The link between LC-NE and sensory-motor decision making and its clinical implications

The link between LC-NE and sensory-motor decision

making and its clinical implications

Melina Bergen

Bachelorthesis

Universiteit van Amsterdam

Clinical Neuropsychology

Supervisor: Tobias Donner

Student number: 10010904

Word count: 6588

Contents

Abstract...03

Introduction...04

The role of the LC-NE system in regulating control state ...05

The link between LC-NE and task performance in behavioral

decision tasks...06

The link LC-NE and task utility...09

The link between LC-NE and pupil diameter...13

The relationship between LC-NE and cortical neurons...16

The link between LC-NE and stress...18

Discussion...19

Abstract

This review shall give an overview of recent findings concerning the locus coeruleus -

norepinephrine (LC-NE) system. It is known to play an important role in arousal and

behavioral decision making. Also disruption in this system may contribute to stress-related

disorders like ADHD or PTSD.The LC-NE plays a particularly important role in modulating

neural gain.There are two modes of LC activity, the phasic and the tonic mode. The phasic

LC mode facilitates task-related behavioral responses, whereas the tonic LC mode makes

subjects more sensitive to task irrelevant stimuli. When a new stimulus is presented, the

firing rate of LC-NE neurons changes from tonic firing to phasic bursts of activity, which

enhance the ‘signal’ to ‘noise’ ratio. This helps to regulate the trade-off between

exploration and exploitation. Furthermore, the LC neurons have widespread connections

to various areas of the cortex, where they influence, for example, motor activity or visual

stimulus processing through neural gain. Finally, the LC-NE system is closely related to

stress and adrenergic blockage can be used to counteract negative effects of stress, for

example, in PTSD.

Introduction

Undeniably, arousal is a natural mechanism that is essential for survival for all living creatures.

Arousal causes you to focus your attention on, or switch your attention to a relevant stimulus. Too low arousal will lead to drowsiness and sleepiness, whereas too high arousal will lead to

distractibility and overstimulation (Aston-Jones et al., 2005). According to the Yerkes-Dodson theory, optimal performance occurs on cognitive and behavioral tasks when the individuals‘s level of arousal is held within a narrow range, neither too high nor too low.

The Locus Coeruleus nucleus, which means blue spot in Latin, is a small collection of noradrenergic neurons, that is located in the dorsorostral pons. It is the only source of the neurotransmitter norepinephrine (NE) in the cortex. Traditionally the LC-NE system has been known to play an important role in arousal. However recent findings in the field of cognitive neuroscience have indicated that the LC-NE system might play a far more complex role in

behavioral modulation than originally suggested (Aston-Jones et al., 2005). It was found that highly salient and arousing stimuli evoke a phasic activation of LC neurons (Aston-Jones & Bloom,

1981b; Grant et al., 1988; Herve-Minvielle & Sara 1995, Rasmussen et al., 1986, retrieved from Aston-Jones et al., 2005). Rather than producing a direct excitatory or inhibitory effect on

postsynaptic cells, NE modulates such effects that are produced by other neurotransmitters such as glutamate or GABA. Aston-Jones et al. (2005) propose that phasic activity of the LC-NE system facilitates task-related responses and suppresses responses to task-irrelevant stimuli, through neural gain. However, when the expected reward for a task decreases, LC activity will become more tonic, which increases the responsiveness to task-irrelevant stimuli. In the phasic mode, brief bursts (1/10 msec) of LC activity are observed. In contrast, during tonic mode LC baseline activity is elevated, but there are no phasic bursts of activity. The phasic mode is associated with high behavioral performance, whereas the tonic mode is associated with poor behavioral performance.

In this review, the link between the LC-NE system and sensory-motor decision making

shall be investigated. As there are indications that the disruption of the LC-NE system may lead to various psychiatric disorders such as ADHD or PTSD. The clinical implications of this system will also be discussed. In the first section the focus shall be on the relation between the LC-NE system

and task performance in behavioral decision tasks, in order to determine how the LC-NE system influences the performance on these tasks. In the second section it will be examined how LC activity modulates the control state (task engagement vs. task disengagement) depending on task utility. Pupil diameter is thought to be a good index for LC activity, however there has not been a lot of direct scientific evidence to proof that. The relationship between the LC-NE system and pupil diameter will be discussed in the third section. Furthermore, two studies on the relationship

between the LC-NE system and its connections with other parts of the brain shall be included, in order explore how exactly it affects the post-synaptic neurons and understand how the LC-NE system can affect cognition. Finally, in the last section the clinical implications of these findings will be discussed.. As there has not yet been much of research on the link between the LC-NE system and Post-traumatic stress disorder (PTSD), the focus will be on establishing this link, in order to possibly create a scientific base for further research.

The role of the LC-NE system in regulating control state

Aston-Jones et al. (2005) have proposed a new theory on LC function, specifying its role in

behavioral modulation. It is suggested that the LC-NE system plays an important role in optimizing behavioral performance by modulating the trade-off between exploitation and exploration.

Exploitation refers to the engagement in one task that is thought to bring a high amount of reward, whereas exploration refers to the disengagement from one task in order to the search for other behavioral strategies that can possibly bring a greater reward. The LC has widespread

connections throughout the cortex. The adaptive gain theory (AGT, Aston-Jones & Cohen, 2005) proposes the LC responds to changes in task utility by altering the its activity mode and thereby influencing the gain of cortical processes. The LC activity ranges from phasic activity, where brief bursts of LC activity are observed, that seem to be elicited by novel, salient stimuli and task-related decision processes, to tonic activity, during which LC baseline activity is elevated, but there are no phasic bursts of activity. Task performance is higher during phasic mode than during tonic mode. During tonic mode behavior seems more distractible. The AGT proposes that phasic LC activity promotes exploitation, while tonic LC activity promotes exploration.

Phasic LC-NE activity and performance of sensorimotor decision tasks

To be able to study the role the LC-NE system plays in behavior, we first need to understand what drives LC responses and what kind of behavior it is involved in. Aston-Jones (2005) stated that phasic LC activity is robustly observed in a monkeys brain when tapping the cage door around feeding time. This is accompanied by a behavioral orienting response and physiological signs of arousal. Stimuli that do not elicit a behavioral response however, do not evoke an LC response. This suggests that LC neurons tend to activate following salient stimuli in many modalities that elicit behavioral responses and that the LC-NE system has a relatively broad, nonspecific effect on cortical information. To examine the link between LC activity and behavioral responsiveness to a certain stimulus, Rajkowski et al. (2004) measured LC-activity in monkeys performing a target detection task. In this experiment, two adult monkeys were trained to release a pedal in response to a certain stimulus. They had to perform a stimulus discrimination task with 2x2 conditions. The easy and the difficult condition were operationalized by varying the similarity of targets and nontargets. A slow and a fast condition were created by varying the intervals between the trials. Every time the monkey responded correctly, it was rewarded with juice. The monkeys performed with a high accuracy of more than 90%.

Results showed that in the easy as well as in the difficult condition LC neurons responded to targets with phasic activation, but not to nontargets. The phasic LC activation consistently preceded the behavioral response. Both LC response latencies and behavioral response times increased in the difficult condition compared to the easy condition. Fast condition trials lead to faster and less accurate performance. Also the LC response latencies were shorter for fast

reaction time trials than in long reaction time trials, regardless of difficulty. Further it was found that responses to a nontarget (false alarms) also produce a phasic LC response, although smaller than for hit trials. The correct omission of a nontarget resulted in a small inhibition in LC activity. A novel early activation of LC neurons was detected that followed every task stimuli and was distinct from the LC response elicited by target cues.

As the LC phasic response was only visible on target trials and not on correct omission trial, we can assume that the LC only responds to decisions that produce a behavioral response.

Furthermore, the phasic LC response is not specific to particular sensory attributes, as suggested by findings of Aston-Jones et al. (1994, retrieved from Rajkowski et al., 2004). The results show that LC responses vary with task difficulty in a target detection task. More difficult discrimination tasks resulted in longer latencies in LC activity, as well as longer behavioral response times. The analysis of LC response latencies and behavioral response times shows that the LC response is constantly activated shortly after stimulus onset and before the behavioral response, which

indicates a close temporal relationship between the two and makes it possible that the NE release directly influences the behavioral response on the same trial. However, the detection of an early, small, non discriminative LC response, indicates that early LC responses are also closely linked to sensory stimuli. These findings suggest that LC responses in a discrimination task reflect neither purely sensory nor motor activities.

Based on these results the authors hypothesize that phasic LC activity facilitates the response to a certain target stimulus and promotes selective behavioral responding. However it remains unclear, what exactly drives the LC response. As lever releases outside of the task did not elicit an LC response, the LC response cannot be purely linked to motor or pre-motor activity. It is more likely that LC activity is linked to the expectation of reward in some way. Additional work is needed to test this idea and further specify the role of phasic LC responses in adaptive behavioral activity. In a forced-choice task, targets and distractors are displayed simultaneously and require more frequent behavioral responses. This allows a more detailed examination of LC responses.

In a study by Clayton et al. (2004) two adult rhesus monkeys executed a forced-choice task. They were trained to depress two pedals in response to a certain stimuli. There were 4 conditions: A simple and a complex condition and a noncue and nonreward condition. Correct responses were rewarded with juice, while incorrect responses were punished with a time-out.

As indicated by Rajkowski et al. (2004) LC activity was more closely related to behavioral response than to stimulus onset. Response times were longer and had a greater variability for

incorrect trials than for correct trials. Non-reward did not have an effect on LC activity, which indicates that LC activity is not dependent on the expectation of reward. .

There was a close temporal relation between LC activity and behavioral responses. Analyses of response-locked versus stimulus-locked PETHs showed that for all conditions LC phasic activation was more closely linked to lever releases than to stimulus presentation. The LC response always preceded the behavioral response by about the same amount of time for all trials. Again there was no LC response for correct omissions or with behavioral responses outside the task performance.

In comparison to the detection task by Rajkowski et al. (2004) we can see that the LC is activated after the attended target stimulus even when the target stimulus is presented frequently (80-100% of the trials) and that these responses do not depend on a go–no go contingency but also occur when behavioral responses are demanded on each trial. Additionally we can see that, the LC activation apparently does not depend on the expectation of reward, as LC activity did not change in the non-reward condition.

So far we have focussed on phasic activation of the LC during a single task, in which the LC responses help to optimize task performance in behavioral decision tasks. In a review by Aston-Jones et al. (2005) it is suggested that there is a second mode of LC response. In contrast to phasic responses the LC tonic mode produces a persistent increase in gain, which makes the system more sensitive to task irrelevant stimuli. The tonic LC mode seems maladaptive at first, as it impairs task performance. However, according to Aston-Jones et al., the tonic mode is important as behavior might have to be changed as the environment changes in order to find new, more valuable opportunities for reward. An animal (including humans) must decide whether to spend most of its time engaged in behavior that proved to bring reward (exploitation) or keep searching for new behavior that might bring even more reward (exploration). This is called trade-off between exploitation and exploration (e.g., Kaelbling et al., 1996, retrieved from Aston-Jones et al., 2005). The adaptive gain theory proposes that the LC-NE system responds to changes in task utility by switching between the two modes of LC activity and thereby regulating the balance between exploitation and exploration. The tonic mode promotes exploration because the baseline firing rate is elevated, which increases the gain of units in the network indiscriminately and makes them more

responsive to any stimuli. In contrast, the phasic mode promotes exploitation by facilitating responses to task relevant stimuli.

The trade-off between exploration and exploitation will be further discussed in the second section.

The link between LC-NE activity and task utility

In order to really understand how the LC-NE system works in humans studies should be made with

human subjects. However to be able to do that a non-invasive measure of LC activity is required.

Rajkowski et al. (1993, retrieved from Gilzenrat et al, 2010) found a correlation between pupil

diameter and LC tonic discharge frequency. They suggested that LC tonic mode is marked by a

relatively large baseline pupil, while LC phasic mode is marked by a relatively small baseline pupil

diameter. It was also shown that task processing is accompanied by rapid pupil dilation (Beatty,

1982a, 1982b; Einhäuser, Stout, Koch, & Carter, 2008; Richer & Beatty, 1987, retrieved from

Gilzenrat et al., 2010), which is consistent with the idea that LC phasic response is associated with

small baseline pupil diameter. Given the strong correlation between pupil diameter and LC activity, measuring pupil diameter could be a useful indirect proxy of LC activity.

Gilzenrat et al. (2010) tested the prediction that trial-to-trial changes in baseline pupil diameter and task evoked pupil dilation are negatively correlated with each other, and that these pupillary responses can be used to track control state (task engagement vs. disengagement) and modes of LC function. 23 Princeton university students took part in the study and had to perform an auditory oddball discrimination task, while the pupil diameter of their left pupil was constantly measured. They were instructed to press a key when they heard a target tone (20% of the trials) and to ignore distractor tones (80% of the trials). There was no feedback during the task.

Large baseline pupils were associated with a higher false alarm rate than were small pupils. Average response times were shorter and tighter for small than for large baseline pupils. Furthermore, average maximal pupil dilations were far larger for detected targets (hits) than for

correctly rejected distractors. Peak dilations to targets were larger for small than for large baseline pupil trials.

These results support the hypothesis, that pupil diameter is a reliable predictor of task performance. Smaller baseline pupils were associated with better performance, and larger task-evoked dilations. The opposite effect was observed for larger baseline pupils. This was in line with the predictions and with previous studies that examined to relationship of LC activity and

behavioral responses in monkeys.

However, the law of initial values would provide an alternative explanation for these

results. According to this law, the size of the physiological response to a stimulus might be affected by the baseline level. That is, if the baseline level is high already, the expect increase will be small, due to a ceiling effect, whereas when the baseline level is low, the expected effect will be relatively greater. The authors tried to eliminate this possible alternative explanation by taking reference measurements of pupil diameter in the darkness for a subset of the participants in a second version of this experiment that was otherwise the same as this one. In this version of the experiment participants were assigned to either a dark or a light condition. Participants in dark condition performed the task in a dark room , whereas participants in the light condition performed the task in a normally lighted room. If the law of initial values was responsible for the results found, it would be expected that the increase in pupil diameter in the dark condition would be relatively greater than in the light condition, due to a ceiling effect in the light condition. As there was no significant difference between the light and the dark condition, we can assume that the findings are not due to variations in baseline level, but a result of the relationship between pupillary response and control state.

In a second experiment, the authors wanted to use the insights about the relationship

between pupil diameter and LC activity gathered in the study above to test the predictions of the adaptive gain theory (AGT). According to AGT, the control state can be influenced by manipulating the task utility, that is the costs and rewards. In the second experiment, the same procedure was used as in the first experiment, except participants were presented two tones and had to say if the second one was higher or lower than the first. The participants also received a feedback sound to

let them know if they were right after each trial. In two of the four blocks, the participants received 6 trials (25%) with impossible-discrimination comparison tones (low-conflict condition). In these trials, the participants had to choose whether the second tone was higher or lower, even though there was actually no difference between the tones. The other two blocks contained 12 (50%)

impossible-discrimination trials (high-conflict condition). In two of the four blocks, the participants always received positive feedback on impossible-discrimination trials (high reward). In the other two blocks, the participants always received negative feedback on impossible-discrimination trials (low reward condition).

As predicted, pupil baseline diameter was smaller and dilation was larger in the low conflict/low reward condition and in the high conflict/high reward condition, while pupil baseline diameter was larger and pupil dilation smaller in the high conflict/low reward condition. No differences in behavioral RTs were observed between the different conditions.

The observed differences were -although significant- quite small and in contrast to the expectations no differences in behavioral RT could be observed. This lead to the assumption of the authors that the manipulations might not have been strong enough. Also there was no option to really disengage from the task. In a third experiment, the authors tried to eliminate these

limitations.

In this experiment the participants had to complete a similar pitch discrimination task as in

Experiment 2. They earned points for each trial that they judged correctly. If a participant

responded correctly on a particular trial, the value of that trial was added to the participant’s total score. In the next trial, the reward would then be increased and the discrimination task would be more difficult. At first the expected value progressively increases from trial to trial, as the reward value gets higher. However, at some point the number of errors made, due to the increasing task difficulty, reduces the expected value even in the face of the increasing reward value. This is expected to lead to task-disengagement, and therefore to a switch in LC mode from phasic to tonic. Prior to each trial the participants had the opportunity to escape the trial. When they did that,

the points that could be earned would be reset to the original value and also the task difficulty would be reset to the easiest level thereby increasing the task utility (expected value) again. It was expected that as task difficulty increases the participants would experience increasing conflict. At first the higher conflict will promote task-engagement (phasic LC mode), however, when conflict gets too high it will slowly lead to task-disengagement (tonic LC mode) and eventually participants will use the escape option. Further it was predicted that baseline and transient pupil diameter would track the changes in control state and would anticipate the escape event. Baseline pupil diameter is expected to increase over the trials leading up to the escape as task utility

increases and rapidly decrease on the trials following the escape. For pupil dilation the opposite pattern is expected. Thus pupillary responses are expected to track changes in the expected value of the task.

A reliable increase in baseline pupil diameters (with a peak on the escape trial) was

observed, and dilations decreased as expected value began to decline and participants were about to escape. After an escape, the baseline pupil diameters decreased and the transient dilations increased. Accuracy slowly decreased during the first four trials of a new epoch, and dramatically decreased just before the escape. Performance was best on the trial immediately following the escape and then gradually decreased as task difficulty increased. Responses were slower and less accurate prior to an escape than afterward. RTs continued to increase over the trials leading up to an escape and peaked on the trial immediately prior to the escape. This indicates tonic LC activity and therefor increasing disengagement from the current task, which would be in line with the predictions of the AGT.

These findings suggest that LC activity regulates the control state, and that changes in control state are driven by assessments of task utility. Also pupil diameter seemed to track LC activity. Experiment 1 showed that there is an inverse relationship between baseline and task-evoked pupil diameter. In Experiment 2 and 3 conflict and reward was manipulated in order to produce changes in control state. The results were in line with the predictions made based on the AGT. Yet these experiments did not fully investigate the exploitation-exploration trade-offs, as there was no possibility for the participants to really disengage from the task and switch to a different task. This option was included in an experiment by Jepma et al. (2011).

The 17 participants performed a four-armed bandit task, while their pupil diameter was

constantly measured. They were presented with pictures of four different colored slot machines. The slot machines appeared for four seconds, then a black fixation cross appeared and

participants could choose one of the slot machines. They had 1.5 seconds to make their choice or otherwise got a time-out. The points the participants earned were displayed after each trial. The number of points paid off by the four slot machines gradually and independently changed from trial to trial. Participants were told that on average €2.50 are earned in this experiment, but not how the points are converted to money. At the end each participant received €3.

In order to classify each choice as exploitative or exploratory a reinforcement-leaning model was used. The mean-tacking rule estimates the payoff of each machine and the choice rule selects a machine based on these estimations The exploitation–exploration balance is adjusted by a

parameter referred to as gain. With higher gain, action selection is determined more by the relative estimated payoffs of the different options, whereas with lower gain, action selection is more evenly distributed across the different options. each choice was classified as exploitative or exploratory according to whether the chosen slot machine was the one with the maximum estimated payoff (exploitation) or not (exploration).

It was found that explorative choices were accompanied by larger pupil diameters than exploitative choices. Within the exploration trials, baseline pupil diameter increased as a function of the degree of exploration. Individual differences in baseline differences in baseline pupil diameter were found to be predictive of exploratory choice behavior. Further, trial to trial changes in baseline pupil diameter surrounding the transition between choice strategies correlated systematically with changes in utility, at least during the transition from exploration to exploitation. The switch between exploitative and exploratory choice strategies was preceded by a gradual decrease in utility, but an abrupt increase in baseline pupil diameter, while the switch back to exploitation was marked by a gradual increase in utility accompanied by a gradual decrease in baseline pupil diameter. This suggests that transition from tonic to phasic mode is rather gradual, whereas transition from phasic to tonic mode is more abrupt.

These results support the hypotheses of the AGT and the idea that patterns of pupillary are closely related to control state. It is also indicated that the LC mode regulated the trade-off between exploration and exploitation and that those changes are driven by assessments of task utility. In conclusion these studies show, that the LC-NE system not only works to facilitate behavioral responses to salient stimuli, but also regulates the balance between exploration and exploitation, depending on the task utility. However, there no direct evidence has been established yet, that pupil diameter tracks LC activity.

The link between LC-NE activity and pupil diameter

Fourteen participants took part in a study by Murphy et al (2014), in which the relationship between

pupil diameter and the BOLD activity in the human LC was investigated. A fMRI scan was

conducted while participants were instructed to think about nothing in particular and maintain

fixation on a fixation cross for 8 minutes (resting-scan). Afterwards they had to perform an oddball

discrimination task, during which there pupil diameter was recorded and an fMRI scan was

conducted.

It was revealed that there is a relationship between pupil diameter and BOLD activity in dorsal pontine cluster overlapping with the LC. This relationship was present during rest and during

oddball task performance. Furthermore, the spatial extent of this pupil/LC relationship guided a

volume-of-interest analysis in which we provide the first demonstration in humans of a fundamental

characteristic of animal LC activity: phasic modulation by oddball stimulus relevance. Taken

together, these findings highlight the potential for utilizing pupil diameter to achieve a more

comprehensive understanding of the role of the LC–NA system in human cognition. The results of

this study establish more direct evidence that pupil diameter can be used as an index for LC

activity.

Given the findings with respect to LC activity and behavioral decisions, Einhäuser et al. (2010) intended to explore whether NE may play an equivalent role in consolidating purely cognitive decisions.

Einhäuser et al. conducted three experiments to examine whether it is possible to predict covert cognitive choices using pupil-dilation measures. In their first experiment, Einhäuser et al. (2010) instructed five male participants to push a button exactly once during a 10 second interval. The participants were promised a 50 cents reward if they pressed the button during the lucky 1 second interval. They were correctly informed that the interval was chosen at random and that there was no strategy to determine the lucky interval. Wining or losing was indicated by a smiley face after each trial.

To be able to separate the effect of the decision to act from the action itself, a second experiment was conducted, in which the five participants were presented five digits consecutively for 2 seconds each. They were asked to covertly choose one of the digits. Only after the sequence had finished were the observers able to indicate their choice. If they chose the randomly defined ,lucky‘ digit, they won a 10 cent reward. Further, to assess whether maybe the expectation of reward had an influence, the same experiment was done again, except neither reward nor feedback was provided.

Finally, to establish whether any observed pupillary responses reflected the pre-decisional cognitive appraisal component of the decision or the post-decisional consolidation of the selected outcome, the researchers administered an “instructed pick” experiment. The experiment was the same as in 2B, except that the participants could not choose freely but were told what to choose.

Pupil dilation showed a general increase during the decision period, which was

independent of the chosen interval. Pupil diameter was found to increase at the time of a decision irrespective of whether the choice related to the voluntary execution of a motor act or the selection of a digit from a series of five numbers.That indicates that pupil dilation can predict an individual‘s choice before it is openly revealed.

The results of the previous studies motivated Eldar et al. (Donner et al., 2013) to study the degree to which an individual‘s learning performance is dependent on attentional predisposition, by measuring pupil diameter. The researchers asked the participants to fill in a questionnaire about the extent to which they were predisposed to process and learn from either sensory or abstract

and/or semantic dimension. They could get rewards by finding out which visual and semantic feature were associated with reward, through trial and error learning. It was shown that the higher the neural gain, thus the smaller the pupils, the more the individual learning style was boosted. These results indicate that neural gain boosts all strong correlations between the activities of local groups of neurons. Second, the findings predict that interactions between neurons should become more tightly clustered under high gain.

There is great evidence that pupil dilation can be used as a reliable index of LC activity, not only in behavioral decision tasks but also in cognitive decision tasks. Additionally the

commentary of Donner et al. (2013) on the study of Eldar et al. (2003) shows that the influence of the LC-NE system is even more complex than just the facilitation of behavioral decisions. It shows that it can also boost individual learning styles, through neural gain.

What remains uncertain is how the LC-NE system manages to influence behavior in this way. In the next section it will be examined, how the LC-NE system can affect the post synaptic neurons and thereby facilitate cognitive and behavioral performance?

The relationship between LC-NE and cortical neurons

As seen in the sections above, there is already quite a great deal of evidence that the LC-NE system affects task performance and facilitates decisions, through neural gain. However it is not completely understood how the release of NE affects other parts of the brain in order to facilitate certain behavioral responses, and how it can affect cognition. In this section, the mechanisms underlying neural gain in information processing and cognition will be examined. Recently new techniques have been found to examine how the LC-NE influences specific parts of the brain, in order to facilitate certain responses.

Carter et al. (2010) have used the new technique of optogenetic stimulation to address the

precise role LC-NE plays in relation to the rest of brain. They found that levels of arousal and

Optogenetic stimulation uses light to excite or inhibit specific neurons in the LC. In contrast, with

electrical stimulation all neurons in the vicinity of the electrode will be activated. Carter et al.

delivered the light to the LC of a free moving mouse through a fiber optic cable. They found that

optogenetic activation restricted to the LC could rapidly awaken a sleeping mouse. Further they

discovered that high- stimulation of the LC caused a loss of muscle tone. High frequency

stimulation of the LC decreased the release of norepinephrine, probably because it depleted the

cells‘ store of norepinephrine.

Polack et al. (2013) studied how gain modulation works in the visual cortex.Visual cortical

neurons fire at higher rates to visual stimuli during locomotion than during immobility, while maintaining orientation selectivity. Polack et al. (2013) examined which processes are critical for brain state-dependent changes in information processing. They measured the membrane potential (Vm) in order to determine subthreshold activity leading to alternation of visual evoked activity. To measure the activity of L2/3 neurons during locomotion versus immobility a mouse was put in a head fixed spherical treadmill. To record the spontaneous activity of the neurons, the mouse was presented with a grey screen. During locomotion the ECoG showed that low frequencies got lower in power and high frequencies got higher in power than during immobility. The Vm was significantly more depolarized and less variable during locomotion than during stationary periods. However the mean firing rate did not change. The onset of depolarization occurred slightly before the onset of locomotion, whereas the repolarization did not significantly precede the onset of immobility. Then, the mouse was presented with a series of sines-wave, enabling the reseachers to obtain a

complete orientation tuning curve of the Vm, the Vm standard deviation (SD), and the firing rates for locomotion and immobility. Those orientation tuning curves revealed that the tonic

depolarization and decrease in Vm variability was associated with an increase in the gain of L2/3 excitatory neurons without a change in orientation tuning.

In oder to find out what causes the tonic depolarization and decrease in Vm variability of L2/3 neurons during locomotion the role of neuromodulatores that are known to play a role in arousal was explored. First the role of cholinergic input was tested and it was found that it is essential for

mediate tonic depolarization during locomotion. When testing the role of norepinephrine, it was shown that the blockage of norepinephrine led to hyperpolarization of the Vm of L2/3 neurons. Also the amplitude and mean depolarization in response to visual stimulation was reduced.

It was shown that actylcholine is essential for maintaining the unimodal and broad distribution of Vm during quiescent periods. Norepinephrine appears to be necessary for depolarizing neurons during locomotion. The modification of Vm dynamics during locomotion enhanced the gain and the signal to noise ratio of the cortical neurons. Polack et at. (2013)

suggest that NE can enhance visual attention by increasing the signal-to-noise ration of excitatory neurons.

These studies provide evidence that attention is associated with an alteration of the gain of sensory tuning and therefore provide first indication of how the LC-NE system can affect cognition through its widespread connections throughout the cortex.

The link between LC-NE and stress

So far we have seen that LC-NE activity modulates neural activity and can enhance attention and arousal, which leads to improved performance in for example discrimination tasks. However, based on the Yerkes-Dodson curve there is an inverted U-shaped relationship between LC activity and task performance, which predicts that too high activity of the LC-NE system will negatively influence task performance (Aston-Jones et al., 2005). In the following section we shall examine how LC-NE activity relates to stress, which can in turn impair task performance.

Alexander et al. (2007) conducted a study to examine the link between LC-NE activity and stress-related impairments of task performance. Sixteen participants took part in that study and four conditions were used: placebo and control, placebo and stress, propranolol and control, and propranolol and stress. The heart rate (HR) and blood pressure (BP) of the participants was recorded for six minutes. Then 40mg of propranolol or a placebo were administered. The Trier Social Stress Test (TSST; Kirschbaum et al., 1993), a mental arithmetic stressor and a public-speaking stressor were used in the stress conditions, whereas similar but non-stressful tasks were

used in the control conditions. During the public-speaking task participants had give a 5-minute soeech on why they should get the job at law office of for admission into graduate business school. Cognitive flexibility was measured with lexical-semantic and associative problem-solving tasks. The control tasks included a visual-spatial memory and motor processing speed task.

The results showed that stress impaired the performance on cognitive flexibility task but not

on control task. Also cognitive flexibility improved in the stress and propranolol condition compared

to the stress and placebo condition.

These results illustrate that while psychological stress impairs the performance on cognitive

flexibility tasks in normal, healthy subjects, the effect disappears when administering a

beta-adrenergic antagonist. This indicates that the LC-NE system may well play a role in stress-related

impairments in cognitive flexibility.

Theses findings have important clinical implications, as they may offer new ways of treating stress

related diseases such as ADHD or PTSD.

According to Ramos et al. (2007) there is evidence that high concentrations of NE impair prefrontal cortex (PFC) function through activation of α1 adrenergic receptors. High amounts of NE are for example released during stress as seen in the study by Alexander et al. In humans as well as animals, traumatic stressors likely lead to excessive NE release and α1 adrenergic receptor engagement. Animal research has indicated that high levels of α1 receptor stimulation weaken the prefrontal cortex (PFC ) inhibitory functions and strengthen the amygdala function, which leads to symptoms observed in PTSD. Reversely, drugs that block those receptors appeared to lessen symptoms of PTSD, by strengthening the inhibitory functions of the PFC. Studies in animals suggest that this drug treatment may prevent amygdala-induced enhancement of the traumatic memories but may not strengthen PFC inhibitory abilities. Specific blockade of the β1 receptor subtype may provide a more powerful therapeutic effect because this strategy strengthens the PFC and could weaken amygdala function. Thus, unlike propanolol, betaxolol may prove useful in treating established PTSD as has been suggested previously (Ramos et al., 2005).

From the findings in this review we can conclude that the LC-NE system plays a complex role in modulating and facilitating behavioral decisions. There are two modes of activity, the phasic and the tonic mode. The research shows that the LC reacts to task relevant, salient stimuli with phasic bursts of activity that facilitate the behavioral response by increasing gain of cortical processing units. In this the LC-NE system works to improve task performance while in phasic mode. In the tonic mode however, task performance decreases and behavior is more distractible. This is due to the fact, that in tonic mode the baseline activity is elevated and the system is more sensitive to any stimulus. It was found that, as predicted by the AGT, the two modes serve to modulate the trade-off between exploitation and exploration, in order to maximize the reward. This trade-off is of course critical in an evolutionary sense, as behavior has to be constantly adapted to the ever changing environment around us. As the paid off reward for a certain behavior decreases, it is necessary to be able to change behavior in order to find a new, more rewarding occupation. In the last section, we can see how an over-responsiveness of the LC phasic mode is just as

maladaptive as an under-responsiveness. A high amount of NE release is associated with stress, which if chronic, can manifest itself in various stress-related disorders, such as PTSD. An

improved understanding of the LC-NE system therefore does not only have great advantages for neuroscience, but also has the potential to facilitate understanding stress-related disorders and finding possible cures. As there has not been a much research on the clinical implications yet, it would be especially important to conduct studies, that include individuals with a stress-related disorder as participants and be able to compare them to healthy participants. Howells et al. (2012) have proposed a new theory of how the LC-NE system is required for optimal performance. In their paper they suggest that, ,hyperarousal‘ disorders, like ADHD, are caused by low tonic firing of the LC-NE system, while ,hyperarousal‘ disorders, like anxiety disorders, are associated with high tonic firing of the LC-NE system. Both of which result in ineffective phasic LC activity and therefore in poor performance. This theory would explain the specific underlying mechanisms

of how the LC-NE system is related, not only to task performance, but also to psychiatric disorders. However, more scientific evidence will be needed to support this hypothesis.

This review shows how complex the function of the LC-NE systems is and that there is great potential in this field for further research.

References

Alexander, J. K., Hillier, A., Smith, R. M., Tivarus, M. E., & Beversdorf, D. Q. (2007). Beta-adrenergic modulation of cognitive flexibility during stress. Journal of Cognitive

Neuroscience, 19(3), 468-478.

Aston-Jones, G., & Cohen, J. D. (2005). An integrative theory of locus coeruleus-norepinephrine function: adaptive gain and optimal performance. Annu. Rev. Neurosci., 28, 403-450.

Clayton, E. C., Rajkowski, J., Cohen, J. D., & Aston-Jones, G. (2004). Phasic activation of

monkey locus ceruleus neurons by simple decisions in a forced-choice task. The Journal of neuroscience, 24(44), 9914-9920.

Donner, T. H., & Nieuwenhuis, S. (2013). Brain-wide gain modulation: the rich get richer. Nature

neuroscience, 16(8), 989-990.

Einhauser, W., Koch, C., & Carter, O. (2010). Pupil dilation betrays the timing of decisions. Frontiers in human neuroscience, 4, 18.

Gilzenrat, M. S., Nieuwenhuis, S., Jepma, M., & Cohen, J. D. (2010). Pupil diameter tracks

changes in control state predicted by the adaptive gain theory of locus coeruleus function.

Cognitive, Affective, & Behavioral Neuroscience, 10(2), 252-269.

Howells, F. M., Stein, D. J., & Russell, V. A. (2012). Synergistic tonic and phasic activity of the

locus coeruleus norepinephrine (LC-NE) arousal system is required for optimal attentional

performance. Metabolic brain disease, 27(3), 267-274.

Jepma, M., & Nieuwenhuis, S. (2011). Pupil diameter predicts changes in the

exploration–exploitation trade-off: evidence for the adaptive gain theory. Journal of cognitive neuroscience, 23(7), 1587-1596.

McGregor, R., & Siegel, J. M. (2010). Illuminating the locus coeruleus: control of posture and

arousal. Nature neuroscience, 13(12), 1448.

Polack, P. O., Friedman, J., & Golshani, P. (2013). Cellular mechanisms of brain

Rajkowski, J., Majczynski, H., Clayton, E., & Aston-Jones, G. (2004). Activation of monkey locus coeruleus neurons varies with difficulty and performance in a target detection task. Journal of neurophysiology, 92(1), 361-371.

Ramos, B. P., & Arnsten, A. F. (2007). Adrenergic pharmacology and cognition: focus on the