Cocaine dependence is a chronically relapsing disorder leading to a variety of medical complications along with devastating psychosocial consequences. It remains a major public health problem bearing enormous societal costs and is currently afflicting over 1.5 million American citizens 1. Thus, epidemiological data presented at the recent Community Epidemiology Work Group meeting indicate that in a number of major US cities, representing 21 geographic areas, hospital admissions for primary cocaine-related problems exceeded those for heroin, methamphetamine and marihuana, altogether 2.

Craving, defined as an intense and often irrepressible urge to seek and consume the drug, resulting in relapse even after extended periods of abstinence 3, is one of the most malignant and treatment resistant features of cocaine dependence 4. Therefore, in the search for innovative and efficient therapeutic approaches, craving has been subjected to an intensive scientific inquiry in both laboratory and clinical arenas. One consistent finding produced by this effort, is a considerable overlap between neural correlates of craving, cognition and motivation 3567. Hence, understanding how cocaine produces its effects necessitates investigation of the cognitive and motivational contexts of craving.

Cocaine craving may result from the desire for pleasurable effects of the drug 8. A closely related concept to this model is that of dopamine as the motivation/cognition-related neurotransmitter, the phasic release of which underlies the anticipation of pleasurable outcomes and motivational behaviors targeted at avoiding the loss of pleasure 910111213141516.

Primate electrophysiological studies of individual dopamine neurons in the brain 91718192021 have begun dissecting these motivational processes that could be potentially broadened to cocaine craving. Neuronal responses to reward and reward-predicating stimuli appear to be dependent on event predictability i.e., neuronal activity is increased by events with reward values which are better than predicted, is uninfluenced by events that are as good as predicted, and is reduced by events that are worse than predicted 132223. Elevated dopamine concentrations may also prime the organism to associate cues around cocaine use with the pleasure experienced when drug is taken 12131518192024. When these cues are encountered again, they may elicit a surge of activity in dopaminergic neurons leading to exaggerated anticipatory craving of the future cocaine reward 25. However, if individuals with cocaine dependence are exposed to stimuli associated with cocaine when they cannot consume it, they may experience a decrease in dopamine activity, which may potentially lead to a dysphoric state of withdrawal 16.

These observations lead Volkow and colleagues 16 to assume the existence of specific neural loci related to the processes of expectancy, predictions of reward and contingency assessment. They assessed this entity by using a two-by-two within subjects design to measure the effects of expectation on subjective effects and regional brain metabolism in cocaine abusers under four conditions: 1) expecting and receiving placebo; 2) expecting placebo and receiving a drug similar to cocaine, methylphenidate; 3) expecting methylphenidate and receiving placebo; 4) expecting and receiving methylphenidate. The authors found that expectation of the drug and subsequent receipt of the drug led to an increased physiological response. This was reflected in a 50% increase in global brain glucose utilization, with specific increases in the thalamic and cerebellar regions and an enhancement in subjective assessments ("drug liking," "high," "feel drug," and "restlessness") compared to expectation of placebo and receipt of drug 16. Thus, after methylphenidate infusion, the presence or absence of the drug-induced subjective effects apparently involved some retrieval and comparison to the "remembered utility" 16, or the emotional memory associated with the prior use.

The present study attempted to extend Volkow et al.'s 16 findings to cocaine itself and to craving. To that end, cocaine dependent subjects' self-reports of craving when they were certain in the receipt of the drug were contrasted to those when such probability stood at only 33%. The primary hypothesis to be tested was that subjects experience stronger craving when they know that cocaine infusion is forthcoming versus a situation in which it is less likely. Other subjective ratings i.e., high, rush and low 1626 were assessed using exploratory analysis. Finally, given the previous finding by our group that subjective expectancy ratings accounted for some variance in the hemodynamic responses to 'unblinded' cocaine 27, we also examined the effects of perceived cocaine availability on heart rate and on mean arterial pressure (MAP).

The major finding of this study is that changing the motivational context from 100% to 33% cocaine expectancy, by matching a potential cocaine infusion to two other potential infusions, significantly modified subjective and objective cocaine responses. To our knowledge, this is the first report of a non-pharmacological effect of perceived cocaine availability on self-reported baseline craving prior to receiving cocaine and on subjective and hemodynamic variables following cocaine administration.

The elevated high ratings in the 'unblinded' group, although did not survive stringent statistical adjustment, are suggestive of an expectancy component in cocaine's rewarding effects and are consistent with the greater "high" reports in the context of 100% vs. 50% methylphenidate expectancy state 37. Although there were important methodological similarities between Volkow et al. 37 and our study (i.e., enrollment of cocaine-dependent subjects, use of subjective self-ratings and acute administration of a psychostimulant), there were also substantial differences in specific definitions of self-ratings, the priming agent, level of uncertainty about the receipt of the drug, intersubject design, and the measures of hemodynamic responses. Thus, these independent results strongly support the validity of the relationship between expectancy context and cocaine effects reflected in self-reports and in physiological changes.

It is notable that notwithstanding similar (to the 'blinded' group) craving scores on a multidimensional assessment several hours before the infusions, 'unblinded' subjects reported higher craving levels on a more proximal evaluation, immediately prior to the infusions. These data are consistent with a substantial body of literature 383940 documenting failure to detect a relationship between measures of craving obtained in hospital and laboratory environments (where expectancy of obtaining drugs is low) and subsequent relapse to seeking and consuming cocaine. Thus, this study suggests that even within a laboratory environment, craving can be exacerbated by the imminent availability of the drug. Furthermore, our data may be indicative that the effect of expectancy is time sensitive such that proximal rather than distant prospects of getting the drug have greater effect on craving. Therefore, in order to draw valid conclusions about clinical outcomes, laboratory-based research may need to take into account motivational and cognitive contexts of craving e.g., perceived availability of the drug.

Probability of 100% for receipt of cocaine obviously presents a stronger drug-related cue, as compared to 33%, for the sensitized dopaminergic system, thus evoking heightened anticipatory craving. This difference in craving response levels off because the next cue i.e., the priming dose of cocaine itself is not different between the groups. Given the intuitively obvious liking of the desired objects 41, subjects with elevated craving for cocaine, as those in the 'unblinded' condition, presented heightened hedonic rush responses. Alternatively, these may be reflective of the hemodynamic effects (e.g., perception of the elevated heart rate).

Subjective and hemodynamic cocaine effects may indeed be connected via conditioned mechanisms. Thus, heart rate elevations repeatedly paired with cocaine use may become a conditioned stimulus and elicit cocaine-like psychological responses 4243, which in turn may produce increases in cardiovascular activity 44. Such a self-sustaining feed-forward loop whereby minor stimuli can trigger escalating reactions may be implicated in both, the chronically deteriorating course of cocaine dependence 45 and in high rates of cardio- and cerebrovascular morbidity associated with this illness 46. Notably, that causality does not run in the opposite direction as pharmacologically induced decrements in hemodynamic indices do not seem to affect subjective effects of cocaine 4748.

Homeostatic regulation of hemodynamic function can be dissected into distinct, but interacting catecholaminergic systems including dopamine, norepinephrine and epinephrine. Dopamine has greater β than α affinity, which explains its predominant heart rate effects. On the other hand, a rather selective α agonist, norepinephrine, primarily increases blood pressure, accompanied by only minimal heart rate changes. Epinephrine binds both α and β adrenoceptors, thus evenly affecting blood pressure and heart rate 49. Hence, different slopes of heart rate increases in the 'unblinded' subjects may reflect an expectancy-induced surge in dopamine 5051, which is the key neurochemical implicated in expectancy 13192052 and in sensitization 53 effects.

An alternative/complimentary explanation may be consistent with an idea that a psychological state, such as cocaine expectancy, could alter the homeostatic baroreflex, which is responsible for slowing the heart rate during increases in blood pressure by relaying neural stimuli generated by the arterial receptors' distortion to the CNS, nucleus of the solitary tract 4954. Future research can evaluate signs of diminished baroreceptor function such as orthostatic hypotension by performing supine and upright blood pressure measurements along with Valsalva maneuver and plasma catecholamine assays.

This study's strengths include fully randomized and counterbalanced cocaine administration and employment of both subjective and objective measures of cocaine response. There are, however, some caveats that have to be considered in interpreting the present data. First, a limitation of the present study is that group differences in age required us to perform statistical adjustments for this potentially confounding variable. Second, the lack of true randomization into the 'unblinded' and 'blinded' groups, along with the modest size of the latter, render our results preliminary pending replication in more rigorously designed trials.

Third, although the crossover design in the 'blinded' group offers certain advantages 5556, a few effects could have carried over from the hydrocortisone session. This is unlikely, though, given only transient and subtle subjective effects of hydrocortisone 34 and similar baseline values of the reported variables (excluding craving). Nonetheless, to fully account for such confounding, a second placebo arm could be an important consideration for future research. Fourth, mostly men participated in the experiment and results may not be easily extrapolated to women, as gender differences in craving have been previously demonstrated 57. Finally, to accommodate for motivational and attentional deficits commonly encountered in cocaine dependent populations 858596061, this study focused on short-term acute subjective responses. As some of the self-ratings did not return to baseline at the conclusion of the 20-minute data collection period, longer study duration may have yielded different results.

In conclusion, rather than being a unitary state characterized by one set of motivations 62, cocaine craving is modulated by its own availability, which also affects other emotions evoked by the drug and cocaine-induced hemodynamic changes.

The inability to correlate craving and subsequent drug taking behavior, as has been shown for laboratory settings, potentially generalizes to hospital and clinical treatment facilities 3463. A better model could be evaluations performed at a site with a greater likelihood of obtaining the drug. Thus, current results suggest that more ecologically valid assessments of craving, such as phone interviews in a patient's home, might be important for understanding the patterns of craving that lead to drug use and subsequently for intervention and prevention techniques as well as for proper disease staging and treatment matching procedures employed in public health policy research and determination 6465.

The author(s) declare that they have no competing interests.

RY carried out the statistical analysis and drafted the manuscript. KK coordinated the study and carried out the procedures. DR contributed cardiovascular expertise and helped drafting the manuscript. SL helped with the conceptualization and interpretation of the findings. IE designed the study and supervised the co-authors in their writing of the manuscript. All authors read and approved the final manuscript.