Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Optogenetic dissection of temporal dynamics of amygdala-striatal interplay during risk/reward decision… Bercovici, Debra Ann 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_september_bercovici_debra.pdf [ 1.54MB ]
Metadata
JSON: 24-1.0354544.json
JSON-LD: 24-1.0354544-ld.json
RDF/XML (Pretty): 24-1.0354544-rdf.xml
RDF/JSON: 24-1.0354544-rdf.json
Turtle: 24-1.0354544-turtle.txt
N-Triples: 24-1.0354544-rdf-ntriples.txt
Original Record: 24-1.0354544-source.json
Full Text
24-1.0354544-fulltext.txt
Citation
24-1.0354544.ris

Full Text

   Optogenetic dissection of temporal dynamics of amygdala-striatal interplay during risk/reward decision making   by   Debra Ann Bercovici   B.Sc., McGill University, 2015    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIRMENTS FOR THE DEGREE OF   MASTER OF ARTS   in   The Faculty of Graduate and Postdoctoral Studies   (PSYCHOLOGY)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   August 2017   © Debra Ann Bercovici, 2017   ii Abstract  Assessing costs and benefits associated with different options that vary in terms of reward magnitude and uncertainty is an adaptive behaviour which allows us to select an optimal course of action. Previous studies using reversible pharmacological inactivations have shown that the basolateral amygdala (BLA) to nucleus accumbens (NAc) pathway plays a key role in promoting choice towards larger, riskier rewards. Neural activity in the BLA and NAc show distinct, phasic changes in firing prior to action initiation and following action outcomes.  Yet, how temporally-precise patterns of activity within BLA-NAc circuitry influence choice behaviour is unclear. We assessed how optogenetic silencing of BLA projection terminals in the NAc altered action selection during probabilistic decision making. Rats that received intra-BLA infusions of an AAV encoding for the inhibitory opsin eArchT were well-trained on a probabilistic discounting task, where they chose between a smaller/certain reward and a larger/riskier reward, with the probability of obtaining the larger reward changing from 50% to 12.5% across two separate blocks of trials. During testing, discrete 4-7 second pulses of light were delivered via fiber optic ferrules into the NAc to suppress activity within BLA terminals during specific task events; during the period prior to choice or during the outcome immediately following a choice. Silencing activity of BLA inputs to the NAc prior to choice reduced selection of the more preferred option, suggesting that at this time, activity within this pathway biases choice towards more preferred rewards.  Silencing during reward omissions increased risky choice during the low-probability block, indicating that activity in this circuit after non-rewarded actions serves to modify subsequent choice behaviour. In contrast, silencing during rewarded outcomes did not reliably affect choice behaviour. Collectively these data demonstrate how patterns of activity in 	 iii BLA-NAc circuitry convey different types of information that guide optimal action-selection in situations involving reward uncertainty.   	 iv Lay Summary  Being able to assess costs and benefits associated with choices allows organisms to make decisions that are profitable in the long-term. The basolateral amygdala and the nucleus accumbens are two interconnected brain regions known to be involved in promoting optimal decision making in situations involving risk. Nonetheless, little is known about whether cellular activity between these two regions mediate choice behaviour differentially at distinct phases of the decision-making process. In well trained rats, we selectively disrupted communication from the basolateral amygdala to the nucleus accumbens during specific task events. We saw that this pathway alters choice behaviour differently during periods prior to making a choice compared to during choice outcomes. These findings will help us understand in more detail how regions within our brain communicate with each other to bias decision making and will serve as a backbone for understanding what drives us to make riskier versus safer decisions.   	 v Preface All experiments were conducted at the University of British Columbia and were planned and carried out by Debra Bercovici. In addition, Debra Bercovici completed all surgeries, behavioural training and optogenetic manipulations. Maric Tse performed the electrophysiology experiment with Debra Bercovici under the consultation of Dr. David Moorman (University of Massachusetts). Histological analysis was performed by Debra Bercovici with the help of Oren Princz-Lebel. Debra Bercovici analyzed the data and wrote this document with the editing help of Dr. Stan Floresco, supervisory author on this project. Research for this thesis was approved by the UBC Animal Care Committee, protocol number A14-0210.   	 vi Table of Contents Abstract .......................................................................................................................................... ii Lay Summary ............................................................................................................................... iv Preface ............................................................................................................................................ v Table of Contents ......................................................................................................................... vi List of Tables .............................................................................................................................. viii List of Figures ............................................................................................................................... ix Acknowledgements ....................................................................................................................... x Introduction ................................................................................................................................... 1 Neural circuitry underlying risk/reward decision making: nucleus accumbens and basolateral amygdala ..................................................................................................................................... 1 Information encoding: discrete patterns of phasic firing in BLA and NAc ................................ 5 Optogenetically inhibiting BLA terminal inputs in the NAc ...................................................... 7 Materials and Methods ................................................................................................................. 9 Subjects ....................................................................................................................................... 9 Stereotaxic surgery ...................................................................................................................... 9 Apparatus .................................................................................................................................. 10 Lever press training ................................................................................................................... 11 Probabilistic discounting training ............................................................................................. 12 Optogenetic inhibition .............................................................................................................. 13 In vivo single unit recordings .................................................................................................... 15 Histology ................................................................................................................................... 17 Data analysis ............................................................................................................................. 19 Results .......................................................................................................................................... 21 Histology ................................................................................................................................... 21 Inhibition prior to choice .......................................................................................................... 21 Inhibition following risky losses ............................................................................................... 23 Inhibition following risky wins ................................................................................................. 24 Inhibition following small/certain wins .................................................................................... 25 Inhibition during the inter-trial interval .................................................................................... 26 Neurophysiological confirmation of BLA-NAc inhibition ....................................................... 27 	 vii Discussion .................................................................................................................................... 30 BLA-NAc activity promotes choice preference during periods prior to choice ....................... 30 BLA-NAc activity after risky losses biases choice away from uncertain options .................... 32 Input from BLA to NAc not influential in biasing choice during rewarded outcomes ............ 35 Control measures and limitations .............................................................................................. 36 Summary and Conclusions ....................................................................................................... 38 References .................................................................................................................................... 39    	 viii List of Tables Table 1. Choice behaviour following optogenetic inhibition of BLA-NAc during discrete periods on probabilistic discounting task.. ........................................................................................ 27   	 ix List of Figures Figure 1.  Histology: acceptable placements of optic ferrules and terminal viral expression within the NAc.. ........................................................................................................... 18 Figure 2.  Performance measures for BLA-NAc optogenetic inhibition prior to choice in probabilistic discounting task.. ..................................................................................... 23 Figure 3.  Performance measures for BLA-NAc optogenetic inhibition after risky losses in probabilistic discounting task.. ..................................................................................... 24 Figure 4.  Performance measures for BLA-NAc optogenetic inhibition after rewarded trials and during inter-trial interval in probabilistic discounting task.. ........................................ 25 Figure 5.  Neurophysiological confirmation of optogenetic inhibition of BLA-NAc.. ............... 29   	 x Acknowledgements  First and foremost, I’d like to thank my supervisor, Dr. Stan Floresco, for his unwavering mentorship and patience. Also, to my fellow lab mates for their incredible friendship and support, and especially Maric, for his help with the recording experiment. A special thank-you to Shaina for helping me perfect my techniques and letting me borrow equipment. Finally, thank-you to my committee members, Dr. Catharine Winstanley and Dr. Luke Clark for their invaluable time and feedback.    1 Introduction 	Reward-related decision making often requires the assessment of costs and benefits associated with different choices. This adaptive behaviour allows us to adopt optimal reward-seeking strategies resulting in outcomes of higher value. This complex action-selection process emerges when a more valuable reward option is associated with some cost. This cost can take the form of a delay in reward delivery, an increase in the effort required to obtain the reward, or, as will be discussed in this thesis, a risk of whether the reward will be delivered or not. Uncertain reward delivery is a situation in which choice of the larger, more desirable reward risks that it may not be delivered, whereas choice of the less desirable reward results in a certain reward delivery. Risk/uncertainty-based decision making requires considerable cognitive processing as it requires integrating information about both reward magnitude (value) and probability (risk of no receipt) (Winstanley & Floresco, 2016) in order to choose options that may be more profitable in the long-term. Neural circuitry underlying risk/reward decision making: nucleus accumbens and basolateral amygdala Studies in both humans and animals have implicated numerous nodes within cortical, limbic and striatal circuitry in mediating different aspects of risk/reward decision making. Among these regions, the nucleus accumbens (NAc) appears to be a focal point where information processed by prefrontal and limbic regions pertaining to reward value, magnitude and history are integrated to bias actions towards more preferred rewards (Floresco, 2015; Mogenson et al., 1980; Nicola, 2007).  	 2 The NAc consists of two primary subregions: the shell and the core (Zahm & Brog 1992) where excitatory glutamatergic input from the hippocampus, basolateral amygdala and prefrontal cortex converge and synapse within these subregions in a topographical manner (Brog et al., 1993). Amygdala and cortical inputs innervate both the core and shell region of the NAc though less pronounced in the medial shell, while hippocampal axons are most pronounced in the medial shell but modest in the other subregions (Britt et al., 2012). This arrangement of incoming signals is thought to create clusters of neuronal ensembles each subserving different roles as defined by their input regions (O’Donnell 1999). Additionally, these pathways interact with dopaminergic input originating from the ventral tegmental area (VTA) which can facilitate or suppress NAc activity (Charara & Grace, 2003; Nicola & Malenka, 1997).   In rodents, the probabilistic discounting task is one manner through which to assess decision making under conditions of reward uncertainty. In this task, rats choose between a small/certain lever that always delivers one sucrose reward pellet, and a large/risky lever that delivers four pellets with decreasing odds of 100, 50, 25, 12.5, and 6.25 percent across blocks of trials in a single session. Normal well-trained animals adjust their choice behaviour across these probability blocks such that they pick the large/risky lever when it is more advantageous (at 100 & 50%), and bias their response towards the small/certain lever when that option becomes more advantageous (at 12.5 & 6.25%). Previous work in our lab shows that reversible pharmacological inactivations of the NAc via infusion of GABA agonists lead to a reduction in preference for the large/risky option when the odds of winning are high (at 100 and 50%) and choosing this option has greater long-term value (Stopper & Floresco, 2011). Thus, when delivery of the more valued reward is uncertain, the NAc continues to bias choice towards larger options. Likewise, excitotoxic lesions to the NAc after training on the probabilistic discounting task initially causes 	 3 rats to be indifferent to the two choices regardless of the probability. However, with extended training, these rats eventually bias their choice towards the small/certain lever during blocks of high probability of reward delivery, despite the large/risky lever being more advantageous at this time (Cardinal & Howes, 2005). It appears from these data that, under conditions of risk, the NAc plays a general role in biasing choice towards larger, more preferred rewards and higher long-term payouts.  These findings are complemented by functional imaging studies in humans. In one experiment, subjects participating in a financial risk task chose between risky stocks, which yielded either a large gain or large loss, and a safe bond, which yielded a certain but small gain. They found that activity in the NAc preceded risky choices but not safer choices (Kuhnen & Knutson, 2005). These findings have been corroborated with numerous other studies correlating activity in the NAc to choice of riskier options that have been associated with larger rewards (Knutson et al., 2008; Matthews et al., 2004).   The basolateral amygdala (BLA) is a key input region to the Nucleus accumbens. This region also plays a critical role in mediating decisions involving different types of risks and rewards. For example, in the Iowa Gambling Task, subjects choose between four decks of cards: two “risky”/disadvantageous decks yielding large gains but also large losses, and two “safe”/advantageous decks yielding small gains and small penalties. Healthy participants adopt the optimal strategy of selecting the more conservative decks and gradually accumulate profit throughout the session. However, patients with selective damage to the amygdala have trouble identifying and maintaining optimal long-term payout strategies and instead continue to choose from the risky decks despite the losses (Bechara et al., 1999). Although human lesion and fMRI studies provide insight into which regions like the BLA are important key players in decision 	 4 making, they lack the neural specificity and experimental control necessary to make causative claims.    To address these issues, the rat gambling task has been developed as a rodent analogue to the Iowa gambling task. Rats try to maximize sugar pellet profits by choosing between safe options associated with smaller rewards and risky options offering larger rewards but long and frequent time-out punishments. During training, rats learn that over the course of a session, selecting the risky option results in less reward over time due to the punishing wait in between trials (Zeeb et al., 2009). Lesions to the BLA performed after training on this task lead to an increase in choice of the risky option (Zeeb & Winstanley, 2011). Similarly, in another rodent risky decision making task where the large/risky option delivers large rewards accompanied by the possibility of punishment in the form of a foot shock (probability of punishment increase across a session), BLA-lesioned rats also display an increase in choice of the risky option (Orsini et al., 2015). In tasks involving the risk of punishment, the BLA appears to encode aversion to negative events and biases action selection away from options associated with punishment (Winstanley & Floresco, 2016). In contrast, in the probabilistic discounting task, inactivation of the BLA decreases choice of the large/risky option (Ghods-Sharifi et al., 2009) suggesting that in tasks involving the risk of uncertain reward delivery, the BLA biases choice to ignore potential losses and seek larger rewards (Winstanley & Floresco, 2016). Taken together these data imply a larger overarching role for the BLA in assigning value to reward options (Balleine & Killcross, 2006) and promoting actions that seek to obtain subjectively more valuable rewards.            The finding that inactivation of both the NAc and BLA reduce preference for larger/riskier rewards during probabilistic discounting suggests that these two regions may form a functional circuit that biases choice towards larger, more preferred rewards. One method 	 5 commonly used to examine such a hypothesis is the asymmetric disconnection procedure. This technique allows researchers to identify whether a neural circuit is based on serial communication from one structure to the efferent pathway on the contralateral hemisphere. Inactivation of the BLA in one hemisphere and the NAc in the opposite hemisphere functionally disconnects this pathway. When performed during the probabilistic discounting task, BLA-NAc inactivation results in a decrease in choice of the large/risky option (St Onge et al., 2012) inducing a similar effect as inactivations of either region by itself and confirming that this pathway is involved in biasing choice towards more valued rewards. Findings from rodent lesion and inactivation studies provide compelling evidence that the BLA-NAc pathway plays a key role in mediating risk/reward decision making however, these techniques suppresses activity in this pathway for an entire session and does not allow researchers to look more specifically at how activity in this pathway influences choice during discrete epochs.        Information encoding: discrete patterns of phasic firing in BLA and NAc  Neurophysiological studies have provided additional insight into how patterns of neural activity corresponding to different task events may be associated with different aspects of risk/reward decision making.  Sugam and colleagues (2014) trained rats on a similar risky decision making task wherein rats chose between smaller/certain (one pellet delivered 100% of the time) and larger/risky (two pellets delivered 50% of the time) rewards.  The authors observed numerous neural correlates associated with reward delivery, but more importantly, observed that certain populations of NAc neurons displayed differential patterns of firing that correlated with choice preference or subjective value.  Specifically, during trials in which both levers were extended and rats were given a choice between the two rewards, a subpopulation of neurons exhibited brief (~2 second) phasic increases in firing during the period prior to choice.  However, this 	 6 phasic activity was greater when they chose their more preferred reward compared to when they chose their less preferred reward.  Furthermore, another population of NAc neurons displayed phasic changes that encoded different action outcomes. In particular, some NAc neurons displayed a robust increase in firing after a non-rewarded risky choice, but not after rewarded ones.  These findings suggest that brief periods of firing of NAc neurons differentially encode choices that are to be made during risk/reward decision making, and the outcomes of those actions.  Although there have been relatively few studies that have investigated neural correlates of decision making in the BLA, it is well established that activity within this nucleus encodes information about reward availability.  For example, on a discriminative stimulus task, onset of one discriminative stimulus predicts that a lever press will deliver reward, whereas presentation of another, non-rewarded stimulus informs the animal that lever presses will not be reinforced.  Recordings from the BLA in rats performing this task revealed that a subset of neurons show an increase in phasic activity to the discriminative stimulus in comparison to the presentation of a non-rewarding stimulus (Ambroggi et al., 2008). This study also found that these phasic responses in the BLA preceded similar patterns of activity in the NAc suggesting further that activity in the amygdala region may mediate activity downstream.  The findings of these neurophysiological studies, in combination with previous lesions and inactivation studies indicate that, during risk/reward decision making, signals from the BLA are related to the NAc to guide choice behaviour. However, a number of issues remain unclear. What behavioural information is encoded in these temporally discrete patterns of neural firing? Do these choice and outcome related phasic increases in NAc neural activity directly contribute to ongoing behaviour and are they driven in part by inputs from the BLA? The present study 	 7 addresses these questions by clarifying how silencing BLA projection terminals in the NAc during distinct task events alters action selection in well-trained rats performing a probabilistic discounting task. Optogenetically inhibiting BLA terminal inputs in the NAc To suppress neural activity during specific task events, we use optogenetics: the use of light sensitive microbial proteins called opsins, expressed in targeted neurons (Zhang et al., 2011) that can either excite or inhibit cells through exposure to a specific wavelength of light. DNA encoding an opsin is packaged into viruses and injected into a region of interest for neuronal infection. Cells subsequently integrate the gene into their neuronal genome, where it is expressed through an endogenous promoter the neuron can already identify (e.g. the CaMKII general excitatory promoter) (Deisseroth, 2011). This combination of genetics and optics allows scientists to achieve gain or loss function in specific cells with unparalleled temporal and spatial precision (Boyden et al., 2005) in freely moving animals (Aravanis et al., 2007).   Optogenetic approaches to manipulating neural activity have several advantages over traditional pharmacological, lesion or inactivation techniques. First, optogenetics allows for a temporal precision which is not possible using injectable drugs or lesions. Light can be turned on or off with millisecond precision in such a way that researchers can target cell manipulation for discrete task events with critical time windows (Tye & Deisseroth, 2012). This way, we are capable of inhibiting cells specifically during periods “prior to choice” or different “choice outcomes” in a modified version of the probabilistic discounting task (adapted from Stopper et al., 2014).   Secondly, the cell-specificity of optogenetics allows for intricate circuit mapping using a viral transduction strategy (Fenno, Yizhar, & Deisseroth, 2010). At the time of the initial viral 	 8 injection, expression of the opsin is limited to the cell bodies localized at the site of the injection. However, after an incubation period, the protein is trafficked downstream to the axon terminals of those cell bodies (Tye & Deisseroth, 2012) where these projections can now be the target of photo-manipulation. This allows for specific cells to be targeted as opposed to manipulating all the cells in one area as is the case in pharmacological, electrophysiological, and lesion techniques. For the purpose of this study, we used the adeno-associated virus (AAV) which travelled anterogradely from the cell bodies in the BLA where it was injected, to terminals in the NAc so we could limit our manipulations only to cells that both originate in the BLA and project to the NAc.   	 9 Materials and Methods Subjects Long Evans rats (Charles River Laboratories) weighing approximately 275-300 g upon arrival were group housed and provided food ad libitum for one week. Following daily handling and acclimatization to the colony, rats underwent stereotaxic viral infusion surgery into the BLA (see Viral infusion) and were subsequently single housed for the remainder of the experiment. Following one week of post-surgical monitoring, rats were food restricted to approximately 85% of their free-feeding weight prior to beginning behavioural training in operant chambers. Their weights were monitored daily and food was adjusted to maintain a weight gain of ~5 grams per week. Roughly 30 days after training, rats received fiber optic implantation surgery into the NAc (see Fiber optic implantation). Following recovery, rats returned to behavioural testing in operant chambers for the remainder of the experiment. All testing was conducted in accordance with the Canadian Council of Animal Care and the Animal Care Committee at the University of British Columbia.   Stereotaxic surgery Forty-eight hours prior to surgery, rats were given food ad libitum to minimize complications arising from surgery. Rats were given a subanasthetic dose of a ketamine (50 mg/kg)/ xylazine (5 mg/kg) cocktail intraperitoneally and maintained on isoflurane for the full procedure. When rats attained a surgical plane of anesthesia, they were placed into a stereotaxic frame and secured with earbars to maintain a flat skull. At this time, analgesia was administered subcutaneously (Anafen, 10 mg/kg). Over the course of this experiment, rats received two surgeries, the first 	 10 involving infusions of virus into the BLA prior to behavioural training and the second entailing implantation of fiber optics into the NAc after training. Viral infusion 1.0 µL of rAAV5-CaMKIIα-eArchT3.0-eYFP (University of North Carolina VectorCore) was infused bilaterally via micro infusion pumps into the BLA (coordinates from bregma: -3.2 mm anteroposterior; ±5.1 mm mediolateral; -7.6 mm dorsoventral from dura) at a flow rate of 0.1 µL per minute. Injectors were left in place for 10 minutes following infusion to allow for virus diffusion in tissue. Incisions were closed with sutures.     Fiber optic implantation Optic fibers consisting of 400 µm cores (Precision Fiber Products) threaded through 2.5 mm-wide metal ferrules (Precision Fiber Products) were implanted over the NAc at a 12º angle (coordinates from bregma:  +1.5 mm anteroposterior; ±1.5 mm mediolateral; -6.7 mm dorsoventral from dura). A head assembly was made to secure the fibers in place using four screws, and dental cement was then used overtop to secure the assembly leaving about half of the optic fibers uncovered. Apparatus Behavioural testing was conducted in operant chambers (30.5 x 24 x 21 cm; Med-Associates) enclosed in sound-attenuating boxes. Each box was equipped with a fan with the purpose of providing ventilation and limiting extraneous sounds. The chamber was fitted with a central food receptacle where sucrose food reward pellets (45 mg; Bioserv) were dispensed. Two retractable levers were located on either side of the food receptacle. The chamber was illuminated by a 100 mA house light located on the top center of the box opposite the food receptacle. Four infrared 	 11 photobeams located just above the grid floors tracked locomotor activity (number of beam breaks), while a fifth beam positioned at the food receptacle tracked food receptacle entries. All data was recorded by a personal computer connected to the operant chambers via an interface. Lasers were controlled by Med PC software which delivered a TTL+ pulse to lasers to initiate light delivery and a TTL- pulse to terminate it.  Lever press training The initial training protocols described below were identical to those described in previous studies conducted in our lab (e.g. St. Onge et al., 2012; Stopper et al., 2014). Following three to five days of food restriction, rats were given approximately 30 sugar pellets in their cage on the day prior to beginning operant training. On the first day of lever training, two sugar pellets were placed in the food receptacle, either the right or left lever was extended, and crushed sugar pellets sprinkled on the extended lever. Animals were trained to lever press for pellets under a fixed ratio-1 schedule until a criterion of 60 presses in 30 minutes were met for both levers. Rats were subsequently trained on a simplified version of the full task. In this task consisting of 90 trials, rats were presented with one of the levers that if pressed within 10 seconds, would deliver one sugar pellet with a 50% probability. If the lever was not pressed within this time, it was retracted and the trial was scored as an omission. Following either a lever press or omission, there came a 40 second inter-trial interval (ITI) before the next lever presentation. Animals trained until a criterion of less than 10 omissions for a minimum of two consecutive days (~4 days of training) and then progressed to the final training step (lasting ~3 days) prior to commencing the full task. This stage aimed at counterbalancing the rats to learn that the large pellet reward will be designated by either the right or the left lever. Each session was made up of 72 trials where rats were given both the choice between (free choice) and were offered one of 	 12 two levers (forced choice). Choice of the large reward lever delivered four pellets at a 50% probability of reward delivery, whereas choice of the small reward lever always delivered one pellet. Criterion was met when rats chose the large lever on more than 60% of the free choices.  Probabilistic discounting training Each daily session consisted of 60 trials separated into two blocks of 30 (10 forced followed by 20 free choice trials), and took 40 minutes to complete. Rats were trained five to seven days per week. One lever was designated the small/certain lever and the other lever was designated the large/risky lever. Lever assignment was based on random assignment during the last phase of training. Trials lasted 40 seconds and began when the house light was turned on followed by the insertion of one (forced choice) or both (free choice) of the levers. Rats were given 10 seconds to press a lever otherwise the lever(s) was/were retracted and the trial scored as an omission. Choice of the small/certain lever resulted in the delivery of one pellet 100% of the time. Choice of the large/risky lever resulted in the delivery of four sugar pellets at changing probabilities. For the first 30 trials, the probability of reward delivery was set at 50% (making it a more advantageous selection over the small/certain lever). For the remaining 30 trials, the probability of reward delivery was set at 12.5% (making it more advantageous to select the small/certain lever for these trials). Following a selection/omission, there is an ITI for the remainder of the 40 seconds, until the start of a new trial. It is important to note that rats may not experience these exact probability values as the actual probability of reward delivery varies from session to session since it is drawn from a set probability distribution. Averaged across multiple sessions, the probability of reward delivery more closely approximates the desired 50% and 12.5%.   Rats were trained until the group demonstrated stable choice behaviour (~ 30 days), evaluated by analyzing data from three consecutive days using a two-way repeated-measures 	 13 analysis of variance (ANOVA), with day and trial block as the two factors. Behaviour was deemed stable when there was no main effect of day and no day x block interaction (at p>.10). Once achieved, rats were given food ad libitum for three days before undergoing fiber optic implantation as described in the preceding section. Following recovery from surgery, rats were retrained on the task while tethered to the laser apparatus in the operant chambers to allow for habituation to the headgear. Once stable behaviour was re-established (~ 15 days), optogenetic test sessions commenced.  Optogenetic inhibition Testing occurred roughly 2 months post viral infusion. This time frame gives the opsins initially infused into the BLA enough time to traffic down to terminals in the NAc where they can be targeted for optogenetic inhibition (see Figure 1A).  Following fiber optic implantation and recovery, animals were tethered to green (532 nm) diode-pumped solid-state lasers (Laserglow Technologies) in the operant chambers for the duration of the behavioural session. The lasers were coupled to a 200 µm core (ThorLabs) followed by a dual-channel optical rotary joint (Doric Lenses) that then split the light so that each channel emitted 50% of the light intensity output. The rotary joint was attached to optic fiber patch cables (ThorLabs) that were then plugged into ferrules on the animal heads such that when the laser was on, it delivered ~30 mW of 532 nm light per split fiber.   Each optogenetic manipulation consisted of a three-day sequence.  The first two days were baseline days, where the animal was connected to the fiber optic cables, but no laser light was delivered over the session.  On the third test day of the sequence, animals received brief pulses of laser light to suppress activity of BLA terminals in the NAc during discrete task events. For each of these tests, behavioural data from each manipulation was averaged across two test 	 14 sequences and was compared against a total of four baseline days (two days prior to each test day). Animals completed all tests for a manipulation before moving on to the next type of test. The order with which testing was presented was counterbalanced across rats. Some experiments had considerable attrition rates as a result of damage to headcaps. Ultimately some rats did not receive all silencing tests, explaining why there are a differing number of subjects in each analysis. In addition to damaged headcaps, two rats did not display typical discounting behaviour (i.e. they did not favor the large/risky lever in the 50% block and the small/certain lever in the 12.5% block) and were therefore removed from data analysis.   Pre-choice silencing In one experiment, we silenced activity within the BLA-NAc pathway prior to animals making a choice. To isolate neural inhibition so that it occurred only during the choice phase of the task, lasers were turned on four seconds prior to lever extension and were turned off either when a choice was made and the levers retracted or in the case of an omission, when the levers were retracted. Lasers were on for 4-10 seconds each trial depending on response latency/omission. Silencing only took place during free choice trials.   Risky “loss” silencing In another experiment, we inhibited activity in this pathway only on trials where animals chose the risky option and did not receive a reward (“risky loss”).  Here, lasers were turned on for all free and forced choice trials when a rat had selected the large/risky lever and did not receive the larger, four pellet reward. Lasers were turned on immediately after these choices and left on for the seven seconds post-lever press.   	 15  Risky “win” silencing Another experiment inhibited activity in this pathway after rewarded risky choices (“risky wins”). On these test days, lasers were turned on for all free and forced choice trials after a rat selected the large/risky lever and received the larger reward. Lasers were left on for the seven seconds post lever press (this included the time it took for pellet delivery and consumption).  Small/certain “win” silencing This experiment inhibited activity in BLA-NAc after a small/certain choice (“small wins”). In this manipulation, lasers were turned on for all free and forced choice trials immediately after a rat selected the small/safe lever. Lasers were left on for the seven seconds post lever press, including the time it took for pellet delivery and consumption.  Inter-trial interval To ascertain that the effects of outcome-associated silencing of the BLA-NAc pathway was attributable to inhibiting neural activity that was temporally linked to these events, a control experiment tested the effects of silencing activity in this pathway for seven seconds during the 40 second ITI. Lasers were initiated randomly 6-14 seconds after the start of the ITI (i.e. 6-14 seconds following a lever press or omission) for all free and forced choice trials. In vivo single unit recordings A separate cohort of rats underwent viral infusion surgery and recovery for electrophysiology experiments. In these experiments, some rats received infusions of the virus encoding for eArchT3.0-eYFP, and others with a control virus that only encoded for the fluorescent protein, e-YFP. All rats were given food ad libitum post-viral infusion before physiological recordings (for 	 16 approximately two months) Before electrophysiological recordings, rats were anesthetized with urethane (1.5 mg/kg subcutaneously) and inserted into a stereotaxic frame. The rat’s scalp was incised, and burr holes were drilled in the skull at coordinate above the BLA and the NAc; stimulating and recording optical microelectrodes (optrodes) were lowered at the coordinates described below.  The electrophysiological cell-searching and extracellular recording procedures are adapted from Floresco et al. (2001) and Floresco & Tse (2007).  Recording microelectrodes were constructed from 2.0 mm outer diameter borosilicate glass capillary tubing (World Precision Instruments) using a vertical micropipette puller (Narishige). The tips of the electrodes were broken back against a glass rod to ~1 µm tip diameter. An optrode was constructed by coupling the microelectrode to a stripped end of a 200 µm core patch cable (ThorLabs) which was connected to a 532 nm diode-pumped solid-state laser (Laserglow Technologies). The signal from the glass microelectrode was amplified and filtered (500–2000 Hz) using an X-Cell3+ microelectrode amplifier (Frederick Haer Company). Action potential data were acquired, discriminated from noise, stored, and analyzed using Spike 2 software (Cambridge Electronics Design) running on an Intel-based personal computer with a data acquisition board interface (micro 1401 mk II; Cambridge Electronics Design). A stimulating electrode connected to an Iso-Flex optically-isolated stimulator (AMPI) that received programmed pulses from a Master-8 pulse generator (AMPI) was lowered into the BLA (coordinates from bregma: -3.0 mm anteroposterior; ±5.1 mm mediolateral; -7.2 mm dorsoventral from dura). Afterwards, a cell-searching procedure began, wherein the optrode was lowered into the NAc with a hydraulic microdrive (coordinates from bregma: +1.4 mm anteroposterior; ±1.2 mm mediolateral; ranges between 5.00 – 8.00 mm dorsoventral from dura 	 17 based on cell search), while single pulse electrical stimulation was delivered to the BLA every eight seconds.  Searching continued until a NAc neuron that exhibited a reliable, monosynaptic action potential in response to BLA stimulation was isolated. Upon isolation of such a cell, stimulation currents were titrated to evoke a baseline firing probability of ~50% (for eYFP group, average current: 1290 ± 297 µA, current range: 300-2000 µA; for eArchT3.0 group, average current: 1217 ± 159 µA, current range: 400-2000 µA). Once a stable baseline was established (40 sweeps over approximately five minutes), a second set of stimulations were administered. Here, lasers were turned on for a duration of four seconds (5 or 10 mW, starting two seconds prior to electrical stimulation of the BLA (40 sweeps). This was to test whether local NAc light application could suppress BLA-evoked firing of medium spiny neurons. Immediately after test sweeps, another set of 40 sweeps were tested without laser delivery, to assess recovery of evoked neural firing. We typically obtained between one to three cells per rat tested. At the end of data collection, rats were euthanized via transcardial perfusion and brains were obtained for histological analysis. Histology Rats were euthanized via transcardial perfusion with 4% paraformaldehyde. Brains were fixed in 4% paraformaldehyde for 24 hours and then stored in one molar phosphate buffered solution (PBS) with sodium azide. Each brain was sliced in 50 µm sections using a vibratome (Leica). Sections were treated with citric acid at 95º C for 10 minutes, incubated for three days in PBS with 10% Triton-X, 3% horse serum and chicken anti-GFP (1:500; GFP1020, Aves Labs). Visualization was performed with anti-chicken488 secondary antibody (Jackson Laboratory) diluted 1:250 in PBS with 10% Triton-X, 3% horse serum for 60 minutes at room temperature. Sections were mounted onto slides, counter-stained and coverslipped using Fluoromount-G with 	 18 DAPI (eBioscience). Placements of fiber implants were localized on a confocal microscope (Leica SP8) using a 40X oil-immersion lens (Figure 1C). Viral expression was verified in the BLA using a 10X objective and terminal expression in the NAc was localized using a 63X objective (Figure 1B). Results from rats whose placements were found to be outside the borders of the nucleus accumbens and basolateral amygdala were determined referencing a neuroanatomical atlas (Paxinos and Watson, 2005) and subsequently removed from data analysis. Figure 1.  Coronal sections of the rat brain displaying experimental design and histology for the probabilistic discounting experiment. A, Optogenetic experimental design. Virus infusions occurred in the BLA, opsin trafficked down to terminals in the NAc. Location of photo-inhibition through fiber optic implants in the NAc. B, Representative slice magnified at 63X showing GFP expression (green) in BLA terminals in the NAc core. Image is counterstained with the nuclear dye DAPI (blue). C, Acceptable fiber optic placements in the NAc (n=16). 	     LV+2.04 +1.80 +1.56 +1.20 Fiber	optic	in	NAc A B C 	 19 Data analysis Probabilistic Discounting For each behavioural manipulation, the primary dependent variable was the proportion of choice of the large/risky option, controlling for any trial omissions. This was calculated by dividing the number of large/risky choices made in a given session by the total number of choices made in that session. These data were analyzed with treatment (optogenetic inhibition) and probability block as the factors in a two-way within-subjects ANOVA. For these experiments, the main effect of probability block is always significant (p<.05) and will not be mentioned further as this demonstrates that rats learn to choose the large reward lever when the odds of winning are high and discount their choice of this lever when the odds of winning are low. Graphs of these data were plotted with percent choice of the risky option on the y-axis as a function of blocks of 10 free choice trials, with two sets of 10 trials per probability block.     If there was a significant effect of optogenetic inhibition on choice behaviour, further choice-by-choice analyses were performed to evaluate whether changes in behaviour could be explained by changes in reward sensitivity (win-stay behaviour) and negative feedback sensitivity (lose-shift behaviour).  The analyses compared each free choice to the outcome of the previous free choice of the risky lever. A win-stay ratio was calculated based on the proportion of times rats chose the risky lever following previous receipt of the large reward (a risky win) over the total number of large rewards obtained. Win-stay behaviour is an index of reward sensitivity. The lose-shift ratio was calculated based on the proportion of times rats chose the safe lever following previous non-rewarded choice (a risky loss) over the total number of non-rewarded choices. Lose-shift behaviour is an index of negative feedback sensitivity. Both these 	 20 values were analyzed together in a two-way ANOVA with either win-stay or lose-shift and treatment as the two within-subject factors.  Additionally, trial omissions, response latencies, and locomotion were analyzed with one-way repeated measures ANOVAs. In vivo single-unit recordings Evoked spike probabilities were calculated by dividing the number of action potentials observed by the number of stimuli administered. Changes in spike probabilities were used as an index of the effect of neural inhibition on the magnitude of change in NAc neuronal activity produced by subsequent BLA stimulation. This was analyzed with one-way ANOVAs with phase (baseline vs laser application) as the within-subjects factor. Only cells that recover during the post-laser phase were included in the analysis.   	 21 Results Histology Confocal confirmation of eYFP fluorescent protein amplified with GFP in the BLA revealed robust GFP expression in cell bodies encompassing the entire anterior – posterior range. Fiber optic placements ranged from +1.20 to +2.04 anterior to posterior based on bregma and were largely found around the border region between the accumbens core and shell subregions, while some fibers were localized primarily in the core or shell (Figure 1B). Terminal GFP expression in the NAc was scattered within both the core and shell subregions and visual inspection shows terminals interacting with NAc neurons (Figure 1C). One rat upon histological analysis, revealed only ipsilateral viral expression possibly due to an error during viral infusion surgery. We chose to include data from this subject in our analyses regardless as we have previously shown that ipsilateral effects exist within the BLA-NAc pathway in risky choice behaviours (St Onge et al., 2012) and statistical results were not altered significantly if this subject was to be removed from this study.      Inhibition prior to choice Data were obtained from 16 rats with acceptable optic fiber placements in the NAc in which the BLA-NAc pathway was optogenetically inhibited for four seconds prior to lever extension and lever press during all free choice trials. Choice behaviour was averaged across two test days and compared to an average baseline encompassing four days. Analysis of the choice data revealed a significant treatment x block interaction (F(1,15)=11.44, p<.01, Figure 2A) but no main effect of treatment (F(1,15)=0.03, p>.10).  Simple main effects analyses revealed that this interaction was driven by a significant (p<.05) decrease in risky choice during the 50% block on silencing test days, where rats showed a strong preference for the risky option under baseline conditions.  In 	 22 contrast, in the 12.5% block, rats displayed an increase in risky choice on silencing test days compared to baseline, where they normally showed a preference for the small/certain option.   To further dissect whether these changes in risky choice were mediated by altered sensitivity to the recent rewarded or non-rewarded choices, we analyzed each choice as a function of the outcome of the previous trial (Figure 2B). This analysis revealed no main effect of treatment (F(1,15)=0.06, p>.10) and no treatment x response type (win-stay/lose-shift) interaction (F(1,15)=0.85, p>.05). Therefore, changes in negative/reward sensitivity feedback was not a mediating factor in this overall change in risky choice. With respect to other performance measures, one-way ANOVAs revealed a small, but significant increase in trial omissions (F(1,15)=4.86, p=.044) but no significant change in locomotion (F(1,15)=3.62, p=.076) and response latency (F(1,15)=1.91, p>.10; see Table 1).  Taken together, these data indicate that optogenetic silencing of BLA inputs to the NAc prior to choices attenuated selection of more preferred options, in that they chose less risky when they normally preferred the risky option and chose more risky when the small/certain option had greater utility. 	         	 23 Figure 2. Inhibition of BLA-NAc pathway prior to choice mediates choice behaviour in the probabilistic discounting task. A, Percentage choice of the large/risky option when BLA-NAc inhibition occurs prior to choice on all free-choice trials. Inhibition reduced choice preference (n=16). B, Win-stay/lose-shift ratios. No reliable change in feedback sensitivity when BLA-NAc inhibition occurred prior to choice. Error bars are SEM, * denotes p<.05 compared to baseline choice behaviour.   	Inhibition following risky losses The BLA-NAc pathway was optogenetically inhibited in 15 rats with acceptable fiber optic placements for seven seconds during a reward omission following selection of the large/risky lever. Analysis of the choice data from this experiment did not yield a significant main effect of treatment (F(1,14)=2.29, p>.10), but did reveal a significant treatment x block interaction (F(1,14)=5.87, p=.03, Figure 3A). Simple main effects analyses revealed that this effect was driven by an increase in choice of the large/risky lever during the low (12.5%) probability block compared to normal (F(1,14)=11.00, p<.01), whereas there was no change in choice behaviour during the 50% probability block (p>.50). Choice-by-choice analyses showed that this increase in risky choice was not driven by changes in win-stay or lose-shift behavior (main effect of treatment (F(1,13)=0.02, p>.50); treatment x response type interaction (F(1,13)=0.149, p>.50, 	 24 Figure 3B)). There were no significant effects of risky loss inhibition on locomotion, trial omission or response latency (all F-values<4.60, all p-values>.10; see Table 1). These findings suggest that optogenetic inhibition of BLA-NAc after non-rewarded actions increased bias for the risky choice when it was more advantageous to select the small/certain option.    Figure 3 Inhibition of BLA-NAc pathway after risky losses mediates subsequent choice behaviour in the probabilistic discounting task A, Inhibition of BLA-NAc pathway during risky losses for free choice trials shows an increase in percent choice of the large/risky option during the low probability block (n=15). B, Win-stay/lose-shift ratios. No reliable change in feedback sensitivity when BLA-NAc inhibition occurred during risky losses. Error bars are SEM, * denotes p<.05 compared to baseline choice behaviour.  	Inhibition following risky wins In 14 rats with acceptable fiber optic placements, the BLA-NAc pathway was optogenetically inhibited on two days during free choice trials. Inhibition lasted seven seconds from the time when rats chose the large/risky lever and received the large reward. As opposed to the effects of silencing after risky losses, inhibition following risky wins revealed no main effect of treatment (F(1,13)=1.66, p>.10) and no treatment x block interaction (F(1,13)=0.24, p>.50, Figure 4A). Furthermore, there were no significant effects of inhibition after risky wins on locomotion, trial 	 25 omission or response latency (all F-values<4.67, all p-values>.10; see Table 1). Thus, silencing neural activity from BLA inputs to the NAc after a risky loss shows no change in subsequent choice behaviour.   	Figure 4. Inhibition of BLA-NAc pathway after rewarded outcomes and during the ITI in the probabilistic discounting task. A, B, Inhibition of BLA-NAc pathway during risky wins/small wins for free choice trials shows no reliable change in percent choice of the large/risky option (n=14 for both). C, Inhibition of BLA-NAc pathway during ITI for all trials shows no reliable change in percent choice of the large/risky option (n=12).  Error bars are SEM, * denotes p<.05 compared to baseline choice behaviour.    	Inhibition following small/certain wins In accordance to the effects of silencing after risky wins, optogenetically inhibiting the BLA-NAc pathway for seven seconds after rats chose the small/safe lever and received the reward also did not affect choice.  Analysis of the choice data revealed no main effect of treatment (F(1,13)=0.52, p>.10) and no treatment x block interaction (F(1,13)=0.02, p>.50, Figure 4B) in 14 rats. As well, no significant changes in response measures were apparent due to inhibition following small wins (all F-values<4.67, all p-values>.10; see Table 1). Ergo, inhibiting this pathway proceeding a small win does not bias future action selection. 	 26 Inhibition during the inter-trial interval To establish the temporal specificity of outcome related effects associated with risky losses, the BLA-NAc pathway was optogenetically inhibited for seven seconds starting at a randomized time point within each ITI in 12 rats. Upon visual inspection of Figure 4C, there appears to be a small increase in overall risky choice. However, looking at individual choice data revealed that this effect was driven primarily by 3 rats that showed an increase in risky choice. All other subjects either displayed no change in behaviour or a decrease in risky choice. As such, the overall analysis of these data failed to yield a reliable significant effect (main effect of treatment (F(1,11)=3.24, p>.10); treatment x block interaction (F(1,11)=0.84, p>.05)). Furthermore, there were no significant effects on locomotion, trial omission or response latency (all F-values<4.84, all p-values>.05; see Table 1).             	 27   Mean (SEM) Probabilistic Discounting Baseline BLA-NAc inhibition Prior to choice          Response Latency (s) 0.50 (0.03) 0.57 (0.05)      Trial Omissions 0.41 (0.12) 0.73 (0.16)*      Locomotion 1371 (83) 1522 (105) Risky loss          Response Latency (s) 0.46 (0.04) 0.50 (0.03)      Trial Omissions 0.22 (0.09) 0.23 (0.14)      Locomotion 1487 (120) 1473 (119) Risky win          Response Latency (s) 0.42 (0.05) 0.42 (0.06)      Trial Omissions 0.12 (0.05) 0.25 (0.13)      Locomotion 1503 (163) 1560 (146) Small win          Response Latency (s) 0.42 (0.04) 0.42 (0.05)      Trial Omissions 0.32 (0.11) 0.18 (0.11)      Locomotion 1531 (145) 1545 (143) Inter-trial-interval          Response Latency (s) 0.40 (0.05) 0.41 (0.07)      Trial Omissions 0.19 (0.12) 0.33 (0.14)      Locomotion 1720 (190) 1661 (180) Table 1.  Performance measures for BLA-NAc optogenetic inhibition during discrete periods of probabilistic discounting. Response latency is measured in seconds, locomotion is indexed by photobeam breaks. Values are displayed as mean (SEM). * denotes p<.05 between baseline and pathway inhibition.    Neurophysiological confirmation of BLA-NAc inhibition  In order to confirm that our optogenetic manipulations were indeed suppressing neural activity in the BLA-NAc pathway, a second experiment used in vivo single-unit recordings of NAc neurons to confirm that 532 nm light application within the NAc could attenuate firing of NAc neurons 	 28 driven by the BLA.  In one group of cells (n=4) obtained from two rats infused with virus encoding for eArchT in the BLA, application of 10 mW light pulses for 4 seconds around the time of BLA stimulation caused a 76 ± 7% reduction in BLA evoked firing compared to baseline (firing probabilities; baseline =60 ± 14%; laser = 16.9 ±7 %; F(1,3)=35.00, p<.01 Figure 5B, C).  Similarly, in another group of cells (n=6) recorded from four rats infused with eArchT, application of 5 mW light during BLA stimulation caused a 67 ±13% reduction in evoked firing of NAc neurons relative to baseline (firing probabilities; baseline =60 ±14%; laser = 16.9 ±7 %; F(1,5)=9.92,  p<.05, Figure 5B, C).  In contrast, data obtained from cells recorded from two rats infused with virus that only encoded for the fluorescent protein eYFP, but not the inhibitory opsin (n=5), application of light around the time of BLA stimulation did not reliably alter evoked firing (firing probabilities; baseline =55 ±13%; laser = 55 ±7 %; F(1,4)=0.002,  p>.90, Figure 5B, C).  These data confirm that application of 532 nm light in the NAc can suppress firing of NAc neurons driven by inputs from the BLA, but only in animals that received intra-BLA infusions of the virus promoting expression of eArchT.   	 29 	Figure 5.  Neurophysiological confirmation of BLA-NAc silencing. A, Experimental design: stimulating electrode in BLA and optrode in NAc allowing for recording of evoked firing of accumbens neurons while suppressing BLA terminal activity. B, Suppression of amygdala evoked firing in the accumbens when stimulation coincides with 10 mW pulses of 532 nm light compared to baseline and recovery firing rate. Peristimulus time histograms represent averaged evoked firing for 4 cell during baseline sweeps (left), during sweeps where 10 mW laser light was applied around the time of BLA stimulation (middle) and recovery (right). C, Group average percent change in BLA-evoked firing probability compared to baseline. eYFP compared to eArchT at 5 and 10 mW light application. 15 neurons recorded from 4 rats. * denotes p<.05 between baseline and pathway inhibition.        Time	(s)BLA	StimBaseline 10	mW	532	nm	light RecoveryA B-100-80-60-40-20020eArchT10	 mW5eYPF%	Baseline	BLA-evokedfiring	probability(6) (4)(5)C* *	 30 Discussion The objective of these studies was to investigate how increases in activity of BLA projection/inputs to the NAc during different task events contribute to decision making in situations involving reward uncertainty. In addition, we aimed to clarify what behavioural information is encoded in the temporally discrete patterns of neural firing within the BLA-NAc circuitry. By optogenetically inhibiting BLA terminal inputs in the NAc, we established that temporal patterns of activity between these nodes carry differing reward-related information depending on task phase, advocating for optimizing choice outcomes. We found that prior to choice, activity in this circuitry promotes actions directed towards more preferred rewards whereas immediately following a reward omission, activity in this pathway informs about non-rewarded actions to modify decision biases. Per contra, proceeding choices resulting in reward, BLA-NAc does not play an influential role in biasing subsequent choice behaviour. BLA-NAc activity promotes choice preference during periods prior to choice In one of the key experiments in this study, we inhibited the BLA-NAc pathway during each free choice trial at the time when rats are in the process of making a choice. We found a significant interaction effect between neural inhibition and probability block. Compared to baseline, there was a shift in choice behaviour away from their preferred option. At a delivery probability of 50%, when rats under baseline condition normally preferred the large/risky reward option, inhibition resulted in a decrease in choice of this option. This effect was opposite in the low 12.5% probability block. Rats played more risky when they normally preferred the safe, low risk option. Choice-by-choice analysis revealed no effect of negative/reward sensitivity feedback in mediating the alteration in preference. Thus, disruption in choice preference was not related to 	 31 the outcome of the previous trial but instead a disruption in long-term strategy for maximizing reward. These results indicate that the BLA encodes preference for different rewards at different contingencies which biases actions in the NAc to seek goals with long-term payouts. In addition, there was also a modest increase in trial omissions. However, upon visual inspection, this significant performance measure effect carries little explanatory power as the change observed is very small.  The finding that suppressing activity in the BLA-NAc pathway prior to choices reduces selection of more preferred rewards complements previous neurophysiological studies demonstrating how activity of NAc neurons relates to action selection during risk/reward decision making.  Sugam et al. (2014) showed that in rats choosing between larger/uncertain and smaller/certain rewards, the period before animals made a response (i.e. before lever press) was associated with higher activity in the NAc at the presentation of a cue indicating the more preferred reward compared to NAc activity at the presentation of a cue indicating the less preferred reward. Our data suggest that during the probabilistic discounting task, preference encoding occurring in the NAc prior to choice is driven by inputs from the BLA as inhibiting the pathway at this time reduces choice preference.  A possible mechanism for how reward-related information encoded by the BLA mediates NAc firing is via the modulatory effects of dopamine. This notion is supported by findings from a cued-instrumental task showing increases in phasic dopamine (DA) release in the NAc during the presentation of a cue predicting reward delivery (Jones et al., 2010). Following BLA inactivation via GABA agonists, this phasic DA release was reduced. Authors subsequently performed fast-scan cyclic voltammetry in the NAc during a risk-based decision making task and revealed cue-evoked DA release signaled the animals preferred reward contingency, regardless 	 32 of future choice. More recent evidence isolating the choice phase of an alternate uncertainty-based decision making task showed that timed phasic stimulation of neurons containing D2 receptors in the NAc during only the decision phase caused risk-preferring rats to instantly become risk-adverse (Zalocusky et al., 2016). We propose that a similar mechanism may operate in the probabilistic discounting task. Although our lab has previously found that inactivation of D2 receptors in the NAc does not alter risky choice in this assay (Stopper et al., 2013),	it is possible that activity at D2 receptors plays a more temporally discrete role in risk/reward-related processes. As such, BLA input to the NAc in our own model of reward uncertainty could be mediating reward preferences via D2 receptor-expressing NAc neurons. Putting these together, perhaps during the choice phase of probabilistic discounting, D2 receptor-expressing NAc neurons are the targets of BLA efferents in order to pass on reward preference biases at a given reward delivery contingency.   Another mechanism through which DA may modulate activity in this circuitry is through activation of D1 receptors. Floresco et al. (2001) showed that D1 receptor activation in the NAc facilitates NAc neuron firing in response to excitatory input from the BLA. Thus, in light of the present findings suggesting that the BLA inputs to the NAc promote choices directed towards more preferred rewards, it is plausible that these signals may be strengthened by D1 receptor activity.   BLA-NAc activity after risky losses biases choice away from uncertain options  Our data show that inhibiting BLA-NAc immediately after rats selected the risky option and did not receive a reward increased choice of the risky option during the low reward probability (12.5%) block. On average, rats increased percent choice of the large/risky option by ~10%. In comparison, when we optogenetically inhibited the BLA-NAc pathway for seven 	 33 seconds starting at a randomized time point within each ITI, we found no reliable change in choice behaviour compared to baseline. These findings suggest that the effects of silencing after non-rewarded choices were contingent on the temporal specificity of BLA activity following a risky loss. Moreover, this effect was found not to be mediated by immediate reward/negative sensitivity feedback as determined by trial-by-trial choice analysis. Rather, it appears that this pathway accumulates information about losses over the course of multiple trials in order to modify subsequent actions. By inhibiting neural activity during these reward omissions, rats behaved as though they were less affected by the increasing number of losses as the contingencies switched and the odds of winning decreased. As such, these data indicate that in addition to biasing choices towards more preferred options, activity in this pathway also encodes non-rewarded actions, which in turn may shape bias away from larger/uncertain rewards when actions are rewarded less frequently.  Knowing that a subset of neurons in the NAc show an increase in phasic activity following a non-rewarded choice (Sugam et al., 2014), we tested the hypothesis that this response is mediated by reward-related information from the BLA. Classic views of amygdala function implicate this region in responding to aversive stimuli such as punishment or even losses/reward omissions (Balleine & Killcross, 2006; Winstanley & Floresco, 2016). The degree to which animals perceive reward omissions as being aversive may influence the degree to which they attempt to avoid these losses (Sugam, et al., 2014). More specifically in decision making, the role of the amygdala has been defined as assigning value to a reward option, and using accumulated information about the absence of reward from multiple trials to guide subsequent behaviour (Averbeck & Costa, 2017; Cardinal et al., 2002). By disrupting the ability of the BLA to inform the NAc that previous choices were not rewarded, the potential for a large reward may 	 34 have become more salient than the negative feedback from losses collected across previous trials (Bechara, 1999). Rats are trained to know that choice of the uncertain option always comes with the possibility of no reward delivery. Therefore, during the block with high probability of reward delivery, rats learn that despite some losses, it is more advantageous to seek the larger reward. However, during the block with low probability of reward delivery/high risk, more frequent aversive losses over time are what informs them that these larger rewards are no longer worth the risk.  It is important to note that our findings contrast with the effects of disruptions of activity in the medial prefrontal cortex during probabilistic discounting. In these instances, when the odds start out high (at 100%) and decrease across blocks of trials, inactivations of the medial prefrontal cortex result in rats continuing to favor the larger/riskier option as they are unable to track the changes in contingencies and therefore unable to update their reward-seeking strategies. Inactivating the medial prefrontal cortex causes rats to maintain their initial bias (St Onge & Floresco, 2010). Moreover, this effect appears to be driven by interactions with the prefrontal cortex and the BLA, as disconnection of these two regions also increases risky choice and reduces lose-shift behavior.  In contrast, we propose that inhibiting BLA inputs to the NAc following a risky loss appears to blunt the negative effects of the loss, but that this effect is not apparent immediately after a choice. Instead, over the course of multiple trials, rats behave as though they are less impacted by the accumulation of losses when inputs to the NAc from the BLA are suppressed over time. As such, the subjective value of the larger/riskier option was higher because it was no longer associated with risk. Consolidating these theories, as losses begin to accumulate across trial blocks, our model suggests that the BLA is responding to these 	 35 more frequent aversive outcomes and is responsible for informing the NAc following these non-rewarded actions. These signals provide a longer-term modulation of future decision-biases.  Input from BLA to NAc not influential in biasing choice during rewarded outcomes  In contrast to the effects of silencing after risky losses, optogenetic inhibition of BLA-NAc activity when rats selected the large/risky lever and won the reward or when rats selected the small/certain lever and received the reward resulted in no change in overall choice behaviour compared to baseline. This null effect was particularly interesting considering that neurons in both the BLA and NAc do show an increase in firing after rewarded actions (Belova et al., 2008; Sugam et al., 2014). Thus, it appears even though increased input to the NAc from the BLA may occur after receipts of rewards, this outcome-related activity does not seem to play a major role in biasing subsequent choice behaviour.  In this regard, other inputs in the mesocorticolimbic circuit may play a larger role during these task events.  For example, the lateral habenula (LHb) exerts inhibitory control over downstream DA neuron activity. Neurons within this region encode a negative reward prediction error meaning that they are inhibited by stimuli that predict the presence of reward (Hikosaka, 2010; Matsumoto & Hikosaka, 2007). Conversely, dopaminergic neurons in the VTA encode a positive reward prediction error meaning that phasic bursts of activity occur in response to reward-predictive stimuli (Schultz et al., 1997). In turn, DA released by the VTA can reach the NAc where it can modulate neuronal activity and affect behavior. Stimulation of the LHb-VTA indirect pathway is one manner in which researchers may inhibit phasic DA neuron activity to ascertain how reward-related signals may influence behavior (Stopper & Floresco, 2014). Data from our lab has previously shown that stimulating the LHb during risky wins in the probabilistic discounting task led to an overall reduction in risky choice. Animals behaved as if they were not 	 36 receiving the larger reward (Stopper et al., 2014). Alternatively, stimulating the LHb after a safe win increased overall risky choice, making it similar to behaviour seen when the small reward is omitted during probe trials. These data suggest that short-term outcome-related phasic bursts or dips in DA provide information about recent rewarded actions in order to guide subsequent choices. Thus, during this form of decision making, phasic increases in dopaminergic inputs to the NAc may inform this system about rewarded actions, whereas glutamatergic input from the amygdala may play a complementary role and inform about non-rewarded choices.  These two signals may work in concert to further refine decision biases when reward probabilities are volatile.   A second theory for explaining these null findings could be that during the consumatory phase of a rewarded action, some neurons in the NAc display a suppression of activity (Nicola et al., 2004; Krause et al., 2010; Taha & Fields, 2005). It is possible that since there is no NAc activity during reward receipt, excitatory BLA inputs to this region alone are not impactful enough to incite accumbens neurons to fire.   Control measures and limitations In order to confirm that our optogenetic manipulation was effective at suppressing activity in the BLA-NAc pathway, we recorded evoked firing of accumbens neurons while suppressing BLA terminal activity. We showed that at baseline, NAc neurons displayed a reliable increase in firing evoked by BLA stimulation. Moreover, when stimulation coincided with four second pulses of 532 nm light, we saw a marked suppression of amygdala evoked firing in accumbens neurons. Similar results were seen for cells receiving 5 mW and 10 mW of light intensity and in addition, these effects were not exhibited in animals infused with the control rAAV5-CaMKIIα- eYFP virus which did not contain the inhibitory opsin eArchT. Although this manipulation did not 	 37 completely suppress all evoked action potentials, we showed a significant and reliable suppression in neural firing. These findings support our claim that our behavioural data resulted from a reduction of phasic firing within the BLA-NAc pathway.   It bears mentioning that the magnitude of changes in choice behavior caused by our optogenetic manipulations were comparatively smaller compared to those observed using more traditional inactivation approaches (Ghods-Sharifi et al., 2009; St Onge et al., 2012; Stopper & Floresco, 2011). Although we did observe significant alterations in choice biases, typical discounting patterns endure (i.e. rats continue to select the larger reward lever above chance level in the 50% block and below chance level in the 12.5% block). While shifts in behaviour appear modest, emphasis remains on the fact that these changes in behaviour are reliable across subjects. Instead, we attribute these outcomes to the strength of our manipulation. As stated previously, our electrophysiological data show that optogenetic inhibition of BLA terminals did not eliminate activity completely, therefore, there remains the possibility that some communication between the BLA and NAc persists.   It is also important to note that inputs from other regions within the mesocorticolimbic circuitry converge in the NAc to guide decision making (Floresco, 2015; St. Onge et al., 2012; Stopper et al., 2014; Jenni et al., 2017). In this regard, temporally inhibiting BLA inputs would not be expected to disrupt inputs from other regions that may communicate with the NAc to guide decision making, such as the prefrontal cortex and ventral tegmental area. Since all these regions also encode decision making information, it is reasonable to assume that the NAc may still receive reward-related information from these other regions to promote somewhat optimal goal-seeking strategies in this case.       	 38  One key control experiment that is required to strengthen our results will involve conducting optogenetic manipulations on the probabilistic discounting task for animals infused with virus not containing the inhibitory opsin. With these data, we will be able to conclusively state that our choice behaviour effects were due to changes in neural firing, and not due to other unlikely physiological side-effects from viral infusion and light application.  Summary and Conclusions To summarize, the present findings provide novel insight into how temporally-precise patterns of activity of BLA inputs to the NAc during distinct phases of decision making influences goal-directed responding (Cardinal et al., 2002; Nicola, 2007). Phasic activity in this pathway prior to choice biases choice towards more preferred options, nudging behavior towards larger/uncertain or smaller/certain rewards, depending on which option has greater utility. Activity following reward omissions informs about the accumulation of non-rewarded actions in order to modify future decisions, and bias choice away from larger rewards when the probability of obtaining them are low. In contrast, activity within this circuitry following a rewarded action does not appear to have a major influence on subsequent choice behaviour suggesting that other inputs to the NAc may be more influential for biasing action-selection at these moments in time. Collectively, these data suggest that temporally-discrete patterns of activity in BLA–NAc circuitry mediates different aspects of decisions involving uncertainty. Results from these experiments help shed light into how information flows within the brain to guide action selection, as well as further our understanding of the neural network dynamics which drive us to make riskier versus safer decisions.  	 39 References Ambroggi, F., Ishikawa, A., Fields, H. L., & Nicola, S. M. (2008). Basolateral amygdala neurons facilitate reward-seeking behavior by exciting nucleus accumbens neurons. Neuron, 59(4), 648-661. Aravanis, A. M., Wang, L. P., Zhang, F., Meltzer, L. A., Mogri, M. Z., Schneider, M. B., & Deisseroth, K. (2007). An optical neural interface: in vivo control of rodent motor cortex with integrated fiberoptic and optogenetic technology. Journal of neural engineering, 4(3), S143.		Averbeck, B. B., & Costa, V. D. (2017). Motivational neural circuits underlying reinforcement learning. Nature Neuroscience, 20(4), 505-512. Balleine, B. W., & Killcross, S. (2006). Parallel incentive processing: an integrated view of amygdala function. Trends in neurosciences, 29(5), 272-279. Bechara, A., Damasio, H., Damasio, A. R., & Lee, G. P. (1999). Different contributions of the human amygdala and ventromedial prefrontal cortex to decision making. Journal of Neuroscience, 19(13), 5473-5481. Belova, M. A., Paton, J. J., & Salzman, C. D. (2008). Moment-to-moment tracking of state value in the amygdala. Journal of Neuroscience, 28(40), 10023-10030. Boyden, E. S., Zhang, F., Bamberg, E., Nagel, G., & Deisseroth, K. (2005). Millisecond-timescale, genetically targeted optical control of neural activity. Nature neuroscience, 8(9), 1263-1268. 	 40 Britt, J. P., Benaliouad, F., McDevitt, R. A., Stuber, G. D., Wise, R. A., & Bonci, A. (2012). Synaptic and behavioral profile of multiple glutamatergic inputs to the nucleus accumbens. Neuron, 76(4), 790-803. Brog, J. S., Salyapongse, A., Deutch, A. Y., & Zahm, D. S. (1993). The patterns of afferent innervation of the core and shell in the “Accumbens” part of the rat ventral striatum: Immunohistochemical detection of retrogradely transported fluoro-gold. Journal of Comparative Neurology, 338(2), 255-278. Cardinal, R. N., & Howes, N. J. (2005). Effects of lesions of the nucleus accumbens core on choice between small certain rewards and large uncertain rewards in rats. BMC neuroscience, 6(1), 37. Cardinal, R. N., Parkinson, J. A., Hall, J., & Everitt, B. J. (2002). Emotion and motivation: the role of the amygdala, ventral striatum, and prefrontal cortex. Neuroscience & Biobehavioral Reviews, 26(3), 321-352. Charara, A., & Grace, A. A. (2003). Dopamine receptor subtypes selectively modulate excitatory afferents from the hippocampus and amygdala to rat nucleus accumbens neurons. Neuropsychopharmacology, 28(8), 1412. Deisseroth, K. (2011). Optogenetics. Nature methods, 8(1), 26-29. Fenno, L., Yizhar, O., & Deisseroth, K. (2011). The development and application of optogenetics. Annual review of neuroscience, 34. Floresco, S. B. (2015). The nucleus accumbens: an interface between cognition, emotion, and action. Annual review of psychology, 66, 25-52. 	 41 Floresco, S. B., Blaha, C. D., Yang, C. R., & Phillips, A. G. (2001). Dopamine D1 and NMDA receptors mediate potentiation of basolateral amygdala-evoked firing of nucleus accumbens neurons. Journal of Neuroscience, 21(16), 6370-6376. Floresco, S. B., & Maric, T. T. (2007). Dopaminergic regulation of inhibitory and excitatory transmission in the basolateral amygdala–prefrontal cortical pathway. Journal of Neuroscience, 27(8), 2045-2057. Ghods-Sharifi, S., Onge, J. R. S., & Floresco, S. B. (2009). Fundamental contribution by the basolateral amygdala to different forms of decision making. Journal of Neuroscience, 29(16), 5251-5259. Jones, J. L., Day, J. J., Aragona, B. J., Wheeler, R. A., Wightman, R. M., & Carelli, R. M. (2010). Basolateral amygdala modulates terminal dopamine release in the nucleus accumbens and conditioned responding. Biological psychiatry, 67(8), 737-744. Knutson, B., Wimmer, G. E., Kuhnen, C. M., & Winkielman, P. (2008). Nucleus accumbens activation mediates the influence of reward cues on financial risk taking. NeuroReport, 19(5), 509-513. Krause, M., German, P. W., Taha, S. A., & Fields, H. L. (2010). A pause in nucleus accumbens neuron firing is required to initiate and maintain feeding. Journal of Neuroscience, 30(13), 4746-4756. Kuhnen, C. M., & Knutson, B. (2005). The neural basis of financial risk taking. Neuron, 47(5), 763-770. 	 42 Larkin, J. D., Jenni, N. L., & Floresco, S. B. (2016). Modulation of risk/reward decision making by dopaminergic transmission within the basolateral amygdala. Psychopharmacology, 233(1), 121-136. Matsumoto, M., & Hikosaka, O. (2007). Lateral habenula as a source of negative reward signals in dopamine neurons. Nature, 447(7148), 1111. Matthews, S. C., Simmons, A. N., Lane, S. D., & Paulus, M. P. (2004). Selective activation of the nucleus accumbens during risk-taking decision making. Neuroreport, 15(13), 2123-2127. Mogenson, G. J., Jones, D. L., & Yim, C. Y. (1980). From motivation to action: functional interface between the limbic system and the motor system. Progress in neurobiology, 14(2), 69-97. Nicola, S. M. (2007). The nucleus accumbens as part of a basal ganglia action selection circuit. Psychopharmacology, 191(3), 521-550. Nicola, S. M., & Malenka, R. C. (1997). Dopamine depresses excitatory and inhibitory synaptic transmission by distinct mechanisms in the nucleus accumbens. Journal of Neuroscience, 17(15), 5697-5710. Nicola, S. M., Yun, I. A., Wakabayashi, K. T., & Fields, H. L. (2004). Cue-evoked firing of nucleus accumbens neurons encodes motivational significance during a discriminative stimulus task. Journal of neurophysiology, 91(4), 1840-1865. O’Donnell, P. (1999). Ensemble coding in the nucleus accumbens. Psychobiology, 27(2), 187-197. 	 43 Orsini, C. A., Trotta, R. T., Bizon, J. L., & Setlow, B. (2015). Dissociable roles for the basolateral amygdala and orbitofrontal cortex in decision making under risk of punishment. Journal of Neuroscience, 35(4), 1368-1379. Paxinos G, Watson C (2005) The rat brain in sterotaxic coordinates, Edition 5. San Diego Elsevier Academic.  Schultz, W., Dayan, P., & Montague, P. R. (1997). A neural substrate of prediction and reward. Science, 275(5306), 1593-1599. St. Onge, J. R., & Floresco, S. B. (2009). Prefrontal cortical contribution to risk-based decision making. Cerebral Cortex, 20(8), 1816-1828. St Onge, J., Stopper, C. M., Zahm, D. S., & Floresco, S. B. (2012). Separate prefrontal-subcortical circuits mediate different components of risk-based decision making. Journal of Neuroscience, 32(8), 2886-2899. Stopper, C. M., & Floresco, S. B. (2011). Contributions of the nucleus accumbens and its subregions to different aspects of risk-based decision making. Cognitive, Affective, & Behavioral Neuroscience, 11(1), 97-112. Stopper, C. M., & Floresco, S. B. (2014). What's better for me? Fundamental role for lateral habenula in promoting subjective decision biases. Nature neuroscience, 17(1), 33-35. Stopper, C. M., Khayambashi, S., & Floresco, S. B. (2013). Receptor-specific modulation of risk-based decision making by nucleus accumbens dopamine. Neuropsychopharmacology, 38(5), 715. 	 44 Stopper, C. M., Maric, T. L., Montes, D. R., Wiedman, C. R., & Floresco, S. B. (2014). Overriding phasic dopamine signals redirects action selection during risk/reward decision making. Neuron, 84(1), 177-189. Sugam, J. A., Saddoris, M. P., & Carelli, R. M. (2014). Nucleus accumbens neurons track behavioral preferences and reward outcomes during risky decision making. Biological psychiatry, 75(10), 807-816. Taha, S. A., & Fields, H. L. (2005). Encoding of palatability and appetitive behaviors by distinct neuronal populations in the nucleus accumbens. Journal of Neuroscience, 25(5), 1193-1202. Tye, K. M., & Deisseroth, K. (2012). Optogenetic investigation of neural circuits underlying brain disease in animal models. Nature reviews. Neuroscience, 13(4), 251. Winstanley, C. A., & Floresco, S. B. (2016). Deciphering Decision Making: Variation in Animal Models of Effort-and Uncertainty-Based Choice Reveals Distinct Neural Circuitries Underlying Core Cognitive Processes. Journal of Neuroscience, 36(48), 12069-12079. Zahm, D. S., & Brog, J. S. (1992). On the significance of subterritories in the “accumbens” part of the rat ventral striatum. Neuroscience, 50(4), 751-767. Zalocusky, K. A., Ramakrishnan, C., Lerner, T. N., Davidson, T. J., Knutson, B., & Deisseroth, K. (2016). Nucleus accumbens D2R cells signal prior outcomes and control risky decision making. Nature, 531(7596), 642-663. Zeeb, F. D., & Winstanley, C. A. (2011). Lesions of the basolateral amygdala and orbitofrontal cortex differentially affect acquisition and performance of a rodent  	 45 Zhang, F., Vierock, J., Yizhar, O., Fenno, L. E., Tsunoda, S., Kianianmomeni, A., ... & Magnuson, J. (2011). The microbial opsin family of optogenetic tools. Cell, 147(7), 1446-1457.  	

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0354544/manifest

Comment

Related Items