Abstract
Foraging decisions involve assessing potential risks and prioritizing food sources, which can be challenging when confronted with changing and conflicting circumstances. A crucial aspect of this decision-making process is the ability to actively overcome defensive reactions to threats and focus on achieving specific goals. The ventral pallidum (VP) and basolateral amygdala (BLA) are two brain regions that play key roles in regulating behavior motivated by either rewards or threats. However, it is unclear whether these regions are necessary in decision-making processes involving competing motivational drives during conflict. Our aim was to investigate the requirements of the VP and BLA for foraging choices in conflicts involving overcoming defensive responses. Here, we used a novel foraging task and pharmacological techniques to inactivate either the VP or BLA or to disconnect these brain regions before conducting a conflict test in male rats. Our findings showed that BLA is necessary for making risky choices during conflicts, whereas VP is necessary for invigorating the drive to obtain food, regardless of the presence of conflict. Importantly, our research revealed that the connection between VP and BLA is critical in controlling risky food-seeking choices during conflict situations. This study provides a new perspective on the collaborative function of VP and BLA in driving behavior, aimed at achieving goals in the face of dangers.
Significance Statement
While searching for food, animals must also be cautious about dangers. This study aimed to investigate how the two subcortical brain regions, the ventral pallidum (VP) and basolateral amygdala (BLA), help individuals make optimal decisions in such conflicting situations. Using a novel task and drugs to silence these regions, we observed how rats made choices. Our findings show that the VP drives the effort required to obtain food, whereas the BLA triggers caution around dangers. Notably, we revealed that cooperation between these regions prevents animals from taking excessive risks when foraging food involves facing threats. Gaining an understanding of how the brain prioritizes caution over reward could aid in the treatment of emotional disorders.
Introduction
Decisions that have life or death consequences demand a rapid weighing of obstacles and opportunities to take appropriate action in continually changing environments (Rangel et al., 2008; Headley et al., 2019; Mobbs et al., 2020). The process of transitioning from secure food acquisition to risk-prone food seeking presents an intriguing challenge in survival decision-making, as it requires actively overcoming defensive behaviors to pursue specific goals. When searching for food, foraging animals often face dangers in nature, which illustrates their ability to overcome defensive reactions to achieve their goal. A novel behavioral tool, the crossing-mediated conflict (CMC) task, was recently developed to investigate this choice strategy in rats (Illescas-Huerta et al., 2021). This novel seminaturalistic task, based on a large literature of approach–avoidance conflict (Elliot, 2008), allows for the assessment of foraging choices that involve rapid transitions from effortful safe food acquisition to food seeking despite threats within dynamic and unpredictable conditions. In the CMC test, individuals are presented with a foraging challenge that requires them to select an approach based on prior associations they have learned. The test involves crossing an alley to obtain food, and rats must decide whether to cross to obtain food safely or confront their fear and cross despite the danger involved. The relatively understudied ability in this task highlights the necessity of researching the neural processes that govern decisions when there is a struggle between rewards and threats.
The ventral pallidum (VP) and basolateral amygdala (BLA) are critical neural substrates involved in orchestrating motivated behaviors driven by rewards and threats (Smith et al., 2009; Moscarello and LeDoux, 2013; Janak and Tye, 2015; Root et al., 2015). The VP has traditionally been thought of as the motor output of the basal ganglia, responsible for guiding reward-seeking behaviors. However, recent research has called this understanding into question. Emerging research has demonstrated that VP neuronal activity exhibits dynamic responsiveness to relative threats (Moaddab et al., 2021) and mediates avoidance behaviors (Saga et al., 2017; Faget et al., 2018; Stephenson-Jones et al., 2020), positioning VP uniquely to detect and process threat-related information. Additionally, VP has been implicated in encoding expected reward values that influence incentive motivation for choice behaviors (Richard et al., 2016, 2018; Fujimoto et al., 2019; Ottenheimer et al., 2020b). Similarly, the conventional notion of BLA in threat-related memory expression has changed over time (Pare and Quirk, 2017), and accumulated evidence has shown that BLA neurons also respond to reward cues (Beyeler et al., 2016). This dual function of VP and BLA highlights their potential roles in decision-making processes shaped by opposing motivational drives. Consistent with this notion, studies employing food-seeking behavior or threat-related responses have shown that VP inactivation reduces drive to seek rewards, even with potential punishment (Farrar et al., 2008; Farrell et al., 2021; Lederman et al., 2021), whereas BLA inactivation increases the drive to approach threat-associated cues, potentially influencing riskier foraging decisions (Choi and Kim, 2010; C. Bravo-Rivera et al., 2014; Burgos-Robles et al., 2017; Orsini et al., 2017). Despite these advancements, the extent to which VP and BLA interact and influence foraging choices when competition occurs remains unknown.
Here, we investigated the individual and collaborative necessity of VP and BLA to influence foraging decisions in conflict situations. Specifically, we aimed to study how these brain regions contribute when animals choose between making minimal effort to safely obtain food and engaging in more demanding actions that involve overcoming defensive behaviors. To investigate this, we pharmacologically inactivated the VP and BLA or disconnected these brain regions before the CMC test. Our findings reveal that BLA is necessary for food-seeking decisions that involve taking risks during conflicts, whereas VP primarily regulates decisions related to the motivational drive to seek food, irrespective of the presence of conflict. Notably, our research underscores the importance of connectivity between VP and BLA in restraining risk behaviors associated with actions to seek food while facing danger.
Materials and Methods
A total of 181 adult male Wistar rats (Institute of Cell Physiology breeding colony), aged 2–3 months and weighing 250–300 g, were individually housed in polyethylene cages. Daily handling aimed at minimizing stress responses, and the rats adhered to a standard 12 h light/dark schedule. Ten rats were excluded from the data analysis due to CMC task learning difficulties, while thirty-four were excluded based on histological findings or insufficient postsurgery recovery. All the experiments were performed during the light phase. To maintain consistent motivation for food-related tasks, the rats underwent food restriction (12 g/d of standard laboratory rat chow, with a 5 g bonus feeding weekly to sustain rats at 85% of their initial weight). The procedures were approved by the Institutional Animal Care and Use Committee of the Universidad Nacional Autónoma de México following the guidelines of the National Ministry of Health for Laboratory Animal Care.
CMC task
The task involved training and testing in straight alleys (100 × 30 × 50 cm) with stainless steel frames situated in a sound-attenuating chamber (150 × 70 × 140 cm). These alleys comprise two safety zones and one danger zone (Fig. 1A). The danger zone (60 cm × 30 cm) had a stainless steel bar floor capable of delivering a scramble footshock (Coulbourn Instruments). The safety zones (20 cm × 30 cm) included a speaker, cue light, lever linked to a pellet dispenser, food plate, and acrylic floor. Each safety zone was outfitted with a lever, which enabled rats to press it and receive a food sucrose pellet as a reward (each lever press was reinforced with one pellet). The reward amount remained consistent across both conflict and non-conflict trials, ensuring that this factor remained constant throughout the experimental design. A computer running MATLAB (MathWorks) custom scripts managed pellet delivery, cue lights, speakers, shock delivery, and recorded lever presses. Between the experiments, the alley's floors and walls were cleaned with soapy water and 70% alcohol and dried with paper towels. Before conflict training, all rats were conditioned to associate lever pressing with sucrose pellet reinforcement (45 mg dustless precision pellets; Bio-Serv). Each session began and concluded with a 5 min context-alone exposure devoid of cue lights, shocks, or noise.
CMC task to study the act of overcoming threats for food. A, Rats were trained and tested in a grid-divided alley foraging between feeding zones, based on cues denoting either conflict or non-conflict situations. B, Training involved three consecutive stages: acquisition of reward learning through a light cue in a safety zone, acquisition of threat learning via a white noise cue in a danger zone, and discrimination between cue-guided crossing trials involving conflict (simultaneous light and noise) and those without conflict (light only). Subsequent testing assessed memory-guided decision-making during both conflict and non-conflict trials. C, Representative rat data indicate the prolonged latency in crossing and lever pressing on the opposite side of the alley during conflict trials (purple) compared with that during non-conflict trials (green) within the same session. D, Example ethogram, movement tracking, and heat map during the first two consecutive non-conflict and conflict trials showing the relative timing of cues (food and threat), crossing, reward-seeking behavior (lever press), and threat-related responses (freezing and risk assessment). E, Analysis of the time spent per zone shows that rats spent more time in start and choice zones but less time in danger and goal zones during conflict compared with those during non-conflict trials. F, Analysis of behavioral events per zone shows increased defensive responses and crossing latency during conflict compared with those during non-conflict trials. **p < 0.01 and ***p < 0.001 in a Student's t test (paired, two-tailed). The error bars in these and subsequent figures represent the SEM, whereas each individual rat is represented by small empty circles.
Conflict training
Rats underwent training for the CMC task as previously described (Illescas-Huerta et al., 2021). Briefly, this training encompassed three stages, reward training, threat training, and discrimination training, to differentiate between non-conflict and conflict trials. The entire conflict training lasted for 34 d and consisted of one training session per day.
Reward training
Rats confined to a safety zone were trained to anticipate food availability with the presentation of a light cue. When the cue light was illuminated, a pellet was delivered to the food plate with each lever pressed (light trial); however, no pellet was delivered in the absence of light (no-light trial). Light trials concluded when a specific number of rewarded lever presses were achieved, which ranged between 5 and 20 presses. On the other hand, no-light trials ended after a predetermined duration of time had passed, which varied between 30 and 180 s. Consequently, light trials required lever pressing, whereas no-light trials did not require any action to be taken. After rats showed an increased and stable lever press during light trials, but not during no-light trials, they were trained to cross from one safety zone to the other safety zone to obtain food cued by light. A reward-crossing trial started when the light turned on, on the side of the alley opposite the location of the rat. The trial ended when the light turned off after either the rat had crossed to that safety zone and pressed the lever or after 180 s without crossing. To encourage goal-directed behavior, rats received one to three reinforced lever presses in the safety zone before each crossing trial, promoting response to the light cue rather than habitual reactions. Up to this point, rats were required to acquire food by crossing to the safety zone in the alley where the light was illuminated and pressing the lever, without encountering conflict.
Threat training
Rats confined to the danger zone experienced five pairings per session [variable intertrial interval (ITI) of 1–3 min] of white noise (85 dB, 30 s) coupled with a mild footshock (0.5 mA, 1 s). Shocks were administered throughout the 30 s noise presentation in a randomized manner (dependent on the variable ITIs), with the aim of preventing the anticipation of the delivery time. The next day, two noise-alone trials in the absence of shocks were presented to test cued threat memory. Next, rats were challenged to cross the alley in the same way as in reward-crossing trials, but in the presence of a threat consisting of noise signaling a footshock of 0.5 mA in the middle of the alley (variable ITIs ranged from 1 to 3 min). The duration of the threat-crossing trials was extended on a gradual basis during the first 5 d of training. On the initial 2 d, the trials lasted for 30 s, while on the third day, they extended to 60 s. On the fourth day, the duration was increased to 90 s, and finally, on the fifth day, the trials lasted for 120 s. It is important to note that the shocks were present throughout the entire trial, regardless of whether the rat successfully crossed over to the other side of the alley or not. Each trial ended either after the time of crossing had elapsed (30–120 s) or when the rats successfully crossed and pressed the lever on the opposite side of the alley. By this point, the rats had learned to cross to obtain food despite threats indicated by the concurrent light and footshock signaled by sound cues during the conflict.
Discrimination training
Rats were trained to discriminate between non-conflict (crossing for food without threat) and conflict (crossing for food despite threat) trials. Short acrylic barriers (9 cm tall) were strategically placed between each safety zone and the grid, defining a choice zone to constrain choice-mediated crossing behavior in both time and space. Each discrimination session comprised three consecutive blocks of 10 crossing trials, with 7 or 9 non-conflict trials and 1 or 3 conflict trials per block (10 and 30% chance of threat trials, respectively). Both types of crossing trials started once the cues were activated on the opposite side of the rat's position. The trials concluded when the rat had crossed the safety zone and pressed the lever or after 180 s elapsed without a crossing. Upon conclusion of the trial, the cues were deactivated, signaling the start of a new trial. The trial types were presented randomly to prevent anticipation. The rats had limited time (180 s) to decide whether to cross the alley in each trial. The proportion of conflict trials and shock intensity increased during the training period. After this stage, the rats underwent a cannula implantation surgery. Upon full recovery, they underwent four additional discrimination sessions, each with three blocks containing seven non-conflict and three conflict trials. Shock intensity increased by 0.1 mA per day, reaching 0.8 mA. To focus on learning and decision-making, we included only the rats that consistently distinguished between trial types by the end of training and completed the task within the allotted time. Rats failing to discriminate trial types at the end of training (evaluated as p value greater than 0.05 in the comparison between the average of conflict vs non-conflict trials in the last block of trials) and those unable to cross within 180 s were excluded from the study.
Conflict test
The conflict test was conducted on the day following the additional discrimination sessions. Rats underwent 10 crossing trials presented under the same conditions as the last day of additional discrimination training (30% chance of threat: three conflict and seven non-conflict randomly presented trials) but without shocks. The test involving conflict was executed in the absence of any aversive consequences associated with shocks, guaranteeing that our manipulations of neural processes were aimed at approach–avoidance decision-making guided by threat memory specifically, thereby precluding any consequences stemming from shock sensitivity. Results were expressed by contrasting the average temporal and spatial behavioral patterns, encompassing cues, crossing, reward-seeking actions, and threat responses between the control and experimental groups during both conflict and non-conflict trials.
Separate behavioral tests
Threat memory test
Training and testing were conducted using conditioning chambers (Coulbourn Instruments). On the first day, threat conditioning involved five trials with a tone (75 dB, 30 s) coterminated with a footshock (0.7 mA, 1 s) delivered through a stainless steel grid floor. On the second day, the threat memory test consisted of two trials with tone presentations in the absence of a footshock. Trials started after 5 min without stimuli and followed a variable ITI of 120 s.
Free food intake test
The test was conducted in standard home cages, where rats were provided with a 20 g plate of Bio-Serv food (45 mg, dustless precision pellets) for 30 min. This test represents an independent experiment aimed at dissociating instrumental responses from the motivation to seek and consume the food reward.
Open-field test
The test involved individually placing the rats in a dimly lit (20 lx) arena and recording their behavior for 10 min. Locomotion was assessed based on the distance traveled, while anxiety-like behavior was evaluated based on the center time in the arena.
Stereotaxic surgery
Rats were anesthetized with isoflurane inhalant gas and positioned in a stereotaxic apparatus. Anesthesia maintenance (2–3% isoflurane) was facilitated using a facemask. For pharmacological inactivation, bilateral implantation of 23 gauge stainless steel guide cannulas (15 mm hypodermic stainless steel; A-M Systems) occurred in the BLA (−2.8 mm AP; ±5.0 mm ML; −7.5 mm DV, from the bregma) and VP (+0.4 mm, AP; ±2.1 mm, −7.5 DV, from the bregma). In the disconnection experiments, guide cannulas were placed contralaterally or ipsilaterally in the BLA and VP, with the hemisphere side counterbalanced. Cannula tips were aimed at 1.0 mm above the target structure, secured with dental acrylic cement, and anchored with two surgical screws on the skull. Rats were allowed a minimum of 7 d for postsurgery recovery before initiating behavioral training.
Local infusions
The day before infusion, the rats were habituated to manipulation to minimize stress, with injectors briefly inserted into the cannulas without infusion. The injector tips were extended by 1 mm beyond the guide cannula. On the test day for each behavioral task, a between-subject design was employed with separate groups of rats receiving muscimol and baclofen (Sigma-Aldrich) GABA receptor agonists, or muscimol alone was used to inactivate target regions (INACT) and saline solution as vehicle (VEH). Muscimol and baclofen (125 ng of each drug/0.5 µl per side; as in Ghods-Sharifi et al., 2009) or VEH was infused into BLA. Muscimol (10 ng/0.5 µl per side; as in Farrar et al., 2008) or VEH was infused into the VP. Infusions were performed 15 min before behavioral testing. The rate of infusion was 0.4 µl/min controlled by a microinfusion pump (KD Scientific) operating 10 µl syringes (Hamilton) connected to the cannulas via polyethylene tubing. The injectors were left in place for 1 min after infusion to allow diffusion.
Histology
After the behavioral experiments, the rats were deeply anesthetized with pentobarbital sodium (150 mg/kg, i.p.) and transcardially perfused with 0.9% saline. Brains were stored in 10% formalin solution (Sigma-Aldrich) for at least 3 d. Formalin was then substituted with 30% sucrose solution until tissue saturation was reached. Brains were cut into 40-μm-thick coronal sections using a cryostat (Leica CM1520), stained for Nissl bodies, and examined under a bright-field microscope (Nikon, H550S) to verify cannula tip placement. Only rats with cannula locations within the borders of the BLA and VP were included in statistical analysis.
Data collection and analysis
All behaviors were recorded using digital video cameras (Provision, model D-380D5) located above each task apparatus. Conflict test video images were analyzed using the video tracking software ANY-maze 7.1 (Stoelting). Temporal and spatial behavioral patterns, including cues, crossing, reward-seeking actions, and responses to potential threats, were assessed by segmenting the 100 cm alley into four tracking zones: start (19 cm), choice (10 cm), danger (42 cm), and goal (29 cm). The start zone marked the safety zone where rats initiated crossing trials, whereas the choice zone represented the area for risk assessment before crossing, encompassing acrylic barriers and the initial three stainless steel bars of the grid. The danger zone comprised the grid where the rats crossed, and the goal zone denoted the safety zone where the rats concluded the crossing trial by pressing a lever. Rat tracking at three body points (head, center, and tail) provided data on the time spent and entries per zone. The center point of the rat was used to determine its presence in the start, danger, and goal zones, whereas the head point was used to assess its presence within the choice zone (with the head point required to be within 10 mm of the zone for a successful visit). The percentage of the time spent per zone was calculated as follows: (time spent per zone × 100) / (total time spent in the start, danger, and goal zones). Additionally, freezing bouts (number of times the rats spent at least 300 ms immobile in the start zone), latency to press (seconds to press a lever on the same side), risk-assessment events (number of entries into the choice zone characterized by approach/avoidance oscillations toward the reward site), and crossing speed [the rate (m/s) at which the rat crosses the danger zone] were analyzed. Finally, we calculated the latency to press in the goal zone, which represented the combined time spent in the start, danger, and goal zones. This measurement determined the reaction time required to complete the crossing from the start zone to the goal zone.
Movement tracking and heat maps were obtained from the rat head position. A maximum of 10 s was used as the hottest color in the heat map. All data, movement tracking, and heat maps were obtained for both the conflict and non-conflict crossing trials. Freezing responses were expressed as the percentage of the time spent without movement (300 ms immobile, except for respiration) during the auditory cue presentation which were automatically calculated using the tracking software (ANY-maze 7.1, Stoelting). For the open-field test, the distance traveled (m) and center time in the open field were calculated automatically using a tracking software. For the free food intake test, food intake was obtained by comparing the food plate weight (g) before and 30 min after food presentation. The percentage of food intake was calculated as follows: (total food intake × 100) / (total amount of food presented). Food intake was manually recorded and scored by an experimenter blinded to the experimental conditions. Experimental groups were compared by using, when appropriate, unpaired Student's two-tailed t tests or analysis of variance (two-way ANOVA) followed by planned comparisons or Tukey's multiple-comparison post hoc test (Statistica, StatSoft, and GraphPad Prism 7). The accepted value for significance was set at p < 0.05.
Results
Rats were trained in the CMC task to discern between safe (non-conflict) and risky (conflict) crossings in order to obtain food in the straight alley apparatus depicted in Figure 1A. The conflict test included non-conflict and conflict trials without shocks (Fig. 1B). By not using shocks, we ensured that our experiments focused on approach–avoidance decision-making guided by threat memory. This eliminated any confounding factors related to shock sensitivity.
Risk-taking is incentivized by food-seeking behavior
While approach–avoidance conflict behaviors are well documented in animal models and human studies (Miller, 1944; Elliot, 2008; Choi and Kim, 2010; C. Bravo-Rivera et al., 2014; Amir et al., 2015; Aupperle et al., 2015; Friedman et al., 2015; Burgos-Robles et al., 2017; Hamel et al., 2017; Piantadosi et al., 2017; Kyriazi et al., 2018, 2020; Walters et al., 2019; Engelke et al., 2021; H. Bravo-Rivera et al., 2021; Herzallah et al., 2022), the specific factors influencing risk-taking behavior during reward seeking remain unclear. To address this knowledge gap, we used a group of unimplanted rats (n = 7) to examine the range of cue-triggered behaviors that emerged during the conflict test of the CMC task. Figure 1C displays data from a representative rat, showing the extended reaction times (latency from start to goal zone) for crossing to lever press in conflict trials compared with those in non-conflict trials (Student's two-tailed paired t test, t(2) = 16.65; p = 0.003). Figure 1D further reveals the temporal (top) and spatial (bottom) patterns associated with cues, crossing, reward-seeking actions, and responses to potential threats in the same representative rat. Notably, when food-seeking actions coincided with potential threats (conflict trials), a cautions decision-making process emerged, involving both passive (freezing) and active defensive (risk-assessment) behaviors, in contrast to non-conflict trials.
Both non-conflict and conflict trials required crossing to access food, with conflict trials involving added complexity, as rats appeared to evaluate their decision to cross despite potential danger. Our analysis of time allocation and behavioral events within delimited zones (start, choice, danger, and goal zones) during both conflict and non-conflict trials revealed distinct temporal and spatial profiles (Fig. 1E). In conflict trials, rats allocated more time to the start and choice zones (start zone, Student's two-tailed unpaired t test, t(6) = 6.96; p < 0.000; choice zone, Student's two-tailed unpaired t test, t(6) = 5.58; p = 0.001) while spending less time in the danger and goal zones (danger zone, Student's two-tailed unpaired t test, t(6) = 9.97; p < 0.000; goal zone, Student's two-tailed unpaired t test, t(6) = 5.95; p = 0.001), suggesting a cautious approach in the face of threats.
Notably, when reaching the choice zone, rodents exhibit characteristic oscillatory conflict behaviors, including hesitantly moving back and forth (“indecision”), head dips, and stretched postures, indicative of risk-assessment behaviors. A comparison of behavioral events showed marked differences between conflict and non-conflict trials (Fig. 1F). Conflict trials revealed the emergence of freezing events in the start zone (Student's two-tailed unpaired t test, t(6) = −6.50; p < 0.000) and the appearance of risk-assessment events in the choice zone (conflict trials, 4.28 events; non-conflict trials, 0.08 events; Student's two-tailed unpaired t test, t(6) = −13.62; p < 0.000). Rats demonstrated faster crossing of the danger zone (Student's two-tailed unpaired t test, t(6) = −10.72; p < 0.000) and longer latency in pressing the lever in the goal zone (Student's two-tailed unpaired t test, t(6) = 4.28; p = 0.005) during conflict trials. In summary, our analysis emphasizes the significant temporal investment required to overcome defensive responses in favor of food-seeking actions during conflict trials, representing a cautious risk-taking strategy when confronted with challenging and potentially threatening situations. Thus, choosing food-seeking actions over defensive reactions involves evaluation of risk during conflict.
BLA restrains food-seeking decisions when faced with threats
To examine the role of the BLA in decision-making between seeking food and defensive reactions in response to threats, we employed pharmacological inactivation (n = 8) or vehicle treatment (n = 5) of this limbic structure before subjecting the rats to a cued memory conflict test (Fig. 2A). Our findings revealed a significant impact of BLA inactivation on the temporal aspects and behavioral events during conflict trials, while no discernible effect was observed during non-conflict trials (Fig. 2B–D). BLA inactivation during conflict trials rendered all behavioral measures similar to those observed in non-conflict trials. Specifically, during conflict trials, BLA inactivation reduced the time spent in the start zone, aligning with the durations observed in non-conflict trials (factorial ANOVA, group, F(1,22) = 33.98; p = 0.000; trials, F(1,22) = 16.71; p = 0.000; interaction, F(1,22) = 8.71; p = 0.007; conflict trials VEH vs INACT post hoc comparison, p = 0.000; conflict trials INACT vs non-conflict trials VEH post hoc comparison, p = 0.61; conflict trials INACT vs non-conflict trials INACT post hoc comparison, p = 0.79). This finding suggests that BLA plays a pivotal role in delaying food-seeking decisions when faced with threats, as reflected by the decrease in time spent in the start zone. Moreover, BLA inactivation significantly diminished defensive responses, including freezing in the start zone and risk assessment in the choice zone (freezing bouts, factorial ANOVA, group, F(1,22) = 10.57; p = 0.003; trials, F(1,22) = 14.45; p = 0.000; interaction, F(1,22) = 10.65; p = 0.003; conflict trials VEH vs INACT, post hoc comparison, p = 0.000; risk assessment, factorial ANOVA, group, F(1,22) = 9.03; p = 0.006; trials, F(1,22) = 13.55; p = 0.001; interaction, F(1,22) = 8.93; p = 0.006; conflict trials VEH vs INACT post hoc comparison, p = 0.001). Notably, BLA inactivation also resulted in increased time spent within the danger zone and slower crossing speed (danger zone, factorial ANOVA, group, F(1,22) = 15.42; p = 0.000; trials, F(1,22) = 9.00; p = 0.006; interaction, F(1,22) = 12.39; p = 0.001; conflict trials VEH vs INACT post hoc comparison, p = 0.000; crossing speed, ANOVA, group, F(1,22) = 2.48; p = 0.12; trials, F(1,22) = 12.74; p = 0.001; interaction, F(1,22) = 19.09; p = 0.000; conflict trials VEH vs INACT post hoc comparison, p = 0.002), indicating that the BLA plays a vital role in promoting the rapid crossing of this dangerous area to achieve a goal. Additionally, rats with BLA inactivation exhibited an extended duration of the time spent in the goal zone but a shorter time to press for food (goal zone, factorial ANOVA, group, F(1,22) = 29.46; p = 0.000; trials, F(1,22) = 14.03; p = 0.001; interaction, F(1,22) = 5.61; p = 0.02; conflict trials VEH vs INACT post hoc comparison, p = 0.000; latency to press in the goal zone, factorial ANOVA, group, F(1,22) = 17.82; p = 0.000; trials, F(1,22) = 19.91; p = 0.001; interaction, F(1,22) = 18.11; p = 0.000; conflict trials VEH vs INACT post hoc comparison, p = 0.000). After the inactivation of the BLA, the decrease in the latency to cross was comparable to that previously observed using the anxiolytic drug diazepam (Illescas-Huerta et al., 2021).
BLA inactivation unmasks reward approach behaviors by impairing threat memory retrieval during conflict. A, Rats were infused with VEH or a cocktail of muscimol and baclofen (INACT) into the BLA before the conflict test. B, Overlaid movement tracking and heat maps of representative rats (VEH and INACT) during the first two consecutive non-conflict and conflict trials of the conflict test. C, Analysis of the time spent per zone showed that BLA inactivation decreased the time spent in the start safety zone while increasing it in the danger and goal zones during conflict trials without affecting non-conflict trials. D, Analysis of behavioral events per zone shows that BLA inactivation impaired defensive responses (freezing and risk assessment) in the start zone, slowed crossing speed in the danger zone, and decreased crossing latency to press in the goal zone during conflict trials without affecting non-conflict trials. BLA inactivation did not affect the latency to press on the same side of the alley (inset). E, BLA inactivation impaired the retrieval of auditory threat conditioning memory, as indicated by low levels of freezing in the inactivation group compared with those in the vehicle group. F, BLA inactivation did not affect anxiety-like behavior and general locomotion in the open-field test, as well as not affect feeding behavior in free food intake test, as indicated by the similar time in the center of the open-field and similar levels in the other behavioral measurements comparing vehicle and inactivation groups. **p < 0.01 and ***p < 0.001 in post hoc comparison after ANOVA and Student's t test.
Overall, our study strongly suggests that the BLA plays a critical role in the cautious decision-making process employed by animals when seeking food in the presence of a threat. Notably, the effects of BLA inactivation on cautious behavior were robust and emerged as early as the very first conflict test trial (time spent per zone, start, factorial ANOVA, group, F(1,22) = 22.92; p = 0.001; trials, F(1,22) = 3.42; p = 0.07; interaction, F(1,22) = 8.35; p = 0.008; conflict trials VEH vs INACT post hoc comparison, p = 0.001; choice, factorial ANOVA, group, F(1,22) = 3.03; p = 0.09; trials, F(1,22) = 11.90; p = 0.002; interaction, F(1,22) = 3.78; p = 0.064; danger, factorial ANOVA, group, F(1,22) = 13.74; p = 0.001; trials, F(1,22) = 0.18; p = 0.673; interaction, F(1,22) = 8.37; p = 0.001; conflict trials VEH vs INACT post hoc comparison, p = 0.000; goal, factorial ANOVA, group, F(1,22) = 16.89; p = 0.000; trials, F(1,22) = 3.54; p = 0.07; interaction, F(1,22) = 5.24; p = 0.032; conflict trials VEH vs INACT post hoc comparison, p = 0.000; behavioral events per zone, start, factorial ANOVA, group, F(1,22) = 8.02; p = 0.009; trials, F(1,22) = 8.69; p = 0.007; interaction, F(1,22) = 7.86; p = 0.01; conflict trials VEH vs INACT post hoc comparison, p = 0.003; choice, factorial ANOVA, group, F(1,22) = 8.05; p = 0.009; trials, F(1,22) = 14.50; p = 0.001; interaction, F(1,22) = 9.54; p = 0.005; conflict trials VEH vs INACT post hoc comparison, p = 0.002; danger, factorial ANOVA, group, F(1,22) = 0.32; p = 0.57; trials, F(1,22) = 0.92; p = 0.3; interaction, F(1,22) = 2.68; p = 0.115; goal, factorial ANOVA, group, F(1,22) = 31.37; p < 0.000; trials, F(1,22) = 23.13; p < 0.000; interaction, F(1,22) = 25.67; p < 0.000; conflict trials VEH vs INACT post hoc comparison, p = 0.000). This specific observation provides evidence to suggest that BLA activity is likely more influential during the deliberation stage of decision-making rather than influencing the evaluation of potential outcomes. However, it is important to acknowledge that definitively differentiating between these two processes requires more precise temporal control over brain manipulations. Future studies utilizing optogenetics rather than pharmacological inactivations could offer the necessary level of precision to distinguish between the role of BLA in deliberation and outcome evaluation during risky food-seeking behavior under threat (Orsini et al., 2017; Bercovici et al., 2018).
Consistent with previous results (Muller et al., 1997; Sierra-Mercado et al., 2011), in a separate group of rats, we found that BLA inactivation impaired threat memory retrieval (Fig. 2E,F; threat memory, Student's two-tailed unpaired t test, t(8) = 3.91; p = 0.004; VEH n = 4; INACT n = 6) while leaving anxiety-like behavior, general locomotion, and consummatory feeding behavior unaffected (anxiety-like behavior, Student's two-tailed unpaired t test, t(10) = 0.71; p = 0.49; locomotion, Student's two-tailed t test, t(10) = 1.12; p = 0.28; feeding behavior, Student's two-tailed t test, t(10) = 0.38; p = 0.70; VEH n = 6; INACT n = 6). Consistent with the lack of effect of BLA inactivation on feeding behavior, cued reward-seeking responses (as depicted in the inset in Figure 2D; Student's two-tailed unpaired t test, t(11) = −1.93; p = 0.078) also remained unaltered. In summary, our results support the notion that BLA inactivation impaired the retrieval of learned threat-related behaviors, thereby allowing the expression of food-seeking behaviors even in the face of potential harm. This suggests that BLA normally inhibits risky behaviors and facilitates a cautious approach when confronted with the challenge of obtaining food. Thus, BLA may restrain the choice of seeking food over defensive reactions during conflict.
VP mediates the motivation drive for food seeking, irrespective of conflict
VP plays a pivotal role in incentive motivation involving the invigorating effort to seek food (Farrar et al., 2008; Lederman et al., 2021), which leads to the intriguing possibility that it also mediates the additional challenge that represents risky food seeking. To evaluate the specific involvement of the VP in confronting threats for food, we employed pharmacological inactivation (n = 7) or vehicle treatment (n = 7) of this ventral basal ganglia structure before the memory conflict test (Fig. 3A,B). Notably, our findings revealed that VP inactivation significantly reduced the time spent and the occurrence of risk-assessment events in the choice zone during conflict trials (choice zone, factorial ANOVA, group, F(1,24) = 18.67; p = 0.000; trials, F(1,24) = 25.23; p = 0.000; interaction, F(1,24) = 18.40; p = 0.000; conflict trials VEH vs INACT post hoc comparison, p = 0.000; risk assessment, factorial ANOVA, group, F(1,24) = 6.80; p = 0.015; trials, F(1,24) = 24.51; p = 0.000; interaction, F(1,24) = 7.04; p = 0.013; conflict trials VEH vs INACT post hoc comparison, p = 0.005; Fig. 3C,D). This finding suggests that VP is not only critical in incentive motivation for food seeking but also in the complex process of weighing potential risks and rewards associated with such choices. Moreover, VP inactivation led to a substantial increase in the time required by rats to obtain food, regardless of whether the task involved confronting a threat (as in conflict trials) or not (as in non-conflict trials; factorial ANOVA, group, F(1,24) = 27.38; p = 0.000; trials, F(1,24) = 6.18; p = 0.02; interaction, F(1,24) = 1.28; p = 0.26; non-conflict trials VEH vs INACT post hoc comparison, p = 0.000; conflict trials VEH vs INACT post hoc comparison, p = 0.03). This VP inactivation effect is comparable to providing free access to food (satiety) 1 d before the CMC test (Illescas-Huerta et al., 2021), indicating that the task involves goal-directed behavior guided by incentive motivation. Nearly half of the rats subjected to VP inactivation (three out of seven) failed to cross within the allotted time (180 s), suggesting a compromised incentive motivation to obtain food, regardless of the trial type. This apparent lack of motivation was also evident in the VP-inactivated rats’ reluctance to press for food within the start zone (without the need to cross), as depicted in the inset of Figure 3D (Student's two-tailed unpaired t test, t(12) = −2.81; p = 0.015). Thus, these findings support the role of VP in incentive motivation for food seeking and risk evaluation.
VP inactivation impairs effort-related actions and conflict-triggered risk–taking. A, Rats were infused with VEH or muscimol (INACT) into the VP before the conflict test. B, Overlaid movement tracking and heat maps of representative rats (VEH and INACT) during the first two consecutive non-conflict and conflict trials of the conflict test. C, Analysis of the time spent per zone showed that VP inactivation decreased the time spent in the start safety zone while increasing it in the danger and goal zones during conflict trials without affecting non-conflict trials. D, Analysis of behavioral events per zone shows that VP inactivation decreased risk assessment related to conflict in the choice zone while increasing the latency to cross irrespective of the trial type. VP inactivation also increased the latency to press on the same side of the alley (inset). E, VP inactivation did not affect the retrieval of auditory threat conditioning memory, as shown by similar freezing levels in the vehicle group as compared with those in the inactivation group. F, VP inactivation did not affect anxiety-like behavior and general locomotion in the open-field test, nor did it affect feeding behavior in the free food intake test, as indicated by the similar time in the center of the open field and similar levels in the other behavioral measurements comparing vehicle and inactivation groups. *p < 0.05, **p < 0.01, and ***p < 0.001 in post hoc comparison after ANOVA and Student's t test.
Consistent with previous research, our findings support the idea that VP plays a pivotal role in incentive motivation (Tindell et al., 2004; Richard et al., 2016; Ottenheimer et al., 2018). However, given the observed decrease in risk assessment in the CMC task following VP inactivation and the recent implications of VP in threat-related processing (Lenard et al., 2017; Faget et al., 2018; Moaddab et al., 2021), we conducted separate experiments in different rat groups to explore whether VP inactivation is necessary for learning threat-related and anxiety-like behaviors (Fig. 3E,F). The results of these additional experiments indicated that VP inactivation did not influence freezing levels during the retrieval of a threat memory (Student's two-tailed unpaired t test, t(9) = 0.03; p = 0.97; VEH n = 5; INACT n = 6). These results are consistent with the lack of VP inactivation effect on freezing bouts in the start zone during the CMC test. Furthermore, VP inactivation had no discernible effect on anxiety-like behavior and general locomotion in the open-field test as well as no difference on feeding behavior in free food intake test (time in center, Student's two-tailed unpaired t test, t(6) = 0.56; p = 0.59; distance traveled, Student's two-tailed unpaired t test, t(6) = −0.46; p = 0.65; percent food intake, Student's two-tailed unpaired t test, t(6) = 0.27; p = 0.79; VEH n = 4; INACT n = 4). In summary, our findings suggest that VP plays a critical role in the motivational control of goal-directed actions, whether associated with rewards or threats.
VP-BLA connectivity mediates restraint of risky reward choice behaviors
Despite the extensively reported roles of VP and BLA in motivated behaviors and their extensive reciprocal anatomical connections (Zaborszky et al., 1984; Mitrovic and Napier, 1998; Mascagni and McDonald, 2009), the functional significance of the communication between these structures in motivation has been sparsely explored. We showed that BLA inactivation promotes risky food-seeking decisions during conflict, whereas VP inactivation restrains effort-based food-seeking behaviors, regardless of the trial type. Consequently, disrupting the exchange of information between the VP and BLA is likely to exert a substantial impact on a rat's behavior, particularly during conflicts involving food seeking and defensive reactions. To investigate this possibility, we conducted a disconnection experiment by comparing the effects of VP and BLA inactivation on the opposite hemisphere side (contralateral, VEH n = 8; INACT n = 7) with their effects on the same hemisphere side (ipsilateral, VEH n = 10; INACT n = 12). Given that most communication between brain structures occurs ipsilaterally, contralateral inactivations were employed to disrupt the communication between the VP and BLA (Fig. 4), whereas ipsilateral inactivations served as controls (Fig. 5), where communication between structures was presumably minimally disrupted. Thus, ipsilateral inactivation spares VP-BLA communication in one hemisphere, whereas contralateral inactivation disrupts VP-BLA communication in both hemispheres.
Contralateral VP-BLA inactivation impairs conflict-induced fear in food-seeking efforts. A, To functionally disconnect VP-BLA communication, we infused the rats into the VP and BLA on the contralateral sides with vehicle saline solution (VEH) or muscimol into VP and a cocktail of muscimol and baclofen into BLA (INACT) before the conflict test. B, Overlaid movement tracking and heat maps of representative rats (VEH and INACT) during the first two consecutive non-conflict and conflict trials of the conflict test. C, Analysis of the time spent per zone showed that VP-BLA contralateral inactivation decreased the time spent in the start zone while increasing it in the danger and goal zones during conflict trials without affecting non-conflict trials. D, Analysis of behavioral events per zone shows that VP and BLA contralateral inactivation impaired defensive responses (freezing and risk assessment) in the start zone, slowed crossing speed in the danger zone, and decreased crossing latency to press in the goal zone during conflict trials without affecting non-conflict trials. VP and BLA contralateral inactivation did not affect the latency to press on the same side of the alley (inset). E, VP and BLA contralateral inactivations did not affect the retrieval of auditory threat conditioning memory as shown by similar freezing levels in the inactivation group as compared with those in the vehicle group. F, VP and BLA contralateral inactivation did not affect anxiety-like behavior and general locomotion in the open-field test, nor did it affect feeding behavior in the free food intake test, as indicated by the similar time in the center of the open field and similar levels in the other behavioral measurements comparing vehicle and inactivation groups. *p < 0.05, **p < 0.01, and ***p < 0.001 in post hoc comparison after ANOVA and Student's t test.
Ipsilateral VP-BLA inactivation did not affect crossing-related behaviors. To allow VP-BLA communication, we infused the rats into the VP and BLA on the ipsilateral side with VEH or muscimol into VP and a cocktail of muscimol and baclofen into BLA (INACT) before the conflict test. B, Overlaid movement tracking and heat maps of representative rats (VEH and INACT) during the first two consecutive non-conflict and conflict trials of the conflict test. C, Analysis of the time spent per zone showed that ipsilateral inactivation of the VP-BLA did not affect the crossing-mediated time spent in any of the zones. D, Analysis of behavioral events per zone shows that VP and BLA ipsilateral inactivation did not affect crossing-mediated behaviors, as well as the latency to press on the same side of the alley (inset).
Consistent with our hypothesis, we observed that contralateral, but not ipsilateral, inactivation of the VP-BLA before the conflict memory test decreased the time spent and evaluated behaviors in most zones during conflict trials, while it left intact the timing and behaviors triggered by non-conflict situations (Fig. 4A,B). Similar to BLA inactivation, VP-BLA contralateral inactivation reduced the time spent in the start zone while increasing the time spent in the goal and danger zones (start zone, factorial ANOVA, group, F(1,26) = 6.34; p = 0.01; trials, F(1,26) = 62.90; p = 0.000; interaction, F(1,26) = 4.07; p = 0.053; conflict trials VEH vs INACT post hoc comparison, p = 0.017; goal zone, factorial ANOVA, group, F(1,26) = 6.78; p = 0.01; trials, F(1,26) = 52.53; p = 0.000; interaction, F(1,26) = 2.60; p = 0.11; conflict trials VEH vs INACT post hoc comparison, p = 0.02; danger zone, factorial ANOVA, group, F(1,26) = 1.96; p = 0.17; trials, F(1,26) = 59.92; p = 0.000; interaction, F(1,26) = 7.73; p = 0.009; conflict trials VEH vs INACT post hoc comparison, p = 0.03; Fig. 4C). Unlike BLA inactivation, however, contralateral inactivation did not affect freezing in the start zone (freezing bouts, factorial ANOVA, group, F(1,26) = 3.71; p = 0.06; trials, F(1,26) = 36.13; p = 0.000; interaction, F(1,26) = 3.88; p = 0.06), yet it slowed down the crossing speed in the danger zone, thus delaying the latency to press in the goal zone (crossing speed, factorial ANOVA, group, F(1,26) = 2.10; p = 0.15; trials, F(1,26) = 15.31; p = 0.000; interaction, F(1,26) = 11.63; p = 0.002; conflict trials VEH vs INACT post hoc comparison, p = 0.01; latency to press, factorial ANOVA, group, F(1,26) = 28.84; p = 0.000; trials, F(1,26) = 67.02; p = 0.000; interaction, F(1,26) = 17.0; p = 0.000; conflict trials VEH vs INACT post hoc comparison, p = 0.000). Notably, similar to both VP and BLA inactivation, VP-BLA contralateral inactivation robustly reduced the risk-assessment behavior in the choice zone (factorial ANOVA, group, F(1,26) = 26.83; p = 0.000; trials, F(1,26) = 65.05; p = 0.000; interaction, F(1,26) = 27.08; p = 0.000; conflict trials VEH vs INACT post hoc comparison, p = 0.000; Fig. 4D).
Next, we conducted separate experiments in different rat groups to explore whether contralateral VP-BLA inactivation was necessary for threat-related responses and feeding behavior. Unlike BLA inactivation, yet similar to VP inactivation, VP-BLA contralateral inactivation did not affect threat memory retrieval (Student's two-tailed unpaired t test, t(11) = −1.08; p = 0.29; VEH n = 6; INACT n = 7; Fig. 4E) while also leaving anxiety-like behavior, general locomotion, and feeding behavior intact (anxiety-like behavior, Student's two-tailed unpaired t test, t(10) = −0.64; p = 0.53; distance traveled, Student's two-tailed unpaired t test, t(10) = 0.71; p = 0.49; food intake, Student's two-tailed unpaired t test, t(10) = 0.79; p = 0.44; VEH n = 5; INACT n = 7; Fig. 4F). The lack of an inactivation effect on the retrieval of threat memory was consistent with the lack of an inactivation effect on freezing bouts in the start zone during the CMC test.
In contrast to contralateral VP-BLA inactivation, ipsilateral inactivation during the conflict test had no effect on any of the temporal and spatial behavioral parameters evaluated (start zone, factorial ANOVA, group, F(1,40) = 0.04; p = 0.83; trials, F(1,40) = 63.22; p = 0.000; interaction, F(1,40) = 0.42; p = 0.5; choice zone, factorial ANOVA, group, F(1,40) = 0.89; p = 0.35; trials, F(1,40) = 39.31; p = 0.000; interaction, F(1,40) = 1.85; p = 0.18; danger zone, factorial ANOVA, group, F(1,40) = 0.76; p = 0.38; trials, F(1,40) = 89.72; p = 0.000; interaction, F(1,40) = 0.66; p = 0.41; goal zone, factorial ANOVA, group, F(1,40) = 0.00; p = 0.97; trials, F(1,40) = 42.32; p = 0.000; interaction, F(1,40) = 0.27; p = 0.6; freezing bouts, factorial ANOVA, group, F(1,40) = 0.50; p = 0.48; trials, F(1,40) = 10.57; p = 0.002; interaction, F(1,40) = 0.47; p = 0.49; risk assessment, factorial ANOVA, group, F(1,40) = 2.72; p = 0.10; trials, F(1,40) = 38.89; p = 0.000; interaction, F(1,40) = 3.01; p = 0.09; crossing speed, factorial ANOVA, group, F(1,40) = 0.02; p = 0.86; trials, F(1,40) = 14.74; p = 0.000; interaction, F(1,40) = 0.01; p = 0.90; latency to press, factorial ANOVA, group, F(1,40) = 0.48; p = 0.48; trials, F(1,40) = 41.03; p = 0.000; interaction, F(1,40) = 0.44; p = 0.50; latency to press in the start zone, Student's two-tailed unpaired t test, t(6) = 0.08; p = 0.93; Fig. 5A–D). Thus, unlike the inactivation of the VP and BLA in the opposite hemispheres (contralateral), which had robust observable effects on various temporal and spatial behavioral parameters during the conflict test, inactivation in the same hemisphere (ipsilateral) did not result in any discernible impact on the parameters assessed. The specific and selective effect of contralateral inactivation suggests that the intricate interplay between VP and BLA is essential for orchestrating conflict-mediated behaviors, highlighting the nuanced and complementary roles of these structures in the regulation of risky actions. In summary our findings indicate that VP-BLA connectivity plays a vital role in controlling cautious approach choice behaviors guided by previous experiences (Fig. 6).
VP-BLA circuit orchestrates cautious reward approach. This diagram illustrates the proposed subcortical circuit underlying decision-making in the CMC test. In non-conflict trials (safe approach), the VP uses learned associations with a safe stimulus (light) to guide behavior toward food retrieval. Conversely, during conflict trials (risky approach), the VP and BLA integrate threat information from a predictive stimulus (sound) with the reward cue (light). This VP-BLA circuit likely facilitates memory-guided choice behavior, ultimately enabling a cautious approach toward the desired food despite the associated threat.
Discussion
We studied the influence of VP and BLA on foraging decisions in conflict situations. By combining a novel foraging choice task and pharmacological inactivation, we showed that VP-BLA connectivity plays a crucial role in decision-making. Our results support previous research, which suggests that BLA promotes cautious choices related to threats and that VP encourages reward-seeking responses. Our findings indicate that VP and BLA provide feedback on the current value of threats and rewards, enabling efficient decision-making during conflicts and preventing overly risky behavior.
Our study found that rats adjust their behavior to cross and reach the food based on the presence or absence of threats, indicating that they weigh the potential risks against the reward of food and make decisions accordingly. Prior research has demonstrated that animals adjust their risk preferences based on factors such as reward value, threat intensity, and type. For example, food deprivation, social competition, and predator cues can lead to increased risk-taking behavior in rats (Davis et al., 2009; Choi and Kim, 2010; Dent et al., 2014; Engelke et al., 2021; Zoratto et al., 2022). The CMC test, which has both a safe start zone and a danger zone, is an effective method for examining risk-taking behavior in rodents. This task requires rats to navigate threats, and previous studies have shown that administering diazepam before the test can encourage them to cross more easily when food is at stake, without affecting their behavior in nonthreatening situations (Illescas-Huerta et al., 2021).
This task could be suitable for exploring active suppression of learned threat-related behaviors during food-related conflicts, potentially distinct from the passive suppression in fear extinction (Sotres-Bayon and Quirk, 2010). Our research uncovered a brain circuit critical for prioritizing food despite danger, indicating that animals can balance risk and reward. Further studies should examine how rats integrate threat and reward motivation and how these factors influence their performance in the CMC task while considering the limitations of inferring emotions from behavior. Our findings provide a foundation for exploring VP's role in various threat responses, such as fear extinction, and could shed light on how animals cope with different threats and make decisions that balance risk and reward.
Distinct BLA and VP roles in foraging decisions
The BLA and VP exhibit dissociable roles in shaping foraging decisions during food-seeking conflicts. Our study reveals that the BLA plays a central part in shaping these decisions, orchestrating delayed food-seeking choices, promoting quick crossings, and signaling caution in response to potential threats. The BLA's multifaceted involvement in food-seeking decisions supports its role in threat and reward processing and its ability to modulate approach–avoidance conflicts (Muller et al., 1997; Shabel and Janak, 2009; Choi and Kim, 2010; Sierra-Mercado et al., 2011; Amir et al., 2015; Kim et al., 2016; Piantadosi et al., 2017; Bindi et al., 2018; Kyriazi et al., 2018, 2020). BLA's importance lies in its ability to encode the importance of both positive and negative stimuli, regulate emotional and motivational aspects of learning and memory, and interact with other brain regions to integrate threat and reward information and regulate goal-directed actions (Stuber et al., 2011; Sotres-Bayon et al., 2012; Namburi et al., 2015; Burgos-Robles et al., 2017).
Our results indicate that VP plays a critical role in food-seeking behavior, particularly in weighing potential risks and rewards. The VP may be a key structure for survival decision-making, as it encodes the value and salience of appetitive and aversive stimuli and regulates decisions related to the motivational drive to seek food (Tindell et al., 2004; Richard et al., 2016, 2018; Saga et al., 2017; Moaddab et al., 2021). VP mediates the effects of motivational manipulations, such as hunger and stress, on effort-related choice behavior (Farrar et al., 2008; Chang and Grace, 2014; Ottenheimer et al., 2020b). VP has been shown to increase the rate and intensity of operant responses such as lever pressing or nose poking, which are contingent on reward delivery (Farrar et al., 2008; Ahrens et al., 2018). VP also mediates the effects of reward anticipation, uncertainty, and magnitude on the vigor of actions (Ottenheimer et al., 2020a; Lederman et al., 2021; Moaddab et al., 2021). Our VP inactivation effects on conflict trials align with recent findings that VP activity mediates punished reward seeking (Farrell et al., 2021). Containing GABAergic and glutamatergic neurons that balance reward seeking and threat avoidance, VP interacts with other brain regions to integrate information about reward, threat, and vigor, guiding goal-directed actions (Mingote et al., 2008; Leung and Balleine, 2013; Tooley et al., 2018; Stephenson-Jones et al., 2020).
Our study demonstrates a complex relationship between the VP function and motivation in conflict resolution. While VP inactivation increased latencies to press on both trial types, suggesting a general motivational deficit, disconnection between BLA and VP facilitated responding on conflict trials. This suggests that other brain circuits may interact with VP to regulate effort allocation and vigor. Disconnection of BLA from VP may disinhibit appetitive drive from other brain regions, potentially overriding the amygdala's role in processing threat. Further research is necessary to elucidate the intricate circuitry involved in this process and to understand the interplay between motivation, risk/reward processing, and effort allocation during food-seeking behavior. Our findings contribute to a growing body of knowledge about the distinct roles of BLA and VP in foraging behavior and offer valuable insights into the complex neural mechanisms underlying survival decision-making in animals facing conflicting motivations under threat.
Collaborative necessity of VP and BLA in foraging choices
VP and BLA have extensive mutual connections and may interact dynamically to balance the costs and benefits of risk-taking behavior and to adjust the motivational drive and vigor for food according to the environmental context (Zaborszky et al., 1984; Mitrovic and Napier, 1998; Mascagni and McDonald, 2009). We found that disconnecting the pathway between VP and BLA impaired the process of assessing and weighing the potential risks and rewards associated with food-seeking choices. This suggests that the connectivity between the VP and BLA plays a crucial role in controlling risky food-seeking choices during conflicts while maintaining the motivation for food seeking in non-conflict situations. Specifically, disrupting the VP-BLA pathway selectively decreased risk-taking behavior in response to food-seeking conflicts, highlighting its importance in assessing and weighing potential risks and rewards. This suggests that VP-BLA connectivity may be required to coordinate conflict-mediated behaviors, reconcile the interplay between reward and threat processing, and regulate risk-taking behaviors. This aligns with previous studies that have implicated both VP and BLA in threat and reward processing as well as in modulating approach–avoidance conflicts.
Our research unveils that the VP-BLA circuitry is essential for risk/reward behavior. Both regions excite each other, and VP's inhibitory and cholinergic projections to BLA suggest it fine-tunes threat responses (Soares-Cunha and Heinsbroek, 2023). The disconnection of VP-BLA mostly mimicked BLA inactivation, implying that BLA's defensive signals might overpower VP's reward modulation. VP may integrate motivational drives with BLA's threat processing for a nuanced risk assessment (Moaddab et al., 2021). This underscores VP-BLA as a key pathway in survival decisions, allowing animals to prioritize caution by potentially suppressing reward-seeking circuits during food seeking.
Caveats and future directions
Our study has several limitations that are currently being addressed in ongoing experiments. First, our study only involved male rats; therefore, future research should examine female rats (Orsini et al., 2016). Second, one of the drawbacks of the pharmacological inactivation method is its lack of cell type and time specificity. GABA agonists, by hyperpolarizing neurons, effectively silence all cell types within a region, making it difficult to target specific cell types or pathways at a particular time point. Furthermore, the use of different drug combinations and doses in our study prevents us from ruling out the possibility that the observed differences reflect variations in the level of inactivation. Finally, one of the difficulties in the CMC task is that it is challenging to disentangle the factors of time, effort, and potential punishment, as rats need to consider all of these variables when deciding whether and when to cross. Future research should employ more refined techniques, such as optogenetic manipulation and single-unit recordings, to gain a deeper understanding of the specific cell and pathway mechanisms that mediate multidimensional foraging choice behaviors which may seem indissociable (Kyriazi et al., 2018, 2020). In particular, it would be intriguing to investigate the unique contributions of specific cell types in the VP and BLA during conflict situations, as certain neuronal types in both of these regions have been linked to opposing approach or avoidance choice behaviors (Beyeler et al., 2018; Stephenson-Jones et al., 2020).
Conclusions
Our study found a new role of the VP-BLA connection in managing foraging decisions during conflicts. We discovered that VP and BLA collaborate to evaluate and signal threats and reward responses, and their communication is crucial for balancing approach and avoidance behaviors. Our results provide novel insights into the neural mechanisms that govern survival decision-making in animals, and they suggest that targeting the VP-BLA connection could potentially aid in treating disorders related to deficits in risk-taking behavior, such as addiction, anxiety, and depression.
Footnotes
This study was supported by the Consejo Nacional de Ciencia y Tecnología (CONACyT, PN2463), as well as by the Dirección General de Asuntos del Personal Académico de la Universidad Nacional Autónoma de México (UNAM, IN214520 and IN214223) and the International Brain Research Organization (Return Home Fellowship) to F.S-B. A.H-J. is a doctoral student at Programa de Doctorado en Ciencias Biomédicas at UNAM and was supported by a CONACyT fellowship (417151). We thank Christian Bravo-Rivera for his comments on a previous version of the manuscript and the Sotres-Bayon laboratory members for their technical assistance and helpful discussions.
The authors declare no competing financial interests.
- Correspondence should be addressed to Francisco Sotres-Bayon at sotres{at}ifc.unam.mx.
