Does phasic dopamine release cause policy updates?

Phasic dopamine activity is believed to both encode reward-prediction errors (RPEs) and to cause the adaptations that these errors engender. If so, a rat working for optogenetic stimulation of dopamine neurons will repeatedly update its policy and/or action values, thus iteratively increasing its work rate. Here, we challenge this view by demonstrating stable, non-maximal work rates in the face of repeated optogenetic stimulation of midbrain dopamine neurons. Furthermore, we show that rats learn to discriminate between world states distinguished only by their history of dopamine activation. Comparison of these results to reinforcement learning simulations suggests that the induced dopamine transients acted more as rewards than RPEs. However, pursuit of dopaminergic stimulation drifted upwards over a time scale of days and weeks, despite its stability within trials. To reconcile the results with prior findings, we consider multiple roles for dopamine signalling.

Dopamine neurons do not constitute an obligatory stage in the final common path for the evaluation and pursuit of brain stimulation reward.

The neurobiological study of reward was launched by the discovery of intracranial self-stimulation (ICSS). Subsequent investigation of this phenomenon provided the initial link between reward-seeking behavior and dopaminergic neurotransmission. We re-evaluated this relationship by psychophysical, pharmacological, optogenetic, and computational means. In rats working for direct, optical activation of midbrain dopamine neurons, we varied the strength and opportunity cost of the stimulation and measured time allocation, the proportion of trial time devoted to reward pursuit. We found that the dependence of time allocation on the strength and cost of stimulation was similar formally to that observed when electrical stimulation of the medial forebrain bundle served as the reward. When the stimulation is strong and cheap, the rats devote almost all their time to reward pursuit; time allocation falls off as stimulation strength is decreased and/or its opportunity cost is increased. A 3D plot of time allocation versus stimulation strength and cost produces a surface resembling the corner of a plateau (the "reward mountain"). We show that dopamine-transporter blockade shifts the mountain along both the strength and cost axes in rats working for optical activation of midbrain dopamine neurons. In contrast, the same drug shifted the mountain uniquely along the opportunity-cost axis when rats worked for electrical MFB stimulation in a prior study. Dopamine neurons are an obligatory stage in the dominant model of ICSS, which positions them at a key nexus in the final common path for reward seeking. This model fails to provide a cogent account for the differential effect of dopamine transporter blockade on the reward mountain. Instead, we propose that midbrain dopamine neurons and neurons with non-dopaminergic, MFB axons constitute parallel limbs of brain-reward circuitry that ultimately converge on the final-common path for the evaluation and pursuit of rewards.

Learning to use past evidence in a sophisticated world model.

Humans and other animals are able to discover underlying statistical structure in their environments and exploit it to achieve efficient and effective performance. However, such structure is often difficult to learn and use because it is obscure, involving long-range temporal dependencies. Here, we analysed behavioural data from an extended experiment with rats, showing that the subjects learned the underlying statistical structure, albeit suffering at times from immediate inferential imperfections as to their current state within it. We accounted for their behaviour using a Hidden Markov Model, in which recent observations are integrated with evidence from the past. We found that over the course of training, subjects came to track their progress through the task more accurately, a change that our model largely attributed to improved integration of past evidence. This learning reflected the structure of the task, decreasing reliance on recent observations, which were potentially misleading.

Valuation of opportunity costs by rats working for rewarding electrical brain stimulation.

Pursuit of one goal typically precludes simultaneous pursuit of another. Thus, each exclusive activity entails an "opportunity cost:" the forgone benefits from the next-best activity eschewed. The present experiment estimates, in laboratory rats, the function that maps objective opportunity costs into subjective ones. In an operant chamber, rewarding electrical brain stimulation was delivered when the cumulative time a lever had been depressed reached a criterion duration. The value of the activities forgone during this duration is the opportunity cost of the electrical reward. We determined which of four functions best describes how objective opportunity costs, expressed as the required duration of lever depression, are translated into their subjective equivalents. The simplest account is the identity function, which equates subjective and objective opportunity costs. A variant of this function called the "sigmoidal-slope function," converges on the identity function at longer durations but deviates from it at shorter durations. The sigmoidal-slope function has the form of a hockey stick. The flat "blade" denotes a range over which opportunity costs are subjectively equivalent; these durations are too short to allow substitution of more beneficial activities. The blade extends into an upward-curving portion over which costs become discriminable and finally into the straight "handle," over which objective and subjective costs match. The two remaining functions are based on hyperbolic and exponential temporal discounting, respectively. The results are best described by the sigmoidal-slope function. That this is so suggests that different principles of intertemporal choice are involved in the evaluation of time spent working for a reward or waiting for its delivery. The subjective opportunity-cost function plays a key role in the evaluation and selection of goals. An accurate description of its form and parameters is essential to successful modeling and prediction of instrumental performance and reward-related decision making.

The Effects of Electrical and Optical Stimulation of Midbrain Dopaminergic Neurons on Rat 50-kHz Ultrasonic Vocalizations.

RATIONALE: Adult rats emit ultrasonic vocalizations (USVs) at around 50-kHz; these commonly occur in contexts that putatively engender positive affect. While several reports indicate that dopaminergic (DAergic) transmission plays a role in the emission of 50-kHz calls, the pharmacological evidence is mixed. Different modes of dopamine (DA) release (i.e., tonic and phasic) could potentially explain this discrepancy. OBJECTIVE: To investigate the potential role of phasic DA release in 50-kHz call emission. METHODS: In Experiment 1, USVs were recorded in adult male rats following unexpected electrical stimulation of the medial forebrain bundle (MFB). In parallel, phasic DA release in the nucleus accumbens (NAcc) was recorded using fast-scan cyclic voltammetry. In Experiment 2, USVs were recorded following response-contingent or non-contingent optogenetic stimulation of midbrain DAergic neurons. Four 20-s schedules of optogenetic stimulation were used: fixed-interval, fixed-time, variable-interval, and variable-time. RESULTS: Brief electrical stimulation of the MFB increased both 50-kHz call rate and phasic DA release in the NAcc. During optogenetic stimulation sessions, rats initially called at a high rate comparable to that observed following reinforcers such as psychostimulants. Although optogenetic stimulation maintained reinforced responding throughout the 2-h session, the call rate declined to near zero within the first 30 min. The trill call subtype predominated following both electrical and optical stimulation. CONCLUSION: The occurrence of electrically-evoked 50-kHz calls, time-locked to phasic DA (Experiment 1), provides correlational evidence supporting a role for phasic DA in USV production. However, in Experiment 2, the temporal dissociation between calling and optogenetic stimulation of midbrain DAergic neurons suggests that phasic mesolimbic DA release is not sufficient to produce 50-kHz calls. The emission of the trill subtype of 50-kHz calls potentially provides a marker distinguishing positive affect from positive reinforcement.

The effect of probability discounting on reward seeking: a three-dimensional perspective.

Rats will work for electrical stimulation of the medial forebrain bundle. The rewarding effect arises from the volleys of action potentials fired by the stimulation and subsequent spatio-temporal integration of their post-synpatic impact. The proportion of time allocated to self-stimulation depends on the intensity of the rewarding effect as well as on other key determinants of decision-making, such as subjective opportunity costs and reward probability. We have proposed that a 3D model relating time allocation to the intensity and cost of reward can distinguish manipulations acting prior to the output of the spatio-temporal integrator from those acting at or beyond it. Here, we test this proposition by varying reward probability, a variable that influences the computation of payoff in the 3D model downstream from the output of the integrator. On riskless trials, reward was delivered on every occasion that the rat held down the lever for a cumulative duration called the "price," whereas on risky trials, reward was delivered with probability 0.75 or 0.50. According to the model, the 3D structure relating time allocation to reward intensity and price is shifted leftward along the price axis by reductions in reward probability; the magnitude of the shift estimates the change in subjective probability. The predictions were borne out: reducing reward probability shifted the 3D structure systematically along the price axis while producing only small, inconsistent displacements along the pulse-frequency axis. The results confirm that the model can accurately distinguish manipulations acting at or beyond the spatio-temporal integrator and strengthen the conclusions of previous studies showing similar shifts following dopaminergic manipulations. Subjective and objective reward probabilities appeared indistinguishable over the range of 0.5 ≤ p ≤ 1.0.

Optimal indolence: a normative microscopic approach to work and leisure.

Dividing limited time between work and leisure when both have their attractions is a common everyday decision. We provide a normative control-theoretic treatment of this decision that bridges economic and psychological accounts. We show how our framework applies to free-operant behavioural experiments in which subjects are required to work (depressing a lever) for sufficient total time (called the price) to receive a reward. When the microscopic benefit-of-leisure increases nonlinearly with duration, the model generates behaviour that qualitatively matches various microfeatures of subjects' choices, including the distribution of leisure bout durations as a function of the pay-off. We relate our model to traditional accounts by deriving macroscopic, molar, quantities from microscopic choices.

Validation and extension of the reward-mountain model.

The reward-mountain model relates the vigor of reward seeking to the strength and cost of reward. Application of this model provides information about the stage of processing at which manipulations such as drug administration, lesions, deprivation states, and optogenetic interventions act to alter reward seeking. The model has been updated by incorporation of new information about frequency following in the directly stimulated neurons responsible for brain stimulation reward and about the function that maps objective opportunity costs into subjective ones. The behavioral methods for applying the model have been updated and improved as well. To assess the impact of these changes, two related predictions of the model that were supported by earlier work have been retested: (1) altering the duration of rewarding brain stimulation should change the pulse frequency required to produce a reward of half-maximal intensity, and (2) this manipulation should not change the opportunity cost at which half-maximal performance is directed at earning a maximally intense reward. Prediction 1 was supported in all six subjects, but prediction 2 was supported in only three. The latter finding is interpreted to reflect recruitment, at some stimulation sites, of a heterogeneous reward substrate comprising dual, parallel circuits that integrate the stimulation-induced neural signals.

Role of dopamine tone in the pursuit of brain stimulation reward.

Dopaminergic neurons contribute to intracranial self-stimulation (ICSS) and other reward-seeking behaviors, but it is not yet known where dopaminergic neurons intervene in the neural circuitry underlying reward pursuit or which psychological processes are involved. In rats working for electrical stimulation of the medial forebrain bundle, we assessed the effect of GBR-12909 (1-[2-[bis(4-fluorophenyl)-methoxy]ethyl]-4-[3- phenylpropyl]piperazine), a specific blocker of the dopamine transporter. Operant performance was measured as a function of the strength and cost of electrical stimulation. GBR-12909 increased the opportunity cost most subjects were willing to pay for a reward of a given intensity. However, this effect was smaller than that produced by a regimen of cocaine administration that drove similar increases in nucleus accumbens (NAc) dopamine levels in unstimulated rats. Delivery of rewarding stimulation to drug-treated rats caused an additional increase in dopamine concentration in the NAc shell in cocaine-treated, but not GBR-12909-treated, rats. These behavioral and neurochemical differences may reflect blockade of the norepinephrine transporter by cocaine but not by GBR-12909. Whereas the effect of psychomotor stimulants on ICSS has long been attributed to dopaminergic action at early stages of the reward pathway, the results reported here imply that increased dopamine tone boosts reward pursuit by acting at or beyond the output of the circuitry that temporally and spatially summates the output of the directly stimulated neurons underlying ICSS. The observed enhancement of reward seeking could be attributable to a decrease in the value of competing behaviors, a decrease in subjective effort costs, or an increase in reward-system gain.

Cannabinoid receptor blockade reduces the opportunity cost at which rats maintain operant performance for rewarding brain stimulation.

There is ample evidence that blockade of CB(1) receptors reduces reward seeking. However, the reported effects of CB(1) blockade on performance for rewarding electrical brain stimulation stand out as an exception. By applying a novel method for conceptualizing and measuring reward seeking, we show that AM-251, a CB(1) receptor antagonist, does indeed decrease performance for rewarding electrical stimulation of the medial forebrain bundle in rats. Reward seeking depends on multiple sets of variables, including the intensity of the reward, its cost, and the value of competing rewards. In turn, reward intensity depends both on the sensitivity and gain of brain reward circuitry. We show that drug-induced changes in sensitivity cannot account for the suppressive effect of AM-251 on reward seeking. Therefore, the role of CB(1) receptors must be sought among the remaining determinants of performance. Our analysis provides an explanation of the inconsistencies between prior reports, which likely arose from the following: (1) the averaging of data across subjects showing heterogeneous effects and (2) the use of methods that cannot distinguish between the different determinants of reward pursuit. By means of microdialysis, we demonstrate that blockade of CB(1) receptors attenuates nucleus accumbens dopamine release in response to rewarding medial forebrain bundle stimulation, and we propose that this action is responsible for the ability of the drug to decrease performance for the electrical reward.

At what stage of neural processing does cocaine act to boost pursuit of rewards?

Dopamine-containing neurons have been implicated in reward and decision making. One element of the supporting evidence is that cocaine, like other drugs that increase dopaminergic neurotransmission, powerfully potentiates reward seeking. We analyze this phenomenon from a novel perspective, introducing a new conceptual framework and new methodology for determining the stage(s) of neural processing at which drugs, lesions and physiological manipulations act to influence reward-seeking behavior. Cocaine strongly boosts the proclivity of rats to work for rewarding electrical brain stimulation. We show that the conventional conceptual framework and methods do not distinguish between three conflicting accounts of how the drug produces this effect: increased sensitivity of brain reward circuitry, increased gain, or decreased subjective reward costs. Sensitivity determines the stimulation strength required to produce a reward of a given intensity (a measure analogous to the KM of an enzyme) whereas gain determines the maximum intensity attainable (a measure analogous to the vmax of an enzyme-catalyzed reaction). To distinguish sensitivity changes from the other determinants, we measured and modeled reward seeking as a function of both stimulation strength and opportunity cost. The principal effect of cocaine was a two-fourfold increase in willingness to pay for the electrical reward, an effect consistent with increased gain or decreased subjective cost. This finding challenges the long-standing view that cocaine increases the sensitivity of brain reward circuitry. We discuss the implications of the results and the analytic approach for theories of how dopaminergic neurons and other diffuse modulatory brain systems contribute to reward pursuit, and we explore the implications of the conceptual framework for the study of natural rewards, drug reward, and mood.

The effects of reinforcer magnitude on timing in rats.

The relation between reinforcer magnitude and timing behavior was studied using a peak procedure. Four rats received multiple consecutive sessions with both low and high levels of brain stimulation reward (BSR). Rats paused longer and had later start times during sessions when their responses were reinforced with low-magnitude BSR. When estimated by a symmetric Gaussian function, peak times also were earlier; when estimated by a better-fitting asymmetric Gaussian function or by analyzing individual trials, however, these peak-time changes were determined to reflect a mixture of large effects of BSR on start times and no effect on stop times. These results pose a significant dilemma for three major theories of timing (SET, MTS, and BeT), which all predict no effects for chronic manipulations of reinforcer magnitude. We conclude that increased reinforcer magnitude influences timing in two ways: through larger immediate after-effects that delay responding and through anticipatory effects that elicit earlier responding.

Prolonged rewarding stimulation of the rat medial forebrain bundle: neurochemical and behavioral consequences.

Extracellular dopamine levels were measured in the rat nucleus accumbens by means of in vivo microdialysis. Delivery of rewarding medial forebrain bundle stimulation at a low rate (5 trains/min) produced a sustained elevation of dopamine levels, regardless of whether train onset was predictable. When the rate of train delivery was increased to 40 trains/min, dopamine levels rose rapidly during the first 40 min but then declined toward the baseline range. The rewarding impact of the stimulation was reduced following prior delivery of stimulation at the high, but not the low, rate. These results support the idea that dopamine tone plays an enabling role in brain stimulation reward and is elevated similarly by predictable and unpredictable stimulation.

Growth of brain stimulation reward as a function of duration and stimulation strength.

The strength of a train of rewarding brain stimulation required to support a criterion level of operant performance declines hyperbolically as the duration is increased. This finding has been attributed to a process of leaky integration. However, the rate at which integration approaches asymptote has been shown to depend on stimulation strength, a finding that differs from the behavior of a simple leaky integrator. The authors replicate both findings and show that they are both well described by a new model that incorporates a hyperbolic strength-duration function, a logistic function mapping stimulation frequency onto reward intensity, and another logistic function mapping reward intensity onto performance.

Operant tempo varies with reinforcement rate: implications for measurement of reward efficacy.

Herrnstein's melioration theory has been used to account for the hyperbolic form of the single operant matching law and to scale the effectiveness of reinforcing brain stimulation. Underlying this scaling method is the assumption that the mean rate of responding during operant bouts (the response 'tempo') is fixed and does not vary with the rate of reinforcement. The validity of this account was assessed by testing the constant-tempo assumption via a survivor analysis of the distributions of inter-response times at different variable-intervals (VIs) in rats responding for rewarding electrical stimulation of the lateral hypothalamus. Contrary to the constant-tempo assumption, response tempo was not fixed but rather decreased as the VI was lengthened. This demonstration challenges Herrnstein's account of single-operant matching and suggests that the reinforcement rate that supports a half-maximal rate of responding on a single VI schedule may not provide a valid scale for the value of brain stimulation. Possible remedies are discussed. Although the conclusions of the study are restricted to experiments on brain stimulation reward in rats, the inter-response time analysis employed can provide the basis for testing the validity of the constant-tempo assumption in other species and for other reinforcers.