured, participants perform four cognitive computer tasks
(lasting about 20 minutes each; cf. Section 7.3); prior to EEG measurements,
participants complete a number of questionnaires (which will take less than an
hour in total). One session will last about four hours and twenty minutes.

Clonidine, 150 µg, single dose, administration p.o. in a capsule. Clonidine is
a centrally acting adrenergic alpha-2 agonist that is used mainly as an
antihypertensive drug, as well as a maintenance drug in cases of migraine.
Clonidine is metabolized by the liver and has an elimination half-life of
ranging from 5.6-12.3 hours (mean 7.7 hours; Keränen et al., 1978).The usual
maintenance dose for antihypertensive indications is between 0.075 and 0.15 mg
thrice a day; for migraine indications, usual maintenance doses are between
0.025 mg and 0.075 mg twice a day. We intend to administer a single oral dose
of 0.2 mg clonidine, the most widely used dose in psychopharmacological
research (e.g. Frith et al., 1985; Coull et al., 2001; Smith et al., 2003).
Because of the anti-hypertensive properties of clonidine, heart rate and blood
pressure will be monitored for subject safety. Measurements will be taken just
prior to ingestion of the tablet, and then every 60 mins for the remainder of
the experimental session.

Scopolamine hydrobromide, 1.2 mg, single dose, administration p.o. in a
capsule. Scopolamine is a cholinergic muscarinic receptor antagonist that is
used mainly as an anti-emetic and a treatment of intestinal cramping.
Scopolamine is generally administered in one of two forms: as a transdermal
patch (indication: travel/motion sickness) or in *pure form* (as tablets or
injections, mainly for research purposes, or occasionally as treatment for
intestinal cramping). Because it is difficult to predict the pharmacokinetics
of transdermal patches, researchers usually administer the pure-form
scopolamine via enteral, intravenal, or intramuscular routes. Because oral
administration is the least aversive method of administration, we have decided
to give scopolamine tablets. A literature search has revealed that in cognitive
research, scopolamine tablets are administered in doses that range from 0.15 to
1.2 mg; higher doses generally produce more pronounced cognitive effects
without eliciting more adverse side-effects. Therefore, we intend to administer
a single oral dose of 1.2 mg, which has measurable effects on cognitive
function (e.g. Callaway et al., 1991; Parrott, 1986; Pompéia et al., 2002), and
induces a similar level of sedation as 0.2 mg clonidine.

Note that participants do NOT receive BOTH clonidine and scopolamine during a
single session of the study: these drugs are administered during separate
sessions.

The number of on-site visits is three (plus a medical screening prior to
participation); in one session placebo is administered; in a second session
clonidine is administered, and in a third session scopolamine is administered.
The risk of a single dose of either of these drugs is minimal, but to further
minimize the remaining risk, participants will be screened for bradycardia
and/or hypotension, as well as glaucoma. The presence of any of these
conditions will preclude participation in the proposed study. Previous studies
have used comparable or higher doses of these drugs without reporting adverse
side-effects (e.g. Tiplady et al., 2005: 300 µg of clonidine; Callaway et al.,
1991: 200 µg of clonidine and 1.2 mg of scopolamine). Saliva will be obtained
from participants for a genetic analysis of specific polymorphisms related to
activity of the noradrenaline and acetylcholine systems; this procedure is not
painful and will take place in a secluded room to guarantee participants*
privacy.