| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Ernest Gallo Clinic and Research Center, University of California, San Francisco, Emeryville 94608; 2 Graduate Program in Neuroscience, University of California, San Francisco 94143; 3 Departments of Neurology and Physiology and Wheeler Center for the Neurobiology of Addiction, University of California, San Francisco, California 94143
Submitted 9 July 2003; accepted in final form 20 November 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
; Ikemoto and Panksepp 1999). For instance, NAc dopamine has been strongly implicated in appetitive behavior but seems not to be involved in reward consumption (Berridge and Robinson 1998; Ikemoto and Panksepp 1999; Salamone 2002). On the other hand, opioid peptides injected into the NAc strongly increase feeding, particularly consumption of fatty, salty, or more palatable foods; furthermore, opioid receptor antagonists injected in the NAc reduce feeding (Kelley et al. 2002). General inhibition of the NAc with GABA receptor agonists or glutamate receptor antagonists also increases feeding (Saper et al. 2002). NAc neurons are therefore likely to regulate both reward seeking and reward consummatory activity.
The mechanisms by which NAc neurons influence feeding are not fully understood. The lateral hypothalamus, an important nucleus for homeostatic regulation, projects to the NAc, and the NAc reciprocates this projection both directly and through the ventral pallidum (Saper et al. 2002). In addition, the NAc receives direct projections from taste sensory areas, such as the nucleus of the solitary tract and the gustatory cortex (Kelley et al. 2002). Therefore because of its projections to motor areas such as the ventral pallidum and substantia nigra (Zahm 2000), the NAc is anatomically well situated to control consummatory motor behavior. Consistent with the idea that the NAc influences both consummatory and appetitive behavior, treatment of the NAc with a µ opioid agonist not only increases food consumption but also increases animals' willingness to work on a progressive ratio schedule for highly palatable food (Zhang et al. 2003). This suggests that NAc neurons utilize information about the palatability of food to promote the appetitive behavior required to obtain it.
Further supporting a role for NAc neurons in consummatory behavior, a number of electrophysiological studies have identified subpopulations of NAc and ventral striatal neurons that fire phasically during reward consumption. For instance, when juice reward is delivered in cue responding tasks, many primate ventral striatal neurons are excited (Apicella et al. 1991; Bowman et al. 1996; Shidara et al. 1998). These excitations are dependent on delivery of reward because no excitation is observed if the reward-associated cues are presented without the reward (Apicella et al. 1991; Hollerman et al. 1998) and the excitations are delayed when the reward is delayed (Apicella et al. 1991). In addition, the excitations depend on the identity and magnitude of the reward (Cromwell and Schultz 2003; Hassani et al. 2001). Although the excitations are not precisely time-locked to licking behavior (Apicella et al. 1991), it is possible that they encode some aspect of the motor behavior of reward consumption.
In rats, many studies have reported both excitations and inhibitions when animals receive several types of reward, including drugs of abuse. For instance, receipt of cocaine during self-administration results in long-lasting inhibition of many NAc neurons (Nicola and Deadwyler 2000; Peoples and West 1996; Peoples et al. 1998) as well as inhibitions or excitations on a shorter time scale (Carelli and Deadwyler 1994; Carelli et al. 1993; Chang et al. 1994, 1996; Peoples and West 1996; Peoples et al. 1997; Uzwiak et al. 1997). Self-administration of other drugs, such as ethanol and heroin, is also associated with excitation and inhibition after each operant response (Chang et al. 1998; Janak et al. 1999; Lee et al. 1999). Several studies have attempted to determine more precisely the information encoded by the NAc neuronal firing changes that occur after the operant response for cocaine. Postoperant excitations and inhibitions have been found to reflect the stimuli associated with cocaine but not necessarily the receipt of cocaine itself (Carelli 2000; Carelli and Deadwyler 1996). In some cases, the magnitude of the cocaine-associated neuronal response was dependent on whether an operant response was required to obtain the cocaine (Peoples et al. 1997). Additional studies have examined how delivery of natural reward (sucrose or water) during both fixed ratio and maze tasks affects the firing of NAc neurons. Natural reward delivery in these tasks is associated with both excitation and inhibition of NAc neurons (Carelli and Deadwyler 1994; Carelli et al. 2000; Hollander et al. 2002; Lavoie and Mizumori 1994; Miyazaki et al. 1998; Roop et al. 2002; Shibata et al. 2001), but the contribution of operant behavior to the natural reward-associated firing responses has not been systematically examined. Thus although recordings from primates have used tasks in which appetitive and consummatory phases of behavior are well separated, this distinction has been more difficult to make in free-run operant tasks performed by rats working for natural reward. The primate literature has focused mainly on excitations of ventral striatal neurons during reward consumption, but the rat literature indicates that many neurons may be inhibited as well. To extend our understanding of rat NAc neuronal responses during reward consumption, we report here an analysis of their firing patterns during the consummatory phase of the discriminative stimulus (DS) task described in the companion paper (Nicola et al. 2004).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
). In the second experiment, we randomly withheld reward from the animals after some correct responses to the cue (random withholding task). Basic DS task
Subjects, training, surgery, recording, and data analysis techniques were identical to those described in Nicola et al. (2004) and are summarized briefly here. In the basic DS task, two cues were presented to the animal on independent schedules with variable intervals averaging 2 min. Actual intervals between DS presentation were randomly selected by the computer from the following list: 60, 90, 104, 112, 116, 118, 119, 119.5, 120, 120.5, 121, 122, 124, 128, 136, 150, and 180 s. The first cue, the DS, indicated that, after the animal performed a nose-poke, a 10% sucrose reward (50 µl) would be delivered into a receptacle located next to the nose-poke. The DS was presented for 
20 s and was terminated by a correct nose-poke. The DS consisted of a 6 kHz (12 rats) or 4 kHz (9 rats) intermittent tone which was on for 200 ms and off for 550 ms; house lights were also dimmed by turning off one of the two lamps. The second cue, the nonrewarded stimulus (NS), was presented for 20 s and responding to it had no consequence. The NS consisted of a 6- or 4-kHz intermittent tone (for each rat, the tone chosen for the NS was the tone that was not chosen to be the DS), which was on for 200 ms and off for 550 ms and accompanied by dimmed houselights. Reward was only delivered in response to nose-pokes made during the DS; uncued nose-pokes had no consequence. A photobeam across the reward receptacle indicated when the animal's head entered and exited the receptacle.
Immediately after a successful response to the DS, a conditioned stimulus (CS) was presented for 20 s. The CS consisted of an 8-kHz intermittent tone (on for 200 ms, off for 550 ms) accompanied by continued dimmed houselights. For most animals, a liquid dipper was used to deliver reward, and therefore dipper activation noise also contributed to the CS. The dipper was activated on completion of a successful nose-poke in response to the DS and remained up for the duration of the 8-kHz intermittent tone/dimmed houselights. For some animals, reward was delivered by a syringe pump into a well located in the reward receptacle; the syringe pump was located outside the sound-attenuating chamber.
Random withholding task
In this experiment, animals were trained on the basic DS task and implanted with NAc electrodes exactly as described in Nicola et al. (2004). After several recording sessions on the basic DS task, animals were placed on the random withholding task. This task was identical to the basic DS task, with one exception. In 40% of trials (randomly chosen), no reward was delivered after the animal made a successful nose-poke response to the DS. In these trials, only the 8-kHz intermittent tone and dimmed houselights were presented; in addition, the dipper was activated for 300 ms, which was too brief to provide the animal with access to sucrose but did provide dipper activation noise. (Only dippers, not syringe pumps, were used to deliver rewards in this experiment.)
Electrophysiology and data analysis
Surgical implantation of microwire electrodes in the NAc, waveform isolation, classification of firing patterns, and data analysis were conducted as described in Nicola et al. (2004). Briefly, electrodes were implanted in trained animals, and animals were allowed 1 wk of recovery before commencement of recording sessions. Waveforms were isolated on-line and re-sorted off-line, and firing patterns were classified according to the criteria described in the companion paper. Because firing rates were not normally distributed, ranks-based statistical tests (Wilcoxon signed-rank) were used to compare firing rate increases and decreases in different conditions. For the basic DS task, firing rate changes associated with cued versus uncued entries into the reward receptacle were compared. Cued receptacle entries refer to entries that occurred after a nose-poke response to the DS. Uncued entries refer to those occurring in the absence of the DS, NS, or CS. Uncued entries often occurred in rapid sequences. To eliminate the potential confound introduced by this, only the first entry in bursts defined by a minimum inter-burst interval of 10 s were used for this analysis. (In some sessions, no uncued receptacle entries were present. Neurons that were recorded only during such sessions were necessarily excluded from the analysis.) When firing rate increases and decreases relative to baseline were computed, the baseline we used was 10 s prior to DS presentation for cued responses and from 15 to 5 s before uncued responses.
Of the 22 rats recorded, 19 were recorded during the basic DS task only, two were recorded during both the basic DS task and the random withholding task, and one was recorded during the random withholding task only.
Histology
After completion of experiments, animals were deeply anesthetized with pentobarbital, and the position of each electrode was marked by passing current through it. This causes an iron deposit that can be visualized with the Prussian Blue stain (Green 1958). After electrode marking, rats were perfused first with saline, then with 10% formalin and last with 3% potassium ferrocyanide/10% formalin to develop the Prussian Blue deposits. Sections (40 µm) were cut on a microtome and stained with Neutral Red, and the location of electrodes determined.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
BEHAVIOR. In the basic DS task, the animal performs a nose-poke response to the DS, which activates the delivery of a sucrose reward into a receptacle nearby. This task is diagrammed in Fig. 1A. In the companion paper (Nicola et al. 2004), we examined NAc firing patterns correlated with DS onset and the nose-poke response. In this paper, we analyze the firing patterns correlated with receptacle entry and exit. These were excitations during receptacle entry, sustained excitation and inhibition during the animal's stay in the receptacle, and excitations after exit from the reward receptacle. The analysis windows used to classify these firing patterns are shown in Table 1 of Nicola et al. (2004), the proportion of neurons exhibiting each firing pattern is described in Table 2 of that paper, and the distribution of these firing patterns in relation to other firing patterns is shown in Table 3 of that paper. During performance of the basic DS task, animals often (roughly once every 2–3 min) made uncued (in the absence of DS, NS, or CS) nose-pokes and entries into the reward receptacle, which were never rewarded. These uncued responses allowed us to determine the degree to which NAc neuronal firing associated with receptacle entry and exit is influenced by whether reward was explicitly predicted by the cue and/or by the presence of reward itself.
|
FIRING PATTERNS. Receptacle entry excitations. Approximately 8.3% of NAc neurons were excited during the act of entering the reward receptacle. These excitations were defined by increased firing from 0.5 s before to 0.5 s after the animal broke the photobeam across the reward receptacle. The excitations depended strongly on whether the entry was uncued or occurred after a nose-poke response to the DS (Fig. 2). The example neuron shown in Fig. 2A exhibited a brief excitation beginning 
0.5 s before the receptacle entry, peaking just after the entry, and receding to baseline 1 s after the entry [while the animal was still in the receptacle; circles on the raster plot indicate receptacle exits (symbols used in raster figures are explained in Table 1.)]. No excitation at all was observed if the receptacle entry was uncued (Fig. 2A2). These results are confirmed by the analysis of median firing rates across 49 neurons with entry excitations (Fig. 2, B and C). The median increase in firing in the 1-s window around the receptacle entry relative to baseline was substantially larger for entries after a response to the DS than for uncued entries (P < 0.001, n = 49; Fig. 2C). Therefore the degree of receptacle entry excitation depended on the presence of either the predictive cue or reward in the receptacle. These possibilities are explicitly differentiated with the random withholding task (see Fig. 7).
|
|
|
Sustained receptacle excitations. In addition to excitations that accompanied the animal's entry into the reward receptacle, we also observed sustained excitations that began soon after entry into the receptacle and continued until the animal exited (Fig. 3, A and B). These excitations were observed in 2.8% of neurons and were defined by excitation from 1 to 5 s after entry into the reward receptacle. Some of these excitations (31.1%) began within 0.5 s of receptacle entry, and therefore these neurons also exhibited receptacle entry excitations; an example is shown in Fig. 3A1. Sustained excitations in neurons that did not also exhibit receptacle entry excitation began later, between 0.5 and 1 s after entry into the reward receptacle (e.g., Fig. 3B1). To avoid including firing that occurred during receptacle entry (i.e., that part of the excitation defined as receptacle entry excitation) in the analysis of sustained excitations, we set the analysis window for sustained excitation to begin 1 s after receptacle entry.
|
), only 28 were used for this analysis because in many sessions all uncued receptacle entries lasted <1 s]. Although this finding is consistent with the idea that sustained receptacle excitation reflects reward consumption, it leaves open the possibility that the firing is also in part determined by the presentation of a reward-predictive cue. This hypothesis is further tested with the random withholding task (see Fig. 9).
|
0.5–1 s after entry into the reward receptacle (e.g., Fig. 4B1), but in many neurons, the inhibitions began earlier: a large proportion (23.5%) of the neurons with sustained inhibitions also had operant inhibitions (see Nicola et al. 2004), defined as inhibition during the brief interval between nose-poke and receptacle entry (e.g., Fig. 4A1). The sustained receptacle inhibitions were similar to sustained receptacle excitations in nearly every way, other than the sign of the firing rate change. The inhibitions began soon after entry into the reward receptacle, continued throughout brief exits from the receptacle, and recovered within 1 s of the final exit from the receptacle (Fig. 4, A–D). Furthermore, as shown in the examples in Fig. 4, A2 and B2, inhibition was not present after uncued receptacle entries. Analysis of median firing rates (Fig. 4C) and median firing rate decreases from 1 s after receptacle entry to receptacle exit (Fig. 4E) confirmed that inhibition was much smaller for uncued than for rewarded receptacle entries (P < 0.04, n = 93). Therefore like sustained receptacle excitations, the sustained inhibitions are likely to encode some aspect of reward consumption and may also encode information regarding whether reward was predicted by the cue.
|
|
|
BEHAVIOR. The excitations and inhibitions described in the preceding text were all significantly smaller when the response was uncued (and unrewarded) compared with when it was cued by the DS (and rewarded). The smaller firing change could be due to the fact that reward was not presented after uncued receptacle entries or to the fact that no cue indicated the presence of reward. To distinguish between these possibilities, we subjected three animals to a task identical to the basic DS task, except that reward was withheld from the animal following a random 40% of successful responses to the DS. This task is diagrammed in Fig. 6A. The animals' behavior demonstrated that they could not determine until after entry into the reward receptacle which trials would not be rewarded. First, the proportion of nose-pokes in response to the DS that were followed by receptacle entry was identical whether or not the nose-poke resulted in reward delivery (t34 = 1.5, P > 0.1; Fig. 6B). Second, the latencies between the nose-poke and receptacle entry were also indistinguishable in the two conditions (t34 = 2.0, P > 0.05; Fig. 6C). These results contrast strongly with the comparison of rewarded receptacle entries and uncued entries (Fig. 1, B and C), in which many fewer receptacle entries followed uncued than cued nose-pokes, and the latency between nose-pokes and entries was higher for uncued nose-pokes. Therefore in the random withholding task, animals were unable to determine until after entry into the reward receptacle whether reward was delivered after a DS-cued nose-poke.
|
Reward withholding task
FIRING PATTERNS. Receptacle entry excitations. Consistent with the behavioral evidence that animals could not determine prior to entry whether reward would be withheld, receptacle entry excitations were identical on rewarded and unrewarded trials of the random withholding task (P > 0.3, n = 10, Fig. 7). Because excitation was equivalent whether or not reward was present, the reduced excitation observed (in the basic DS task) during uncued entries compared with entries after a DS-cued nose-poke (Fig. 2) was not due to the absence of reward in the receptacle in uncued trials. Instead, receptacle entry excitations are dependent on reward-predictive cues.
Operant inhibitions. In Nicola et al. (2004), we showed that the magnitude of operant inhibitions was affected by the predictive value of the cue that elicited the operant response. Inhibitions were greatest when the operant response was to the DS, smaller when it was in response to the NS, and smallest when the operant response was uncued. Because operant inhibition is defined by inhibition between the operant response and receptacle entry and because these inhibitions usually continued until well past receptacle entry, it is conceivable that one reason why the inhibition is smaller after uncued responses is because reward is absent from the receptacle. To test this, we compared operant inhibition during rewarded and unrewarded trials of the random withholding task (Fig. 8). A typical neuron with operant inhibition exhibited no difference in this inhibition in rewarded and unrewarded trials (Fig. 8A). However, comparison of the median inhibition in the 1 s around receptacle entry (relative to the pre-DS baseline) showed a significant difference between these trial types (P < 0.04, n = 15, Fig. 8, B and C). The difference in median inhibition was exceedingly small (.15 Hz). Therefore operant inhibitions do not strongly encode the presence of reward, but depend on the degree to which reward is predicted by environmental stimuli.
|
Sustained receptacle inhibitions. Similar results were obtained for sustained inhibitions as for sustained excitations. Figure 10, A and B, shows that in typical neurons with sustained inhibitions, the inhibitions were reduced in duration in unrewarded trials due to the earlier exit from the receptacle. The magnitude of the inhibition was, however, indistinguishable for rewarded and unrewarded trials. These results are confirmed by the median histogram (Fig. 10C), which again shows that the magnitude of the inhibition is unchanged, but the time course is substantially shorter for unrewarded trials. When the median firing rate decrease from 1 s after receptacle entry to receptacle exit was compared, no significant difference was found between rewarded and unrewarded trials (P > 0.1, n = 17; Fig. 10D). Therefore similar to sustained receptacle excitations, the magnitude of sustained receptacle inhibitions was not affected by the presence of reward in the receptacle or the animal's location during receptacle entry. Rather inhibitions are likely a function of the motor behavior of reward consumption. Like sustained excitations, the sustained inhibitions may also reflect combinations of cues (i.e., the tone CS coupled with the reward receptacle itself, neither of which alone is sufficient to drive the inhibition).
|
Sustained receptacle exit excitations are correlated with reduced uncued operant responding
Although receptacle exit excitations were defined by excitation in the 1 s after receptacle exits, many receptacle exit excitations continued for substantially longer periods. This is illustrated by the histograms in Figs. 5C and 11C, which show that the median firing rate of these neurons does not return to the pre-DS baseline until many seconds after the receptacle exit. We therefore isolated a subpopulation of receptacle exit excited neurons with sustained excitations. This subpopulation was defined by significant (paired t-test P < 0.05 across trials) excitation from 10 to 11 s after the receptacle exit, relative to the 10 s pre-DS baseline; 27 neurons matched this criterion. An example of such a neuron recorded during the basic DS task is shown in Fig. 12A. In this case, the excitation remained above the pre-DS baseline for >60 s after receptacle exit.
|
35 s after receptacle exit (Fig. 12B). To determine the behavioral correlate of sustained receptacle exit excitation, we reasoned that uncued operant and receptacle entry behavior may be influenced by the timing of DS presentation. Figure 12C shows that the rate of uncued nose-poke responses across all basic DS task sessions was sharply reduced immediately after the receptacle exit compared with the period just before DS presentation. The uncued nose-poke rate then gradually increased to reach a peak before the next DS. This pattern can also be observed in the rasters in Fig. 12A (open triangles denote uncued nose-pokes; these are nearly absent in the minute or so after receptacle exit but occur with relatively high frequency in the minute before DS presentation). The rate of uncued receptacle entries tended to be somewhat higher just before the DS and after exits from the reward receptacle after DS-cued nose-poke-entry sequences (Fig. 12D). To determine if sustained receptacle exit excitation was correlated with the rate of uncued nose-poking or uncued receptacle entries, we computed the median firing rate of neurons with these excitations in 10-s bins in the 90 s after receptacle exits and the 90 s before DS presentation. We also computed the average rate of uncued nose-pokes and receptacle entries in the same 10-s bins across all sessions of the basic DS task. We found a strong linear correlation between the interval between uncued nose-pokes (the inverse of the average nose-poke rate in each bin) and firing rate (r2 = 0.66, P < 0.001; Fig. 12E). There was a weaker correlation in the opposite direction between the firing rate of these neurons and the interval between uncued receptacle entries (r2 = 0.32, P < 0.02; Fig. 12F). These results raise the possibility that sustained receptacle exit excitations inhibit nose-poking and facilitate receptacle entries after reward has been obtained. Histology
Figure 13 shows the approximate center of the electrode arrays; the arrays extended for 0.5 mm in both the anterior and posterior directions and 0.25 mm in the both medial and lateral directions from the center point. All electrodes were verified to be in the NAc.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Information encoded by receptacle entry excitations and operant inhibitions
Receptacle entry excitations began just before receptacle entries and usually ended within 1 s afterward (although some excitations that began at receptacle entry continued until receptacle exit). Operant inhibitions, which were described in Nicola et al. (2004), were maximal near receptacle entry as well. Both of these firing patterns were smaller in magnitude for uncued receptacle entries than for DS-cued entries. In contrast, neither were substantially different for rewarded and unrewarded receptacle entries in the random withholding task. Both rewarded and unrewarded entries in the random withholding task were elicited by the same cue. Therefore receptacle entry excitations and operant inhibitions are not well correlated with whether reward is present in the receptacle. Rather they depend strongly on the degree to which a cue has predicted that reward will be available in the receptacle.
Information encoded by sustained receptacle excitations and inhibitions
Sustained receptacle excitations and inhibitions began within 1 s of entry into the reward receptacle and continued until exit from the receptacle. These firing patterns therefore occurred precisely when the animal was engaged in consummatory behavior. The magnitude of the excitations and inhibitions cannot be accounted for by the sensory stimulus of reward (e.g., its taste) because the magnitude was the same in both rewarded and unrewarded trials of the random withholding task. It is possible that some residual, unconsumed sucrose remained in the receptacle and that the sensory stimulus resulting from consumption of this reward drove the excitations and inhibitions when we did not deliver reward. However, the magnitude of the firing rate change was much smaller during uncued stays in the reward receptacle even though residual sucrose could have been present in these cases as well. This finding rules out the possibility that certain sensory cues drove the firing patterns, such as the animal's location (i.e., in the reward receptacle) or the odor of sucrose in the dipper reservoir located below the reward receptacle. Unrewarded entries in the random withholding task differed from uncued entries in the basic DS task in two ways: an explicit tone CS was present during all entries in the random withholding task and the animal remained in the receptacle for substantially longer periods of time during unrewarded entries in the random withholding task even though in neither case was any reward available for consumption. Although the fact that the CS was always present when the firing changes were observed raises the possibility that sustained receptacle excitations and inhibitions depend on the presence of the CS, the excitations and inhibitions ended when the animal exited the receptacle even though the CS continued well past this point in most trials.
Another possible explanation is that these firing patterns encode information about the motor behavior of reward consumption. This is supported by the evidence that animals remained longer in the receptacle during unrewarded (but cued) trials than uncued (and unrewarded) trials. It is possible that the longer stay in the receptacle was due to attempts to consume reward (for instance, by licking the area of the receptacle where sucrose would otherwise have been delivered). Whether or not animals actively engaged in consummatory behavior during unrewarded entries, the longer stay is certainly a function of the fact that reward was predicted by the DS, of the animal's earlier operant response to it, and/or of the continuing CS. The sustained receptacle excitations and inhibitions may therefore also be influenced by these same factors. One possibility is that these firing patterns facilitate or drive attempts to consume reward based on the information that reward should be present. In this sense, the firing patterns present during reward consumption may be similar to the excitations and inhibitions elicited by cues such as the DS, which are significantly larger when the cue predicts reward than when it does not, even when the behavioral latency to respond to the cue is the same (Nicola et al. 2004). Additional experiments showing a direct relationship between NAc neuronal firing and specific consummatory muscle movements (and showing an impact of predictive cues on this relationship) are required to support this suggestion. Nevertheless, this interpretation is consistent with the idea that one general function of NAc neurons is to facilitate specific motor behaviors in response to the environmental stimuli that predict that those behaviors will result in reward (Chang et al. 1994; Mizumori et al. 1999; Nicola et al. 2004; Shidara et al. 1998).
Our findings extend those of others who have recorded from rat NAc neurons during the performance of fixed ratio tasks. When the behavior was reinforced with natural reward (sucrose solution, food pellets, or water), both excitations and inhibitions immediately after the operant response were found (Carelli and Deadwyler 1994; Carelli et al. 2000; Hollander et al. 2002; Roop et al. 2002), consistent with our observations. However, the earlier studies did not determine when animals began and ended reward consumption, and therefore it is difficult to determine whether the neural responses were more related to operant responding or consumption. We find that different populations of NAc neurons are excited or inhibited during operant responding, entry into the reward receptacle, and reward consumption. Combinations of these firing patterns are often observed in the same neuron, suggesting that the postoperant excitations and inhibitions described by Carelli and colleagues were composed of both operant- and reward-related firing. Postoperant excitations and inhibitions often lasted for several sec (Carelli et al. 2000; Roop et al. 2002), and others have observed similar long-lasting firing changes time-locked to delivery of (and presumably consumption of) ethanol reward (Janak et al. 1999). These findings suggest that the later stages of the firing changes observed after operant responses in FR1 tasks may in large part reflect reward consumption.
Interestingly, previous studies showed that neurons that displayed one type of phasic firing (e.g., excitation after the operant response) for one reward (sucrose, food, or water) tended to display the same phasic firing pattern for the other types of reward even though responses on different levers were required, different CS's accompanied delivery of each reward, and different motor actions (licking vs. chewing) were required to consume reward (Carelli et al. 2000; Roop et al. 2002). On the other hand, completely different populations of NAc neurons were excited and inhibited after operant responses for intravenous cocaine than for food or water (Carelli and Deadwyler 1994; Carelli et al. 2000). Because one of the key differences between cocaine and natural reward is that the animal must perform consummatory behavior after the operant response only for natural reward, the findings of Carelli and colleagues are most consistent with the idea that the postoperant phasic firing they observed in natural-reward tasks encoded information about the motor activity of reward consumption. Even though the specific motor behaviors required to consume food and water (but not liquid sucrose and water) are different, much of the required behavior was similar: for all of these natural rewards (but not for cocaine), the animal had to turn, move to the reward delivery site, stop, begin consumption, swallow, and maintain the posture required for obtaining reward. Consistent with this interpretation is the finding that postoperant excitations and inhibitions were absent during extinction (Hollander et al. 2002), when consummatory behavior should be absent or much reduced. However, reward itself is also absent during extinction, leaving open the possibility that postoperant firing encoded detection of the different rewards rather than consummatory motor behavior.
The idea that excitations and inhibitions during the animal's stay in the reward receptacle encode information about consummatory motor behavior is supported by a large number of behavioral pharmacology studies in which consumption or consummatory behaviors were measured after local injection of drugs into the NAc. Inhibition of the NAc with GABA receptor agonists (Basso and Kelley 1999; Reynolds and Berridge 2001; Soderpalm and Berridge 2000; Ward et al. 2000) or glutamate receptor antagonists (Kelley and Swanson 1997; Stratford et al. 1998) strongly increases consumption of freely available food or sucrose solution, whereas excitation of the NAc with a glutamate agonist decreases consumption in food-deprived animals (Stratford et al. 1998). Furthermore, µ opioid receptor agonists, which have generally been found to inhibit NAc neurons (Chieng and Williams 1998; Hjelmstad and Fields 2001; Hoffman and Lupica 2001; Yuan et al. 1992), also cause increased feeding when injected into the NAc (Kelley et al. 2002). These results indicate that the firing of some NAc neurons exerts an inhibitory effect on consumption that can be released by inhibition of these neurons. In our rats, many more NAc neurons were inhibited during consumption than were excited. A suggestive hypothesis is therefore that the firing of these neurons inhibits consummatory behavior; when their firing is reduced (either by injection of NAc-inhibiting compounds or when the animal encounters a situation where environmental cues indicate that reward should be available), consummatory behavior is disinhibited. Testing this hypothesis would require determining whether receptacle-associated inhibitions do in fact promote consummatory behavior.
The possibility that the firing of neurons with sustained receptacle inhibitions inhibits consumption has implications for how the firing of neurons active during appetitive behavior is interpreted. This is because about half of the neurons that were excited by cues and during the operant response were inhibited during reward consumption (Nicola et al. 2004, Table 3). Thus the firing of neurons with both appetitive excitation and consummatory inhibition may serve two roles: the increase in firing promotes a specific appetitive behavior and inhibits consummatory behaviors, whereas the decrease in firing inhibits the appetitive behavior and promotes (or disinhibits) consummatory behavior. This biphasic pattern of activity could facilitate switching from appetitive to consummatory behaviors. Interestingly, one of the most prevalent firing patterns in the NAc of rats during cocaine self-administration is a long-lasting inhibition that begins on cocaine delivery and slowly recovers before the animal begins to work for the next injection (Nicola and Deadwyler 2000; Peoples and West 1996; Peoples et al. 1998). The time course of inhibition parallels that of cocaine metabolism and is dependent on dopamine (Nicola and Deadwyler 2000), and the same neurons tend to show excitations just before and during operant behavior (Nicola and Deadwyler 2000; Peoples and West 1996). Inhibition of NAc neurons by cocaine may therefore mimic the inhibition that occurs during natural reward consumption and could explain how goal-directed behaviors other than cocaine-taking are inhibited during self-administration: the decreased firing of many NAc neurons after cocaine delivery may inhibit a broad range of possible appetitive behaviors.
The electrophysiological and pharmacological evidence from studies using rats supports the idea that the phasic firing of NAc neurons (particularly inhibitions) correlated with reward consumption promotes consummatory motor behavior. Inhibitions during consumption in primate ventral striatum have not been reported in detail, although striatal tonically active neurons (TANs) can be inhibited by reward delivery, particularly if the reward is unexpected (Apicella et al. 1997). Recordings
, rats used for the basic DS task or for both the basic DS task and the random withholding task (and for the data in Nicola et al. 2004
, rat that was recorded only during the random withholding task. Symbols are placed at the estimated center of the electrode array. Arrays were 0.75–1 mm long and 0.3–0.5 mm wide and were placed lengthwise along the anterior-posterior axis. Numbers on each coronal section indicate the distance anterior to bregma (Paxinos and Watson 1998