2.1. Subjects

Forty-eight male Holtzman Sprague–Dawley rats (Harlan Inc., Indianapolis, IN) were used. The rats were 3 months old and weighed 250 g at the beginning of the experiment. They were housed in pairs in plastic cages (42.5 cm×22.5 cm×19.25 cm). Lights were on in the colony room from 07:00 to 19:00. All behavioral testing occurred during the light phase of the light/dark cycle. Food (Harlan Teklad Laboratory Diet #8604, Harlan Inc., Indianapolis, IN) was continuously available. Access to water was restricted to 20 min per day given immediately after testing sessions. The rats were tested 7 days per week. Twenty-four rats were trained on delay discounting, while the remaining 24 were trained on probability discounting. Within each condition, half the animals received excitotoxin lesions, while the rest were given sham lesions. This study was conducted in accordance with the guidelines set up by the Institutional Animal Care and Use Committee of The University at Buffalo, The State University of New York.

2.2. Apparatus

2.3. Surgery

Subjects were randomly assigned to receive excitotoxin or sham lesions aimed at the NAC. Rats were anesthetized with Avertin (2% [w/v] 2,2,2-tribromoethanol, 1% [w/v] 2-methylbutan-2-ol, and 8% [v/v] ethanol in phosphate-buffered saline, 10 ml/kg ip) and placed in a stereotaxic frame (Stoelting CO, Wood Dale, IL). The skull was exposed, and a Dremel tool (Dremel, Mount Prospect, IL) was used to remove the bone directly above the injection sites. The dura mater was broken with the tip of a needle to avoid damaging the underlying venous sinuses. Lesions were performed in accordance with the atlas of Paxinos and Watson [30] using bregma as the origin, with the incisor bar set to ensure the surface of the skull was level.

Fiber-sparing excitotoxic lesions aimed at the NAC were made with 0.15 M quinolinic acid (Sigma, St. Louis, MO) in 0.1 M phosphate buffer (final pH = 7.2–7.4) injected through a 30-gauge cannula mounted on a stereotaxic frame. At each site, 0.5 µl of toxin was infused over 3 min, after which 2 min was allowed for diffusion before the injector was removed. Lesion coordinates for the NAC (in millimeters from the skull surface at bregma) were AP +1.6, ML ±1.5, DV −7.1. The cannula was connected by a polythene tube to a 10 µl Hamilton syringe (Hamilton Company, Reno, NV). Sham lesions were made in the same manner, except that vehicle was infused. After surgery the rats were given 1 week to recover with free access to food and water and were not tested. During this period, rats were weighed daily. Rats showing signs of dehydration were given subcutaneous injections of lactated Ringer’s solution (Webster Veterinary Supply Inc., Sterling, MA).

2.4. Histology

At the end of the experiments, subjects were deeply anesthetized with Fatal Plus (pentobarbitone sodium, 200 mg/ml, >1.5 ml ip) and perfused transcardially with 0.01 M phosphate-buffered saline followed by 4% (w/v) paraformaldehyde in phosphate-buffered saline. Their brains were removed and post-fixed in paraformaldehyde overnight before being cryoprotected in 20% (w/v) sucrose; they were sectioned coronally at 50 µm thickness on a freezing microtome, and every fifth section was mounted and allowed to dry. Sections were then passed through a series of ethanol solutions of descending concentration (3 min in each of 100%, 95%, and 70% [v/v] ethanol in water) and stained for ~5 min with cresyl violet (0.05% [w/v] aqueous cresyl violet, 2mM acetic acid, and 5 mM formic acid in water). After staining, sections were rinsed in water and 70% ethanol before being differentiated in 95% ethanol. Finally, they were dehydrated and delipidated in 100% ethanol and Citrisolve (Organic Specialties Inc., East Providence, RI) before being cover-slipped and allowed to dry.

2.5. Immunocytochemistry

A one-in-five series adjacent to the cresyl violet stained sections was used for NeuN immunostaining. Free-floating sections were rinsed in 0.1 M phosphate-buffered saline (PBS) and treated with 0.5% H₂O₂ in 0.1 M PBS for 30 min to suppress endogenous peroxidase activity. Sections were then incubated for 30 min with blocking solution consisting of 10% normal horse serum (Sigma, St. Louis, MO) and 0.3% Triton X-100 (Sigma, St. Louis, MO) in PBS (PBST). The sections were then incubated for 24 h at 4°C in the monoclonal anti-NeuN serum (1:5000, Chemicon, Temecula, CA) diluted in PBST. After rinsing in PBST, sections were incubated for 1 h at room temperature in biotinylated horse antimouse serum (1:1000, Vector Laboratories, Burlingame, CA) in PBST. After more rinses in 0.1 M PBST, sections were incubated for 1 h in avidin-biotin-horseradish peroxidase complex (1:200, ABC-Elite, Vector Laboratories) in PBST. Sections were then rinsed in 0.1 M PBS followed by rinses in 0.1 M tris buffered saline (TBS) and placed in a chromagen solution consisting of 0.05% diaminobenzidine (Sigma, St. Louis, MO) and 0.01% H₂O₂ for approximately 5 min. The reaction was visually monitored and stopped in rinses of 0.1 MTBS, and the sections were mounted and dried on gelatin-coated slides.

2.6. Procedures

2.7. Phase 1: the effects of NAC lesions on discounting with constant delay and probability

2.8. Phase 2: the effects of NAC lesions on delay and probability discount curves

2.9. Phase 3: the effects of NAC lesions on responsiveness to changes in reward parameters

3.1. Histology

Lesion damage was confirmed by areas devoid of NeuN staining, indicating no neurons were present [33]. Lesioned animals showed enlargement of the lateral ventricles and general reduced volume of the ventral portion of the brain around the lesion site, indicating major tissue loss and subsequent reduction in brain volume. All rats included in the analysis showed significant damage to both the NAC core and shell (Fig. 1). In the delay condition, four rats that did not meet these criteria were excluded from the analysis. Additionally, one sham lesioned rat that showed substantial enlargement of the lateral ventricles with no clear evidence of localized lesion damage was dropped from the analysis. In the probability discounting group one rat did not learn the discounting task and was dropped from the experiment prior to surgery. In addition, one rat that did not have bilateral NAC core and NAC shell damage and two that had extensive damage to the dorsal striatum were dropped from the analysis. In the rats retained for the analysis, a wide range of lesion damage was detected with the NeuN staining, a highly sensitive technique for detecting excitoxin damage months after lesion induction [33,34]. In some cases, as indicated in Fig. 1, lesion damage was observed to extend beyond the NAC to the ventral pallidum, medial septum, and striatum. Half of the rats in the delay and probability groups had minor lesion damage to the ventral pallidum. One rat in the delay group had damage that extended into the striatum above the NAC. In the probability group, two rats had minor damage to the striatum above the NAC and one rat had damage to the medial septum. The only areas of lesion damage common to all animals were the NAC core and shell and damage in other areas was not related to the behavioral impairments described below.

3.2. Phase 1: the effects of NAC lesions on discounting with constant delay and probability

The purpose of phase 1 was to determine the effects of NAC lesions on how much rats valued delayed and probabilistic rewards when the delay and probability were held constant. The rats were trained and tested with a single delay (4 s) or a single probability (0.4) of reward. Training the animals prior to surgery reduced the likelihood that any observed lesion effects would be the result of impaired task acquisition.

3.3. Phase 2: the effects of NAC lesions on delay and probability discount curves

In phase 2 we tested a range of delays or probabilities to derive discount curves which could be quantitatively characterized using the hyperbolic discount function [27,31,32].

3.4. Phase 3: the effects of NAC lesions on responsiveness to changes in reward parameters

The purpose of phase 3 was to assess the animals’ sensitivity to acute changes in parameters associated with reward delivery. The results of phase 2 may have been related to an inability to adapt to changes in the delay conditions, rather than specifically decreasing delay discounting. To test this, we assessed the animals’ sensitivity to changes in reinforcer delay, probability, magnitude and intertrial interval (ITI).

The rats were first tested daily for 4 weeks with either a 4 s delay to reward (delay discounting rats) or a 0.4 probability of reward (probability discounting rats). Training with constant delay and probability of reward lasted for 4 weeks and followed the phase of training in which the discount curves were obtained. No differences in discounting were found at the baseline values (4 s delay to reward, and 0.4 probability of reward).

3.5. Effects of NAC lesions on bodyweight and locomotor activity

NAC lesioned rats exhibited more activity in the activity monitors (lesion 6052 ± 721.13 cm versu sham 4861.23 ±254.39 cm) [t(37) = 1.766, p = 0.042] and had significantly lower bodyweights at the conclusion of the experiment (lesion 601.42 g ± 18.68 versus sham 673.5± 12.31 g) [t(37) = 3.417, p < 0.001]. Increased locomotor activity and decreased rates of weight gain are behavioral indicators of NAC core damage [35].

4.1. Impairment specific to delay

The absence of an effect on probability discounting indicates that NAC lesions did not cause a general impairment in the ability of the animals to perform the adjusting amount procedure. Both the delay and probability versions of the adjusting amount procedure require the rats to track the changing value of the immediate reward across the session, compare its value to the delayed or probabilistic large reward, and then make a decision about which alternative has the greatest value. Furthermore, there was no evidence that the lesion caused rats to generally per-severate in responding for either option on either the delay or probability discounting versions of this task. Because both the delay and probability versions of the task had the same general performance requirements and the NAC lesioned rats were not impaired on the probability version of the task, it does not seem likely that NAC lesions generally disrupted performance on the adjusting amount procedure.

The overall lack of effects on probability discounting observed in the present study contrasts with Cardinal and Howes [36] findings in which NAC core lesions decreased choice of uncertain larger rewards over smaller certain rewards on a different probability discounting task. Although we found no significant effects of lesions on probability discounting, there was a tendency for lesioned rats to discount slightly more in all phases of the study, therefore providing some supportive evidence that the NAC may play some role in probability discounting. Undoubtedly this is a matter that warrants further investigation.

The differential effects of NAC lesions on delay and probabilistic reinforcement suggest that these processes require different underlying neural circuits. Previously it has been suggested that delay and probability discounting are governed by similar mechanisms. In natural settings, rewards that are delayed for long periods are often less certain and consequently may be perceived as less likely than rewards delivered immediately [37–39]. Conversely, it has been hypothesized that a series of probabilistic rewards is experienced as a series of delayed rewards [40]. The finding that NAC lesions affected only delayed reinforcement provides evidence that perception of delay and probability are governed by dissociable processes. Further evidence for the hypothesis that delay and probability discounting are separate processes is provided by the fact that varying the ITI in the current experiment had opposite effects with decreasing ITI durations resulting in higher indifference points on the delay task and lower indifference points on the probability task.

4.2. Relation of present study to Cardinal et al. [13]

While the effects of NAC lesions on delay discounting in the present study appear opposite to those reported by Cardinal et al. [13], there are actually strong similarities between the findings of the two studies. The findings of Cardinal et al. [13] could also be explained by an effect of NAC lesions on responsiveness to changes in delay within the test sessions. In that study, lesioned animals showed a strong bias toward the immediate small reward, choosing it approximately 70% of the time even when there was no delay to the larger reward. Cardinal et al. interpreted their findings as indicating that NAC core lesions decreased the ability of rats to tolerate delays to appetitive reinforcement. However in the present study NAC lesions appeared to increase the ability of rats to tolerate delays to appetitive reinforcement when the delay to reward was varied between sessions. It does not appear that we truly affected delay discounting as we found no changes in discounting when delay to reinforcement was held constant across sessions. In contrast, the discounting procedure used by Cardinal et al. always involved varying the delay within a single session and animals were never tested with a reward held to a constant delay across sessions. Based on our results that NAC lesions affected performance only when the delay was changed across sessions and not when the delay was held constant, it appears that lesions of the NAC may impair learning about or adapting to changes in delayed reinforcement but do not affect the ability to tolerate delays to appetitive reinforcement.

When contrasting the results of Cardinal et al. [13] with the present study, differences in the behavioral test procedures need to be taken into account. The adjusting amount procedure used in the current study was designed to measure the value of delayed or probabilistic rewards. Rats are given a choice between a delayed or probabilistic constant volume of water and an immediate adjusting amount of water. The value of the adjusting alternative is increased after a choice of the delayed or probabilistic alternative and decreased following a choice of the immediate, certain adjusting alternative. When the rat chooses the delayed or probabilistic alternative and the adjusting immediate alternative with equal frequency we infer that the subjective value of the amount of water on the adjusted alternative matches the subjective value of the standard alternative. This is referred to as the indifference point. Smaller indifference points indicate delayed rewards are valued less and are thought to indicate increased impulsivity. The primary measure in the procedure developed by Evenden and Ryan [41] and used in the Cardinal et al. study is percent choice of the larger delayed alternative. The delay to the large alternative is set at zero at the beginning of the test session and is systematically incremented to 10, 20, 40 and 60 s during the session. Theoretically, the point at which the animals in the Evenden and Ryan procedure choose the delayed alternative 50% of the time is equivalent to the indifference point measure used in the adjusting amount procedure. However, it is notable that in the Cardinal et al. study no such indifference point was obtained because the lesioned animals chose the large alternative approximately 30% of the time even when it was not delayed. Whether this bias toward the immediate alternative was due to an “intolerance of delayed reward” or an inability to discriminate between varying delays and adapt behavior to changes in delay is debatable.

In support of their interpretation of the data, Cardinal et al. [42], subsequently, reanalyzed their data to include only those animals that chose the large reward at least 90% of the time when retrained with no delays to this option. When these selected animals were tested with varying delays in subsequent test sessions, steeper discount functions were obtained. However it was notable that the NAC lesioned animals still appeared to make fewer choices of the immediate large reward than control animals when the delays were varied. It is difficult to reconcile the Cardinal et al. interpretation of hypersensitivity to the effects of delaying reinforcement with the seeming insensitivity to the effects of delay observed in the present study. However, an inability to discriminate between varying delays and adapt behavior to changes in delay may explain both of these effects.

Both the findings of the present study and those of Cardinal et al. [13] could be the result of the NAC lesioned rats’ being less sensitive to changes in delay. In the present study, during the determination of the delay discount curve three shorter delays were added (0, 1, and 2 s) but only one longer delay was added (8 s). The lesioned rats may have formed an association that the delayed reward was now available sooner and adapted their behavior accordingly. However, in the Cardinal et al. study it appears that intact subjects preferred the large reinforcer over the smaller immediate reinforcer at 0 and 10 s and not the three other delays (20, 40 and 60 s). Consequently, lesioned animals in the Cardinal et al. study may have associated that the large reinforcer with an unappealing delay and consequently were biased away from this option.

While an insensitivity to changes in delays is a plausible explanation for the results of the present study as well as those observed by Cardinal et al. [13], it is necessary to consider other alternatives. It is possible that the NAC lesion did in fact make rats more tolerant to delays to reward, and that the 4 s delay used in phase 1 was simply not long enough to observe this effect. It is notable that a significant effect of delay was found only at 8 s in both the discount curve analysis and in the acute challenge data. This interpretation seems less likely because it would be completely opposite the results observed by Cardinal et al. [13], which used delays much longer than 8 s. Alternatively, as all the delays tested in the present study were shorter than those used by Cardinal et al., it is possible that we would have observed different or even opposite effects had we tested animals on longer delays. Further studies with longer delays using the adjusting amount procedure will be needed to conclusively rule out these interpretations.

It is also important to emphasize that both the core and shell regions of the NAC were damaged in the present study while the Cardinal et al. study reported that only the core region was damaged. It cannot be ruled out that the additional NAC shell damage in the present study may have played a role in the results observed in the present study. However, a recent study by Pothuizen [43] reported no effect of shell lesions on preference for immediate uncertain versus delayed certain rewards, suggesting that the shell is not important for choosing between immediate and delayed reinforcement. Clearly further studies are needed to clarify these issues.

4.3. Supporting evidence for a role of the NAC in adapting to temporal contingencies

More evidence for a major role of the NAC in delayed reinforcement can be obtained from previous studies on the effects of NAC lesions on operant responding in rats [44–46]. NAC core lesions impaired the acquisition of responding for delayed but not immediate reinforcement [46]. In another study, NAC lesions prevented the increases in latency to resume responding on a progressive ratio schedule, as work requirements and consequently delay to reinforcement increased [44]. This finding is consistent with insensitivity to changes in delay to reinforcement. Finally, NAC lesions prevented decreases in run speeds that are normally observed when reinforcers were omitted on a T-maze [45], also suggesting insensitivity to changes in delayed reinforcement contingencies. These reports, considered in the context of the present study, suggest that the NAC may be important for adapting behavior in response to delayed consequences, when these relationships are initially learned or when contingencies of delayed reinforcement are changed.

Numerous electrophysiological and imaging studies on rodents, humans and non-human primates have implicated the NACas part of a neural circuit responsible for predicting delayed rewards [14–26]. The results of the present study provide strong behavioral evidence that the NAC is an essential part of this reward prediction circuit. The NAC may be necessary for precise temporal prediction of delayed rewards, although impaired temporal prediction would not necessarily increase or decrease delay discounting.

4.4. Potential implications

NAC dysfunctions could potentially be an important component in impulsive disorders. An intact NAC appears to be necessary for behavior to be well-governed by delayed reinforcement, not by differentially affecting the discounting of delayed rewards but by acting as part of the neural substrate for temporal associations. Therefore, dysfunctions in the NAC could be expected to lead to maladaptive behavior due to an impaired ability to learn the relationship between a behavior and delayed consequences.