Effects of corticosterone administration on memory consolidation and decision making in a place conditioning task in rats

Student: Gerben Kraaij

Supervisor: Silviu Rusu

Effects of corticosterone administration on memory consolidation

and decision making in a place conditioning task in rats

Abstract

In this study we examined the effects of corticosterone treatment on memory consolidation and decision making in rats. Behavioral performance on a place conditioning paradigm was used to quantify these effects. Rats received a 1.0 mg/kg corticosterone injection at different time points relative to task onset for 2 weeks in total. Hormone administration at: i) 3 h before task, ii) 30 min before task, or iii) 30 min after task, was used to asses differential effects of timing of the injection. During the last three to five training sessions reward contingencies were reversed. In a first batch we found that rats developed a stable preference for the safe option, this preference was preserved in all injection conditions with no additional effect of injections on choice behavior. In the reversal phase it seemed that corticosterone injections, 30 min before task, were enhancing reversal learning. This could indicate that corticosterone enhances consolidation enabling animals to acquire new reward contingencies faster. In a second batch a general observation was that rats were not able to develop a stable preference on the original protocol. They were able to discriminate between reward conditions if contrast was maximized. Probe trials suggested that suboptimal consolidation in this batch might have caused these results. However, results from the first batch have to be confirmed in subsequent batches. In the future, electrophysiological measurements are needed to unravel the neuronal mechanisms underlying these behavioral results.

Theoretical background

Extensive research has taught us a lot about the effects of stress on the central nervous system, reviewed by McEwen (2007). The increasingly prominent role that stress is taking in daily life in western societies is one of the main reasons why over the past decades so much scientific effort has been spend to unravel the mechanisms underlying stress. Stress has a direct impact on behavior and the choices we make. This can have profound consequences, not only in financial sense (Porcelli and Delgado, 2009) but often also in terms of human welfare (Kowalski-Trakofler et al., 2003). Increasing insight into how stress affects memory and decision making through glucocorticoid concentrations could also be of value for the understanding of related medical conditions such as PTSD (Yehuda, 2006).

Stress and the central nervous system

In the human body stressful situations induce HPA-axis activation which leads amongst others to increased levels of glucocorticoids in the brain. However, it is often hard to say what effects this rise in corticosterone will elicit. Multiple other factors like stress history (Karst et al., 2010), individual differences in risk preference or anxiety (Bellani et al., 2006; Roitman et al., 2010), brain area (Vyas et al., 2002; Barsegyan et al., 2010), age (Bodnoff et al., 1995; Lupien et al., 2009) and time of the day (Butte et al., 1976) might account for heterogeneity of the observed effects. In the literature these widely varying factors can lead to contradictory or confusing results. Stressful situations also activate the sympatho-adrenal response which causes secretion of adrenaline and noradrenaline from the adrenal glands. In the brain noradrenaline is also released from the locus coeruleus. As will be addressed later on, noradrenaline might play a pivotal role in mediating the effects of corticosterone on hippocampal and prefrontal areas in the central nervous system.

Distinct effects are observed following acute and chronic stress. Acute effects arise via non-genomic or non-genomic mechanisms as a direct result of a rise in corticosterone levels after a single stressful event. Corticosterone levels are at their peak level 30 minutes after stress initiation and return to baseline within two hours. Acute genomic effects can last up to five hours meaning that the effects might still be there after the corticosterone itself is removed. Chronic effects include

permanent changes in cell structure, cell functioning and behavior, prominently through genomic action (Joëls et al., 2007; Dias Ferreira et al., 2009). Typical chronic stress paradigms involve daily stress administration for three weeks prior to behavioral testing (Dias Ferreira et al., 2009; McEwen, 2007).

Acute corticosterone effects are mediated by both membrane-bound mineralocorticoid receptors (MR’s) and glucocorticoid receptors (GR’s) in the cytoplasm. These receptors can directly facilitate neurotransmission by inhibiting hyperpolarizing potassium currents both in pre- and postsynaptic neurons. Or by increasing vesicle release probability in the pre-synaptic neuron. membrane-bound GR’s can have the opposite effect on neurotransmission but in general they operate slower than MR’s, which are activated first (Groeneweg et al., 2011). Nuclear MR’s and GR’s which are present in the cytoplasm induce genomic effects by altering gene transcription. Taken together corticosterone effects are highly dependent on brain area because of the large variation in presence and

concentration of above mentioned types of receptors receptors between areas (Groeneweg et al., 2011).

Hippocampus and memory consolidation

The hippocampus is an area which is highly sensitive to changes in corticosterone levels because of its high concentration of GR’s as well as the presence of MR’s. The hippocampus is involved in acquisition, consolidation and retrieval of information. For all these processes long-term potentiation (LTP) is widely believed to be the main underlying mechanism. LTP is affected by stresshormones in several ways. Non-genomic effects include enhanced LTP in the hippocampus mainly through membrane-bound MR’s, although GR’s might be involved in certain aspects as well (Groeneweg et al., 2011). Genomic effects are impaired LTP and an increase in long-term depression (LTD) (Joëls, 2007). There is increasing evidence that changes in glutamatergic transmission are at least partly underlying these long term changes (Popoli et al., 2012). Opposing effects on memory consolidation can be obtained depending on the type of receptor that was activated (Avital et al., 2006). Due to differences in binding affinity MR’s are activated at lower concentration than GR’s.Also the timing of the stressor relative to the task is crucial in determining the effects corticosterone will have on plasticity and learning.

Another important role is played by the amygdala. An intact amygdala is required for

glucocorticoid effects on consolidation (Kim et al., 2001). More specifically noradrenergic activation is required (Roozendaal et al., 2006) and shows a dose-dependent increase in memory retention performance (McIntyre et al., 2002). Through opposing effects on excitatory and inhibitory neurons in the basolateral amygdala (BLA), the signal to noise ratio of outward projecting neurons is

increased (Roozendaal et al., 2009). This changed signal from the amygdala is thought to induce LTP processes at synapses outside the amygdala. In this way they can prone active inputs towards synaptic potentiation as has been demonstrated for corticostriatal inputs (Popescu et al., 2010). Acute stress can already affect acquisition of memories elsewhere in the brain (Shackman et al., 2011; Kajiwara et al., 2002), but most robust effects are found on consolidation of already formed memories. Consolidation can be defined as “the process whereby a memory becomes increasingly resistant to interference from competing or disrupting factors with the continued passage of time” (McGaugh, 2000). During this process memories can stabilize, enhance, reorganize and restructure (Stickgold, 2005). This is a critical process since newly formed memories are often very malleable and prone to disruption by other incoming information.

High frequency ripple events that occur during slow-wave sleep might contribute to this

consolidation. During ripples highly synchronized activation is observed between hippocampus and neocortex (Ji and Wilson, 2007). Not only do they allow hippocampo-neocortical communication (Buzsaki,, 1996), they also allow associations to form between for example spatial and reward

information as described by Lansink et al, (2009). Importantly Lansink and colleagues found that the hippocampus is the leading structure in this process, which underscores its importance.

Using different protocols, both evidence for enhancement (Thompson et al., 2004; Nathan et al., 2007) and impairment (Kim et al., 2007) of consolidation has been obtained following stress or glucocorticoid administration.

Place cells and stress

Regarding the hippocampus and memory consolidation, much research has focused on place cells. The place cell network in the hippocampus, which was originally discovered by O’keefe and Dostrovsky (1971), has now been extensively studied. Place cells fire high frequency bursts,

superimposed on the normal theta rhythm, depending on the animal’s location in space. In this way they enable the animal to determine its position. The formation of stable place cells is of great importance for performance on spatial tasks. The more certain and precise an animal is able to determine its position, the better it can guide itself through the environment.

The functioning of place cells is believed to depend on plasticity properties of the underlying circuits in the hippocampus. Inhibition and excitation at different frequencies and at different locations along the dendrites can create rhythmic patterns of activity in hippocampal pyramidal neurons. By changing the ratio of inhibitory and excitatory input the frequency of a place cell can be adjusted. This mechanism can cause a place cell to fire earlier with regard to the theta rhythm as the animal approaches the center of a place field (Magee, 2001).

Kim et al, (2007) suggest that LTP impairment, as induced by corticosterone, can affect the formation of stable environment specific representations. They exposed rats to 2 hours of acute, unpredictable stress through playing loud white noise at irregular intervals. In hippocampal slices they found that place cell firing locations were stable, but the rate maps displayed greater variation than normal. An impairment in hippocampal LTP was thought to have caused these changes. However, as noted before, plasticity in the hippocampus might also be enhanced depending on the nature and timing of the stressor. The effect of corticosterone on place cell functioning is important since Lansink et al, (2009) demonstrated that place field information is coupled to reward

information in the striatum. A destabilization or enhanced stabilization of place cells can thus have consequences, not only for consolidation in hippocampus, but for the functioning of other areas as well.

Value coding

Hippocampus does not only have connections with the striatum but also interacts with prefrontal neocortical areas like medial prefrontal cortex (mPFC) in spatial memory tasks (Benchenane et al., 2011; Jones and Wilson, 2005). mPFC appears to be integrating both spatial and reward information to guide behavior (Wallis, 2009). Much of this reward information is received from orbitofrontal cortex (OFC), which is involved in the encoding of reward value and decision making. As indicated by lesion studies (reviewed by Wallis, 2009), OFC also is important for reversal learning and updating valuation based on the outcome of choices. Although reward sensitive neurons are also found in areas other than OFC (Wallis, 2009), the cleanest value signal is encoded in the OFC (Wallis, 2009). Value is a combined concept arising from the combination of reward properties like magnitude and probability. Both parameters are encoded in the OFC (Van Duuren et al., 2008) Firing patterns are most correlated with expected reward magnitude and probability during expectancy period (Van Duuren et al., 2008).

Despite our knowledge about encoding of reward properties, not much is known about how this coding is affected by stress hormones. In general it has been found that chronic stress causes structural reorganization in frontal and striatal circuits in rats, leading to habitual behavior (Dias Ferreira et al., 2009; Packard, 2009). OFC activation is found to decrease in humans after acute stress (Ossewaarde et al., 2010), but it is not certain what the electrophysiological correlates of this

decrease are. Glucocorticoids can impact on neuronal coding either at PFC level via hippocampo-prefrontal interactions or directly via MR’s or GR’s. Nongenomic effects like facilitation of

oppositely impair OFC functioning and disrupt decision making processes. The same time dependency as for the hippocampal effects might apply here.

Research questions and hypotheses

Drawing from these observations we would like to assess what the effect of acute glucocorticoid administration will be on memory consolidation and decision making in a place conditioning task. To formulate specific expectations the time point of injection relative to task onset is of great

importance. If corticosterone is administrated long enough before task for genomic mechanisms to exert their effects, consolidation will be impaired. Administration directly before the task will impair memory retention but enhance learning due to acute effects. When administered after the task, corticosterone will not influence acquisition but will solely enhance consolidation (Nathan et al., 2004; Thompson et al., 2004). We expect that glucocorticoid administration after task will induce the strongest enhancement of memory consolidation.

Since the formation of a stable firing pattern of place cells is of crucial importance to

consolidation of information in hippocampus we expect to find effects on place cells that are in line with our expectations on memory consolidation. Place cell firing is expected to be destabilized if glucocorticoids are administered 2 hours before task onset (Kim et al., 2007). Acute effects of corticosterone can oppositely enhance place cell stabilization either when administered shortly before or shortly after the task.

With regard to decision making expectations are less clear and we strive to find the time point of administration with the greatest effect on decision making. At behavioral level we expect the rats to prefer safe options before administering corticosterone. Corticosterone administration can enhance or disrupt neural signaling in reward related areas like OFC. On a behavioral level this can be

reflected in a change to risky choice behavior or indifferent behavior. Once a change in behavior has been identified electrophysiological correlates for this change will be looked for. Reversal learning will also be assessed; here we expect that corticosterone administration will drive rats towards habitual behavior (Dias Ferrera et al., 2009) leading to slower acquisition of new reward contingencies.

Material and methods

Rats from two different strains have been used in this project. For the first batch 5 young Wistar Unilever rats were obtained from Harlan Laboratories. For the second batch 12 young Wistar

Hannover rats were obtained from the same breeder. Rats were housed in cages pairwise. They were handled and weighed on a daily basis. Water and food were available ad libitum until rats reached approximately 85% of free feeding adult weight. From that moment rats were placed on a food restriction diet and kept constant at 85% of free feeding weight or

more, depending on individual task performance. The second batch was split up in a traditional food restriction and a diet board group. This was done to test whether the diet board could replace traditional food restriction as a feeding method (Kasanen et al., 2009) in our behavioral paradigm. Rats were housed in a temperature-controlled (21°C) room with a 12 h light/dark cycle (lights on 9:00 AM to 9:00 P.M.). Training was performed during the light phase between 9:00 AM and 5:00 P.M. Due to the circadian and ultradian corticosterone variation, testing was performed at the beginning of the light phase between 9:00 A.M. and 13:00 P.M. when corticosterone levels and ultradian variations are minimal.

Fig 1. Schematic representation of

the used maze with the black triangle representing the folder used in the first step of training.

All experiments were approved by the Animal Experimentation Committee of the University of Amsterdam and were performed in accordance with the National Guidelines for Animal

Experimentation. Apparatus

For all training and testing procedures a rotatable hexagonal-shaped maze was used with small upstanding walls attached to the runway (Fig 1). Six identical reward ports were located along the track. In the first phase of training rats only had access to one of the ports. In all other phases rats could clearly see all ports regardless of their position on the maze. Poking at a reward well delivered a reward with a certain magnitude and probability depending on the protocol that was used, as described below. Two lights on top of each port were used to indicate whether poking would be rewarded. Sensors on the maze recorded the location of the animal and ongoing events. These signals were transferred to a neuralynx data acquisition system.

Training and testing were conducted in a symmetrical room with dimed light (50 lux at the maze level) and white noise. Two cues were located on the walls to allow rats to determine their location. The maze was cleaned with ethanol solution (70%) after each training session to avoid choice bias due to odors and residue left from previous rats.

Fig 2. A) Timeline for the first batch of rats. In this batch 1 rat was sacrificed during training because it did not

learn the task properly. Two more rats were excluded during testing due to performance problems. B) Timeline for the food restriction group. N=5 during training period, N=4 from start of testing C) Timeline for the diet board group. N=5 from start of training until present. Numbers on the y-axis indicate dates.

Protocols

Common training protocol

The first three training steps have been the same for all three batches used in this study. In this initial training rats are accustomed to the maze and in several steps they are trained to perform 120 trials on the maze in 45 minutes with a poking time of 3000ms. At the first step of training rats were restricted to one port by placing a folder over that port. Lights went on after a variable inter-trial interval and 75 µl sucrose solution (15%) was released from the reward well after every coincidental poke. After performing 30 trials in 30 minutes the time rats needed to poke for a reward was

increased stepwise from 150 to 1000ms. Once they completed 30 trials in 30 min with 1000 ms poke time, rats were given access to the rest of the maze. After doing a trial at a port, only the lights at the next port in clockwise direction went on in order to teach the rat to do clockwise trials. This phase is called the cue rotate phase. In subsequent training steps in this phase, poking time or number of trials was increased every time the animal reached test criteria for a training step. After doing 60 trials in 30 minutes with 1500ms poking time, reward probability for each port was set to 75% to habituate rats to uncertainty in reward outcome. At the end of this habituation phase rats could do 120 trials in 45 minutes with a poke time of 3000ms which ended the initial training.

During this whole training rats had to go from one port to the next. They were not allowed to skip ports, so all these trails were forced choice trials. From here the protocol diverges for the different batches, therefore their protocols will be described separately.

Batch 1

Additional training After initial training rats from the first batch were trained on a place

conditioning protocol. This protocol involved forced choice trials with three different reward conditions. Two ports were assigned to one of these reward conditions: i) 75 µl with 75% reward probability (75%L), ii) 75 µl with 25% probability (25%L), iii) 225 µl with 25% probability (25%H). Over blocks of four trials per port, 25%H and 75%L conditions yielded the same reward amount (225 µl), while the 25%L condition yielded only 75 µl. The 25%L option is considered the indifferent option, 25%H is considered the risky option (large infrequent rewards), and 75%L the safe option. After this rats were trained with the same reward conditions but now they were free to do a trial at any port regardless of the port they last visited, as long as they traveled in clockwise direction. This allowed rats to develop a pattern of preferences with respect to the reward conditions.

Testing In batch 1 testing started after a stable preference (5 consecutive days) had formed on the

free choice place conditioning protocol. In the testing phase, in which the animals were receiving injections, the same place conditioning protocol was used (Fig 2a). After two weeks of injections three testing sessions followed in which reward conditions for the ports were reversed in the first session and maintained for the other two. This was done in order to determine behavioral flexibility. Batch 2

Food restriction group

Testing In the food restriction (FR) group the testing phase started directly after rats had

completed the habituation to uncertainty phase, before they could develop a preference on the place conditioning protocol. In this group the reversal was delayed until modifications to the protocol enabled rats to develop a stable preference.

Additional protocols An extra protocol was introduced after the initial testing phase to assess

whether rats were able to skip ports and discriminate between ports with different reward conditions. Four adjacent ports were rewarded with 100 % probability; the other two ports were unrewarded. This protocol is referred to as the spatial memory task. Later on, another simplified protocol was introduced to speed up and stabilize the development of a preference in this group. In this protocol the indifferent option (25%L) was excluded and the other two conditions were each assigned to three adjacent ports. After 4 sessions reward amount on the 25% ports was lowered

from 225 µl to 150 µl to match relative outcomes in IOWA gambling task. This protocol was maintained for 10 sessions.

Probes Starting at the spatial memory protocol, 24 probe trials were introduced before and after

training sessions. During probe trials all ports delivered a low magnitude reward with 50% probability. This was done to assess memory consolidation and working memory.

Diet board group

Additional training Both the FR group and the diet board (DB) group were trained at the same

time. Because time constraints did not allow both groups to be injected and tested at the same time, the DB group was given remainder sessions twice a week as long as they kept reaching performance criteria. Because performance dropped at some point, reward probability on all ports was set to 75%. Injections

The adrenocortical hormone corticosterone (HBC complex, Sigma) was injected intraperitonealy into the abdominal cavity of the rat. Used concentration was always 1 mg/kg. This dose has

previously been shown to affect memory consolidation (Roozendaal et al., 2006) and resemble natural corticosterone level increases after a forced swim test (Droste et al., 2008). Drug solutions were prepared in advance and stored at -20°C. Maximum injected amount was 1 ml.

Three different timepoints of injection relative to task onset were used: i) 3 hours before task (-3h cort), to study acute genomic effects of corticosterone, ii) 30 min before task (-30m cort) to assess acute nongenomic effects, or iii) directly after task (a.t. cort), to assess effects on memory

consolidation. Blood collection

Rats were restrained with a towel and a tail vein incision was made with a single edge razorblade. Blood was collected within two minutes after removal from the cage to prevent rises in

corticosterone levels to reach the blood stream. Collection was done using a microvette° (300 µl). Microvettes were centrifuged at 7500 rpm for 10 min at 4°C. The supernatant was removed and stored at -20°C for further processing. Blood collection was performed at different time points; i) 24 hrs before the first testing to assess baseline corticosterone levels, ii) immediately after task during the testing phase to assess corticosterone levels during testing, and iii) 24 hrs after the last injection to check for changes in baseline corticosterone levels following treatment.

Analysis

Analysis of behavioral data was performed using custom written routines in MATLAB. Figures of obtained results were also made using MATLAB.

For training and testing phases three different parameters were analyzed; i) number of valid trials per port. A trial was considered valid if the rat was running clockwise and if the rat poked longer at a reward well than the preset poking time while the light at that port was on. ii) approach latency, and iii) number of licks per port. The number of trials per port was normalized with respect to the total number of trials. For chosen comparisons significance values were either calculated via one tailed t-tests in excel or via one-way ANOVA’s (SPSS Statistics 17.0). In all comparisons a probability level of < .05 was considered to be significant. The same method was used for analysis of approach latency for each port. The normalized percentage of licks per reward condition was calculated by dividing the percentage of licks per reward port by the total number of licks at all reward ports. To quantify the preference of the rats for different reward conditions a preference index (p.i.) was calculated. This was done by subtracting the total number of trials per port for one condition from the total number of trials per port in a second condition. The difference was then divided by the total number of trials on all ports and multiplied by 100 (Hokosawa et al., 2007).

To compare performance of the diet board and food restriction group, a learning index was calculated. This was done by calculating the reciprocal of the number of days needed to reach

performance criteria for the next training step. The learning index was calculated by taking the average of these numbers for all training steps.

Results

For the first batch three rats entered the testing phase. Two different injection conditions were used for each rat, these conditions were alternated between days. The first rat received either; i) saline 3 hours before task, or ii) saline 30 min before task. This rat was never injected with

corticosterone and is therefore referred to as the control rat. The second rat received either i) saline 30 min before task, or ii) corticosterone 30 min before task and is referred to as the -30m rat. The third rat followed the pattern of the -30m rat, but was always injected 3 hours before task and is therefore referred to as the -3h rat. Due to poor performance of the fourth rat we were not able to inject directly after task during the testing phase. During the reversal sessions, the same injection conditions were applied.

Training After switching from the forced choice to the free choice protocol, rats developed a clear

preference for the two ports representing the safe option (BP040

p < .01, 75% ports vs. both 25%L ports and one 25%H port; BP042 p < .01, 75% ports vs both 25%L and both 25%H ports; BP044 p < .001, 75% ports vs. both 25%H ports and one 25%L port). After this preference had been stable for five days (Fig 3.) we moved to the testing protocol.

Testing Figure 4 displays the normalized

number of trials per port for both injection

conditions per rat and for the last five days before injections. Sessions in which animals did less than 48 trials were excluded. Preference for the safe option (75% ports) is preserved in the testing phase. Again the number of trials per port significantly differs between 75% ports and ports from other conditions (BP040 =< .01, 75% ports vs. both 25%H ports and one 25%L port; BP042 p < .05, 75% ports vs. both 25%L ports and one 25%H port; BP044 p < .01, 75% ports vs. both 25%H and both 25%L ports). Between injection conditions within animals there are no structural significant differences. No effect of corticosterone injection on choice preference was observed.

Fig 3. Preference indexes for all three animals over training sessions. Preference index of 0 shows

indifferent behavior, indicated by dotted grey line. Black arrow indicates switch from forced to free trials

Fig 4. Normalized number of trials per port, per

injection condition over whole testing phase. The no inj. condition displays the average values for the last five sessions before injections were started.

Reversal The results from the second and third session of the reversal phase are displayed in

figure 5. Due to previously described constraints on the amount of days we could inject, only three reversal sessions were done. Both animals startedwith an approximately equal preference for the former 75% ports in session 1. The rat which received corticosterone injection 30 min before task had largely extinguished this preference in the second session (Fig 6) (p.i. 75-25H = -2.7; p.i. 75-25L = -5.3) and developed a trend towards a preference for the new 75% ports in the third session (p.i. 75-25H = 6.7;

p.i. 75-25L = 14.3). The saline injected rat still preferred the former 75% ports in the second session (p.i. 75-25H = -22.2; p.i. 75-25L = -50.9) and had only extinguished

it in the third session (p.i. 75-25H = -7.1; p.i. 75-25L = 0). Concluding, it seems like the corticosterone injected rat was faster in picking up the new contingencies.

Food restriction group:

For the food restriction group, four rats were selected for the testing phase and each rat was assigned to one of the above mentioned conditions: i) -3h cort, ii) -30 min cort, iii) a.t. cort, or iv) control and are referred to as such. Injections were administered for a maximum of 13 days to minimize chronic corticosterone effects.

During the reversal sessions, the same injection conditions were applied.

Testing In total 13 days of injections have been performed. There was no difference in reward

conditions on the forced choice trials before the testing phase; therefore no preference was present at the start of the testing phase. Comparing the normalized number of port visits per reward

condition over the whole testing period showed that two rats were avoiding the 25%H ports (Fig 7) (p < .05 25%H vs. both 25%L and 75% ports for the control rat, and the a.t. cort rat). A third animal was visiting the 25%L ports less (-30 min rat, p < .05 25%L vs. both 25%H and 75%L ports).

Comparing the normalized average number of trials per port for the first and the last four days of testing yielded two significant results. The -3h rat visited a 25% low reward port less during the last four days (p < .05). The -30 min rat visited a 75% low reward port more in the last four days Fig 5. Normalized number of trials per port for second

and third reversal session for corticosterone and saline injected rat.

Fig 6. Preference indexes for both

animals tested on three reversal sessions.

Fig 8. Averaged approach latency per reward condition for first

and last four days of injection period.

Fig 7. Normalized number of port visits per

reward condition over the whole injection period

compared to the first four (p < .05).

We next measured reward port approach latency. Rats are thought to approach a port faster if they prefer that port. Results for this comparison are summarized in figure 8. The control rat became faster in approaching the 25% low ports (430 – 398 ms, p < .05),

which were grouped together for this comparison. The -3h cort rat showed the opposite behavior and was slower to approach 25% low reward ports (458 – 522 ms, p < .05). The -30min cort was approaching faster on all three reward conditions (25% low 528 – 425 ms, p < .001; 25% high 531 – 443 ms, p < .05; 75% low 492 – 415 ms, p < .05). For the latter rat the decrease in approach latency for all ports reveals an effect of training but not of reward condition. Since no stable preference pattern developed in the initial testing period,

extinction training on two ports was performed for three consecutive days. No injections were given during this phase.

Spatial memory task The normalized number of trials per port is displayed in figure 9A. After two

sessions all four rats had developed a preference for the 100% ports and were regularly avoiding the 0% ports. In three out of four rats this preference already started developing the first day. In probe trials which were introduced before and after training, on the third session no preference was observed in the first probe. In the second probe all four animals were showing a preference for the 100% ports (Fig 9B). After this demonstration of development of a preference pattern, rats were placed back on the original protocol.

Fig 9A. Average normalized number of trials per

port for three days of spatial memory training.

Fig 9B. Average normalized number of trials

Retraining on original protocol For six sessions rats were placed back on the original place

conditioning protocol. The average percentage of visits per port over this period is displayed in figure 10. It was found that the control rat visited the 75% ports more often than a 25%L port (p < .01). The a.t. cort rat visited one 75% port more often than one 25%L port (p < 0.05). For the -3h cort rat both 75% ports were visited more than one 25%L port (p < .05) and one 25%H port (p < .05). Although these differences were found overthe whole period, there was little stability from day to day. Because rats did not develop a stable preference during the retraining on the original protocol we introduced a simplified version where 25%L was omitted, as detailed in the “Methods” section of this

report.

Simplified protocol & IOWA task The normalized average number of trials per port was calculated

for four sessions on the simplified protocol (Fig 11). In three out of four animals there was a significant difference between the 75% and 25% reward conditions (p < .05 for the control rat and a.t. cort rat; p < .001 for -3h rat). Variability was still high from one day to another. After adjustment of the protocol by lowering the reward magnitude at 25% ports all four animals reached statistical significance for the same comparison (Fig 12) (control, p < .001; a.t. cort, p < .01; 3h cort, p < .001; -30 min cort, p < .05).

The total number of licks at the reward well was also analyzed (Fig 13).

During the first probe three rats were licking more at the 75% ports (Fig 13A. control, p < .05; a.t. cort, p < .05; -30 min cort, p < .001). During the second probe all four rats where licking more at the 75% ports (Fig 13B. p < .001 for all four rats). Since at this step rats did develop a stable preference we went on to reverse the reward contingencies as described in the “Methods” section.

Reversals During the reversal sessions, in

which the animals were injected again, there were no significant differences in the number of port visits between the two reward conditions. Looking at the percentage of licks the same comparisons were made for the probes of the Fig 10. Averaged normalized number of trials per

port on original protocol after spatial memory task.

Fig 11. Averaged normalized number of trials

per reward condition on simplified protocol

Fig 12. Averaged normalized number

of trials per port per reward condition on IOWA protocol

reversal sessions as for the IOWA-like protocol (Fig 14). While in the first probe two rats were licking more at the 25% ports (Fig 14A. a.t. cort, p < 0.01; -3h cort, p < .05), in the second probe two rats were licking significantly more at the 75% ports (Fig 15B. a.t. cort, p < .01; -3h cort, p < . 05) and one rat showed a trend towards significance (control, p = 0.07).

Diet board group

For both the food restriction and the diet board group a learning index was calculated as described before. In figure 15 the average performance index is displayed per group. The learning index of the diet board group was lower compared to the food restriction group although this did not reach significance (DB 0.78 ± 0.06 vs. FR 0.85 ± 0.02, p = 0.23,), but also more variable (Fig 15). Diet board animals were heavier than food restricted animals on average, and had greater weight Fig 13A. Normalized percentage of licks per reward

condition for first probe (24 trials) on IOWA.

Fig 13B. Normalized percentage of licks per reward

condition for second probe (24 trials) on IOWA.

Fig 14A. Normalized percentage of licks per reward

condition for the first probe (24 trials) during reversals.

Fig 14B. Normalized percentage of licks per reward

Fig 15. Average learning index for

both diet board and food restriction group. Empty squares represent individual animals (LI = Learning Index).

variability as well. Additionally, weight seemed to correlate negatively with performance during training (Fig 16).

.

Discussion

Rats in batch one were able to develop a stable preference pattern on the place conditioning protocol in which they favored the safe option. In the reversal phase corticosterone administration appeared to aid extinguishing of the old preference pattern and the development of a new

preference. Rats from the food restriction group did not develop a stable preference during the injection period. Although rats showed an improvement in performance within a single session this improvement was largely gone at the start of the next session. After modification of the protocol rats were able to develop a preference for the safe option.

First batch

Training and testing Figure 4 shows that all three rats were having a clear preference for the 75%

condition over the other two conditions. The differences between preference indexes for 75% over 25%H, and for 75% over 25%L were minor, which means that rats preferred the safe option and were indifferent between the other two conditions. After the first two sessions displayed in figure 4 the protocol changed from forced choice to free choice trials. Within several sessions the preference is clearly starting to develop and it stays present until the end of training. This demonstrates that animals were able to discriminate between reward conditions and that they could develop a preference for the safe option in a stable manner.

The number of visits on the 75% ports differed significantly from all ports except two. These two ports lie adjacent to a 75% port. It is reasonable to assume that the preference for the 75% ports increases the number of visits on these adjacent ports. Therefore, to maximize differences between reward conditions and reduce proximity bias, in the second batch all three conditions were assigned to adjacent ports.

During testing of the first batch we found little to no effect of corticosterone injection on choice behavior. Preference for the safe option was preserved. There are two possible explanations that could have led to this result. In the first place it is known that in humans the effect of stress on choice behavior is also dependent on the nature of the outcomes. When reward outcomes are negative, people tend to make risky choices under stress. When outcomes are positive, people tend to prefer

Fig 16. Correlation between

performance and weight in two animals from diet board group. Grey bar indicates the required weight.

the safe option (Porcelli and Delgado, 2009). To the extent that these results apply to rats, it could be that corticosterone biases rats towards choosing the safe option, which they were already doing. It might be that this was masking the effect of the injections.

Another additional option is that the used dosage of corticsoterone was too low. At least two studies found no effect on memory performance of 1 mg/kg, while they did find an effect of higher doses of corticosterone on either classical fear conditioning (Hui et al., 2004), or retention of fear memory (Cai et al., 2006). The required dosage might also be dependent on age as suggested in a study by Bodnoff et al, (1995). They found that chronic corticosterone treatment (0.3 mg/kg) for 3 months did cause spatial memory deficits in mid-aged but not in young (2 to 5 months old) rats. These findings were attributed to strong feedback on the HPA-axis in early life phases. Since rats from the first batch were within the range of 2-5 months this feedback might mask the expected effects.

Another study which did not find an effect of corticosterone injections on behavior was done by Graham et al, (2010). They found that injections of 3.0 mg/kg corticosterone did not impair learning behavior on a water foraging task, while severe stress induced by tail shocks did impair behavior up to several days. They also show that amygdalar inactivation also blocks the observed stress effects. This is in line with observations from Kim et al, (2005), Nathan et al, (2004), Roozendaal et al, (2006) and McIntyre et al, (2002) that an intact amygdala, and more specifically noradrenergic activity is required for glucocorticoid effects on memory processes. This leaves increasing the dosage of corticosterone and adding a noradrenaline-injected group as possibilities for future stages of the project.

Reversal During reversal learning we found that the corticosterone injected rat was somewhat

faster to extinguish his old preference and developed a trend towards a new preference in the third session. Two explanations can be given for this. In the first place this can reflect increased

consolidation caused by corticosterone. It could be argued that the observed difference is due to increased acquisition, however, judging from the literature this is unlikely. If any, corticosterones effect on acquisition would be impairing instead of facilitating preference extinguishing or development (Graham et al., 2010). This is because corticosterone is known to impair working memory (Barsegyan et al., 2010). Therefore the assumption is made that the corticosterone has increased consolidation. This is done in the knowledge that this finding needs replication because it has only been found in one animal.

Another possibility would be that acute effects of corticosterone optimize OFC functioning. OFC plays an important role in reversal learning, with lesions of OFC causing inabilities to adapt behavior to reversed reward contingencies (reviewed by Wallis, 2009). If the opposite is true it might be that optimization of OFC functioning by glucocorticoids facilitates reversal learning. Adding to this is the fact that the corticosterone injected rat was injected 30 min before task. This means that peak levels of corticosterone are present at task onset.

Food restriction group

Testing Opposed to the first batch, in the second batch rats did not develop a stable preference.

Over the testing phase three rats were showing significant differences between reward conditions. However, preference was somewhat unstable. Although the control and a.t. cort rat appear to prefer the safe option over the risky option, they also prefer the indifferent option over the risky option without discriminating between the safe and the indifferent option. The a.t. injection rat appears to avoid the indifferent option, which could indicate that he is indeed maximizing the net amount of reward. But besides the lack of stability from session to session, this preference is unstable when analyzing approach latency. Approach latency yielded a learning effect for this rat which approached all ports faster at the end of the testing phase, without discriminating between reward conditions. In general the preference pattern is observed in the first batch of rats did not match the second. One of the first things to note is that due to an infection problem at the breeder, a different substrain of rats has been used for the second batch. It appears unlikely that this is the cause of our observations because genetic differences between the strains are minimal. Supporting this claim is a

study done by Belda et al, (2004). In this study they compared Sprague Dawley rats and Wistar rats on two different tasks in which they determined HPA-axis responsiveness. They found no evidence of different responses in these strains. On the contrary a similar behavioral problem was observed in a study performed at the UMC in Utrecht (The Netherlands), where the same switch in used strain was necessary.

We considered the possibility that rats were not using the distal cues in the room to navigate, but were in fact using proximal cues from the maze to guide themselves. We rotated the maze every day while keeping the locations of the reward conditions stable in space. In this way the cues on the wall were the only reliable predictors of at which port a certain reward condition could be obtained. If rats were using the proximal instead of the distal cues for spatial orientation, than this might explain their inability to develop a preference. We tested this hypothesis by keeping the maze position stable for four days. No preference developed over the course of these four days.

Spatial memory task

To find out if rats were able to discriminate between reward contingencies at all we performed three sessions of extinction training in which two ports never delivered a reward. Over these three sessions all four rats were avoiding the 0% ports. Noteworthy, the -30 min cort rat showed this avoidance on the first session. This tells us the rats are able to discriminate based on reward probability and that rats were not insensitive to reward outcome. On the last day of extinction training probe trials were introduced before and after training. During the first probe session none of the rats was avoiding the 0% ports. During task, avoidance was clearly present in all four rats and was retained in the second probe. Two things become clear from this. In the first place it seems that working memory is functioning correctly. Rats pick up the changed contingencies within a session. Secondly it appears that the rats show suboptimal consolidation. When looking at the normalized number of trials, in the first probe rats seem to have no preference at all even though the day before all rats displayed a preference for 100% ports. This could either be due to a problem in consolidation or retrieval. Further analysis will be needed in order to clearly state which explanation is most likely. This analysis could also include within session analysis of sessions on earlier protocols in which probes were not present.

Although it was shown that working memory is functioning correctly on this protocol, it might still be the case that insufficient working memory capacity prevented rats from developing a preference on the original protocol. Rats have to integrate 24 trials with three conditions over six different ports to get a proper expectation of the value of each port. If they fail to retain this information correctly, the derived expected values might be distorted, leaving the rats with little possibility to discriminate between ports. No probe trials were performed on the original protocol. Therefore, to verify this possibility, comparisons in the normalized number of port visits must be made between the beginning and end of each session.

Retraining on original protocol After having determined that the rats were able to discriminate

between ports they were placed back on the original protocol. After six sessions no real preference was observed in any of the animals, probably due to suboptimal consolidation and task complexity compared to the spatial memory task. A few significant differences were found as described in the results, there were however two problems which make it hard to draw a clear conclusion. In the first place the significant differences did not seem to be consistent. For example, in the a.t. cort rat only one of two 75% ports is preferred while there is no difference for the other port. The other problem is that there was high variability in number of visits from session to session. Forming a stable preference would probably take a few weeks under these circumstances, which is too long for our project since the number of corticosterone days is limited. Therefore the protocol had to be simplified so that animals could easily develop a stable preference within the range of days we can inject. The protocol was modified by excluding the 25%L condition.

Simplified protocol The simplified protocol caused animals to acquire a preference at a faster pace

than the original protocol and is therefore more suitable for our project. Still one animal had not developed a preference after 5 sessions. Another adjustment was made by lowering the reward

magnitude of the 25% ports. This task approximates the IOWA gambling task in which the risky option, with high infrequent rewards, yields a net loss compared to the safe option.

IOWA protocol Over 9 sessions all four animals visited the safe option ports more than the risky

option ports. Because of a faster and more reliable preference development, this protocol is suitable for the rest of the project. Analysis of the percentage of licks during the probes showed a substantial difference between preference on the first and second probe. This was taken as support for our previous hypothesis that animals are showing suboptimal consolidation.

Reversal Since a clear preference for all four animals had formed on the IOWA task,

corticosterone treatment was administered during reversal sessions. Remarkably the number of trials per port did not reveal a preference throughout the task, while consistent significant

differences were observed in the number of licks per reward condition. The results from the analysis of the number of licks during the probes resembles the results from the probes during the spatial memory task. Although three rats were showing a small preference during the first probe, this preference was always much higher during the second probe. This means that what was learned during a session was only partly remembered during the first probe the next morning. This argues again for a problem in consolidation in this batch of rats. The results in Fig 14 show that during the second probe the -3h cort and a.t. cort rat both licked significantly more at the 75% ports than at the 25% ports. In general all rats started licking more at the 75% ports and less at the 25% ports

compared between the first and second probe. So all animals seemed to be learning, but differences in learning could be related to the injection conditions. At this point it is hard to explain why

corticosterone administration 3 hours before task would aid acquisition compared to the control rat. While administration 30 minutes before task would not. Individual differences in learning speed might also play a role here and investigation in further batches is needed to verify if these results are consistent. An alternative explanation, also regarding the fact that the control rat showed a trend towards significance (p = 0.07), might be that corticosterone administration 30 minutes before task impairs learning during the task and that all other rats just show the effect of normal learning. Diet board group

As a feeding method traditional food restriction appeared to have an advantage over the diet board. A more natural and stable food intake would increase animal welfare. Motivation to perform on the task was thought to be unaffected. We found that the diet board caused more variation in weight which was due to the frequent replacement of the diet boards. Rats gained heavily after replacement and lost weight before replacement. The diet boards caused the rats to be heavier on average than the food restricted rats. And the correlation between body weight and learning performance indicates that motivation is not maintained during the use of the diet board. As reflected by the learning index, rats from the diet board group were slower to learn the task than rats from the food restriction group. Several observers noticed that rats in the diet board condition were more stressed. This could be due to the continuous presence of the food in the cage without the rats being able to access it directly. Severe dental problems in one rat and developing problems in two others brought additional evidence against a possible advantage of the diet board as a feeding method. Added together the DEC considered the evidence we found enough to discard the diet board as a feeding method and to allow us to continue using traditional food restriction.

Consequently rats from the diet board group were placed on food restriction. Due to their alternative feeding history caution must be taken when interpreting and using results from experiments

performed on this group. General discussion

Summarizing, the results so far have been rather variable and due to changes in rat substrain and training protocol it is difficult to draw a complete picture. Further experiments and an increase of animal number for each experimental group might yield interesting results. For most observations described here it is possible that individual differences between rats could have a great influence as well. Therefore replication in future batches is crucial.

Future plans

Since no consistent effect of corticosterone on memory consolidation or decision making has been observed yet, additional behavioral data will have to be gathered using the latest protocol. Afterwards the project can advance to the electrophysiology phase. Tetrode recordings will be made from both hippocampal and orbitofrontal areas. At the start of the project changes in firing rates of hippocampal place cells were expected as a result of corticosterone injections. However, our expectations might have to be slightly adjusted according to a recent study by Wikenheiser and Redish (2011). Recording from hippocampal place cells in a similar place conditioning task they found the firing rate of place cells to be dependent on the location of the reward conditions. For long it has been known that the majority of place cells was not coding for spatial location alone. In general, other parameters like head direction, velocity or reward also determine the firing frequency of a place cell. According to this study we might also add reward location. In the experimental procedure as it is right now the injections start at the same time as reward contingencies are changed. However this can be overcome by comparing the sham-injected and corticosterone injected group.

References

Avital, A., Segal, M., Richter-Levin, G. (2006). Contrasting roles of corticosteroid receptors in hippocampal plasticity. Journal of neuroscience, 26(36), 9130-9134.

Barsegyan, A., Mackenzie, S. M., Kurose, B. D., McGaugh, J. L., Roozendaal, B. (2010). Glucocorticoids in the prefrontal cortex enhance memory consolidation and impair working memory by a common neural mechanism. PNAS, 107(38), 16656-16660.

Belda, X., Marquez, C., Armario, A. (2004). Long-term effects of a single exposure to stress in adult rats on behavior and hypothalamic–pituitary–adrenal responsiveness: comparison of two outbred rat strains. Behavioral brain research, 154, 399-408.

Bellani, R. Luecken, L. J., Conrad, C. D. (2006). Peripubertal anxiety profile can predict predisposition to spatial memory impairments following chronic stress. Behavioral brain research, 166, 263-270. Benchenane, K., Tiesinga, P. H., Battaglia, F. P. (2011). Oscillations in the prefrontal cortex: a gateway to memory and attention. Current opinion in neurobiology, 21, 1-11.

Berke, J. D., Breck, J. T., Eichenbaum, H. (2009). Striatal versus hippocampal representations during win-stay maze performance. Journal of neurophysiology, 101, 1575-1587.

Bodnoff, R. (1995). Enduring effects of chronic corticosterone treatment on spatial learning, synaptic plasticity, and hippocampal neuropathology in young and mid-aged rats.

Butte, J. C., Kakihana, R., Noble, E. P. (1976). Circadian rhythm of corticosterone levels in rat brain.

Journal of endocrinology, ;68(02):235-9.

Buzsáki, G. (1996). The hippocampo-neocortical dialogue. Cerebral cortex, 6, 81-92.

Cai, W. H., Blundell, J., Han J., Greene, R. W., Powell, C. M. (2006). Postreactivation glucocorticoids impair recall of established fear memory. Journal of neuroscience, 26(37), 9560-9566.

Dias Ferreira, E., Sousa, J. C., Melo, I., Morgado, P., Mesquita, A. R., Cerquiera, J. J., et al. (2009). Chronic stress causes frontostriatal reorganization and affects decision making. Science, 325, 621-625.

Droste, S. K., De Grootte, L., Atkinson, H. C., Lightman, S. L., Reul, J. M. H. M., Linthorst, A. C. E. (2008). Corticosterone levels in the brain show a distinct ultradian rhythm but a delayed response to forced swim stress. Endocrinology, 149(7), 3244–3253.

Van Duuren, E., van der Plasse, G., Lankelma, J., Joosten, R. N. J. M. A., Feenstra, M. G. P., Pennartz, C. M. A. (2008). Single-cell and population coding of expected reward probability in the orbitofrontal cortex of the rat. Journal of neuroscience, 29(28), 8965–8976.

Graham, L. K., Yoon, T., Kim, J. J. (2010). Stress impairs optimal behavior in a water foraging choice task in rats. Learning and memory, 17, 1-4

Groeneweg, F. L., Karst, H., De Kloet, E. R., Joëls, M. (2011). Rapid non-genomic effects of

corticosteroids and their role in the central stress response. Journal of endocrinology, 209, 153-167. Hosokawa, T., Kato, K., Inoue, M., Mikami, A. (2007). Neurons in the macaque orbitofrontal cortex code relative preference of both rewarding and aversive outcomes. Neuroscience research, 57(3), 434-445.

Hui, G. K., Figueroa, I. R., Poytress, B. S., Roozendaal, B., McGaugh, J. L., Weinberger, N. M. (2004). Memory enhancement of classical fear conditioning by post-training injections of corticosterone in rats. Neurobiology of learning and memory, 81, 67-74.

Ji, D., Wilson, M. A. (2007). Coordinated memory replay in the visual cortex and hippocampus during sleep. Nature neuroscience, 10, 100-107.

Joëls, M., Kruger, H. J. (2007). LTP after stress, up or down. Neural plasticity, 93202.

Joëls, M., Karst, H., Krugers, H. J., Lucassen, P. J. (2007). Chronic stress: Implications for neuronal morphology, function and neurogenesis. Frontiers in neuroendocrinology, 28, 72-96.

Jones, M. W., Wilson, M. A. (2005). Theta rhythms coordinate hippocampal–prefrontal interactions in a spatial memory task. Plos biology, 3(12), 2187-2199.

Kajiwara, R., Takashima, I., Mimura, Y., Witter, M. P., Iijima, T. (2002). Amygdala input promotes spread of excitatory neural activity from perirhinal cortex to the entorhinal-hippocampal circuit.

Journal of neurophysiology, 89, 2176-2184.

Karst, H., Berger, S., Erdmann, G., Schütz, G., Joëls, M. (2010). Metaplasticity of amygdalar responses to the stress hormone corticosterone. PNAS, 107(32), 14449-14454.

Kasanen, I. H. E., Inhilä, K. J., Vainio, O. M., Kiviniemi, V. V., Hau, J., Scheinin, M., et al. (2009). The diet board: welfare impacts of a novel method of dietary restriction in laboratory rats. Laboratory

animals, 43, 215-223.

Kim, J. J., Koo, J. W., Lee, H. J., Han, J. S. (2005). Amygdalar inactivation blocks stress-induced impairments in hippocampal long-term potentiation and spatial memory. Journal of neuroscience, 25(6), 1532–1539.

Kim, J. J., Lee, H. J., Han, J. S., Packard, M. G. (2001). Amygdala Is critical for stress-induced modulation of hippocampal long-term potentiation and learning. Journal of neuroscience, 21(14), 5222–5228.

Kim, J. J., Lee, H. J., Welday, A. C., Song, E. Y., Cho, J., Sharp P. E., et al. (2007). Stress-induced alterations in hippocampal plasticity, place cells, and spatial memory. PNAS, 104(46), 18297-18302. Kowalski-Trakofler, K.M., Vaught, C., Scharf, T. (2003). Judgement and decision making under stress; an overview for emergency managers. International Journal of Emergency Management, 1(3), 278-289

Lansink, C. S., Goltstein, P. M., Lankelma, J. V., McNaughton, B. L., Pennartz, C. M. A. (2009). Hippocampus leads ventral striatum in replay of place-reward information. Plos Biology, 7(8), 1-11. Lupien, S. J., McEwen, B. S., Gunnar, M. R., Heim, C. (2009). Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nature neuroscience, 10, 434-445.

Magee, J. C. (2001). Dendritic mechanisms of phase precession in hippocampal CA1 pyramidal neurons. Journal of neurophysiology, 86:528-532.

McEwen, B. (2007). Physiology and neurobiology of stress and adaptation: central role of the brain.

Physiology Review, 87: 873–904.

McGaugh, J. L. (2000). Memory; a century of consolidation. Science, (287) 248–251

McIntyre, C. K., Hatfield, T., McGaugh, J. L. (2002). Amygdala norepinephrine levels after training predict inhibitory avoidance retention performance in rats. European Journal of Neuroscience, 16, 1223-1226.

Nathan, S. V., Griffith, Q. K., McReynolds, J. R., Hahn, E. L., Roozendaal, B. (2004). Basolateral

amygdala interacts with other brain regions in regulating glucocorticoid effects on different memory functions. Annals of the New York academy of sciences, 179-182.

O’Keefe, J., Dostrovsky, J. (1971). The hippocampus as a spatial map: preliminary evidence from unit activity in the freely-moving rat. Brain research, 34, 171-175.

Ossewaarde, L., Qin, S., van Marle, H. J. F., van Wingen, G. A., Fernandez, G., Hermans, E. J. (2010). Stress-induced reduction in reward-related prefrontal cortex function. Neuroimage, 55(1), 345-352. Packard, M. G. (2009). Anxiety, cognition, and habit: A multiple memory systems perspective. Brain

research, 1293, 121-128.

Popeli, M., Yan, Z., McEwen, B. S., Sanacora, G. (2012). The stressed synapse: the impact

of stress and glucocorticoids on glutamate transmission. Nature neuroscience reviews, 13, 22-37. Popescu, A. T., Saghyan, A. A., Nagy, F. Z., Paré , D. (2010). Facilitation of corticostriatal plasticity by the amygdala requires ca2+ induced ca2+ release in the ventral striatum. Journal of neurophysiology, 104, 1673-1680.

Porcelli, A.J., Delgado, M.R. (2009). Acute stress modulates risk taking in financial decision making.

Psychological science, 20(3), 278-283.

Roitman, J. D., Roitman, M. F. (2010). Risk-preference differentiates orbitofrontal cortex responses to freely chosen reward outcomes. European journal of neuroscience, 31, 1492-1500.

Roozendaal, B., McEwen, B. S., Chatarrji, S. (2009). Stress memory and the amygdala. Nature

neuroscience, 10, 423-433.

Roozendaal, B., Okuda, S., Van der Zee, E. A., McGaugh, J. L. (2006). Glucocorticoid enhancement of memory requires arousal-induced noradrenergic activation in the basolateral amygdala. PNAS, 103, 6741–6746.

Shackman, A. J., Maxwell, J. S., McMenamin, B. W., Greischar, L. L., Davidson, R. J. (2011). Stress potentiates early and attenuates late stages of visual processing. Journal of neuroscience, 31(3), 1156-1161.

Stickgold, R. (2005). Sleep-dependant memory consolidation. Nature, 437, 1272-1278.

Thompson, B. L., Ericson, K., Schulkin, J., Rosen, J. B. (2004). Corticosterone facilitates retention of contextually conditioned fear and increases CRH mRNA expression in the amygdale. Behavioral brain

research, 149, 209-215.

Vyas, A., Mitra, R., Shankaranarayana, B. S., Chattarji, S. (2002). Chronic stress induces contrasting patterns of dendritic remodeling in hippocampal and amygdaloid neurons. The journal of

neuroscience, 22(15):6810–6818.

Wallis, J. D. (2007). Orbitofrontal cortex and its contribution to decision-making. Annual reviews of

neuroscience, 30, 31-56.

Wikenheiser, A. M., Redish, A. D. (2011). Changes in reward contingency modulate the trial to trial variability of hippocampal place cells. Journal of neurophysiology, ahead of print.

Yehuda, R. (2006). Advances in understanding neuroendocrine alterations in PTSD and their therapeutic implication. Annals of the New York Academy of Sciences, 1071, 137–166