Abstract
Decision making is a complex cognitive process that recruits a distributed network of brain regions, including the basolateral amygdala (BLA) and nucleus accumbens shell (NAcSh). Recent work suggests that communication between these structures, as well as activity of cells expressing dopamine (DA) D2 receptors (D2R) in the NAcSh, are necessary for some forms of decision making; however, the contributions of this circuit and cell population during decision making under risk of punishment are unknown. The current experiments addressed this question using circuit-specific and cell type-specific optogenetic approaches in rats during a decision making task involving risk of punishment. In experiment 1, Long–Evans rats received intra-BLA injections of halorhodopsin or mCherry (control) and in experiment 2, D2-Cre transgenic rats received intra-NAcSh injections of Cre-dependent halorhodopsin or mCherry. Optic fibers were implanted in the NAcSh in both experiments. Following training in the decision making task, BLA→NAcSh or D2R-expressing neurons were optogenetically inhibited during different phases of the decision process. Inhibition of the BLA→NAcSh during deliberation (the time between trial initiation and choice) increased preference for the large, risky reward (increased risk taking). Similarly, inhibition during delivery of the large, punished reward increased risk taking, but only in males. Inhibition of D2R-expressing neurons in the NAcSh during deliberation increased risk taking. In contrast, inhibition of these neurons during delivery of the small, safe reward decreased risk taking. These findings extend our knowledge of the neural dynamics of risk taking, revealing sex-dependent circuit recruitment and dissociable activity of selective cell populations during decision making.
SIGNIFICANCE STATEMENT Until recently, the ability to dissect the neural substrates of decision making involving risk of punishment (risk taking) in a circuit-specific and cell-specific manner has been limited by the tools available for use in rats. Here, we leveraged the temporal precision of optogenetics, together with transgenic rats, to probe contributions of a specific circuit and cell population to different phases of risk-based decision making. Our findings reveal basolateral amygdala (BLA)→nucleus accumbens shell (NAcSh) is involved in evaluation of punished rewards in a sex-dependent manner. Further, NAcSh D2 receptor (D2R)-expressing neurons make unique contributions to risk taking that vary across the decision making process. These findings advance our understanding of the neural principles of decision making and provide insight into how risk taking may become compromised in neuropsychiatric diseases.
Introduction
Decision making consists of multiple cognitive operations that occur in synchrony with one another to give rise to choice behavior (Fobbs and Mizumori, 2017; Orsini et al., 2019). For example, when faced with a decision, an individual must evaluate the available options and weigh their benefits against their costs (“deliberation”). Once an individual makes a choice, the outcome of their action is evaluated, and this feedback is integrated into the subsequent deliberative process. Thus, what might appear to be a homogenous process is actually multifaceted and requires significant cognitive resources. The architecture of a decision is particularly important to understand in the context of neuropsychiatric conditions associated with impaired decision making. For instance, individuals with substance use disorder display increased risk-taking behavior (Chen et al., 2020), which is thought to be related to diminished sensitivity to risks of adverse consequences (Bechara and Damasio, 2002; Lejuez et al., 2005; Myers et al., 2017). This deficit suggests that the ability to use negative feedback to guide future decisions, rather than the deliberative process, may be compromised in these individuals.
These distinct components of decision making are mediated by temporally-specific recruitment of regions within the mesocorticolimbic circuit (Orsini et al., 2019). For example, activity within the basolateral amygdala (BLA) encodes behaviorally relevant information about rewarding (Courtin et al., 2022) and aversive outcomes (Jean-Richard-Dit-Bressel et al., 2022) in a temporally dynamic manner. Our recent work has further shown that BLA activity critically contributes to decision making involving risk of punishment differentially across the decision process (Orsini et al., 2017). Whereas activity of BLA glutamatergic neurons during deliberation biases choices toward larger, riskier rewards, activity of these neurons during evaluation of punished rewards biases future choices away from larger, riskier rewards. Activity in the BLA may influence risk taking in such a dissociable manner through divergent projections to other brain regions. For instance, BLA projections to prefrontal cortical areas may be important for deliberative processes of risk-based decision making, while BLA projections to the shell subregion of the nucleus accumbens (NAcSh; BLA→NAcSh) may be necessary for evaluation of punished rewards. Support for the latter possibility comes from work showing that inactivation of either the BLA or NAcSh increases lever pressing for rewards associated with footshock (Piantadosi et al., 2017), and the disconnection of these two brain regions decreases the number of attempts to avoid a footshock in an active avoidance task (Ramirez et al., 2015). Although previous studies have shown a temporally-specific role for the BLA→NAcSh in decision making involving risk of reward omission (Bercovici et al., 2018), the temporal dynamics of this pathway in decision making involving risk of punishment remains unknown. This is an important consideration as the neurobiological mechanisms of risk-based decision making differ as a function of the type of risk involved (Orsini et al., 2015).
Within the NAcSh, dopamine (DA) neurotransmission influences decision making under risk of punishment by binding to DA receptors (Orsini et al., 2015; Winstanley and Floresco, 2016). In particular, this form of decision making appears to depend on D2 receptors (D2Rs) in the NAcSh. Levels of D2R mRNA in the NAcSh are negatively correlated with performance in a decision making task involving risk of punishment (Mitchell et al., 2014). Further, systemic or intra-NAcSh administration of D2R agonists decrease choice of large, risky rewards (Simon et al., 2011; Mitchell et al., 2014). The use of pharmacological methods, however, precludes determination of the specific components of the decision process for which these receptors, and the cells on which they reside, are necessary. Like with BLA glutamatergic neurons, the activity of NAcSh D2R-expressing neurons may subserve distinct aspects of decision making. Indeed, a recent study demonstrated that NAc D2R-expressing neurons selectively encode loss sensitivity during deliberation, but not outcome evaluation, in a decision making task involving risk of reward omission (Zalocusky et al., 2016). Hence, engagement of NAcSh D2R-expressing neurons may be similarly selective to deliberation during decision making involving risk of punishment.
The current set of experiments was designed to evaluate the roles of the BLA→NAcSh and NAcSh D2R-expressing neurons during distinct components of decision making involving risk of punishment. Collectively, the results of these experiments advance our understanding of the neural dynamics underlying decision making.
Materials and Methods
Subjects
For experiments involving manipulations of the BLA→NAcSh, Long–Evans rats were obtained from Charles River (University of Florida (UF): Kingston, NY; The University of Texas (UT) at Austin: Hollister, CA) at postnatal day (P)50 and individually housed on arrival. For experiments involving manipulations of D2R-expressing neurons in the NAcSh, male and female hemizygous D2-Cre rats (LE-Tg(Drd2-iCre)1Ottc) were procured from the Optogenetic Transgenic Technology Core (NIDA) through the Rat Research and Resource Center (RRRC#768). Hemizygous male and female rats were initially paired to begin the breeding colony, but after the colony was established, hemizygous male rats were mated with wild-type Long–Evans female rats obtained from Charles River (P90). Upon weaning (∼P21), hemizygous D2-Cre male and female rats were housed in groups of three to four (separately for each sex) until P60, at which point they were individually housed and underwent surgery. Experiments were conducted at both UF and UT Austin, but there were no differences in baseline behavioral performance or in the effects of manipulations; consequently, datasets were merged and analyzed together. All rats were maintained on a reversed light cycle (lights on at 7 P.M.), housed in ventilated cages with Sanichip bedding and had ad libitum access to water. During initial behavioral testing, rats were food-restricted to 85% of their free-feeding weight, with weights adjusted upward by 5 g per week to account for growth. Once fully grown (∼250 g for females, ∼350 g for males), rats were fed 10 g of food (soy-free; Envigo Teklad Irradiated Global 19% Protein Extruded Rodent Diet, #2919) each day after behavioral testing. Environmental enrichment in the form of nylabones was provided in each cage and replaced as needed. All procedures were conducted in accordance with the regulations and policies stipulated by the Institutional Animal Care and Use Committees at UF and UT Austin and adhered to ethical guidelines set forth by the National Institutes of Health.
Apparatus
Behavioral testing was conducted in standard operant chambers (Coulbourn Instruments; eight chambers, UF; six chambers, UT Austin), each of which was housed in a sound-attenuating cabinets. A centrally located food trough projected 3 cm out from the front wall of each chamber and was equipped with photobeams to enable detection of entries into the trough. Each food trough was connected to a feeder, from which 45 mg soy-based food pellets (UF: Test Diet, 5TUL, Richmond, IN; UT Austin: Lab Supply, 5TUL, Houston, TX) were delivered into the trough. A nosepoke was positioned directly above the food trough and two retractable levers flanked each side of the food trough. The floor of each chamber consisted of stainless-steel rods through which scrambled footshocks were delivered via the floor's connection to a shock generator (Coulbourn Instruments). A rotary joint (1 × 2, 200-µm core, Doric Lenses) was mounted on the ceiling of each operant chamber for use during optogenetic manipulations. Each chamber was interfaced with a computer running Graphic State 4.0, which controlled chamber hardware (e.g., lever extension, laser delivery) and concurrently recorded behavioral output (e.g., nosepokes, lever presses).
Laser delivery
For optogenetic inhibition of the BLA→NAcSh (experiment 1), laser light (560 nm, 12- to 16-mW output, Shanghai Laser & Optics Century) was delivered bilaterally through optic fibers (200-µm core, 0.22 NA, Precision Fiber Products) implanted in the NAcSh of rats that received intra-BLA injections of halorhodopsin (eNpHR3.0 group) or mCherry (control group). For optogenetic inhibition of D2R-expressing neurons in the NAcSh (experiment 2), laser light (560 nm, 8- to 10-mW output) was delivered bilaterally through optic fibers implanted in the NAcSh of D2-Cre rats that received intra-NAcSh injections of halorhodopsin (eNpHR3.0 group) or mCherry (control group). In all rats, light was transmitted from the laser to the rotatory joint through a patch cord (200-µm core, Thor Labs) and then from the rotary joint to the implanted optic fibers through two additional patch cords (200-µm core, 0.22 NA, Thor Labs) encased in stainless steel. The laser was interfaced with the computer running Graphic State 4.0 software to enable precise control of the timing of light delivery during specific task events.
Surgical procedures
Rats were anesthetized with isoflurane gas (1–5% in O2) and were administered subcutaneous injections of buprenorphine (0.03 mg/kg), meloxicam (2 mg/kg) and 10 ml of 0.9% saline. Upon reaching a stable plane of anesthesia, rats' hair on the superior aspect of the scalp was clipped, after which they were secured in the stereotax (Kopf Instruments). The scalp was disinfected with chlorohexidine and isopropyl alcohol swabs and ophthalmic ointment was applied to the eyes. A sterile adhesive surgical drape was then placed over the body.
Validation experiment surgeries
Before the behavioral manipulations in experiment 2, RNAscope was used to verify that viral transduction of D2R-expressing neurons was selective to these neurons. In addition, in vitro electrophysiology experiments were conducted to confirm that optogenetic inhibition of D2R-expressing neurons had the intended physiological consequences. For rats in these validation experiments, an incision was made in the scalp and the skin was retracted with hemostats. After clearing the fascia, the skull was leveled to ensure that bregma and lambda were in the same horizontal plane. Two burr holes were drilled in the skull for bilateral injections of AAV5-EF1α-DIO-eNpHR3.0-mCherry (University of North Carolina Vector Core, 4 × 10^12 vg/ml) into the NAcSh (anterior/posterior (AP): +1.7; medial/lateral (ML): ±0.9; dorsal/ventral (DV): −7.2 from skull surface). Injectors were slowly lowered to the target depth and 0.5 µl of the viral vector was infused at a rate of 0.3 µl/min. Injectors were connected to 10-µl Hamilton syringes mounted on an infusion pump (Harvard Apparatus) via PE-20 tubing. After each injection, the injectors remained in place for an additional 5 min to allow diffusion of the virus in the brain tissue. Bone wax was placed in the burr holes and the skin incision was sutured closed. Rats were administered an additional 10 ml of saline and allowed to recover from surgery on a heating pad.
Behavioral experiment surgeries
For rats in behavioral experiments, an incision was made in the scalp and the skin was retracted with hemostats. After clearing the fascia, five to six small burr holes were made in the skull for placement of jeweler's screws, with at least one positioned anterior to bregma and at least four positioned between bregma and lambda. Such a configuration was used to ensure that the cranial implant would be evenly secured across the skull. In experiment 1 (optogenetic manipulation of BLA→NAcSh), the skull was leveled and then two additional burr holes were drilled in the skull for bilateral injections of AAV5-CAMKIIα-eNpHR3.0-mCherry (University of North Carolina Vector Core; 5.8 × 10^12 vg/ml) or AAV5-CAMKIIα-mCherry (4.9 × 10^12 vg/ml) in the BLA (AP: −3.3; ML: ±4.9; DV: −8.5, −8.1 from skull surface). Injectors were slowly lowered to the target depth and the viral vector was infused at a rate of 0.3 µl/min (0.4 µl at the most ventral DV coordinate followed by 0.2 µl at the more dorsal DV coordinate). Injectors were left in place for an additional 5 min after each injection to allow diffusion of the virus through the brain tissue. After viral injections were completed, bone wax was placed into each burr hole. Two more burr holes were then created for bilateral implantation of guide cannulae (22 gauge, Plastics One) above the NAcSh (AP: +1.5; ML: ±3.1; DV: −6.4 from skull surface at a 20° angle). Once both cannulae were lowered to their target depth, a thin layer of Metabond (Parkell) was applied over the skull and jeweler's screws. After the Metabond had cured, dental cement was then used to anchor the guide cannulae in place. Sterile stylets were inserted into the guide cannulae and the incision around the cranial implant was sutured closed. Once Vetricyn antibacterial ointment was applied to the incision, rats were given an additional 10 ml of saline and allowed to recover from surgery on a heating pad. Rats were allowed to recover for one week before being food-restricted in preparation for behavioral training.
In experiment 2 (optogenetic manipulation of D2R-expressing neurons in the NAcSh), the surgical procedures were identical to those described above, with the exception that viral infusions were in the NAcSh. After leveling the skull, two burr holes were drilled for bilateral implantation of guide cannulae above the NAcSh (AP: +1.5; ML: ±3.1; DV: −6.4 from skull surface at a 20° angle). Once the cannulae were secured with dental cement, injectors were inserted into each cannula, through which AAV5-EF1α-DIO-eNpHR3.0-mCherry (University of North Carolina Vector Core; 4 × 10^12 vg/ml) or AAV-EF1α-DIO-mCherry (5.3 × 10^12 vg/ml) was injected into the NAcSh (0.5 µl) at a rate of 0.3 µl/min. Injectors remained in place for an additional 5 min and were then replaced with sterile stylets. The incision around the cranial implant was sutured closed and Vetricyin was applied to the skin around the implant. Rats were administered 10 ml of saline and were placed on a heating pad to recover. After one week of recovery from surgery, rats were food-restricted in preparation for behavioral training.
In vitro electrophysiology
Brain slice preparation
Rats were anesthetized using a ketamine/xylazine cocktail (100 mg/kg ketamine, 10 mg/kg xylazine) to achieve a surgical plane of anesthesia (evaluated by the absence of response to tail or hind paw pinch) and transcardially perfused with ice-cold sucrose-laden artificial CSF (ACSF) containing (in mm): 205 sucrose, 10 dextrose, 1 MgSO4, 2 KCl, 1.25 NaH2PO4, 1 CaCl2, and 25 NaHCO3, oxygenated with 95% O2/5% CO2. Rats were decapitated using a small animal guillotine, brains were extracted and submerged in perfusion solution, and then sectioned horizontally to produce 300-µm slices using a Leica VT1000 vibratome. Slices were transferred to incubation ACSF containing (in mm): 124 NaCl, 10 dextrose, 3 MgSO4, 2.5 KCl, 1.23 NaH2PO4, 1 CaCl2, and 25 NaHCO3, oxygenated with 95% O2/5% CO2 and maintained at 35°C. After 30 min, the incubation chamber was permitted to passively equilibrate to room temperature for at least 30 min before recording. Recordings were performed in ACSF containing (in mm): 126 NaCl, 11 dextrose, 1.5 MgSO4, 3 KCl, 1.2 NaH2PO4, 2.4 CaCl2, and 25 NaHCO3, oxygenated with 95% O2/5% CO2, and maintained at 28°C flowing at 2 ml/min through a perfusion chamber. Patch pipettes with an open tip resistance of 4–6 MΩ were prepared from borosilicate glass (Sutter Instrument BF150-86-10) using a Flaming/Brown pipette puller (Sutter Instrument SU-P97). Cells were visualized with an Olympus BX51WI upright stereomicroscope using IR-DIC optics, a 12-bit CCD camera (QImaging Rolera-XR), and µManager software (Edelstein et al., 2010). A TTL-controlled LED light source (X-Cite 110LED, Excelitas Technologies) paired with a green emission filter (Omega Optical XF414) was used to deliver green light to the sample through a 40× water-immersion objective lens. Electrophysiological recordings were performed using a CV7-B patch-clamp headstage, MultiClamp 700B amplifier, DigiData 1440A digital acquisition system, and pClamp 11 software (Molecular Devices/Molecular Devices). Data were acquired at 20 kHz. Voltage clamp recordings were lowpass filtered using a 2-kHz Bessel filter. Patch pipettes were filled with a physiological internal solution containing (in mm): 125 potassium gluconate, 10 phosphocreatine, 1 MgCl2, 10 HEPES, 0.1 EGTA, 2 Na2ATP, 0.25 Na3GTP, adjusted to pH 7.25 and 295 mOsm. Data are presented uncorrected for the liquid junction potential.
Electrophysiology procedures
Neurons expressing eNpHR3.0 were targeted for whole-cell recording by their mCherry fluorescence. Immediately following establishment of whole-cell configuration, passive membrane properties (holding current, membrane resistance, and whole-cell capacitance) were evaluated in voltage clamp using a brief −10 mV hyperpolarizing step from a holding potential of −70 mV. Sensitivity to green light was evaluated in current-clamp configuration at the resting potential. Light-induced hyperpolarization was quantified as the mean voltage over a 500-ms period of green light exposure relative to the resting voltage, which was measured over a 1-s period before onset of the green light. To demonstrate eNpHR3.0-mediated suppression of firing, neurons were depolarized above action potential threshold using continuous positive current injection and 1-s green light was applied to demonstrate reduction or elimination of firing. In a subset of experiments, the D2R agonist quinpirole (Tocris #1061) was delivered into the perfusion system using a syringe pump to achieve a bath concentration of 10 μM. Changes in holding current required to voltage-clamp neurons at −70 mV were monitored and the effect of quinpirole was reported as the mean holding current observed during a 2-min period during quinpirole exposure minus the mean holding current observed in a 3-min baseline period immediately before application. Membrane resistance was evaluated from the same voltage-clamp step using the same time periods and is reported as the resistance during quinpirole exposure normalized to the baseline resistance. Effects of quinpirole on action potential firing frequency were evaluated using a current-clamp protocol that applied a 1-s excitatory current pulse (sufficient to evoke five or more action potentials in baseline conditions) every 10 s. Quinpirole-induced changes in firing frequency were evaluated in a 2-min period during quinpirole exposure normalized to a baseline period, resulting in a baseline mean of 1.0. For all electrophysiological experiments, significance of effects observed in baseline subtracted or normalized data were evaluated using a one-sample Student's t test (with null hypothesis that mean = 0, or 1, respectively).
Behavioral procedures
Before training in the Risky Decision making Task (RDT), rats were shaped to perform separate components of the task, such as lever pressing and nosepoking to initiate trials as described previously (Orsini et al., 2017; Hernandez et al., 2019). Briefly, rats were first trained to associate the sound of a food pellet being deposited in the food trough with food delivery, after which they learned to press each of the two levers to receive a single food reward. Finally, rats were trained to nosepoke into the nosepoke hole to trigger the extension of a single lever, a press on which resulted in the delivery of one food pellet. Upon completion of these shaping protocols, rats began training in a Reward Discrimination task. This task consisted of three blocks of 28 trials and lasted 56 min. The beginning of a trial was signaled with the illumination of the houselight and nosepoke light. A nosepoke into the nosepoke hole extinguished the nosepoke light and triggered the extension of one lever (forced choice trial; randomly presented) or both levers (free choice trial). If rats failed to nosepoke within 10 s, both lights were extinguished and the trial was scored as an omission. A lever press on one lever resulted in the delivery of one food pellet (small reward), whereas a press on the other lever resulted in the delivery of two food pellets (large reward). The identity of the lever (small vs large reward) was counterbalanced across rats and sexes and remained consistent throughout the entire experiment. Failure to press levers within 10 s led to the retraction of the levers and the termination of lights in the chamber, and the trial was scored as an omission. Once a lever press occurred, levers were retracted and the light in the food trough was illuminated and remained so until food was collected by the rat or after 10 s had elapsed, whichever occurred first. Each block of 28 trials began with eight forced choice trials in which a single lever was extended into the chamber (four presentations of each lever, randomized across the eight trials) and ended with 20 free choice trials in which both levers were extended into the chamber and rats were free to choose between them. Unlike the RDT (see following paragraph), each block was identical in structure and served to teach the rat about the overall design of the task before footshock punishment was introduced. Rats were trained on the Reward Discrimination task until they selected the large reward on at least 80% of the free choice trials for two consecutive days, after which they progressed to training in the RDT. On average, rats required no more than five sessions to reach these criteria.
The task structure for the RDT was identical to that of the Reward Discrimination task, with the exception that delivery of the large reward was accompanied by the possibility of a mild footshock punishment (Fig. 1A). The probability of footshock delivery was 0% in the first block of trials and increased to 25% and 75% in the second and third block, respectively. Rats learned which probability of footshock delivery was in effect for each block during the forced choice trials that preceded the free choice trials. In the forced choice trials, the probability of receiving a footshock was dependent across the four trials in which the large, “risky” lever was available. For example, in the 25% trial block, one and only one of the forced choice trials would lead to the delivery of a footshock. In contrast, in the 75% trial block, three out of the four forced choice trials would lead to the delivery of a footshock. In the free choice trials, however, the probability of footshock on a single trial was independent of the outcomes on other trials within that trial block. Hence, the probability of shock delivery was equivalent across all free choice trials, regardless of whether previous trials resulted in shock deliveries. Shock intensities were initially set at 0.15 mA for females and 0.25 for males, but were individually adjusted for each rat during training to ensure that their baseline performance was in the middle of the parametric space before optogenetic manipulations. Shock intensities never exceeded 0.55 mA and never fell below 0.075 mA. Rats were trained on the RDT until behavioral stability was achieved (see below, Experimental design and statistical analyses, for definition of behavioral stability).
Schematic of the Risky Decision making Task and timing of light delivery. A, In the Risky Decision making Task (RDT), rats choose between two levers that differ in their reward magnitude and the risk of punishment. A press on one lever yields a small reward with no risk of punishment, whereas a press on the other lever yields a large reward associated with risk of punishment. The probability of punishment delivery systematically increases across the three trial blocks. B, Light was delivered during discrete epochs of the decision process in separate test sessions. These epochs included (1) deliberation; (2) delivery of the small, safe reward; (3) delivery of the large, unpunished reward; (4) delivery of the large, punished reward; and (5) the intertrial interval (ITI). Using a within-subjects design, the order of the test sessions was counterbalanced across rats, and successive optogenetic test sessions were separated by nonoptogenetic test sessions to re-establish baseline performance. Figures created with BioRender.com.
Once behavioral stability on the RDT was reached, rats were lightly anesthetized and optic fibers were bilaterally inserted into the cannulae in the NAcSh, extending 1 mm beyond the tip of the cannulae. Fibers were cemented into place and dust caps were placed on the fibers to protect them from debris accumulation. From this point on, stainless steel patch cords were attached to the implanted fibers during each test session. Rats were tested in this manner to habituate them to being tethered before optogenetic manipulations. After behavioral stability was once again reached (2–3 d; tethering rats tended to induce variability in performance for several days), optogenetic manipulations began. In these sessions, laser delivery only occurred during free choice trials and was specific to distinct periods of the trial (Fig. 1B): (1) deliberation, (2) reward outcome, and (3) intertrial interval (ITI). On each free choice trial during deliberation sessions, the laser delivered light specifically during the period between the illumination of the nosepoke hole and a lever press. To ensure that full inhibition was in effect before the beginning of the deliberation phase, the onset of laser delivery preceded the illumination of the nosepoke hole by 0.5 s. If a rat did not press a lever within 10 s, light delivery was terminated. With respect to reward outcome optogenetic sessions, there were three different laser delivery sessions: (1) delivery of the small, safe reward (2) delivery of the large reward without punishment (large unpunished reward), and (3) delivery of the large reward with punishment (large punished reward). In each session, laser delivery was initiated as soon as a rat pressed a lever and was terminated 5 s after food delivery. Finally, serving as a behavioral control for laser delivery, the ITI optogenetic test session consisted of the delivery of laser light 8–15 s after each reward delivery, with each period of light delivery lasting 5 s. A randomized within-subjects design was used to determine the order in which the optogenetic test sessions were presented (across multiple days). After each optogenetic test session, rats were tethered and tested in the RDT until their behavior re-stabilized, after which they proceeded to the next optogenetic test session.
Determination of shock reactivity threshold
Upon completion of testing in the RDT, a subset of eNpHR3.0 rats in experiment 1 (n = 4) underwent a procedure to evaluate the shock intensities at which specific motor responses were generated. This assay, which was based on methods developed by Bonnet and Peterson (1975), was conducted twice for each rat over two consecutive days, with one session involving laser delivery (i.e., optogenetic inhibition) and the other involving no laser delivery (i.e., no optogenetic inhibition). Rats were placed in an operant chamber distinct from that in which they underwent RDT testing and given 2 min to acclimate to the chamber. A 0.40-mA unsignaled footshock (1 s) was then delivered to reduce spontaneous locomotor activity and facilitate observations of motor responses on subsequent trials at lower shock intensities. The shock intensity was then reduced to 0.05 mA and a series of five 1-s footshocks was delivered every 10 s. After each set of footshocks, the shock intensity was increased by 0.025 mA and another series of five footshocks were delivered. This pattern continued until all predetermined motor responses of interest were identified. A shock reactivity threshold for a specific motor response was established when three out of the five shock deliveries in a series elicited the specific motor response. The motor responses for which shock threshold intensities were determined were (1) flinch of one paw or a startle response (2) elevation of one or two paws (3) quick movement of more than two paws. During the session in which laser delivery occurred, light was delivered bilaterally through the implanted optic fibers with the same system and parameters (i.e., wavelength, light intensity) used for optogenetic manipulations during the RDT. Laser onset occurred concurrently with the onset of the footshock (1 s) and was terminated after an additional 4 s (5 s in total duration, which matches the duration of laser delivery during the large, punished reward session). During sessions in which no laser delivery occurred, rats were still tethered to equate behavioral conditions. The order of the test conditions (i.e., laser delivery vs no laser delivery) was counterbalanced across rats.
Fluorescent in situ hybridization (FISH), immunohistochemistry, and histology
FISH
To confirm that halorhodopsin expression was selective to neurons that expressed D2Rs within the NAcSh in D2-Cre transgenic rats, FISH (RNAscope, v1, ACD Bio) combined with immunohistochemistry was performed on brain tissue from rats (n = 4) that received viral infusions of AAV5-EF1α-DIO-eNpHR3.0-mCherry. Three weeks after surgery, rats were overdosed with intraperitoneal injections of Euthasol and transcardially perfused with ice-cold RNAase-free 0.9% NaCl followed by 4% paraformaldehyde (PFA) in 0.1 m PBS. Brains were extracted and postfixed in 4% PFA for 12 h, after which they were transferred to 30% sucrose in 0.1 m PBS. Tissue was sectioned using a freezing cryostat at 12 µm and mounted directly onto Superfrost Plus Gold slides (Fisher Scientific) and then stored at −80°C. The modified procedure that combines RNAscope FISH with immunohistochemistry was previously described in Shallcross et al. (2019) with the following adjustments. The target probe used for detection of D2R mRNA was DRD2 (ACDBio: 315641-C2). Expression of mCherry was detected using anti-mCherry antibody (AbCam, ab167453, 1:1000) and visualized using donkey anti-rabbit secondary antibody conjugated to Alexa Fluor 488 (Life Technologies, A32790, 1:500). Finally, sections were counterstained with DAPI (Life Technologies), coverslipped using ProLong Gold antifade mounting reagent (Life Technologies) and sealed with nail polish.
D2R and mCherry signal (Z-stacks with 1-µm step size) were acquired using a Zeiss LSM70 confocal microscope and a 63× oil immersion objective. The procedures used for the quantification of D2R puncta and co-localization of D2R and mCherry signal have been described previously (Shallcross et al., 2019).
Immunohistochemistry
At the completion of behavioral testing, rats were overdosed with intraperitoneal injections of Euthasol and transcardially perfused with ice-cold 0.1 m PBS, followed by 4% PFA in 0.1 m PBS. Brain tissue was extracted, postfixed in 4% PFA and then transferred to a 30% sucrose in 0.1 m PBS solution. Tissue was sectioned (35 µm) on a cryostat maintained at −20°C. Sections were collected (1-in-6 series for experiment 1; 1-in-4 series for experiment 2) into wells filled with 0.1 m PBS, after which they were transferred to cryoprotectant.
To confirm viral expression, immunohistochemistry was performed to amplify the mCherry fluorescent tag. Free-floating sections were washed three times for 10 min in 0.1 m Tris-buffered saline (TBS) and then incubated in 0.1 m TBS with 5% blocking buffer (normal donkey serum, Jackson ImmunoResearch, NC9624464) and 0.3% Triton X-100 for 2 h at room temperature. Tissue was then transferred into wells filled with primary antibody (rabbit anti-mCherry, AbCam, ab167453, 1:1000) diluted in 0.1 m TBS with 5% blocking buffer and 0.3% Triton X-100. Sections were incubated in the primary antibody solution for 72 h at 4°C and then were washed three times for 10 min in 0.1 m TBS. After the last wash, tissue was transferred into secondary antibody (donkey anti-rabbit conjugated to Alexa Fluor 488, Life Technologies, A32790, 1:300) diluted in 0.1 m TBS with 5% blocking buffer and 0.3% Triton X-100 and allowed to incubate in this solution for 2 h at room temperature. Finally, tissue was washed three additional times for 10 min in 0.1 m TBS and then mounted onto slides. Once sections on the slides were dry, slides were coverslipped with ProLong Gold Antifade mounting medium (Invitrogen, P36934) and sealed with nail polish. Processed tissue was visualized with a Zeiss AxioImager 2 microscope at 10× to confirm viral expression and determine optic fiber placement. Representative images were acquired using Zen Pro software at 5× with the same microscope.
Experimental design and statistical analyses
Sample sizes for each group (eNpHR3.0 vs mCherry) were determined a priori using power analyses in G*Power software. These analyses revealed that sample sizes of at least n = 6 were necessary to detect significant differences between baseline and inhibition conditions in choice performance with effect sizes of ≥ 0.6 (assuming an α of 0.05). Additional rats were added to each group to account for potential attrition during the study and/or missed viral and/or cannula placements. Although both males and females were included in each experiment, group sizes were not powered to specifically test for sex differences as this was not the primary objective of the study. Sex was included as a between-subjects factor in all analyses, however, to inform experimental design for future studies (i.e., if there were sex differences, future studies would be sufficiently powered to explicitly examine sex differences). In all analyses, a p-value ≤ 0.05 was considered statistically significant. If parent ANOVAs yielded main effects or significant interactions, additional post hoc ANOVAs or t tests were conducted to determine the source of the significance and p-values were adjusted to account for multiple comparisons (Bonferroni's correction). Effect sizes are reported as η2 for ANOVAs and as the absolute value of Cohen's d for independent sample's t tests.
Raw files were extracted from Graphic State 4.0 using customized analysis templates and exported to Microsoft Excel. Excel was used to isolate behavioral measures of interest, including lever presses, latencies to lever press, latencies to initiate a trial (nosepoke), latencies to collect food rewards, and omissions. Data were analyzed using SPSS 27.0 and figures were created using GraphPad 9.4.1. The primary dependent variable was the percentage of free choice trials on which a rat chose the large, risky reward. Individual rats were trained in the RDT until they displayed stable choice performance, which was defined as choice behavior with a coefficient of variation of ≤30% for percent choice of the large, risky reward in each block for a sliding window of at least two consecutive test sessions. Upon reaching stability, rats began the series of optogenetic test sessions. Following each optogenetic test session, rats were re-tested in the RDT until the stability criterion was once again achieved, after which they advanced to the next optogenetic test session. For each optogenetic test session, the corresponding baseline session consisted of performance in the RDT averaged across at least two consecutive days of stable behavior immediately preceding the optogenetic session. A two-factor repeated-measures ANOVA was used to assess the effects of optogenetic inhibition on choice behavior, with session (baseline vs inhibition/light delivery) and trial block as within-subjects factors. For this and all subsequent analyses, sex was included as a between-subjects factor; if there was no main effect of sex or a sex × session interaction, data were collapsed across sexes and analyzed and presented accordingly. If there was a main effect of session and/or a session × trial block interaction, trial-by-trial analyses were conducted to determine whether optogenetic inhibition altered the extent to which the outcome of a previous trial affected subsequent choice. Providing a measure of sensitivity to rewarding outcomes, win-stay trials were defined as trials on which a rat continued to choose the large, risky lever after receiving a large, unpunished reward. This variable was calculated by the dividing the number of trials on which the rat selected the large, risky lever after receipt of the large, unpunished reward by the total number of free choice trials on which the rat received the large, unpunished reward. In contrast, lose-shift trials provided a measure of sensitivity to negative feedback (i.e., punishment) and were defined as trials on which a rat shifted its choice to the small, safe lever after receiving a large, punished reward. This variable was calculated by dividing the number of trials on which the rat selected the small, safe lever after receipt of the large punished reward by the total number of free choice trials on which the rat received the large punished reward. Effects of optogenetic inhibition on win-stay and lose-shift trials were determined using a repeated-measures ANOVA, with trial type (win-stay vs lose-shift) and session (baseline vs inhibition) included as within-subjects factors.
If optogenetic manipulations altered choice behavior, ancillary behavioral measures, including latencies to lever press, nosepoke, and collect food rewards, were also compared between baseline and optogenetic test sessions. Latency to lever press during free choice trials was defined as the period of time between the nosepoke to trigger lever extension and a lever press. Because rats displayed a near-exclusive preference for the large reward in Block 1, latencies to press levers during free choice trials were constrained to Blocks 2 and 3 (i.e., there were insufficient data for latencies to press the small reward in Block 1, precluding its comparison to latencies to press the large reward). A repeated-measures ANOVA was used to assess the effects of optogenetic inhibition on latencies to press levers, with session (baseline vs inhibition), lever identity (small, safe lever vs large, risky lever) and trial block as within-subjects factors. Because onset of the laser began 0.5 s before the illumination of the nosepoke hole in test sessions in which the light was delivered during deliberation, latencies to nosepoke (i.e., the interval of time between the illumination of the nosepoke and a nosepoke) were also analyzed to determine whether light delivery altered this behavioral response. Specifically, a repeated-measures ANOVA was used to compare latencies to nosepoke between baseline and the deliberation test condition, with session and trial block as within-subjects factors. In sessions in which light was delivered during the reward delivery period, latencies to collect food were analyzed using a repeated-measures ANOVA, with session and trial block as within-subjects factors. Finally, omissions during free choice trials were calculated by dividing the number of omitted free choice trials by the total number of possible free choice trials (60) and multiplying this number by 100. For the large, punished reward optogenetic test sessions, analysis of omissions was limited to Blocks 2 and 3 as inhibition never occurred during Block 1. Omissions were then compared between baseline and optogenetic test sessions using a repeated-measures ANOVA (the inclusion of sex as a between-subjects factor necessitated the use of an ANOVA over a t test). Notably, the percentage of omitted free choice trials was analyzed for each optogenetic manipulation, regardless of whether there was an effect on choice behavior.
Shock reactivity thresholds in eNpHR3.0 rats were compared between light delivery (i.e., inhibition) and no light delivery sessions using a repeated-measures ANOVA, with session and the type of motor response (one paw flinch, two paw flinches, rapid movement of more than two paws) included as within-subjects factors.
Results
For behavioral experiments, descriptions of statistical results are restricted to those directly related to the effects of optogenetic inhibition (or light delivery in control rats). Unless otherwise noted, there were no sex differences in risk taking (collapsed across optogenetic inhibition/light delivery conditions; all p-values > 0.05). Such an absence of sex differences was likely a result of shock intensities being titrated individually for each rat. For all analyses in which trial block was included as a within-subjects factor, there was a main effect of trial block (p-values < 0.05); this effect will therefore not be reported further.
Experiment 1
Histology
A total of 54 Long–Evans rats (n = 29, male; n = 25, female) were used in experiment 1. Of the 54, 29 rats (n = 16, male; n = 13, female) received intra-BLA infusions of the viral vector containing eNpHR3.0. Four females and four males in this group lost their cranial implants before any optogenetic manipulations and were therefore excluded from the study. In addition, three females and two males were excluded because of lack of viral transduction (i.e., no evidence of viral injection in the BLA and no terminal expression in the NAcSh). Twenty-five rats (n = 13, male; n = 12, female) received intra-BLA infusions of the viral vector containing mCherry. Five females and three males in this group lost their cranial implant before any optogenetic manipulations and were therefore excluded from the study. Three females and three males were also excluded as a result of poor viral transduction in the BLA and BLA terminals in the NAcSh. Consequently, the final sample sizes were n = 16 for the eNpHR3.0 group (n = 10, male; n = 6, female) and n = 11 for the mCherry group (n = 7, male; n = 4, female). Figure 2 depicts the spread of eNpHR3.0 (Fig. 2A) or mCherry alone (Fig. 2C) in the BLA and the placements of the optic fibers implanted in the NAcSh for rats in the eNpHR3.0 (Fig. 2B) and control (Fig. 2D) groups. Representative eNpHR3.0 expression in the BLA and optic fiber placement in the NAcSh are displayed in Figure 2E,F, respectively. Importantly, some rats did not undergo every optogenetic test condition because of illness or loss of their cranial implant.
Virus expression in the basolateral amygdala (BLA) and optic fiber placement and BLA terminal expression in the NAcSh. A, Dark gray shading represents maximum spread of eNpHR3.0-mCherry, and light gray shading represents minimal spread of eNpHR3.0-mCherry in the BLA. B, Circles represent the location of the tip of the optic fiber in the nucleus accumbens shell (NAcSh) of rats with BLA terminal expression of eNpHR3.0. C, Dark gray shading represents maximum spread of mCherry alone (control group), and light gray shading represents minimal spread of mCherry in the BLA. D, Circles represent the location of the tip of the optic fiber in the NAcSh of rats with mCherry terminal expression. E, Representative image of the injection site in the BLA in rats injected with eNpHR3.0-mCherry. F, Representative image of the tip of the optic fiber and BLA terminal expression of mCherry in the NAcSh of an eNpHR3.0 rat. Images were taken at 5× magnification.
Optogenetic inhibition of BLA terminals in the NAcSh during decision making in eNpHR3.0 rats
Inhibition during deliberation
Optogenetic inhibition of the BLA→NAcSh during deliberation (n = 9, male, n = 4, female) significantly increased choice of the large, risky reward (increased risk taking) in both males and females (session, F(1,11) = 5.80, p = 0.04, η2 = 0.35; session × sex, F(1,11) < 0.01, p = 0.98, η2 < 0.01; session × trial block, F(2,22) = 4.44, p = 0.02, η2 = 0.29; session × sex × trial block, F(2,22) < 0.01, p = 0.99, η2 < 0.01; Fig. 3A). Despite the effects of inhibition on choice performance, there was no effect of inhibition on the percentage of win-stay and lose-shift trials (session, F(1,11) = 0.41, p = 0.54, η2 = 0.04; session × sex, F(1,11) = 0.11, p = 0.74, η2 = 0.01; session × trial type, F(1,11) = 0.70, p = 0.42, η2 = 0.06; session × sex × trial type, F(1,11) = 0.03, p = 0.86, η2 < 0.01; Fig. 3B). To determine whether inhibition during deliberation had any lasting effect on task performance (beyond the session in which inhibition took place), baseline performance was compared with performance in the RDT 24 h after the optogenetic test session using a three-factor repeated-measures ANOVA. This analysis revealed that there was no difference in performance between baseline and post-inhibition sessions (session, F(1,10) = 0.88, p = 0.37, η2 = 0.08; sex × session, F(1,10) = 0.56, p = 0.47, η2 = 0.05; session × trial block, F(2,20) = 0.93, p = 0.41, η2 = 0.09; session × sex × trial block, F(2,20) = 1.02, p = 0.38, η2 = 0.09). Considered together, these results reveal that activity in the BLA→NAcSh during the deliberation phase of the decision process biases subsequent choice toward small, safer options.
Optogenetic inhibition of BLA projections to the NAcSh during deliberation. A, Inhibition of the BLA→NAcSh during deliberation increased choice of the large, risky reward. Data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Block 1) is because the size of the SEM was smaller than the point displayed. Asterisks indicate a significant difference between baseline (no laser delivery) and optogenetic inhibition (p < 0.05). B, There was no effect of inhibition of the BLA→NAcSh during deliberation on the percentage of win-stay or lose-shift trials. Open circles represent data points for individual rats.
The onset of laser delivery during the deliberation test condition was 0.5 s before the start of each trial. Hence, to determine whether inhibition altered the latency to nosepoke to trigger lever insertion into the chamber, nosepoke latencies during free choice trials were compared between baseline sessions and the optogenetic test sessions. Despite an overall main effect of sex, with females taking longer to nosepoke than males (F(1,11) = 7.35, p = 0.02, η2 = 0.40), there was no effect of inhibition on this behavioral measure in males or females (session, F(1,11) = 0.08, p = 0.79, η2 < 0.01; session × sex, F(1,11) = 1.29, p = 0.28, η2 = 0.11; session × trial block, F(2,22) = 1.13, p = 0.34, η2 = 0.09; session × sex × trial block, F(2,22) = 1.19, p = 0.32, η2 = 0.10). Similarly, inhibition did not affect latencies to press levers during the free choice trials (session, F(1,8) = 0.10, p = 0.76, η2 = 0.01; session × sex, F(1,8) = 0.23, p = 0.64, η2 = 0.03; session × lever identity (small, safe vs large, risky lever), F(1,8) = 0.89, p = 0.37, η2 = 0.10; session × lever identity × trial block, F(1,8) = 1.11, p = 0.32, η2 = 0.12; session × sex × trial block, F(1,8) = 1.59, p = 0.24, η2 = 0.17; session × sex × lever identity × trial block, F(1,8) = 1.99, p = 0.20, η2 = 0.20). Although this analysis did yield a significant interaction between session, sex and lever identity (F(1,8) = 5.81, p = 0.04), the results of post hoc ANOVAs comparing latencies to press each lever between sessions separately for males and females did not reach statistical significance when correcting for multiple comparisons. Finally, optogenetic inhibition of the BLA→NAcSh did not impact the percentage of omitted free choice trials (session, F(1,11) < 0.01, p = 0.96, η2 < 0.01; session × sex, F(1,11) = 1.73, p = 0.22, η2 = 0.14; Table 1).
Mean (±SEM) omissions
Inhibition during delivery of the small, safe reward
A three-factor, repeated-measures ANOVA revealed that optogenetic inhibition of the BLA→NAcSh during delivery of the small, safe reward (n = 9, male; n = 5, female) had no effect on choice of the large, risky reward (session, F(1,12) = 1.64, p = 0.23, η2 = 0.12; session × sex, F(1,12) = 0.86, p = 0.37, η2 = 0.07; session × trial block, F(2,24) = 2.40, p = 0.11, η2 = 0.17; session × sex × trial block, F(2,24) = 0.80, p = 0.46, η2 = 0.06; Fig. 4A). Consistent with a lack of an effect on choice behavior, there was no effect of inhibition on percentage of omitted free choice trials (session, F(1,12) = 1.28, p = 0.28, η2 = 0.10; session × sex, F(1,12) = 0.45, p = 0.56, η2 = 0.04; Table 1). Collectively, these data reveal that activity in the BLA→NAcSh is not necessary for evaluation of small, safe rewards to guide subsequent risk-taking behavior.
Optogenetic inhibition of BLA projections to the NAcSh during other phases of the RDT. A, There was no effect of BLA→NAcSh inhibition during delivery of the small, safe reward on choice of the large, risky reward. B, There was no effect of BLA→NAcSh inhibition during delivery of the large, unpunished reward on choice of the large, risky reward. C, There was no effect of BLA→NAcSh inhibition during the intertrial interval on choice of the large, risky reward. Data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Block 1) is because the size of the SEM was smaller than the point displayed.
Inhibition during delivery of the large, unpunished reward
Optogenetic inhibition of the BLA→NAcSh during delivery of the large, unpunished reward (n = 9, male; n = 5, female) did not affect choice of the large, risky reward (session, F(1,12) = 2.15, p = 0.17, η2 = 0.15; session × trial block, F(2,24) = 2.31, p = 0.12, η2 = 0.16; Fig. 4B). Despite a main effect of sex (i.e., greater risk taking in males; F(1,12) = 15.29, p < 0.01, η2 = 0.56), there were no significant session × sex (F(1,12) = 1.10, p = 0.32, η2 = 0.08) or session × sex × trial block (F(2,24) = 1.46, p = 0.25, η2 = 0.11) interactions, indicating that inhibition during this task phase was ineffective in altering risk taking in males or females. Although females omitted significantly more free choice trials overall (F(1,12) = 15.98, p < 0.01, η2 = 0.57), there was no main effect of session (F(1,12) = 0.71, p = 0.42, η2 = 0.06) nor a significant sex × session interaction (F(1,12) = 2.22, p = 0.16, η2 = 0.16) on percentage of omitted free choice trials (Table 1). Together, these results indicate that activity in the BLA→NAcSh is not necessary for evaluation of large, unpunished rewards to guide subsequent risk-taking behavior.
Inhibition during delivery of the large, punished reward
In contrast to the effects of inhibition during delivery of the large, unpunished reward, optogenetic inhibition of the BLA→NAcSh during delivery of the large, punished reward (n = 8, male; n = 5, female) increased choice of the large, risky reward (session, F(1,11) = 8.56, p = 0.01, η2 = 0.44; session × trial block, F(2,22) = 10.34, p < 0.01, η2 = 0.48). Unexpectedly, in addition to a main effect of sex (F(1,11) = 11.24, p < 0.02, η2 = 0.51), there were significant session × sex (F(1,11) = 14.18, p < 0.01, η2 = 0.56) and session × sex × trial block (F(2,22) = 9.47, p < 0.01, η2 = 0.46) interactions. To determine the source of the significant interaction, subsequent two-factor repeated-measures ANOVAs were conducted to compare performance between baseline and inhibition sessions separately for males and females. These analyses revealed that optogenetic inhibition increased risk taking in males (session, F(1,7) = 31.85, p = 0.01, η2 = 0.82; session × trial block, F(2,14) = 40.09, p < 0.01, η2 = 0.85; Fig. 5A), but did not affect performance in females (session, F(1,4) = 0.25, p = 0.64, η2 = 0.06; session × trial block, F(2,8) = 1.16, p = 0.36; η2 = 0.22; Fig. 5B). Although inhibition increased risk taking in males, a two-factor repeated-measures ANOVA revealed that there was no effect of this manipulation on the percentage of win-stay and lose-shift trials (session, F(1,7) = 3.05, p = 0.12, η2 = 0.30; session × trial type, F(1,7) = 2.09, p = 0.19, η2 = 0.23; Fig. 5C). Inspection of Figure 5C, however, suggests that inhibition may have increased the percentage of win-stay trials in males; a paired-samples t test confirmed this effect (t(5) = −3.77, p = 0.01, d = −1.54). Nevertheless, because the parent ANOVA did not yield a significant session × trial interaction, the results of this t test should be interpreted with caution. Effects of inhibition on risk taking in males did not persist beyond the optogenetic session as there were no differences between baseline performance and performance on the RDT 24 h after the manipulation (session, F(1,7) = 0.95, p = 0.37, η2 = 0.12; session × trial block, F(2,14) = 1.15, p = 0.35, η2 = 0.14). Collectively, these data show that activity of the BLA→NAcSh in males, but not females, is necessary for evaluation of large rewards that are accompanied by punishment to guide subsequent choice toward safer options. It is important to note that although it is possible that differences in shock intensities used for males and females could have contributed to these divergent effects of BLA→NAcSh inhibition on risk taking, this explanation is unlikely as shock intensities were titrated individually for each rat such that their choice performance was neither at the floor nor at the ceiling of the parametric space. In doing so, shock intensities ranged across the sexes, with some males tested with lower shock intensities than females.
Optogenetic inhibition of BLA projections to the NAcSh during delivery of the large, punished reward. A, Inhibition of the BLA→NAcSh during delivery of the large, punished reward increased choice of the large, risky reward in males. B, There was no effect of BLA→NAcSh inhibition on choice of the large, risky reward in females. For A and B, data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Block 1) is because the size of the SEM was smaller than the point displayed. Asterisks indicate a significant difference between baseline (no light delivery) and optogenetic inhibition (p < 0.05). C, There was no effect of BLA→NAcSh inhibition on the percentage of win-stay and lose-shift trials in males. D, Inhibition of the BLA→NAcSh did not alter shock reactivity thresholds at which shock delivery elicited a paw flinch, elevation of one or two paws or rapid movement of all four paws. For C and D, open circles represent data points for individual rats.
To determine whether inhibition affected other behavioral measures in males, latencies to press levers during free choice trials and collect food were analyzed. These analyses were constrained to data from the blocks in which inhibition occurred (i.e., Blocks 2 and 3). Analysis of latencies to press levers revealed no effect of inhibition on latencies to press either lever (session, F(1,6) = 0.65, p = 0.45, η2 = 0.10; session × lever identity, F(1,6) = 3.59, p = 0.11, η2 = 0.37; session × lever identity × trial block, F(1,6) = 0.11, p = 0.75, η2 = 0.02). Similarly, latencies to collect food were also not impacted by optogenetic inhibition (session, F(1,6) = 0.79, p = 0.41, η2 = 0.12; session × trial block, F(1,6) = 0.08, p = 0.79, η2 = 0.01). Finally, inhibition during delivery of the large, risky reward had no effect on the percentage of omitted free trials in Blocks 2 and 3 in males (t(7) = −0.81, p = 0.45, d = 0.29) or in females (t(3) = 1.38, p = 0.26, d = 0.69; Table 1).
Inhibition during the intertrial interval
To confirm that the effects of inhibition of the BLA→NAcSh during deliberation and delivery of the large, punished reward were not because of nonspecific effects of eNpHR3.0 activation, light was delivered during the intertrial interval (ITI; n = 9, male; n = 4, female). Unsurprisingly, there were no effects of optogenetic inhibition on choice of the large, risky reward (session, F(1,11) = 0.11, p = 0.75, η2 = 0.01; session × sex, F(1,11) = 1.38, p = 0.27, η2 = 0.11; session × trial block, F(2,22) = 0.02, p = 0.98, η2 < 0.01; session × sex × trial block, F(2,22) = 0.34, p = 0.72, η2 = 0.03; Fig. 4C). Similarly, although females made more omissions overall (F(1,11) = 8.33, p = 0.02, η2 = 0.43; Table 1), there was no effect of inhibition during the ITI on the percentage of omitted free choice trials (session, F(1,11) = 0.17, p = 0.69, η2 = 0.02; session × sex, F(1,11) = 0.31, p = 0.59, η2 = 0.03).
Inhibition during assessment of shock reactivity thresholds
One interpretation of the effects of inhibition of the BLA→NAcSh during delivery of the large, punished reward is that inhibition decreased sensitivity to footshock, and that the increase in risk taking was therefore secondary to this effect. To address this possibility, a subset of male rats expressing eNpHR3.0 (n = 4) was tested in a behavioral assay used to determine thresholds at which specific motor responses were elicited by the delivery of footshock (see Materials and Methods for a description of the assay). Because the effects of inhibition on risk taking were only observed in males, female rats were not included in this part of the experiment. Results of a three-factor repeated-measures ANOVA showed that shock reactivity thresholds for each of the preselected motor responses did not differ between light delivery (i.e., inhibition) and no light delivery sessions (session, F(1,3) = 2.78, p = 0.19, η2 = 0.48; session × motor response, F(2,6) = 0.62, p = 0.57, η2 = 0.17; Fig. 5D). Hence, effects of inhibition during delivery of the large, punished reward cannot be attributed to inhibition-induced alterations in sensitivity to footshock.
Light delivery to BLA terminals in the NAcSh during decision making in control rats
To confirm that light delivery alone did not cause the observed changes in risk taking (e.g., via local tissue heating), another group of rats received intra-BLA injections of an AAV containing mCherry alone and were implanted with optic fibers in the NAcSh (n = 7, male; n = 4, female). They were then trained in the RDT and, on achieving stable baseline performance, underwent optogenetic test sessions. These test sessions were restricted to those task phases in which optogenetic inhibition of the BLA→NAcSh in the eNpHR3.0 group significantly altered risk taking (i.e., deliberation and delivery of the large, punished reward). The order of optogenetic test sessions was counterbalanced across rats, and these sessions were separated by nonoptogenetic test sessions to re-establish baseline performance.
Light delivery during deliberation
Light delivery during deliberation (n = 7, male; n = 4, female) did not affect choice of the large, risky reward (session, F(1,9) = 2.55, p = 0.15, η2 = 0.22; session × sex, F(1,9) = 0.1.25, p = 0.29, η2 = 0.12; session × trial block, F(2,18) = 1.26, p = 0.31, η2 = 0.12; session × sex × trial block, F(2,18) = 0.95, p = 0.41, η2 = 0.10; Fig. 6A). There was also no effect of light delivery on latency to nosepoke to trigger lever extension (session, F(1,9) = 2.43, p = 0.15, η2 = 0.21; session × sex, F(1,9) = 1.64, p = 0.23, η2 = 0.15; session × trial block, F(2,18) = 0.05, p = 0.95, η2 < 0.01; session × sex × trial block, F(2,18) = 2.46, p = 0.11, η2 = 0.22). Similarly, light delivery did not affect latencies to press levers during free choice trials (session, F(1,8) = 0.08, p = 0.79, η2 = 0.01; session × sex, F(1,8) = 1.18, p = 0.31, η2 = 0.13; session × lever identity, F(1,8) = 3.09, p = 0.12, η2 = 0.28; session × lever identity × sex, F(1,8) = 1.96, p = 0.20, η2 = 0.20; session × lever identity × trial block, F(1,8) = 1.90, p = 0.21, η2 = 0.19; session × lever identity × sex × trial block, F(1,8) = 2.44, p = 0.16, η2 = 0.23). Lastly, light delivery during deliberation did not affect the percentage of omitted free choice trials (session, F(1,9) = 0.18, p = 0.69, η2 = 0.02; session × sex, F(1,9) = 0.73, p = 0.41, η2 = 0.08; Table 1). Considered together, these results indicate that light delivery alone did not account for the increased risk taking that was observed when the BLA→NAcSh was inhibited during deliberation.
Light delivery to BLA terminals in the NAcSh in control rats. A, Light delivery to BLA terminals in the NAcSh during deliberation in control rats (mCherry only) did not affect choice of the large, risky reward. B, Light delivery to BLA terminals in the NAcSh during delivery of the large, punished reward in control rats did not affect choice of the large, risky reward. Data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Block 1) is because the size of the SEM was smaller than the point displayed.
Additional analyses compared performance between eNpHR3.0 (n = 13) and mCherry control (n = 11) rats during light delivery test sessions. Sex was not included as a between-subjects factor as it was not a significant factor in the analyses of the data from each vector group separately. A three-factor repeated-measures ANOVA (session × vector group × trial block) revealed a main effect of session (F(1,22) = 8,06, p = 0.01, η2 = 0.25) as well as a significant session × trial block interaction (F(2,44) = 5.28, p < 0.01, η2 = 0.19). Although there was neither a main effect of vector group (F(1,22) = 0.02, p = 0.90, η2 < 0.01) nor a session × vector group interaction (F(1,22) = 0.99, p = 0.33, η2 = 0.04), the session × vector group × trial group interaction was significant (F(2,44) = 1.43, p = 0.05, η2 = 0.12). This significant interaction confirms that, relative to baseline performance, light delivery to BLA terminals in the NAcSh increased risk taking only in eNpHR3.0 rats.
Light delivery during delivery of the large, punished reward
Because inhibition of BLA→NacSh during delivery of the large, punished reward affected risk taking specifically in eNpHR3.0 males, analysis of effects of light delivery during delivery of the large, punished reward was conducted only for control male rats (n = 6). Light delivery during delivery of the large, punished reward did not affect choice of the large, risky reward (session, F(1,5) = 0.06, p = 0.82, η2 = 0.01; session × trial block, F(2,10) = 0.56, p = 0.59, η2 = 0.10; Fig. 6B). There were also no effects of light delivery on latencies to press levers during free choice trials (session, F(1,5) = 3.25, p = 0.13, η2 = 0.39; session × lever identity, F(1,5) = 1.03, p = 0.36, η2 = 0.17; session × lever identity × trial block, F(1,5) = 0.57, p = 0.48, η2 = 0.10). Similarly, light delivery had no effect on latencies to collect food (session, F(1,5) = 1.50, p = 0.28, η2 = 0.23; session × trial block, F(1,5) = 0.79, p = 0.41, η2 = 0.14) nor did it impact the percentage of omitted free choice trials (t(5) = −1.01, p = 0.36, d = 0.41; Table 1). Collectively, these data show that, similar to light delivery during deliberation, light delivery during delivery of the large, punished reward also has no effect on risk taking in control male rats.
A three-factor repeated measures ANOVA was used to compare the effects of light delivery between eNpHR3.0 (n = 8) and control male rats (n = 6; this analysis was not conducted in females as there was no effect of inhibition in eNpHR3.0 females). Not only were there main effects of session (F(1,12) = 18.14, p < 0.01, η2 = 0.60) and vector group (F(1,12) = 6.63, p = 0.02, η2 = 0.36), but there were also significant session × vector group (F(1,12) = 18.14, p < 0.01, η2 = 0.63), session × trial block (F(2,24) = 15.17, p < 0.01, η2 = 0.56) and session × vector group × trial block (F(2,24) = 23.02, p < 0.01, η2 = 0.66) interactions. Results of these analyses confirm that the effects of light delivery during delivery of the large, punished reward on risk taking are specific to eNpHR3.0 male rats.
Experiment 2
Histology
A total of 52 D2-Cre transgenic rats (n = 23, male; n = 29, female) were used in experiment 2. Of the 52 rats, 33 rats (n = 14, male; n = 19, female) received intra-NacSh infusions of the viral vector containing the Cre-dependent eNpHR3.0. Three (n = 2, male; n = 1, female) received intra-NAcSh infusions of the viral vector containing Cre-dependent eNpHR3.0 and were subsequently used for in vitro electrophysiology experiments to confirm that optogenetic inhibition of these neurons leads to a reduction in their activity (Fig. 7A–F). An additional four (n = 2/sex) rats injected with eNpHR3.0 were used for validation of selective viral transduction of D2R-expressing neurons using RNAscope (Fig. 7G–J). The remainder were used for the in vivo optogenetic experiments. Four females and two males in the eNpHR3.0 group lost their cranial implants before any optogenetic manipulations and were therefore removed from the study. Six females and one male were also excluded due lack of viral transduction (n = 3), missed cannula placement (n = 3), or infection around the site of the optic fiber (n = 1). Nineteen rats (n = 10, male; n = 9, female) received intra-NAcSh infusions of the viral vector containing Cre-dependent mCherry. Of these rats, three females and one male lost their cranial implant before optogenetic manipulations and were therefore excluded from the study. Only one male was excluded because of a lack of viral expression in the NAcSh. After accounting for attrition, the final sample sizes were n = 13 (n = 7, male; n = 6, female) for the eNpHR3.0 group and n = 14 (n = 7, male; n = 7, female) for the mCherry group. Figure 8 depicts the locations of the tips of the optic fibers in the NAcSh in the eNpHR3.0 (Fig. 8A) and control (Fig. 8B) groups. As in experiment 1, not all rats underwent every optogenetic test condition because of illness or loss of their cranial implant during the course of the experiment.
Functional and anatomic validation of the use of D2-Cre rats to optogenetically inhibit D2R-expressing neurons in the NAcSh. Green light inhibited eNpHR3.0-mCherry-positive D2R-expressing neurons in the NAcSh. A, NAcSh neurons expressing eNpHR3.0-mCherry were targeted by their red fluorescence (scale bar: 20 µm). B, Excitatory current delivered through the patch pipette induced action potential firing. C, Representative NAcSh neuron firing regularly in response to continuous 250-pA current injection is silenced by 1-s green light exposure. D, Excitatory current injection (1 s, 250 pA) was applied to induce continuous action potential firing, after which quinpirole was washed onto the tissue. E, Quinpirole (3 min, 10 μm) reduced the number of action potentials in response to the same current step. F, Normalized action potential frequency during the current step over time, as observed in all mCherry-expressing cells tested with quinpirole (n = 18). The baseline mean of normalized data (1.0) is scaled to 100% in this panel. G–I, Representative images depicting mCherry (G), D2R mRNA (H), and selective transduction of D2R-expressing neurons in the NAcSh (I). Green, mCherry; red, D2R mRNA; blue, DAPI. J, There were significantly more D2R+/mCherry+ neurons than D2R+ alone (non-mCherry; t(23) = 15.08, p < 0.001, d = 3.08) and mCherry+ alone (non-D2R; t(23) = 38.36, p < 0.001, d = 7.83) neurons in the NAcSh of rats injected with Cre-dependent eNpHR3.0-mCherry. Data are represented as mean ± standard error of the mean. Asterisks indicate a significant difference (p < 0.05).
Optic fiber placement and eNpHR3.0 expression in the NAcSh. A, Circles represent the location of the tip of the optic fiber in the NAcSh of rats with expression of eNpHR3.0 in D2R-expressing neurons. B, Circles represent the location of the tip of the optic fiber in the NAcSh of rats with mCherry expression in D2R-expressing neurons. C, Representative image of the tip of the optic fiber in the NAcSh and expression of mCherry in D2R-expressing neurons in the NAcSh of an eNpHR3.0 rat. Image was taken at 5× magnification.
In vitro electrophysiology
Halorhodopsin inhibits D2Rexpressing neurons in the NAcSh
To directly evaluate light-induced inhibition of D2R-expressing neurons in the NAcSh of D2-Cre rats, cells expressing mCherry were targeted for analysis using whole-cell patch-clamp recordings (Fig. 7A,B). On average, the membrane resistance and whole-cell capacitance of patched D2R-expressing neurons was 308.80 ± 39.67 MΩ and 62.59 ± 4.29 pF, respectively (n = 32 cells). Exposure to green light reliably hyperpolarized cells and reduced or eliminated tonic firing (mean ΔV: −26.37 ± 3.67 mV; n = 32, p = 4.34−8; Fig. 7C). Together, these findings demonstrate that green light robustly inhibits D2R-expressing neurons (i.e., those expressing Cre-dependent eNpHR3.0/mCherry) in the NAcSh.
Quinpirole inhibits D2R-expressing neurons in the NAcSh
To evaluate the electrophysiological action of the D2R agonist quinpirole on activity of D2R-expressing neurons in the NAcSh, cells were targeted for study by their red fluorescence (i.e., mCherry). In neurons evaluated in voltage-clamp configuration at −70 mV, 3-min exposure to 10 μm quinpirole produced an excitatory shift in holding current of 3.41 ± 1.4 pA and concomitantly reduced membrane resistance by 10.66 ± 2.30% (n = 10, p < 0.01). In neurons evaluated in current-clamp configuration, 3-min exposure to 10 μm quinpirole reduced the frequency of action potentials observed during a suprathreshold depolarization by 39.1 ± 7.6% (n = 18, p < 0.001; Fig. 7D–F). Together, these findings provide evidence that eNpHR3.0-mCherry is functionally expressed in D2R-expressing NAcSh neurons.
Optogenetic inhibition of D2R-expressing neurons in the NAcSh during decision making in eNpHR3.0 rats
Inhibition during deliberation
Inhibition of D2R-expressing neurons in the NAcSh during deliberation (n = 5, male; n = 6, female) increased choice of the large, risky reward (session, F(1,9) = 20.13, p < 0.01, η2 = 0.69; session × sex, F(1,9) = 0.15, p = 0.71, η2 = 0.02; session × trial block, F(2,18) = 11.15, p < 0.01, η2 = 0.55; session × sex × trial block, F(2,18) = 0.18, p = 0.83, η2 = 0.02; Fig. 9A). Analysis of the percentage of win-stay and lose-shift trials did not yield a main effect of session (F(1,9) = 0.93, p = 0.36, η2 = 0.09), but did reveal a significant session × trial type interaction (F(1,9) = 4.85, p = 0.05, η2 = 0.35; Fig. 9B). Similar to the effects of inhibition on percent choice of the large, risky reward, the effect of inhibition on trial type did not differ between males and females (session × sex, F(1,9) = 0.65, p = 0.44, η2 = 0.07; session × sex × trial type, F(1,9) = 0.02, p = 0.90, η2 < 0.01). Consequently, data were collapsed across males and females, and post hoc analyses were conducted to determine the source of the significant session × trial type interaction. Although inhibition had no effect on the percentage of lose-shift trials relative to baseline (t(10) = 0.94, p = 0.37, d = 0.28), inhibition caused a significant increase in the percentage of win-stay trials (t(10) = −4.33, p < 0.01, d = 1.31). The effects of inhibition on choice performance were specific to the optogenetic test session, as performance in the RDT 24 h after the manipulation was no longer significantly different from baseline performance (session, F(1,9) = 0.32, p = 0.59, η2 = 0.03; session × sex, F(1,9) = 0.71, p = 0.42, η2 = 0.07; session × trial block, F(2,18) = 0.01, p = 0.99, η2 < 0.01; session × sex × trial block, F(2,18) = 1.60, p = 0.23, η2 = 0.15). Together, these data reveal that activity of D2R-expressing neurons in the NAcSh during the deliberation phase of decision making biases choice toward small, safe rewards, potentially via modulating reward sensitivity.
Optogenetic inhibition of D2R-expressing neurons in the NAcSh during deliberation. A, Inhibition of D2R-expressing neurons in the NAcSh during deliberation increased choice of the large, risky reward. Data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Block 1) is because the size of the SEM was smaller than the point displayed. B, Relative to baseline, inhibition of D2R-expressing neurons during deliberation increased the percentage of win-stay trials but had no effect on the percentage of lose-shift trials. Open circles represent data points for individual rats. Asterisks indicate a significant difference between baseline (no light delivery) and optogenetic inhibition (p < 0.05).
To determine whether inhibition of these neurons affected other behavioral measures in the RDT, latencies to nosepoke, latencies to lever press, and percentage of omissions were analyzed. Latencies to nosepoke to trigger lever extension were not affected by inhibition during deliberation (session, F(1,9) = 2.01, p = 0.19, η2 = 0.18; session × sex, F(1,9) = 2.45, p = 0.15, η2 = 0.21; session × trial block, F(2,18) = 0.46, p = 0.64, η2 = 0.05; session × sex × trial block, F(2,18) = 0.14, p = 0.87, η2 = 0.02). Similar to latencies to nosepoke, there was no effect of inhibition on latencies to press levers during free choice trials (session, F(1,6) = 1.72, p = 0.24, η2 = 0.22; session × sex, F(1,6) = 0.36, p = 0.57, η2 = 0.06; session × lever identity, F(1,6) = 1.11, p = 0.33, p = 0.16; session × lever identity × trial block, F(1,6) = 0.56, p = 0.48, η2 = 0.09; session × sex × trial block, F(1,6) = 0.08, p = 0.79, η2 = 0.01; session × lever identity × sex × trial block, F(1,6) = 2.16, p = 0.19, η2 = 0.26). Finally, inhibition did not affect percentage of omitted free choice trials (session, F(1,9) = 0.04, p = 0.86, η2 < 0.01; session × sex, F(1,9) = 0.47, p = 0.51, η2 = 0.05; Table 1).
Inhibition during delivery of the small, safe reward
Optogenetic inhibition of D2R-expressing neurons in the NAcSh during delivery of the small, safe reward (n = 6, male; n = 5, female) caused a significant decrease in choice of the large, risky reward (Fig. 10A). Although there was no main effect of session (F(1,9) = 1.27, p = 0.07, η2 = 0.32), there was a significant interaction between session and trial block (F(2,18) = 9.35, p < 0.01, η2 = 0.51). This effect did not differ between males and females (session × sex, F(1,9) = 0.18, p = 0.69, η2 = 0.02; session × sex × trial block, F(2,18) = 0.08, p = 0.93, η2 < 0.01). Inspection of Figure 10A shows that the significant interaction between session and trial block was mainly driven by an inhibition-induced decrease in choice of the large risky reward in the 25% trial block. Subsequent comparisons between baseline and optogenetic conditions for each block separately (correcting for multiple comparisons) confirmed that inhibition decreased risk taking only in the second block of trials (Blocks 1 and 3: session, p-values > 0.05, η2s < 0.20; Block 2: session: F(1,9) = 9.55, p = 0.01, η2 = 0.52). Analysis of the proportion of win-stay and lose-shift trials revealed no main effect of session (F(1,6) = 2.14, p = 0.19, η2 = 0.26) and no significant session × sex (F(1,6) = 5.12, p = 0.07, η2 = 0.46) or session × trial type (F(1,6) = 0.20, p = 0.67, η2 = 0.03) interactions. Despite these null effects, there was a significant session × sex × trial type interaction (F(1,6) = 8.75, p = 0.03, η2 = 0.59), suggesting that inhibition may have altered feedback processing differently between males and females. Sample sizes for each sex, however, were not sufficiently powered to perform post hoc analyses within each sex separately to identify the source of this significant interaction (three males were not included in the analysis because of the lack of one or both of the trial types during the optogenetic session). Additional trial-by-trial analyses were subsequently conducted to determine the extent to which inhibition during the delivery of the small, safe reward affected subsequent choice. Trial types were categorized as “stay” or “shift,” with the former representing trials on which a rat chose the small, safe lever after receipt of a small reward (with optogenetic inhibition) and the latter representing trials on which a rat chose the large, risky lever after receipt of a small reward (also concurrent with inhibition). Because inhibition affected choice behavior specifically in the 25% risk block (Block 2), this trial-by-trial analysis was conducted only on data from this block of trials. A three-factor repeated-measures ANOVA (session × trial type × sex) showed that although there was no main effect of session (F(1,9) = 1.07, p = 0.33, η2 = 0.11) nor significant session × sex (F(1,9) = 0.77, p = 0. 40, η2 = 0.08) or session × trial type × sex (F(1,9) = 0.22, p = 0.65, η2 = 0.02) interactions, there was a significant session × trial type interaction (F(1,9) = 10.63, p = 0.01, η2 = 0.54; Fig. 10B). Post hoc t tests (collapsed across sex) subsequently compared the proportion of each trial type between baseline and inhibition conditions. These analyses revealed that, relative to baseline, inhibition significantly increased the proportion of “stay” trials (t(10) = −3.48, p < 0.01, d = 1.05), but significantly decreased the proportion of “shift” trials (t(10) = 3.34, p < 0.01. d = 1.01). These results indicate that when D2R-expressing neurons were inhibited during delivery of the small, safe reward, rats were more likely to continue to choose the lever associated with this reward rather than shifting to the lever associated with the more rewarding, yet riskier, option.
Optogenetic inhibition of D2R-expressing neurons in the NAcSh during delivery of the small, safe reward. A, Inhibition of D2R-expressing neurons in the NAcSh during the delivery of the small, safe reward decreased choice of the large, risky reward. Data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Block 1) is because the size of the SEM was smaller than the point displayed. B, Inhibition of D2R-expressing neurons during the delivery of the small, safe reward increased the percentage of “stay” trials and decreased the percentage of “shift” trials relative to baseline. Open circles represent data points for individual rats. Asterisks indicate a significant difference between baseline (no light delivery) and optogenetic inhibition (p < 0.05).
Surprisingly, the effect of inhibition during delivery of the small, safe reward on risk taking persisted beyond the optogenetic test session. When baseline performance was compared with performance 24 h after the optogenetic test session (collapsed across sex as there were no sex-dependent effects of inhibition on risk taking), there was a main effect of session (F(1,10) = 9.10, p = 0.01, η2 = 0.48), but no session × trial block interaction (F(2,20) = 1.11, p = 0.35, η2 = 0.10). When baseline performance was compared with performance 48 h after the optogenetic session, there was still evidence of the effects of inhibition on risk taking (session, F(1,10) = 2.98, p = 0.12, η2 = 0.23; session × trial block, F(2,20) = 11.43, p < 0.01, η2 = 0.53). Together with the effects observed during the inhibition session itself, these results reveal that activity of D2R-expressing neurons in the NAcSh is important for the evaluation of less rewarding, albeit safer, options. Consequently, activity of these neurons during this evaluation period will bias choice toward more rewarding, yet costly, options. Further, disruption of activity in these neurons during this evaluation period may alter the way in which rats learn and encode reward contingencies in the task and therefore lead to longer term effects on risk-taking behavior (i.e., performance during subsequent non-inhibition test sessions).
Comparison of latencies to press levers during free choice trials between baseline and the optogenetic test session yielded no main effect of session (F(1,7) = 0.36, p = 0.57, η2 = 0.05) nor a significant session × lever identity interaction (F(1,7) = 0.32, p = 0.59, η2 = 0.04). There was a significant interaction between session, lever identity and trial block (F(1,7) = 7.28, p = 0.03, η2 = 0.51), but additional post hoc analyses to determine the source of the significance did not reveal any main effects or significant interactions (all p-values > 0.05). Although females displayed overall longer latencies to press levers (F(1,7) = 6.00, p = 0.04, η2 = 0.46), there were no effects of inhibition on latencies to press levers in either males or females (session × sex, F(1,7) < 0.01, p = 0.98, η2 < 0.01; session × lever identity × sex, F(1,7) = 0.12, p = 0.74, η2 = 0.02; session × sex × trial block, F(1,7) = 2.30, p = 0.17, η2 = 0.25; session × lever identity × sex × trial block, F(1,7) = 0.04, p = 0.84, η2 < 0.01). Similar to latencies to press levers, analyses of latencies to collect food were limited to data from Blocks 2 and 3 of the optogenetic test session because of rats' near-exclusive preference for the large reward in Block 1. Inhibition during the delivery of the small, safe reward did not affect latencies to collect food (session, F(1,9) = 2.43, p = 0.15, η2 = 0.21; session × sex, F(1,9) = 0.05, p = 0.83, η2 < 0.01; session × trial block, F(1,9) = 1.88, p = 0.20, η2 = 0.17; session × sex × trial block, F(1,9) = 3.76, p = 0.08, η2 = 0.30). Finally, there was no effect of inhibition on the percentage of omitted free choice trials (session, F(1,9) < 0.01, p = 1.00, η2 < 0.01; session × sex, F(1,9) = 0.10, p = 0.76, η2 = 0.01; Table 1).
Inhibition during delivery of the large, unpunished reward
Optogenetic inhibition of D2R-expressing neurons during delivery of the large, unpunished reward (n = 5, male; n = 5, female) did not affect choice of the large, risky reward (session, F(1,8) = 0.17, p = 0.69, η2 = 0.02; session × sex, F(1,8) = 0.01, p = 0.92, η2 < 0.01; session × trial block, F(2,16) = 1.19, p = 0.33, p = 0.13; session × sex × trial block, F(2,16) = 0.15, p = 0.86, η2 = 0.02; Fig. 11A). Although females omitted significantly more free choice trials (F(1,8) = 6.68, p = 0.03, η2 = 0.46; Table 1), there was no effect of inhibition on omissions in males or females (session, F(1,8) < 0.01, p = 0.96, η2 < 0.01; session × sex, F(1,8) < 0.01, p = 0.97, η2 < 0.01). Hence, activity of D2R-expressing neurons in the NAcSh does not contribute to the evaluation of large, unpunished rewards during risk-taking behavior.
Optogenetic inhibition of D2R-expressing neurons in the NAcSh during other phases of the RDT. A, Inhibition of D2R-expressing neurons in the NAcSh during delivery of the large, unpunished reward did not affect choice of the large, risky reward. B, Inhibition of D2R-expressing neurons during delivery of the large, punished reward did not affect choice of the large, risky reward. C, Inhibition of D2R-expressing neurons during the intertrial interval did not affect choice of the large, risky reward. Data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Baseline Block 1) is because the size of the SEM was smaller than the point displayed.
Inhibition during delivery of the large, punished reward
Similarly, there were no effects of inhibition of D2R-expressing neurons during delivery of the large, punished reward (n = 6, male; n = 5, female) on choice of the large, risky reward (session, F(1,9) = 0.23, p = 0.64, η2 = 0.03; session × sex, F(1,9) < 0.01, p = 0.97, η2 < 0.01; session × trial block, F(2,18) = 1.30, p = 0.30, η2 = 0.13; session × sex × trial block, F(2,18) = 0.44, p = 0.65, η2 = 0.05; Fig. 11B). Consistent with previous results, females did omit significantly more free choice trials compared with males (F(1,9) = 7.90, p = 0.02, η2 = 0.47; Table 1), but inhibition did not affect omissions in either sex (F(1,9) = 1.29, p = 0.29, η2 = 0.13; session × sex, F(1,9) = 1.90, p = 0.20, η2 = 0.17). These results suggest that, similar to the lack of involvement in the evaluation of unpunished rewards, activity of D2R-expressing neurons in the NAcSh does not appear to contribute to the evaluation of larger rewards that are accompanied by punishment.
Inhibition during the intertrial interval
As in experiment 1, rats in experiment 2 (n = 4, male; n = 6, female) also underwent an additional optogenetic test session in which light was delivered during the ITI. Not surprisingly, there was no effect of inhibition during the ITI on choice of the large, risky reward (F(1,8) = 1.19, p = 0.31, η2 = 0.13; session × sex, F(1,8) = 0.18, p = 0.68, η2 = 0.02; session × trial block, F(2,16) = 2.12, p = 0.15, η2 = 0.21; session × sex × trial block, F(2,16) = 0.87, p = 0.44, η2 = 0.10; Fig. 11C). Additionally, there was no effect of inhibition during the ITI on the percentage of omitted free choice trials (session, F(1,8) < 0.01, p = 0.96, η2 < 0.01; session × sex, F(1,8) < 0.01, p = 0.98, η2 < 0.01; Table 1). These results confirm that the inhibition-induced changes in risk taking during deliberation and delivery of the small, safe reward are not attributable to nonspecific effects of eNpHR3.0 activation.
Light delivery to D2R-expressing neurons in the NAcSh during decision making in control rats
Similar to experiment 1, a separate cohort of transgenic rats (n = 8, male; n = 6, female) was used to confirm that the effects of inhibition of D2R-expressing neurons in the NAcSh were not because of light delivery alone. These rats were implanted with bilateral cannula in the NAcSh through which an AAV containing mCherry alone was infused and optic fibers were implanted. Like the eNpHR3.0 group, control rats were then trained in the RDT until they achieved stable baseline performance, after which optogenetic test sessions began. These sessions were limited to those in which inhibition of D2R-expressing neurons had significantly affected risk taking in the eNpHR3.0 group (i.e., deliberation and delivery of the small, safe reward). The order of these test sessions was counterbalanced across rats, and rats were required to re-acquire behavioral stability between successive optogenetic test sessions.
Light delivery during deliberation
In contrast to eNpHR3.0 rats, light delivery to D2R-expressing neurons in the NAcSh during deliberation (n = 8, male; n = 6, female) had no effect on choice of the large, risky reward in control rats (session, F(1,12) = 0.04, p = 0.85, η2 < 0.01; session × sex, F(1,12) = 0.72, p = 0.41, η2 = 0.06; session × trial block, F(2,24) = 0.14, p = 0.87, η2 = 0.01; session × sex × trial block, F(2,24) = 0.21, p = 0.81, η2 = 0.02; Fig. 12A). Light delivery did not affect latencies to nosepoke (session, F(1,12) = 1.95, p = 0.19; session × sex, F(1,12) = 0.08, p = 0.78; session × trial block, F(2,24) = 2.63, p = 0.09, η2 = 0.18; session × sex × trial block, F(2,24) = 0.89, p = 0.42, η2 = 0.07) nor did it alter latencies to press levers during free choice trials (session, F(1,8) = 0.46, p = 0.52, η2 = 0.06; session × sex, F(1,8) = 0.12, p = 0.74, η2 = 0.02; session × lever identity, F(1,8) = 0.28, p = 0.61, η2 = 0.03; session × sex × lever identity, F(1,8) = 0.36, p = 0.57, η2 = 0.04; session × lever identity × trial block, F(1,8) = 0.45, p = 0.52, η2 = 0.05; session × sex × lever identity × trial block, F(1,8) = 0.52, p = 0.49, η2 = 0.06). Finally, there were no effects of light delivery on the percentage of omitted free choice trials (session, F(1,12) = 1.86, p = 0.20, η2 = 0.13; session × sex, F(1,12) = 1.21, p = 0.29, η2 = 0.09; Table 1). Together, these data indicate that light delivery in control rats had no effect on risk-taking behavior.
Light delivery to D2R-expressing neurons in the NAcSh in control rats. A, Light delivery to the NAcSh of control rats during deliberation did not affect choice of the large, risky reward. B, Light delivery to the NAcSh of control rats during the delivery of the small, safe reward did not affect choice of the large, risky reward. Data are represented as mean ± standard error of the mean (SEM). Absence of error bars (e.g., Block 1) is because the size of the SEM was smaller than the point displayed.
As in experiment 1, additional analyses were conducted to compare effects of light delivery during deliberation between eNpHR3.0 (n = 11) and control rats (n = 14). Because there were no sex-dependent effects of light delivery on risk taking in either group of rats, data were collapsed across sexes. Although there was no main effect of vector group (F(1,23) = 0.20, p = 0.66, η2 < 0.01), there was a main effect of session (F(1,23) = 16.23, p < 0.01, η2 = 0.41) and significant session × vector group (F(1,23) = 13.24, p < 0.01, η2 = 0.36), session × trial block (F(2,46) = 8.05, p < 0.01, η2 = 0.26) and session × vector group × trial block (F(2,46) = 5.54, p < 0.01, η2 = 0.19) interactions. Hence, these results confirm that, relative to baseline performance, light delivery during deliberation increased risk taking only in eNpHR3.0 rats.
Light delivery during delivery of the small, safe reward
Because of attrition, only two males remained in the control group that received intra-NacSh light delivery during delivery of the small, safe reward. Given this small sample size and the fact that there were no sex-dependent effects of inhibition in eNpHR3.0 rats, data were collapsed across sexes (n = 2, male; n = 4, female), yielding a sample size of n = 6. Light delivery during delivery of the small, safe reward had no effect on choice of the large, risky reward (session, F(1,5) = 1.20, p = 0.23, η2 = 0.19; session × trial block, F(2,10) = 0.99, p = 0.41, η2 = 0.17; Fig. 12B). Light delivery did not alter latencies to press levers during free choice trials (session, F(1,3) = 0.04, p = 0.85, 0.01; session × lever identity, F(1,3) = 0.39, p = 0.58, η2 = 0.12; session × trial block, F(1,3) = 0.14, p = 0.74, η2 = 0.04; session × lever identity × trial block, F(1,3) = 0.02, p = 0.89, η2 < 0.01) nor did it affect latencies to collect the small food reward (session, F(1,4) = 3.59, p = 0.13, η2 = 0.47; session × trial block, F(1,4) = 0.02, p = 0.89, η2 < 0.01). There was also no effect of light delivery on the percentage of omitted free choice trials (t(10) = −0.03, p = 0.97, d = 0.01; Table 1).
A comparison of the effects of light delivery between eNpHR3.0 (n = 11) and control rats (n = 6; collapsed across sex) revealed that although there were no main effects of session (F(1,15) = 0.85, p = 0.37, η2 = 0.05) or vector group (F(1,15) = 1.41, p = 0.25, η2 = 0.09), there were significant session × vector group (F(1,15) = 4.23, p = 0.05, η2 = 0.22), session × trial block (F(2,30) = 3.31, p = 0.05, η2 = 0.18) and session × vector group × trial block (F(2,30) = 6.13, p < 0.01, η2 = 0.29) interactions. These results confirm that the effects of light delivery during delivery of the small, safe reward were specific to rats in the eNpHR3.0 group.
Discussion
By leveraging the temporal precision afforded by optogenetics, the current study dissected the roles of the BLA→NAcSh and neurons that selectively express D2Rs in the NAcSh in different phases of decision making involving risk of punishment. Our findings reveal sex-dependent engagement of the BLA→NAcSh in processing risk-related information and dissociable roles for NAcSh D2R-expressing neurons in guiding risk-taking behavior.
Roles of BLA→NAcSh in risk-based decision making
We previously demonstrated that the BLA is differentially recruited during distinct components of decision making involving risk of punishment (Orsini et al., 2017). Dissociable engagement of the BLA during risk taking may be because of recruitment of distinct BLA cell populations with divergent projections to downstream regions during different phases of the decision process (Fig. 13). For example, based on work showing a temporally-specific role for BLA→NAc in a probability discounting task involving risk of reward omission (Bercovici et al., 2018), BLA→NAcSh may be particularly important for evaluating punished rewards to provide negative feedback that guides future choices. Consistent with this hypothesis, BLA→NAcSh inhibition during delivery of large, punished rewards increased risk taking in the RDT. This effect was only observed in males, however, with inhibition failing to alter risk taking in females. Such an effect in males is consistent with increased risk taking in male rats when BLA inhibition occurred during the same decision making phase (Orsini et al., 2017) along with augmented activity in the BLA of male rats in response to punished, but not unpunished, rewards (Jean-Richard-Dit-Bressel et al., 2022). Our data are also congruent with previous work demonstrating roles for the BLA and NAc (as well as communication between these regions) in punishment-based instrumental behavior. For example, pharmacological inactivation of either the BLA or NAcSh increases lever pressing for food associated with punishment (Jean-Richard-Dit-Bressel and McNally, 2015; Piantadosi et al., 2017). Similarly, inhibition of projections from the BLA to NAc increases risky choice when such inhibition occurs following non-rewarded risky choices in the probability discounting task (Bercovici et al., 2018). Together with our results, these studies suggest the BLA→NAc is important for evaluating negative outcomes (either punishment or lost reward opportunity) to guide future choices toward safer or more certain rewards.
Conceptual schematic of the neural circuits and cell populations recruited during distinct phases of the decision process. During deliberation, there are two separate populations of neurons in the basolateral amygdala (BLA) that contribute to risk taking in a dissociable manner depending on their projections. Whereas projections from the BLA to the nucleus accumbens (NAcSh) promote risk aversion, projections from the BLA to areas of the prefrontal cortex (PFC) may promote greater risk taking (although this remains to be determined). After reaching the NAcSh, BLA input may drive risk aversion by directly modulating activity of dopamine D2 receptor (D2R)-expressing neurons, the result of which may be suppression of GABA release onto dopamine D1 receptor (D1R)-expressing neurons (Dobbs et al., 2016). During outcome evaluation, PFC-projecting neurons in the BLA are not recruited; in contrast, NAcSh-projecting neurons in the BLA convey information about large, risky rewards to guide future risk taking. Input from the BLA may drive risk aversion by directly modulating an as yet to be determined population of NAcSh neurons, candidates which include fast-spiking interneurons (FSIs). In contrast, D2R-expressing neurons in the NAcSh are recruited during the evaluation of smaller, safer rewards, the outcome of which is to bias choice toward riskier choices. Because BLA projections to the NAcSh are not recruited during this outcome evaluation phase, information about the small, safe reward is likely transmitted to D2R-expressing neurons from a distinct afferent brain region. An asterisk indicates that this circuit function is specific to males. Figure created with BioRender.com.
The observation that BLA→NAcSh inhibition during delivery of large, punished rewards had no effect in females suggests the neural mechanisms by which males and females evaluate punished rewards are different. Previous work shows that BLA activity is greater in females compared with males (Lebron-Milad et al., 2012; Blume et al., 2017, 2019; Keiser et al., 2017; Hodges et al., 2022), and a c-fos mapping study showed greater NAcSh activity and greater BLA-NAc functional connectivity in females compared with males when placed in contexts perceived as negative or aversive (Hodges et al., 2022). Hence, rather than indicating a lack of BLA→NAcSh involvement in evaluation of negative feedback during risk taking in females, the absence of effects of BLA→NAcSh inactivation in females may instead reflect greater overall activity in this circuit in females than males during this decision making phase. Consequently, optogenetic inhibition using our current parameters, which were based on prior work in which BLA→NAcSh inhibition effectively altered motivated behavior (Stuber et al., 2011), may not have been sufficient to disrupt this process in females. Alternatively, the absence of an effect in females could be because of a weaker BLA →NAcSh projection in females relative to males (although to our knowledge there is no evidence to support such an anatomic difference). It is also conceivable that the BLA, and therefore its projections to the NAcSh, may simply not be involved in this phase of decision making in females at all. Our previous work, which formed the basis for the hypotheses driving the current experiments, showed that inhibition of the BLA during delivery of the large, risky reward increased risk taking (Orsini et al., 2017), similar to inhibition of the BLA→NAcSh. Importantly, however, only male subjects were used in this experiment. Hence, this aspect of decision making in females may instead be governed by a distinct set of NAcSh afferents. Although a lack of involvement of the BLA in this phase of decision making in females could reasonably explain our results, it would be incongruent with numerous studies in rats and humans that have shown greater BLA activation in response to a potentially aversive stimulus in females compared with males (Lebron-Milad et al., 2012; Cover et al., 2014; Keiser et al., 2017; Blume et al., 2019; Hodges et al., 2022). Nevertheless, future studies are warranted to determine whether how activity in the BLA, regardless of its projections, contributes to evaluation of negative consequences in both males and females.
Contrary to our initial hypothesis, there was a modest though significant increase in risk taking following BLA→NAcSh inhibition during deliberation in both sexes. These results were surprising for several reasons. First, BLA glutamatergic neurons do not display discernable patterns of activity (e.g., increases or decreases) before choosing between punished and unpunished food rewards (Jean-Richard-Dit-Bressel et al., 2022). Second, optogenetic inhibition of BLA glutamatergic neurons during deliberation decreases, rather than increases, risk taking (Orsini et al., 2017). Finally, BLA→NAc inhibition during deliberation decreases risky choice when risk of reward omission is low (Bercovici et al., 2018). These prior results would predict that inhibition would decrease, rather than increase, risk taking, particularly in the 25% risk block. This discrepancy can be reconciled, however, after considering the target of inhibition and the structure of the decision process itself. In contrast to inhibition of all BLA glutamatergic neurons, which would silence numerous BLA efferent projections, inhibition of a specific BLA projection may result in distinct changes in behavior. More broadly, decision making is not a linear process; thus, information processed in one decision making phase may be integrated in other phases. In support of this, a recent study found that BLA activity selective for a naturally rewarding outcome can actually be observed before the initiation of the action leading to the outcome and is maintained until the outcome is delivered (Courtin et al., 2022). Accordingly, BLA→NAcSh inhibition during deliberation might prevent the integration of outcome information (from the previous evaluation phase) into the deliberative process, resulting in effects on risk taking similar to those observed when inhibition occurs during delivery of the large, punished reward. These findings also suggest the decrease in risk taking following BLA inhibition during deliberation (Orsini et al., 2017) may result from inhibition of BLA projections to downstream targets distinct from the NAcSh (Fig. 13).
Roles of NAcSh D2R-expressing neurons in risk-based decision making
Decision making involving risk of punishment appears to recruit D2R signaling in the NAcSh (Mitchell et al., 2014). This is supported by findings that systemic and intra-NAc D2R agonists attenuate risk taking in the RDT (Simon et al., 2011; Mitchell et al., 2014). Further evidence comes from studies reporting a robust relationship between lower levels of D2R mRNA and higher levels of risk taking (Simon et al., 2011; Mitchell et al., 2014). These studies are limited, however, by their inability to dissect the role of D2Rs in a temporally-specific manner. The current study circumvented this limitation by using optogenetics in a transgenic rat line that enabled selective manipulation of D2R-expressing neurons. Results of validation experiments (Fig. 7) supported the premise that optogenetic inhibition of these neurons is a tractable approach to mimic effects of D2R agonist administration, but in a more temporally-precise manner.
Inhibition of these neurons during deliberation caused an increase in risk taking, consistent with findings that optogenetic excitation of NAc D2R-expressing neurons during deliberation decreased risky choice (i.e., choice of a reward that varied in its magnitude and frequency of delivery; Zalocusky et al., 2016). Together, these results suggest activation of D2R-expressing neurons during deliberation biases choices away from large, risky rewards. In the current study, inhibition-induced increases in risk taking were accompanied by selective increases in the percentage of win-stay trials, indicating that increased risk taking was driven by augmented sensitivity to the larger, more rewarding outcome. In contrast, Zalocusky et al. (2016) attributed their effects to alterations in sensitivity to losses. This discrepancy may be because of the difference in the nature of the risks associated with the “risky” option: footshock in the current study versus reward omission in Zalocusky et al. (2016). Indeed, whereas D2R agonists reduce risk taking under risk of punishment, they increase risk taking under risk of reward omission (St Onge and Floresco, 2009; Simon et al., 2011). Hence, the contribution of these neurons to the integration of feedback information during deliberation may differ depending on the type of risk associated with the available options during decision making.
Effects of inhibition of D2R-expressing neurons during deliberation were similar to effects of BLA→NAcSh inhibition during this same task phase, suggesting BLA neurons may communicate directly with NAcSh D2R-expressing neurons during deliberation to modulate choice behavior (Fig. 13). Indeed, BLA projections to NAcSh provide direct input onto D2R-expressing medium spiny neurons (MSNs; MacAskill et al., 2014; Barrientos et al., 2018; Baimel et al., 2019; Zinsmaier et al., 2022), and D2R antagonists reduce excitatory responses in the NAc evoked by electrical stimulation of the BLA (Liang et al., 1991). Intriguingly, BLA-evoked activity in these same cells can be attenuated by dopamine release from the VTA, and this mechanism also appears to depend on D2Rs (Yim and Mogenson, 1988). Recent work has shown that activation of projections from the VTA to NAc (thus elevating DA levels in the NAc) renders rats insensitive to footshock punishment paired with reward delivery (Verharen et al., 2018), yielding behavioral effects reminiscent of the increase in choice of the large, punished reward following optogenetic inhibition of D2R-expressing neurons or BLA→NAcSh during deliberation. At the synaptic level, modulation of BLA-targeted NAcSh neurons by dopamine via activation of D2Rs could consequently lead to a suppression of GABA release from D2R-expressing neurons onto neighboring D1R-expressing neurons, thereby increasing the excitability of these MSNs (Dobbs et al., 2016). Future studies are therefore needed to determine the behavioral consequences of manipulating D1R-expressing MSNS in the NAcSh as they relate to choice behavior. Notably, recent work has shown that BLA projections to NAcSh predominantly synapse onto fast-spiking interneurons (FSIs), and activation of FSIs by BLA input precedes activation of MSNs (Yu et al., 2017). Further, activation of FSIs by optogenetic stimulation of BLA (and prefrontal) projections to the NAcSh results in feedforward inhibition of MSNs. Collectively, these findings suggest that rather than directly modulating D2R-expressing MSNs, BLA input could indirectly influence this population of MSNs via intermediary FSIs. Alternatively, similar effects of inhibition during deliberation across the BLA→NAcSh and D2R-expressing neurons could be coincidental, such that activity of D2R-expressing neurons is regulated by alternate afferent regions such as the ventral tegmental area (Gerfen et al., 1987). Future experiments will distinguish between these potential explanations for how D2R-expressing neurons are regulated during deliberation to guide subsequent risky choice.
Surprisingly, NAc D2R neuron inhibition during small, safe reward delivery decreased risk taking, suggesting that activity in these neurons during this decision making phase is important for biasing choice toward more rewarding options (Fig. 13). This interpretation is consistent with work showing optogenetic activation of NAc D2R-expressing MSNs enhances motivation for food (Soares-Cunha et al., 2016, 2018). If changes in food motivation were the sole driver of the inhibition-induced reduction in risk taking, however, then inhibition of D2R-expressing neurons during delivery of large rewards should have also decreased risk taking, which was not the case. Alternatively, activity in D2R-expressing neurons during evaluation of small, safe rewards may be important for balancing the relative reward of the alternate option against its associated risk. When the risk is low (e.g., 25% trial block), the risk is perceived as minimal and therefore it may be more adaptive to shift choice toward the larger reward. When these neurons are inhibited, however, the risk overshadows the value of the larger reward, resulting in continued choice of the safer, albeit less rewarding, option.
Additional considerations
Collectively, these data reveal temporally specific roles for activity in the BLA→NAcSh circuit and D2R-expressing neurons in the NAcSh in decision making under risk of punishment. It is possible, however, that optogenetic inhibition altered other aspects of behavior, the effects of which could be challenging to differentiate from changes in risk taking. For example, rather than increased risk taking, the effects of optogenetic inhibition of BLA→NAcSh during delivery of the large, risky reward could instead be a result of inhibition-induced augmentation of motivation to work for the large reward. This interpretation can be discounted, however, by the fact that there were no effects of inhibition on choice behavior when inhibition occurred during delivery of the large reward in the absence of punishment. Further, inhibition during any decision making phase had no effect on choice behavior in the 0% risk block. Hence, it is very unlikely that elevated risk taking following optogenetic inhibition of the BLA→NAcSh during the delivery of the large, punished reward is driven by greater reward motivation. Alternatively, this increase in risk taking could be a consequence of inhibition-induced impairments in behavioral flexibility, leading to perseverative choice of the large reward. The outcome-specific effects of inhibition on choice behavior argue against this interpretation. If inhibition of the BLA→NAcSh were to lead to broad behavioral perseveration (i.e., inflexibility), one would expect to observe an increase in risk taking for every reward outcome session in which this circuit was inhibited. As the data show, however, inhibition only increased risk taking when inhibition occurred during delivery of the large, risky reward.
With respect to the increased risk taking following inhibition of either the BLA→NAcSh or D2R-expressing neurons during deliberation, it is more challenging to dissociate the effects of inhibition on flexibility versus risk taking. Although this does not negate the argument for impaired flexibility, the fact that rats still demonstrated discounting of the large reward as risk of punishment increased (as evidenced by a main effect of trial block) in all optogenetic test sessions suggests that the ability to flexibly adjust choice behavior was at least partially intact. Nevertheless, experimental design of future studies will be refined to more readily distinguish between these different interpretations. For example, including an additional group of rats that are trained and tested in a version of the RDT in which the probability of punishment is reversed (i.e., 75%, 25%, 0%) would enable dissociation between alterations in risk taking and impairments in flexibility because effects of inhibition could be directly compared between the descending and ascending versions (Orsini et al., 2018; Bercovici et al., 2023).
In conclusion, these findings advance our understanding of the neural substrates governing risk-based decision making by providing evidence that circuits and selective cell populations involved in this process can be recruited in a temporally-specific and/or sex-dependent manner. These results provide insight into the neural mechanisms by which risk-based decision making may become compromised in psychiatric diseases and suggest approaches to mitigate these cognitive impairments.
Footnotes
This work was supported by the National Institutes of Health Grant F31DA057112-01 and a Bruce/Jones Graduate Fellowship (L.M.T.), the National Institutes of Health Grant R00DA041493 (to C.A.O.), the National Institutes of Health Grant RF1AG060778 (to J.L.B., B.S., and C.J.F.), and the National Institutes of Health Grant R03DA050118 (to M.S.). We thank Ms. Bonnie McLaurin and Ms. Merrick Garner for their technical assistance.
The authors declare no competing financial interests.
- Correspondence should be addressed to Caitlin A. Orsini at caitlin.orsini{at}austin.utexas.edu
