Rats and other vertebrates will work avidly to trigger delivery of electrical stimulation to certain brain sites. The potent rewarding effect produced by the stimulation ("brain stimulation reward," BSR) can compete with, summate with, and substitute for the rewarding effects of natural goal objects, such as food and water. Thus, the stimulation appears to inject a signal into the central nervous system (CNS) that mimics those produced by natural rewards. Given that the electrically induced rewarding effect originates as a volley of observable action potentials in axons coursing past identifiable CNS sites, the phenomenon of BSR has long been regarded as a gateway to tracing the neural circuitry involved in the pursuit of natural rewards. It has also been proposed that dependence-inducing drugs gain their grip over
Intracranial Self-Stimulation. Fig. 1. Sagittal outline of a rat brain showing the approximate location of regions supporting ICSS. (Reproduced from Wise RA (1995) Annu Rev Neurosci 19:319-340.) Please see original for the numeric key.
Intracranial Self-Stimulation. Fig. 2. Variation of self-stimulation threshold as a function of electrode depth. (Reproduced from Forgie ML, Shizgal P (1993) Behav Neurosci 107(3):506-524. Data on the left are from Rompre PP, Miliaressis E (1985) Brain Res 359:246-259.)
reward signal is represented by a rate code, which sums impulse flow across the population of activated neurons within a time window defined by the stimulation train. The resulting spike count is then transformed nonlinearly into a single time-varying quantity representing the intensity of the reward; growth of the reward signal decelerates over time and eventually levels off (darker curve). Within a time window of given duration, the reward signal grows as a function of the stimulation-induced spike count and eventually saturates (lighter curve). The peak reward intensity achieved during a stimulation train is recorded in memory (not shown) and translated ultimately into behavior.
A vertical gray line in Fig. 3 demarcates the boundary between the BSR-specific and more general processes involved in the control of the reward-seeking behavior. The elements to the left of the gray line include the directly stimulated neurons and the circuitry that integrates their output over time and space, compressing the effects of the electrically triggered volley into a single quantity recorded in memory. To the right of the gray line, this stored reward-intensity signal is combined with information about a) the probability that a reward will be delivered once the response requirement (e.g., pressing the lever a given number of times) has been satisfied, b) the physical effort (''effort cost'') and time
Intracranial Self-Stimulation. Fig. 3. Translation of the volley of action potentials triggered by the electrode into reward-seeking behavior. (Based on Fig. 1 in Arvanitogiannis and Shizgal (2008).)
(''opportunity cost'') required to do this, and c) the delay (not shown) between meeting the response requirement and delivery of the reward. In keeping with the generalized form of the ► matching law and behavioral decision theories in general, the subjective values of all these variables are combined in scalar fashion to yield an estimate of the payoff the subject can expect in return for satisfying the response requirement. Finally, in the spirit of the generalized single-operant matching law, the payoff from BSR (UB, suitably transformed) is compared with the sum of all the (suitably transformed) payoffs available in the test environment, which include the payoffs (UE, suitably transformed) from behaviors such as grooming, exploring, resting, etc. The result determines the proportion of time (''time allocation'' (TA)) devoted to the pursuit of the electrically induced reward. Time allocation grows as the payoff from BSR is increased (darker curve) and as the payoff from competing activities falls (lighter curve).
Preliminaries: measuring the effects of drugs on BSR. The first psychopharmacological study of BSR was published within 3 years of the discovery of the phenomenon by Olds and Milner in 1953. The rate of lever pressing served as the behavioral measure of drug action. The deficiencies of that measurement method rapidly became clear. As can be seen at the right of Fig. 3 (''behavioral allocation function''), the payoff from BSR is translated into observable behavior by a saturating function. Thus, different levels of payoff can result in the same maximal level of time allocation, response rate, etc. For example, once the stimulation current or frequency are sufficiently high, response rates are not boosted by further
Intracranial Self-Stimulation. Fig. 4. Cancelation between the effects of amphetamine and pimozide on ICSS. (Reproduced from Gallistel CR, Karras D (1984) Pharmacol Biochem Behav 20(1):73-77.)
increments, but when offered a choice, rats will exhibit a preference for the stronger stimulation. A more general problem stemming from the use of response rates as the dependent variable can be seen in Fig. 4. As is typically the case, sigmoidal functions translate stimulation strength into behavioral output; a third variable, doses of two drugs that exert opposite influences on dopaminergic neurotransmission, displaces the sigmoidal psychometric functions laterally along the logarithmic axis representing stimulation strength. If only a single value of stimulation strength (in this case, the pulse frequency) had been employed, the measured effects of the ► neuroleptic drug and the combination of the neuroleptic with a ► psycho-motor stimulant would have depended on the arbitrary choice of pulse frequency. For example, had 160 pulses s_1 been selected, little influence of the drugs would have been apparent, but a profound influence would have been seen had 63 pulses s_1 been chosen instead. In contrast, if the curves are sectioned horizontally, rather than vertically, the measured effects of the drugs are essentially independent of the chosen criterion; similar shifts are apparent at behavioral criteria of 25%, 50%, and 75% of the maximal response rate. Thus, by measuring responding for the stimulation over a broad range of pulse frequencies, the countervailing influences of the two drugs can be discerned.
Quantifying the influence of drug treatments on ICSS by measuring lateral displacements of psychometric functions is called the "curve-shift" method (Miliaressis et al. 1986). In addition to the benefits described above, it is also argued that this method can distinguish changes in performance caused by disruptive effects of drugs, such as sedation or stereotypy, from bona fide influences on BSR. Performance disruption is said to alter the maximal rate of responding and the slope of the psychometric functions, whereas, changes in reward effectiveness are said to produce lateral shifts. If correct, this statement would confer a most convenient property on the curve-shift method. Unfortunately, there is contradictory evidence demonstrating that performance variables, such as alterations in the force required to depress the lever, produce systematic lateral curve shifts. Figure 3 argues that such effects should be expected: increases in effort cost decrease payoff but can be offset by increases in reward intensity.
Arvanitogiannis and Shizgal (2008) have delineated a related problem with the curve-shift method. They show that a given curve shift can arise from changes in different variables controlling reward pursuit. In effect, the same shift produced by an action of a drug to the left of the vertical gray line in Fig. 3, such as potentiation of spatio-temporal integration, can be mimicked by an action to the right of the vertical gray line, such as a reduction in subjective effort cost.
The curve-shift method has become the dominant means of assessing the influence of drugs on BSR, and it was employed in most of the work summarized in this essay. However, a new method is required to distinguish actions of drugs on the two sides of the vertical gray line in Fig. 3. Arvanitogiannis and Shizgal (2008) have proposed that by varying both the strength and cost of reward, the influence of drugs on components of the reward circuitry prior to the output of the spatiotemporal integrator in Fig. 3 can be distinguished unambiguously from influences brought to bear on downstream components. Successful application of their method would bring the field into the third turn of a spiral course of progress.
Experiments performed using the rate measurement proved useful in identifying drugs that are effective in altering performance for BSR. The curve-shift method provided a clear methodological advance by eliminating the dependence of drug-induced changes on arbitrarily selected stimulation parameters. The three-dimensional method developed by Arvanitogiannis and Shizgal (2008) promises to better distinguish the influence of drugs on different components of the circuitry underlying pursuit of the rewarding stimulation and to better distinguish effects of drugs on performance capacity from effects on reward integration. However, Fig. 3 points to a limitation of this new method: numerous interacting variables (depicted to the right of the vertical gray line) influence performance in ways that will be indistinguishable unless the mapping of the objective to the subjective values of these variables is nonlinear and these nonlinearities can be exploited experimentally. Thus, there will be a need for continued methodological advances in order to fully account for the effects of drugs on BSR and to maximize the contribution of such experiments in determining the neurochemical basis of reward.
The neurochemical basis of BSR. In the following sections, we summarize the contributions of the most extensively studied neurotransmitter systems to ICSS.
► Dopamine is the neurotransmitter most closely associated with BSR (Wise and Rompre 1989). Reductions in dopaminergic neurotransmission lower the effectiveness of the electrical stimulation in supporting ICSS, leading to rightward displacements of psychometric functions measured by the curve-shift method; as can be seen in Fig. 4, a higher stimulation strength is required to produce a given rate of lever pressing following administration of a dopamine receptor blocker ("► Pimozide 0.3 mg/kg''). Conversely, drugs that enhance dopaminergic neurotransmission increase the effectiveness of the electrical stimulation in supporting ICSS and thus, produce leftward shifts ("► Amphetamine 2 mg/kg''); such drugs reduce the simulation strength required to produce a given level ofbehavioral output. When two drugs that exert opposing influences on dopaminergic neurotransmission were coadministered, their effects on performance for BSR canceled ("Pimozide 0.3 mg/kg and Amphetamine 2 mg/kg''). Similar cancelation has been achieved by pitting the selective ► dopamine transporter blocker, GBR-12909 against the D-1 receptor blocker, SCH-23390. Drugs that block D-2 receptors have also been shown to produce rightward curve shifts, and there is evidence that D-1 and D-2 receptors exert synergistic influences in this regard.
Dopamine neurons manifest multiple activity states. Under resting conditions, roughly 50% of the midbrain population is quiescent. Two active states have been described: sustained low-frequency (tonic) firing and intermittent high-frequency (phasic) bursting. Self-stimulation of the MFB is accompanied by a prolonged increase in tonic output, which is manifested in an increased extracellular concentration of dopamine, as observed by means of in vivo microdialyis probes in dopamine terminal fields. Phasic release, as measured by means of fast-scan cyclic voltam-metry, is also observed in self-stimulating rats. In an early report, these brief increases in the local extracellular concentration of dopamine were described as transient, falling below the detection limit within a minute or so of the onset of self-stimulation. However, in more recent experiments that incorporate a 10-s post-reward time-out, a dopamine transient was recorded in the nucleus accumbens following each and every pulse train delivered via electrodes at the ventral tegmental level of the MFB.
As the evidence reviewed above suggests, there is wide agreement that dopaminergic neurotransmission is deeply implicated in ICSS. Nonetheless, a lively debate continues over exactly what role(s) dopaminergic neurons and their different activity states play. For example, many authors assume that the rewarding effect ofstimulating the MFB is due to direct activation of dopaminergic fibers. However, the properties of these fibers are largely incompatible with this idea. The axons of dopamine neurons are unmyelin-ated and of fine caliber; their thresholds for activation by extracellular currents are high. Thus,relatively few such fibers should be excited directly under the typical conditions of BSR experiments, which entail the use of stimulation electrodes with large exposed tips and currents that are low with respect to the thresholds of dopaminergic fibers at the short-pulse durations commonly employed. The refractory periods of dopaminergic fibers are long and their conduction velocities low in comparison with the estimated values for the directly stimulated axons mediating self-stimulation of the MFB. Given the limited overlap between the excitability properties of dopaminergic fibers and those of the directly stimulated fibers mediating BSR, it would appear that activation of midbrain dopaminergic neurons during ICSS is achieved largely via a trans-synap-tic route (Shizgal 1997). As predicted by this proposal, blockade of ► glutamate receptors in the VTA decreases the magnitude of ventral striatal dopamine transients elicited by rewarding VTA stimulation.
Moisan and Rompre (1998) have proposed a way to reconcile the influence of dopaminergic manipulations on ICSS with the mismatch between the properties of dopa-minergic neurons and those of the directly activated neurons underlying BSR. They first varied the current and pulse frequency of rewarding MFB stimulation so as to determine two sets of stimulation parameters that produced the same level of behavioral responding: one that activated many directly stimulated neurons at low frequency and a second that activated fewer directly stimulated neurons at higher frequency. They then showed that putative midbrain dopamine neurons trans-synaptically activated by the rewarding stimulation fired at similar rates in response to the two different sets of stimulation parameters. Thus, the firing of the dopamine neurons reflects the "counter property'' of spatiotemporal integration previously described in behavioral studies of BSR (depicted by the S symbol in Fig. 3). On this basis, Moisan and Rompre proposed that midbrain dopamine neurons may compose an integral part of the spatiotemporal integrator or relay its output to efferent stages of the circuit.
Salomone has developed an alternative perspective (Salamone 2002). In his view, dopaminergic neurons influence the proclivity to invest effort in the pursuit of reward; changes in the activity of these neurons do not alter reward intensity. Thus, Salomone's view is compatible with an influence of dopamine on subjective effort costs (Fig. 3) or the motivation to pay such costs to obtain reward.
Hypotheses concerning the contribution of dopamine to reward intensity or investment of effort in meeting work requirements are based directly or implicitly on curve-shift data, obtained using either stimulation strength or the work requirement as the independent variable. Arvanitogiannis and Shizgal (2008) have argued that such data cannot link the influence of dopamine unambiguously to processes operating at either side of the vertical gray line in Fig. 3; data obtained by varying both the strength and cost of reward are required to do so. The results of such an experiment have yet to be published as of this writing.
The following sections summarize the contributions of various other neurotransmitters to ICSS. The centrality of midbrain dopamine neurons to the phenomenon remains evident, as several other neurotransmitters appear to exert their influence on ICSS by means of their interactions with dopaminergic neurons.
► Noradrenaline figured heavily in early psychophar-macological research on BSR. The early interest waned after the reductions in response rate produced by agents that decrease noradrenergic neurotransmission were attributed to sedation, and early claims that self-stimulation of sites in the vicinity of the locus coeruleus were due to activation of noradrenergic neurons were disputed. Nonetheless, neurons in the locus coeruleus and lateral tegmental A7 cluster do show increased double labeling for the rate-limiting enzyme in noradrenalin synthesis, tyrosine hydroxylase, and the immediate early-gene product, Fos, following self-stimulation of the MFB. Injection of the a1 receptor antagonist, terazosin, into the locus coeruleus has been shown recently to produce rightward shifts in rate-frequency curves obtained from rats working for electrical stimulation of the MFB. Given the evidence that activation of a1 receptors excites noradrenergic neurons in the locus coeruleus, this finding suggests that the firing of these neurons contributes in some way to the pursuit of rewarding MFB stimulation.
► Acetylcholine has been implicated in self-stimulation by experiments entailing manipulation of projections to midbrain dopamine neurons from cholinergic cell bodies in the pedunculopontine and lateral dorsal teg-mental nuclei (Yeomans et al. 1993). Activation of these excitatory projections potentiates MFB self-stimulation and drives dopamine release in the ► nucleus accumbens. Neurotransmission in the cholinergic projections to the VTA is suppressed by the action of cholinergic agonists at autoreceptors on or near the cholinergic somata or by the action of cholinergic antagonists in the VTA terminal field. These manipulations reduce self-stimulation of the MFB (i.e., they cause rightward curve shifts). Disinhibition of the cholinergic projections by administration of cholinergic antagonists in the vicinity of the cholinergic cell bodies potentiates MFB self-stimulation, as evinced by leftward curve shifts. Enhanced release of acetylcholine is observed during self-stimulation of the MFB, both in the vicinity of the cholinergic cell bodies and in the VTA terminal field. Although modest effects on MFB self-stimulation have been reported following nicotinic manipulations of the cho-linergic projections to the VTA, muscarinic receptors, the M5 sub-type in particular, appear to mediate most of the effect of the cholinergic drive on MFB self-stimulation and on dopamine release in the nucleus accumbens. Administration into the VTA of ► antisense oligonucleotides for the M5 receptor suppresses MFB self-stimulation. The potent modulation of MFB self-stimulation by choliner-gic agents suggests that the effects of activating MFB fibers are relayed to VTA dopamine neurons, at least in part, by constitutively active cholinergic afferents.
► Serotonin. An important role in emotional and behavioral control has been attributed to serotonergic neurons. However, the multiplicity of serotonergic receptors, the widespread distribution of the serotonergic projections, and the action of serotonin both at the cell bodies and in the terminal fields of dopamine neurons make it challenging to build a comprehensive account of the action of serotonin on brain reward circuitry. Nonetheless, there is good agreement on the overall pattern of the results obtained to date in studies of the role of serotonin in ICSS: Release of this neurotransmitter generally exerts a suppressive influence on ICSS and opposes the influence of dopamine release (Harrison and Markou 2001). For example, stimulation of inhibitory cell-body autoreceptors decreases the activity of serotonergic neurons in the rostral raphe nuclei and potentiates self-stimulation of sites along the LH-VTA segment of the MFB. The effects of systemically administered agonists vary as a function of dose, stimulation site, and affinity for different subtypes of serotonin receptors. That said, rightward curve shifts or related increases in ICSS thresholds have been observed following administration of agonists for the 5-HT1A, 5-HT1B, and 5-HT2C receptors. Systemic administration of antagonists for these receptors and the 5-HT3 receptor usually leaves ICSS unaltered but can reverse changes produced by concurrent administration of serotonergic or dopaminergic agonists.
Histamine. A reward-inhibiting role has been proposed for histamine-containing neurons in the tubero-mammillary regions of the posterior hypothalamus (Wagner et al. 1993). These neurons project to the nucleus accumbens, where blockade of H1 receptors increases the extracellular concentration of dopamine. Ipsilateral damage to the tuberomamillary histamine neurons or blockade of H1 receptors in the nucleus accumbens increases the rate of LH self-stimulation. It remains to be determined whether leftward curve shifts can be produced by such manipulations.
Glutamate and GABA. Given the ubiquity of these amino-acid neurotransmitters in the brain, it would be surprising indeed if they did not play important roles in the rewarding effects of electrical brain stimulation.
► Glutamate is released in the VTA during MFB self-stimulation. Numerous nuclei provide glutamatergic input to the VTA, and it is not yet known which subset of glutamatergic neurons are responsible for the release of this neurotransmitter during ICSS. Identifying these neurons is of substantial potential interest because they may well contribute to the directly activated stage of the circuit responsible for BSR. The notion that directly activated MFB fibers provide excitatory input to dopa-mine cells is compatible with the abovementioned hypothesis of Moisan and Rompre.
Experience with ICSS of the MFB has been shown to downregulate the expression of the GluR1 subunit of
AMPA receptors, a phenomenon that has been proposed as an explanation of the lack of sensitization observed over the course of long periods of ICSS testing. Viral-induced increases in the expression of GluRl in the shell region of the nucleus accumbens produce rightward curve shifts, whereas, increased expression of the GluR2 subunit in the above region shifts the psychometric curves leftward (Todtenkopf et al. 2006).
Microinjection of GABAergic agonists or antagonists into the VTA or into basal forebrain regions such as the sublenticular extended amygdala can produce systematic shifts in rate-frequency curves obtained from rats working for MFB stimulation (Waraczynski 2006). In the case of VTA injections, the level of activity of the local dopami-nergic cell bodies appears to determine the sign of the effect. For example, the GABAA agonist, muscimol, produces rightward curve shifts when injected alone, but can reinstate self-stimulation after it has been abolished by intra-VTA injection of a large dose of ► morphine in rats pretreated with the dopaminergic receptor blocker, pimo-zide. This effect has been interpreted to reflect the restoration of firing in dopamine cell bodies that had been driven into depolarization block by the combination of autoreceptor blockade and strong opioidergic excitatory drive; the ► GABAA stimulation is posited to have hyper-polarized the dopamine cells sufficiently to restore their ability to generate action potentials (Wise and Rompre 1989).
Endorphins. The role of ► endogenous opioids in reward has been investigated extensively, and studies of ICSS have played an important role in this endeavor. Peripherally administered opiates and ► opioids to drug-na'ive rats exert a biphasic influence on performance for BSR: an initial decrease in the vigor of responding is followed by an increase above baseline levels. The initial depression tolerates with repeated administration of the drug whereas the enhancement of performance does not, and thus the potentiation of ICSS by systemically administered opiates and opioids emerges as the principal effect of these drugs as a regimen of repeated administration proceeds.
Injection of opioid receptor agonists into either the VTA or the nucleus accumbens terminal field of VTA dopamine neurons can produce leftward shifts in psychometric curves obtained from rats working for rewarding MFB stimulation. In the case of the nucleus accumbens injections, such effects are observed following administration of agonists for either m- or S-opioid receptors. In contrast, systemic administration of the ► k opioid agonist, U-69,593 produced right-ward curve shifts and counteracted the left-shifting influence of cocaine.
The modulation of ICSS by opiates and opioids is linked strongly, but not exclusively, to the effects of these drugs on dopaminergic signaling. For example, GABAergic inter-neurons in the VTA are hyperpolarized by ► m-opioid agonists, thus disinhibiting dopaminergic cell bodies. Opioid agonists have also been shown to increase release of dopa-mine in the nucleus accumbens. That said, opposite effects on dopamine tone in the core and shell subregions of the nucleus accumbens were observed following local administration of m- and S agonists. Given that opioid receptors are found both pre- and postsynap-tically in the nucleus accumbens and have been identified on dopaminergic, cholinergic, glutamatergic, and GABAergic neurons, there are multiple ways that opiates and opioids could influence the processing of reward-related signals in the ventral striatum.
► Cannabinoids. An abundant literature links the endogenous cannabinoid system to the pursuit and evaluation of rewards, and there is evidence that cannabinoid agonists activate both dopaminergic and opioidergic neurons. Within this literature, the data on ICSS are anomalous: in the hands of different investigators, drugs that alter ► cannabi-noid signaling have been observed to enhance, suppress, or fail to alter pursuit of BSR. Methodological issues could be at the root of these conflicting reports, and application of the 3D model described in Fig. 3 may help shed light on these issues. Given the steep slope of the "intensity-growth function'' (lighter curve in the left-hand graph), substantial changes in the values of variables on the right-hand side of the figure can produce only modest shifts in 2D projections of the 3D surface, such as rate-frequency curves, which can prove hard to discern through the measurement noise, individual differences in drug sensitivity, etc. However, such changes should be readily detectable when the 3D-measurement method is used.
Concluding remarks. Together with the conditioned place-preference and drug self-administration paradigms, ICSS has been, and continues to be, a mainstay of research on the psychopharmacology of reward. As this essay suggests, rather a lot has been learned from ICSS experiments about the roles of different neurotransmitter systems in brain reward circuitry. Nonetheless, much additional work will be required to fully account for the powerful influence of drugs on ICSS. Advances in behavioral measurement methods promise to tie the effects of pharmacological manipulations to specific psychological processes that contribute to the pursuit of BSR, and new methods, such as techniques for optical stimulation and inhibition of neurons expressing particular neurotransmitter-related genes, promise to refine our understanding of reward processing at the cellular and circuit levels.
Was this article helpful?
Have you ever been envious of people who seem to have no end of clever ideas, who are able to think quickly in any situation, or who seem to have flawless memories? Could it be that they're just born smarter or quicker than the rest of us? Or are there some secrets that they might know that we don't?