In the rate-frequency curve-shift procedure, the frequency of the stimulation is varied systematically, whereas the other parameters are held constant, so that the population of neurons that is stimulated is held constant. Animals typically press a lever manipulandum, nosepoke in a hole, turn a wheel manipulandum, or walk a runway to receive a train of stimulation. The completion of the operant response (e.g., lever-press) triggers a constant current generator to deliver a train of rectangular cathodal pulses of constant duration and intensity and variable frequency. The pulse ­frequency (i.e., number of pulses within a train) is progressively increased up to 40–50 pulses per stimulation train until the ­animal shows vigorous self-stimulation. As the frequency of the stimulation increases, the response rates rise accordingly until a maximal point of stimulation is delivered over which no further increases in response rates are produced and response rates reach an asymptote (15, 16). During the acquisition phase, the animals are trained to self-stimulate for at least 3 consecutive days, using stimulation parameters that maintain near maximal rates of responding. After the self-stimulation behavior has been acquired and stabilized for a given pulse frequency that supports maximal response rates, animals are trained to self-stimulate using four alternating series of ascending and descending pulse frequencies. The pulse frequency is varied by steps of approximately 0.1 log units. Each frequency is tested within trials of 50–60 s duration, followed by a timeout period of 5–30 s. These parameters can be varied between studies. At the beginning of each trial, the animals receive three to five trains of priming stimulation at the frequency of the stimulation that is available for that trial. The mean number of the four alternating ascending and descending pulse frequencies in each step leads to the generation of a sigmoidal frequency-response function, very similar to the dose-response function seen in pharmacological studies. A frequency-response function is established daily until the self-stimulation indices (i.e., threshold and asymptote measures) stabilize. Each of these daily sessions has a duration of approximately 40–50 min. Threshold and response rate asymptote measures are estimated based on the response function. Threshold is defined as the frequency, intensity, pulse width, or train duration that supports an arbitrary level of performance (see below for methods of estimation). The asymptote is the maximal response level exhibited by the subject. Thresholds reflect the rewarding efficacy of the stimulation, and response rate asymptotes are an estimate of an animal’s ability to perform the task and reflect motor or performance effects. Different quantitative measures are used in the rate-frequency curve-shift procedure to obtain threshold and asymptote ­(performance capability) estimates. The most reliable and commonly used measures are the M ₅₀ (half-maximum) (15, 18), the Gompertz sigmoid growth model (which resembles the M ₅₀) (38), and the Θ₀ (39) (Fig. 1)_. The M ₅₀ measure is analogous to the Effective Dose 50 (ED ₅₀) used in pharmacology and refers to the stimulation frequency (or intensity or duration) that sustains responding at 50% of the maximal rate. The Θ₀ measure refers to the theoretical point at which the stimulation becomes rewarding (i.e., response rates become higher than 0) (39). The M ₅₀ measure can show artificial shifts attributable to increases or decreases in maximal response rates, whereas the Θ₀ measure has been shown to be less affected by performance manipulations (8, 17, 18, 39). The Gompertz sigmoid growth model is a psychophysical method applied to the description of ICSS behavior:

In this equation, α represents the maximal rate (asymptote), and Xi (X at inflection) represents the threshold frequency. The latter is the pulse number producing 36.7% of the asymptotic rate (i.e., the rate lying on the fastest-accelerating region of the curve). Parameter b represents an index of the slope, and e is the base of natural logarithms. The advantage of this model compared with other estimates is that it accounts for every data point in the dose-response curve to measure the threshold and asymptote values (38).

Thus, this method enables distinguishing between reward and performance and quantifying the drug effect (8, 11, 18, 37, 39–41). The quantitative scaling of changes in reward is important for two reasons. First, the size of the reward-facilitating or reward-attenuating effect is useful when comparing the effects of different manipulations. Second, even small changes in brain stimulation reward or performance capability can be detected because a quantitative measure is less likely to have ceiling or floor effects (42). The same procedure may be used while varying the current-intensity (43) or train duration (12), generating rate-intensity or rate-duration curve functions, respectively, while keeping the other two parameters constant.

The discrete-trial current-intensity threshold procedure is a modification of a task initially developed by Kornetsky and Esposito (19). This procedure consists of discrete trials. Each trial begins with the delivery of a non-contingent electrical stimulus followed by a 7.5 s response window within which the subject can make a response to receive a second contingent stimulus identical to the initial non-contingent stimulus. Similar to all other ICSS procedures, this response may be a lever press, nose poke, or turning a wheel manipulandum, with the latter being the most commonly used response with this procedure. A response during this time window is a positive response, and the lack of a response is a ­negative response. Additional responses during a 2 s period ­immediately after a positive response have no consequence and are counted as extra responses. The inter-trial interval that follows either a positive response or the end of the response window (in the case of a negative response) has an average duration of 10 s (ranging from 7.5 to 12.5 s). Responses that occur during the inter-trial interval are timeout responses and result in a further 12.5 s delay of the onset of the next trial. During training on the discrete-trial procedure, the duration of the inter-trial interval and delay periods induced by timeout responses are gradually increased until animals perform consistently for a fixed stimulation intensity at standard test parameters. The animals are subsequently tested on the current-intensity threshold procedure in which stimulation intensities are varied according to the classical psychophysical method of limits (44). A test session consists of four alternating series of descending and ascending current intensities starting with a descending series (Fig. 2). Blocks of three to five trials are presented to the subject at a given stimulation intensity, and the current intensity is changed by steps of 5–10 μA between blocks of trials. The initial stimulus intensity is set approximately 40 μA above the baseline current-intensity threshold for each animal.

Each test session typically lasts 30–40 min and provides four dependent variables for behavioral assessment: threshold, response latency, extra responses, and timeout responses. The current-threshold for each descending series is defined as the stimulus intensity between the successful completion of a set of trials (i.e., positive responses during two or more of the three trials) and the stimulus intensity for the first set of trials, of two consecutive sets, during which the animal fails to respond positively on two or more of the three trials (in the case when three trials are presented at each current intensity). The current-threshold for each ascending series is defined as the stimulus intensity between a current intensity for which the animal fails to respond positively on two or more of the three trials and the first set of trials, of two consecutive sets, during which the animal responds positively on two or more of the three trials. The mean of the four series’ thresholds is defined as the threshold for the session. The time interval between the beginning of the non-contingent stimulus and a positive response is recorded as the response latency. The response latency for each session is defined as the mean response latency on all trials during which a positive response occurs. The extra responses and timeout responses for each test session are defined as the total number of extra responses and timeout responses that occur in the session, respectively. The response latency measure detects performance effects (11), extra responses primarily reflect the force with which a subject turns the wheel, and ­timeout responses reflect response inhibition (45). The main ­advantage of this ­procedure is that it allows for a rate-independent measurement of reward thresholds. Thresholds are defined independently of rate of responding which can be low because timeout responses result in postponement of the next trial (11). The discrete-trial current-intensity procedure has been used primarily with variations in current intensity, although other stimulation parameters can also be varied while keeping the current-intensity constant (8, 11).

In both of the above procedures (rate-frequency curve-shift and discrete-trial current-intensity threshold), data obtained for the threshold and response asymptote or response latency are usually expressed as a percentage of baseline to avoid the complexity of individual subjects having different baseline values for each of the measures (39).

Psychomotor stimulants, such as amphetamine, cocaine, and methamphetamine, are monoamine reuptake inhibitors or releasers, which exert their effects by activating mainly the dopaminergic, noradrenergic, serotonergic, and glutamatergic neurotransmitter systems of the brain (85–89). These brain systems are localized, among others, in brain regions highly implicated in brain reward functions, such as the NAcc, VTA, and PFC (90, 91).

Amphetamines and cocaine potentiate brain stimulation reward when they are administered either intracranially or systemically (3, 19, 20, 22, 35, 51, 53, 92–100). Specifically, a plethora of studies have shown that acute cocaine administration potentiates ICSS behavior reflected in increased response rates and lowering of ICSS thresholds in rats and mice (Fig. 3), whereas during cocaine withdrawal, decreased response rates and elevated ICSS thresholds were observed (19, 20, 89, 93, 101–128) (Fig. 4). Using different ICSS methods and stimulus parameters, acute administration of a variety of cocaine doses, ranging from 0.6 to 20 mg/kg, dose-dependently increased response rates (101) or lowered ICSS thresholds (19, 20, 93). Extending these findings, Markou and colleagues reported that repeated cocaine administration (10 mg/kg for 20 consecutive days) lowered ICSS thresholds, with no tolerance or sensitization to this effect (116). This latter procedure led to classical conditioning of the reward-facilitating effects of cocaine, such that after cessation of cocaine administration, a conditioned lowering of reward thresholds was observed upon re-exposure to the same testing conditions and following a saline injection (116). Kelley and Hodge shed more light on the reward-facilitating effects of cocaine by showing that these effects are mediated through 5-HT₃ receptor activity (89). Other research groups demonstrated that other systems in the brain, such as hypocretin, γ-aminobutyric acid, and cannabinoid, may also affect the reward-facilitating effects of cocaine (119, 121, 123, 124). Interestingly, using the discrete-trial current-intensity method, Markou and colleagues found that self-administration of low doses of cocaine (10 and 20 injections of 0.25 mg/kg) lowered reward thresholds, an effect that only lasted for a short time (less than 2 h). Higher doses and testing in the ICSS procedure at later time points revealed no reward-enhancing effect of cocaine (129). These effects of self-administered cocaine were similar to those seen after intraperitoneal administration of 10 mg/kg cocaine (see above) and did not differ between male and female rats (130).

Other studies using the rate-intensity or rate-frequency ­curve-shift methods also showed that cocaine leads to facilitation of brain stimulation reward, indicated by lowering of ICSS thresholds (110, 111, 115). Frank and colleagues (103) and van Wolfswinkel and colleagues (104) used even higher doses of cocaine (5–30 mg/kg, IP or SC) and tested the animals in a varied train duration or current-intensity ICSS procedure, respectively, with the ICSS electrode implanted at the level of the VTA. Cocaine significantly lowered train duration thresholds, again demonstrating a reward-facilitating effect of cocaine regardless of the ICSS method used (103, 104). Similar effects were also seen after acute or subchronic administration of cocaine when the electrode was located in the ventral pallidum (112), PFC (106), or medial PFC (107) or when the operant response varied (i.e., nosepoking or lever-pressing) in a rate-dependent ICSS procedure (108). Specifically, cocaine (1.5–10 mg/kg, IP) dose-dependently lowered ICSS thresholds of the ventral pallidum or lateral hypothalamus in the rate-frequency curve-shift procedure (112, 114), and this effect was not as pronounced when the animals were treated with methylphenidate during early developmental stages (120).

Using the rate-frequency curve-shift ICSS procedure in mice, acute cocaine administration (2.5–20 mg/kg, IP) dose-­dependently lowered brain stimulation reward thresholds (115), whereas repeated administration of high-dose cocaine (40 mg/kg) decreased response rates and elevated thresholds (105). Consistent with one of the advantages of ICSS is the finding that these effects of cocaine are independent of the psychomotor stimulant effects of cocaine. Doses that induce enhancement of brain stimulation reward do not lead to significant effects on the performance capability of the animals (111, 112).

During cocaine withdrawal, the impact of rewarding brain stimulation is attenuated, quantified by elevations in ICSS thresholds, indicating a depression-like state (53, 105, 109, ­ 131–135) (Fig. 4). In a rate-intensity ICSS procedure, rats received repeated administration of a high cocaine dose (40 mg/kg, IP) immediately after each daily ICSS session for 7 days and were tested 24 h later (105). This repeated cocaine administration led to decreased response rates and elevated thresholds in the ICSS testing of each subsequent day (24 h post-injection). Repeated cocaine administration (30 mg/kg, twice daily) for 18 days led to lowering of thresholds and decreases in response rates when animals were tested for ICSS 5 days after discontinuation of cocaine administration (105). Importantly, in studies by Markou and Koob, in which the discrete-trial current-threshold ICSS procedure was used and cocaine (0.5 mg/kg/injection) was intravenously self-administered by rats under a fixed-ratio 5 schedule of reinforcement for 3–72 h continuously, thresholds were elevated for up to 72 h post-cocaine self-administration for the longest cocaine exposure (53, 109, 131). This anhedonic effect of cocaine withdrawal has been shown to be associated with decreased ­dopaminergic (109, 136), decreased noradrenergic (131), and increased adenosine (134, 135) neurotransmission. Furthermore, Markou and colleagues showed that repeated administration of the tricyclic antidepressant desmethylimipramine (DMI) (10 mg/kg, IP, twice daily for 5 days) significantly downregulated β-adrenergic receptors and shortened the duration of post-cocaine anhedonia reflected in reversal of the threshold elevations 1, 3, and 6 h after the termination of a 12 h cocaine self-administration session (0.5 mg/kg/injection) in the discrete-trial current-threshold ICSS procedure (131). In this study, the level of β-adrenergic receptor downregulation correlated significantly with the degree of effectiveness of DMI treatment in reversing post-cocaine anhedonia (131). Baldo and colleagues showed that repeated administration of cocaine (15 mg/kg, IP, eight injections over 9 h) produced elevations in thresholds 4, 8, and 12 h after cocaine administration, and this effect was reversed by administration of the adenosine-2 (A₂) receptor-selective antagonist 3,7-dimethyl-1-propargylxanthine (DMPX) (3 and 10 mg/kg) administered before both the 8 and 12 h post-cocaine self-stimulation testing sessions (134). Goussakov and colleagues recently showed using the rate-frequency curve-shift ICSS procedure that cocaine withdrawal attenuated the impact of rewarding brain stimulation, reflected in elevated ICSS thresholds, an effect that was evident after 3 and 5 days of a 7-day regimen of “binge” cocaine treatment (15 mg/kg/injection, IP, 3 daily injections at 1 h intervals) (133).

Similar to cocaine, D-amphetamine administration induces reward-enhancing effects in the ICSS procedure in rats, with ­electrodes located in the lateral hypothalamus, dorsal ­noradrenergic bundle, PFC, substantia nigra, and VTA (22, 96, 137–159) (Fig. 3). Withdrawal from D-amphetamine results in elevations in ICSS reward thresholds (51, 160–172) (Fig. 4).

Cazala assessed the effects of D- and L-amphetamine in three different mouse strains with electrodes implanted in dorsal and ventral hypothalamic areas (0.25–8 mg/kg, IP) and found that both isomers increased self-stimulation response rates in all three strains, although with different sensitivity for each mouse strain (157). Using a modified hole-board task for self-stimulation in mice, Kokkinidis and Zacharko found that low D-amphetamine doses increased, and high doses decreased, ICSS response rates (51). A recent study by Elmer and colleagues, using the rate-­frequency ICSS procedure in dopamine D₂ receptor deficient mice, showed that amphetamine (1–4 mg/kg, IP), in contrast to morphine, potentiated brain stimulation reward across the three genotypes used in the study (D₂ wildtype, heterozygous, and knockout mice), indicated by equal leftward shifts in the rate-frequency functions (139).

D-amphetamine appears to elicit differential anatomical sensitivity, measured by selective increases in response rates for ICSS on a fixed-ratio 1 schedule of reinforcement in rats prepared with electrodes in various brain sites (154). Specifically, rats prepared with electrodes in the septal area, the anterior and posterior hypothalamus (0.62, 2.5, and 5 mg/kg), or the ventromedial tegmentum and locus coeruleus (0.5–4 mg/kg) exhibited increased responding after D-amphetamine administration, with the highest responding when electrodes were implanted in the ventromedial tegmentum and posterior lateral hypothalamus (154). Another study showed that a low amphetamine dose (0.1 mg/kg) induced higher facilitation of brain stimulation reward when the electrode was placed more dorsally or medially in the substantia nigra than when it was placed close to the dorsal border of the substantia nigra (155). Additionally, self-stimulation of the PFC was facilitated by chronic D-amphetamine administration (1.5 mg/kg for 9 days), indicated by increases in response rates, whereas self-­stimulation of the NAcc and supracallosal bundle remained unchanged or was suppressed, respectively (152).

In a study that used a rate-intensity ICSS procedure, 2 and 4 mg/kg of amphetamine shifted the curve to the left, indicating enhanced rewarding efficacy of brain stimulation (96). Consistent findings were observed in the autotitration procedure, in which D-amphetamine (0.03–1 mg/kg, IP) dose-dependently lowered thresholds (138). In a variant train duration procedure, amphetamine, administered at three different doses (0.33, 0.66, and 1 mg/kg, IP) counterbalanced over 3 consecutive days, dose-dependently lowered ICSS thresholds, indicating a reward-enhancing effect (145).

Using the discrete-trial current-intensity procedure, Markou and colleagues showed that exposure of rats to chronic mild stress for a period of 19 days potentiated the enhancement of lateral hypothalamic brain stimulation reward induced by acute administration of amphetamine (1–6 mg/kg, IP, for 2 days) compared with nonstressed animals (143). In a recent study using the rate-frequency curve-shift procedure, acute administration of D-amphetamine into the NAcc shell (1 μg/side) significantly lowered thresholds measured by decreases in M ₅₀ values and did not affect maximal rates of responding (141).

Withdrawal from D-amphetamine decreases brain stimulation reward. Amphetamine administered three times per day for 5 days had an anhedonic effect, indicated by decreases in response rates of electrical brain stimulation when rats were tested 24 h following discontinuation of amphetamine administration and for at least 5 days (171). In studies by Markou and colleagues using the discrete-trial current-intensity procedure, continuous amphetamine exposure using 6-day subcutaneous osmotic minipumps (5, 10, and 15 mg/kg/day) dose-dependently lowered brain reward thresholds and decreased response latencies during two consecutive amphetamine exposures (163). By contrast, cessation of administration of the same amphetamine doses delivered via minipumps for 6 days dose-dependently elevated thresholds, an effect that lasted for 5 days for the highest amphetamine dose (163). Termination of intraperitoneal administration of a variety of amphetamine doses (1–5 mg/kg, three times daily for 1, 2, 4, or 6 days) led to ICSS threshold elevations (173). In a study by Markou and colleagues (164), amphetamine was administered intraperitoneally three times per day for 4 days in a rising-dose regimen starting from 1 mg/kg and stabilizing at 5 mg/kg. Thresholds and response latencies were determined for up to 156 h after the last administration in the discrete-trial current-intensity threshold ICSS procedure. Amphetamine withdrawal resulted in decreased reward, reflected in elevated brain reward thresholds, and these effects were alleviated by co-administration of the selective serotonin reuptake inhibitors fluoxetine or paroxetine and the serotonin 5-HT_1A receptor antagonist p-MPPI, indicating that decreased serotonergic function may mediate the effects of amphetamine withdrawal (164, 167). Using a rate-intensity ICSS procedure and stabilizing amphetamine doses at 10 mg/kg (1–10 mg/kg, IP, three times per day for 4 days), Barr and colleagues found that rats exhibited decreased levels of ICSS responding for up to 60 h after the last amphetamine administration, an effect that was counteracted by repeated administration of electroconvulsive shock (172). Previous studies have also reported similar effects of amphetamine withdrawal after cessation of repeated intraperitoneal administration of amphetamine (162, 168, 170, 173).