As noted above, one of the most popular notions in behavioral neuroscience has been that DA, particularly in nucleus accumbens, mediates the reinforcing impact of stimuli such as food, water, sex and drugs of abuse. Indeed, in a recent review article by Smith (1995) it was claimed that the DA hypothesis of reward had been "proved". Yet despite the popularity of this hypothesis, there are enormous difficulties with the idea that accumbens DA mediates the primary reinforcing characteristics of natural stimuli such as food (see Salamone et al., 1997).

Although it often is stated that accumbens DA depletions affect "reward" because they blunt stimulant self-administration, it should be emphasized that depletions of accumbens DA that severely affected cocaine self-administration had little effect on food reinforced responding (Caine and Koob, 1994; Roberts et al., 1977). Although Smith (1995; see also Schneider et al., 1989, 1990) emphasized that sucrose consumption can be affected by intra-accumbens injections of DA antagonists, there are numerous problems with these studies. Very high doses (i.e., nearly systemic doses) of SCH 23390 and raclopride were injected directly into the accumbens, so the anatomical specificity of the drug injections is questionable. Although Smith (1995) argued that the fact that the local lick frequency is not affected by intracranial DA antagonists proves that there is no motor effect being produced, such an argument is extremely spurious. In fact, the local lick frequency is set by brainstem pattern generators rather than the basal ganglia, and this frequency is not affected by even cataleptogenic doses of DA antagonists (Fowler and Das, 1994). Other motor parameters are affected by systemic and intra-accumbens DA antagonists, including lick efficiency, lick duration, lick force, lap volume and tongue extension (Fowler and Das, 1994; Fowler and Mortell, 1992; Gramling and Fowler, 1985; Hsiao and Smith, 1995; Jones and Mogenson, 1979; Schneider et al., 1990; Schneider et al., 1989).

Fundamental aspects of food reinforcement and food motivation are intact after interference with accumbens DA transmission. Appetitive taste reactivity to sucrose is unaffected by accumbens DA depletions (Berridge, 1996). Accumbens DA depletions did not affect the discrimination of reinforcement magnitude, and failed to alter response selection based upon reinforcement magnitude, in a food-reinforced T-maze task (Cousins et al., 1996; Salamone et al., 1994a). Several lines of evidence indicate that accumbens DA depletions do not generally suppress appetite. It has been reported that accumbens DA depletions or intra-accumbens injections of DA antagonists do not affect food intake (Bakshi and Kelley, 1990; Koob et al., 1978; Salamone et al., 1993a). Accumbens DA depletions have little or no effect upon the total amount of food obtained by FR1 lever pressing (Aberman and Salamone, submitted; McCullough et al., 1993; Salamone et al., 1995); the effects on response patterning that have been reported previously (i.e., initial slowing and slowing of the interresponse time distribution) did not resemble those of extinction (McCullough et al., 1993a; Salamone et al., 1995).

Indeed, the presumed relation between extinction and interference with DA systems appears to be a gross oversimplification; as emphasized in a recent review (Salamone et al., 1997), several articles have shown that, upon closer examination, DA antagonists and DA depletions produce effects that are distinct from those of extinction (Asin and Fibiger, 1984; Ettenberg and Carlisle, 1995; Evenden and Robbins, 1983; Faustman and Fowler, 1981, 1982; Gramling et al., 1984, 1987; Mason et al., 1980; McCullough et al., 1993a; Phillips and Fibiger, 1979; Salamone, 1986; Salamone et al., 1995; Spivak and Amit, 1986; Tombaugh et al., 1980, 1983; Willner et al., 1988).

As noted above, with rats on a concurrent FR5/chow feeding schedule, accumbens DA depletions or intra-accumbens injections of haloperidol decreased lever pressing but increased chow consumption (Cousins et al., 1993; Cousins and Salamone, 1994; Salamone et al., 1991; Sokolowski and Salamone, 1998). Thus, it seems untenable to maintain that accumbens DA depletions suppress lever pressing on some schedules because they produce a general reduction in food motivation. In fact, pre-feeding to reduce food motivation was shown to suppress both lever pressing and chow consumption on the concurrent lever pressing/chow feeding task (Salamone et al., 1991). As previously described, the effects of accumbens DA depletions do not closely resemble the effects of extinction. In this context, it is worth emphasizing that the effects of accumbens DA depletions also bear little resemblance to the effects of pre-feeding. A recent study (Aberman and Salamone, submitted) employed FR schedules with various ratio requirements (FR1, FR4, FR16, FR64), and it was observed that the effects of accumbens DA depletions were determined by the schedule; there was no significant effect of accumbens DA depletions on FR1 responding, and as ratio value increased, accumbens DA depletions caused greater suppression of responding.

A second experiment studied the effects of pre-feeding to reduce food motivation. Pre-feeding for 24 hours suppressed lever pressing on all four schedules tested, including the FR1 schedule. Across all schedules, if data are expressed as a percent of control responding, pre-feeding suppressed responding by approximately 50-60%. Thus, it can hardly be argued that the effects of accumbens DA depletions closely resemble the effects of pre-feeding.

In summary, several lines of evidence indicate that rats with accumbens DA depletions remain directed towards the acquisition and consumption of food, provided that the work requirement is relatively low. Several studies have shown that the effects of accumbens DA depletions do not resemble those of either extinction or pre-feeding. In view of the behavioral research and theory emphasizing that motivation is a critical aspect of primary reinforcement (Bindra, 1978; Thorndike, 1911; Timberlake and Allison, 1974; Timberlake, 1993; see reviews by Salamone, 1992, and Salamone et al., 1997), these results indicate that accumbens DA depletions do not interfere with the primary or unconditioned reinforcing properties of food.

As described above, the effects of accumbens DA depletions on food-reinforced lever pressing depend markedly upon the nature of the task being assessed. Lever pressing on some schedules (e.g., FR1, VI30) is relatively unaffected by accumbens DA depletions, while other schedules (e.g. FR64) appear to be highly sensitive to the loss of accumbens DA. Studies with the concurrent FR5/feeding procedure indicate that interference with accumbens DA, either by DA depletions or local injections of DA antagonists, affects the relative allocation of instrumental responses with different kinetic requirements.

For several reasons, which were reviewed above, it does not appear as though this pattern of effects is consistent with the DA hypothesis of reward. An alternative hypothesis that we have put forth is that rats with accumbens DA depletions are very sensitive to the kinetic requirements of the instrumental response being performed. In particular, it has been argued that accumbens DA depletions reduce the propensity for expending energy or exerting effort (Cousins and Salamone, 1994; Neill and Justice, 1981, Salamone, 1987, 1988, 1991, 1992; Salamone et al., 1994, 1997; Szechtman et al., 1994; see also Hsiao and Chen, 1995).

An important manifestation of this hypothesis is that instrumental tasks characterized by a high work output should be more greatly affected by accumbens DA depletions than tasks with a low work output. Indeed, it is possible to use this idea to explain the variability in the effects of accumbens DA depletions across different operant schedules. Schedules that generate only 300-600 responses per 30 min (e.g., CRF, fixed and variable interval 30 sec) show little or no effect after accumbens DA depletions (Cousins et al., in press; McCullough et al., 1993a; Salamone et al., 1995; Sokolowski et al., 1998). Schedules that generate moderately high rates (e.g. FR5, FR16, progressive ratio) are substantially impaired by accumbens DA depletions (Aberman et al., in press; Hamill et al., submitted; Salamone et al., 1993b), and schedules generating very high rates (e.g., FR64) are severely impaired (Aberman et al., submitted). In order to illustrate this principle, we have constructed a figure relating the extent of impairment induced by accumbens DA depletions to the baseline rates of responding generated by each schedule (Figure 1).

Fig.1: This is a scatterplot, showing the relation between baseline response rate per 30 min (i.e., responding in the vehicle-treated control group) and suppression of responding shown in rats with accumbens DA depletions. Data in this figure are taken from several different studies (i.e., each point represents a separate study; see text). Line represents least-squares regression line, which significantly fit these data. For schedules in which baseline response rate is higher, the suppressive effects of accumbens DA depletions are much greater.

This figure includes data from several separate studies from this laboratory. FR1 data are from Aberman et al. (submitted), and data on FI30 performance are from Cousins et al. (in press), while progressive ratio data are from Hamill et al. (submitted). Aberman et al. (submitted) studied FR4, FR16 and FR64 schedules. As shown in figure 1, there is a linear relation between the baseline rate of responding generated by a schedule and the degree to which that schedule can be suppressed by accumbens DA depletions. Higher baseline rates are associated with greater suppression by accumbens DA depletions, while lower baseline rates are associated with little or no suppressive effects. To emphasize that this precise relation is specific to the effects of accumbens DA depletions, figure 2 also shows an additional line for the effects of pre-feeding (data from Aberman et al., submitted).

Fig.2: This depicts two scatterplots, one of which shows the relation between baseline response rate per 30 min (i.e., responding in the control condition) and suppression of responding shown in rats that were pre-fed for 24 hrs (squares). Data in this figure are additional analyses performed on data reported in Aberman and Salamone (submitted). Line represents least-squares regression line, which significantly fit these data. Data from figure 1, for DA depleted rats, also are shown (circles). It can be seen that the relation between baseline responding and suppression of responding is substantially different in pre-fed animals than in DA-depleted ones.

Although pre-feeding also suppresses responding, it does so in a manner that is substantially different from that of accumbens DA depletions. An examination of these two sets of data clearly demonstrates that accumbens DA depletions produce effects that are quite distinct from those produced by pre-feeding, and also illustrates that the effects of accumbens DA depletions depend highly on the baseline rate generated by the schedule.

The relations depicted in figures 1 and 2 also serve to rebut another point that often is raised by proponents of the anhedonia hypothesis. It is sometimes claimed that the motor effects of interfering with DA systems "mask" the blunting of positive reinforcement that also is produced by the same manipulations. A detailed examination of figures 1-2 would not support that. Accumbens DA depletions do not simply produce severe motor impairments that more greatly affect responding than the "subtle" motivational effects would do. Accumbens DA depletions produce more of a suppressive effect than pre-feeding on high-rate schedules, but actually produce less of an effect than pre-feeding does on low-rate schedules. The FR1 schedule, which is the most fundamental example of a simple, continuous, primary reinforced task, is greatly affected by pre-feeding but is little altered by accumbens DA depletions.

The studies reviewed above involve investigations in which the behavioral effects of accumbens DA depletions were assessed. The development of in vivo neurochemical methods has allowed for the possibility of studying behavior and neurotransmitter activity concomitantly (e.g., Salamone et al., 1982, 1989, 1994; see review by Salamone, 1996). Several microdialysis studies have demonstrated that accumbens DA release is positively correlated with lever pressing rate. Responding on either FR1 or avoidance lever pressing schedules increased extracellular DA levels in accumbens, and significant positive linear correlations between the number of responses and the increases in DA were reported (McCullough et al., 1993a, 1993b). During FR5 responding, there also were substantial increases in accumbens extracellular DA over baseline, and a significant hyperbolic relation between responding and increases in DA (Salamone et al., 1994b).

The FI30 schedule is characterized by a higher rate of responding than the FI120 schedule, and also is accompanied by greater increases in DA release (Cousins et al., in press). A recent study (Sokolowski et al., 1998) examined DA release in accumbens core and shell during three operant tasks in the rat, in order to investigate in more detail the relation between DA release and lever pressing rate. Rats were trained to lever press on a fixed-ratio 5, variable-interval 30 sec, or a tandem variable time 30 sec/fixed ratio 5 schedules (referred to as a tandem VI/FR in the article); these three schedules were chosen because they generate a wide range of response and reinforcement rates. Attaching a ratio requirement to the variable time interval generated a very high response rate, but the density of food presentation was approximately the same as the VI 30 sec. After several weeks of training, dialysis probes were implanted into nucleus accumbens core or shell subregions. A single 30 min behavioral test was conducted during the dialysis test session. Rats lever pressing on each of the three operant schedules showed a significant increase in extracellular DA relative to the food-deprived control group during the behavioral session.

In addition, increases in extracellular DA in accumbens shell were found to be significantly greater than in the core during the lever pressing period. Across all three schedules, extracellular DA in the nucleus accumbens was significantly correlated with the number of lever presses performed, but was not correlated with the number of food pellets delivered. Analysis of covariance, which used amount of food consumed as the covariate, showed an overall group difference, indicating that DA levels increased in lever pressing animals even if one corrected for the amount of food consumed. These results indicate that DA release was more responsive in the accumbens shell than in the core during operant responding, and that increases in extracellular DA in nucleus accumbens shown by experienced animals are related to response rate rather than reinforcement magnitude.

From an energetic standpoint, individual cells, as well as complex multicellular organisms, are non-equilibrium open systems. Complex organisms must obtain energy by ingesting substances from their environment. Of course, in order to acquire energy in the form of nutrients, animals must expend energy in the form of muscle contractions. The acquisition of food involves more than just consummatory behaviors; it also involves instrumental actions that increase the proximity and availability of food. Because organisms are separated from significant stimuli such as food by environmental constraints or obstacles (i.e., response or procurement "costs"), these instrumental behaviors often are characterized by a high degree of vigor, persistence and work output.

The notion that motivated behaviors have an energetic or activational component is an old one, and there are numerous examples of this idea from the literatures of psychology (e.g., Cofer, 1972; Duffy, 1963; Killeen et al., 1978) and ethology (e.g. Hinde, 1970; Marler and Hamilton, 1966). In addition, this view is consistent with more recent "economic" models of operant conditioning, and the idea that response procurement "costs" affect operant responding (Allison, 1980; Collier and Jennings, 1969; Collier et al., 1986; Gannon et al., 1983; Hursh et al., 1988; Kaufman, 1980; Lea, 1978; Rashotte and Henderson, 1988; Staddon, 1979, 1983). In mammals, the brain is a critical structure for the regulation of complex behavior, and thus it is important to investigate the brain mechanisms involved in activational aspects of instrumental behavior. The major conclusion that can be drawn from the work reviewed above is that release of DA in nucleus accumbens is an important part of the neural process that enables organisms to overcome work-related response costs. In economic terms, nucleus accumbens DA is involved in the elasticity of demand for food.

As described above, studies involving response choice tasks or various schedules of reinforcement have demonstrated that the effects of accumbens DA depletions interact strongly with the work output typically seen on that particular task. Rats with accumbens DA depletions are impaired on operant schedules characterized by high levels of work output, and, on choice procedures, accumbens DA depletions cause animals to shift their relative response allocation in the direction of the task with lower work requirements. In addition, release of DA in nucleus accumbens is correlated with operant response output. Although it is difficult to identify precisely the behavioral functions of nucleus accumbens DA, it does appear that depletions of DA reduce the propensity for expending energy or effort, and that accumbens DA release could be one of the mechanisms through which response output is regulated. The regulation of global aspects of work output can be conceived of as a higher-order sensorimotor process, but also as an aspect of motivation (for reviews, see Salamone, 1987, 1992; Salamone et al., 1997).

It is possible that nucleus accumbens, through its anatomical connections, participates in frontal cortical control of behavioral regulation. Although nucleus accumbens may not directly assess reward value, or perform cost/benefit analyses, it is possible that accumbens DA transmission sets constraints on energy expenditure that profoundly influence the relative allocation of instrumental responses toward various alternatives. As well as shedding light on the neural control of motivated behavior, research into the functions of nucleus accumbens DA also could have important implications for clinical studies of energy-related disorders, such as anergia or apathy (Campbell and Duffy, 1997; Duffy and Kant, 1997).