Drug-naïve rats with no prior experience with refined sugar or artificial sweetener were allowed to choose 8 times per day between two mutually exclusive levers (): responding on one lever (lever C) was rewarded by a behaviorally effective dose of cocaine (0.25 mg, i.v.) while responding on the other lever (lever S) was rewarded by a 20-s access to water sweetened with saccharin (0.2%) (see Materials and Methods
). Importantly, each day before making their choices, rats were allowed to alternatively sample each lever 2 times to learn their respective reward value (). Different groups of animals were tested under 3 reward conditions. Under the S-/C+ condition (N
30), only responding on lever C was rewarded (+) by cocaine delivery; responding on lever S was not rewarded (-). Under the S+/C- condition (N
9), only responding on lever S was rewarded by saccharin access; responding on lever C was not rewarded. Finally, under the S+/C+ condition (N
43), both levers were rewarded by their corresponding rewards. There was more rats in the S-/C+ or S+/C+ condition than in the S+/C- condition because more experiments were conducted in these former conditions to assess the determinants of choice between saccharin and cocaine (dose, delay, effort, reversal, calorie input, thirst).
Choice between saccharin and cocaine.
On day 1 and whatever the reward conditions, rats were indifferent to both levers, showing that there was no preexisting bias or preference in our setting. As expected, however, with repeated testing, reward conditions considerably influenced the evolution of lever choice [Condition×Day: F
<0.01] (). Under the S-/C+ condition, rats displayed no preference until day 9, when they shifted toward preferring lever C. This preference became statistically reliable on day 11. Similarly, under the S+/C- condition, rats rapidly acquired a preference for lever S which became statistically reliable on day 7. More surprisingly, under the S+/C+ condition, rats immediately developed a strong and stable preference for lever S which became statistically significant on day 2. This preference was indistinguishable from that exhibited by rats in the S+/C- condition [F
0.41, NS] (). In addition, after stabilization of behavior, the latency to select lever S in the S+/C+ condition (14.5±5.0 s, means±SEM of the last 3 stable days) was similar to that in the S+/C- condition (6.5±2.4 s) [t
(50)<1], showing that rats chose saccharin over cocaine without hesitation, as if lever C was not rewarded by cocaine.
The strong preference for saccharin under the S+/C+ condition was not due to a failure to learn the value of lever C. Indeed, from day 7 onward, rats sampled lever C almost maximally, though slightly less than lever S, before being allowed to make their choices (). Thus, despite near maximal cocaine sampling, rats under the S+/C+ condition acquired a preference for lever S as quickly as rats under the S+/C- condition. This finding also shows that cocaine had no positive or negative influence on saccharin acceptance and/or preference in the present choice setting. Finally, after stabilization of behavior, the latency to sample lever C (48.5±10.2 s, means±SEM of the last 3 stable days) was significantly greater than the latency to sample lever S (5.6±1.7 s) [F
<0.01]. This difference shows that animals have effectively learned that each lever is associated with a different outcome.
It is important to note that the preference for saccharin was not attributable to thirst or drinking behavior per se because rats preferred cocaine over mere water (). Finally, the preference for saccharin was not due to its unnatural ability to induce sweetness without calories because the same preference was also observed with an equipotent concentration of sucrose (4%) ().
Choice between lever C and no fluid (N), water (W), saccharin (Sac, 0.2%) or sucrose (Suc, 4%).
To directly assess the behavioral efficacy of cocaine in the discrete-trials choice procedure, we measured the ability of the first cocaine self-injection of the day to induce locomotion on day 1, 5 and 15. As expected, in rats which acquired a preference for lever C under the S-/C+ condition, cocaine induced a rapid increase in locomotion which peaked 1 min post-injection and then returned gradually to baseline within the 10-min inter-trial interval (). This psychomotor effect increased even further after repeated cocaine exposure [Day×Intervals: F
<0.01], a well-established phenomenon, called behavioral sensitization. Sensitization to cocaine was maximal as soon as day 5 and remained stable until the end of the experiment, despite additional cocaine exposure (). Importantly, a behavioral sensitization of a similar magnitude was also observed in rats which acquired a strong preference for lever S under the S+/C+ condition [Day×Intervals: F
<0.01] (). To test the specific contribution of saccharin consumption to the induction of sensitization in the S+/C+ condition, rats initially tested under the S+/C- condition were tested under the S+/C+ condition on day 16. These rats were much less sensitive to cocaine than rats initially trained under the S+/C+ condition [Group×Intervals: F
<0.05] (). This observation clearly shows that saccharin consumption per se
has little impact on sensitization under the S+/C+ condition and thus that the very few doses of cocaine consumed in the S+/C+ condition (mostly during sampling) were sufficient in themselves to induce sensitized responding. Thus, rats preferred saccharin over cocaine despite being fully responsive and sensitized to (and by) cocaine.
Cocaine-induced locomotion in rats tested under.
It is possible that though efficacious in inducing locomotion and sensitization, the dose of cocaine was nevertheless too low to surpass the rewarding effects of saccharin. To address this question, a subgroup of rats (N
11) trained under the S+/C+ condition was tested with increasing i.v. doses of cocaine (0.25-1.5 mg). The highest dose was near but lower than the convulsive dose (i.e., 3 mg) in our conditions. As expected, increasing the dose of cocaine induced a dose-dependent increase in locomotion, as measured during 10 min after the first cocaine self-injection of the first day of each dose substitution [F
<0.01] (). However, regardless of the dose available, rats continued to prefer lever S over lever C [F
0.07, NS] (). Thus, rats preferred saccharin despite a near maximal level of cocaine stimulation. Though the intravenous route of administration allows for rapid and intense drug effects–which explains why this route is often selected by heavy drug users–there is still a brief, incompressible delay between lever pressing and onset of cocaine effects. This delay of action was estimated at 6.2±0.2 s in the present study (see Materials and Methods
). Similarly, the neurochemical effects of cocaine peak between 4 and 20 s after the onset of an intravenous injection 
. In contrast, the delay between response and onset of saccharin drinking was less than 2 s. This difference of delay, though small, could nevertheless explain the preference for saccharin whose rewarding effects are more immediate than those of cocaine. To test the contribution of this factor, saccharin delivery was systematically delayed after selection of lever S (0–18 s) in a subgroup of rats (N
11) while the delay of cocaine delivery remained constant. Increasing the delay of saccharin delivery induced a slight decrease in selection of lever S [F
<0.01] (). This increase was not sufficient, however, to reverse the preference for lever S in favor of lever C. Thus, rats preferred saccharin even when its delay was equal to or above the delay of cocaine effects. Finally, we assessed in another subgroup of rats (N
10) the effects of the reward price (i.e., the number of lever presses required to obtain a reward) on choice. In some cases, increasing reward price can induce a shift in preference 
. However, increasing reward price from 2 to 8 responses/reward did not reverse but instead increased the preference for lever S [F
<0.01] (). Thus, regardless of the price, rats preferred saccharin over cocaine.
Pharmacological and economic determinants of cocaine choice.
The previous series of experiments involved initially drug-naïve individuals with no prior history of cocaine self-administration. To determine whether drug history influences the choice between saccharin and cocaine, a subgroup of rats (N
24) which had acquired a stable preference for lever C under the S-/C+ condition were subsequently tested under the S+/C+ during 10 days. Despite an initial, stable preference for lever C, rats rapidly reversed their preference in favor of lever S when both levers were rewarded (). The proportion of rats that preferred lever C (i.e., mean selection of lever C of the last 3 days>60%) after preference reversal did not differ significantly from that recorded in initially drug-naïve rats (8.3 versus 2.3%, z
<1.96). In addition, the preference for saccharin developed even in rats (N
11) with a long history of cocaine self-administration (6 h per day, during 3 weeks). In the present study, despite 3 weeks of extended access to cocaine self-administration and a large escalation of cocaine consumption [from 7.34±2.50 to 26.04±1.21 mg/day; F
<0.01], rats rapidly acquired a strong and stable preference for lever S over lever C (). The proportion of rats with prolonged access to cocaine that preferred lever C after 10 days of choice did not differ from that recorded in initially drug-naïve rats (0.0 versus 2.3%, z
<1.96). Despite a small decrease in selection of lever S at the highest dose, the preference for lever S in rats pre-exposed to prolonged cocaine self-administration was not surmountable by increasing doses of cocaine (, insert). Finally, the preference for lever S was so strong that it also emerged in rats under the influence of cocaine during choice (N
10). In this experiment, rats had continuous access to lever C alone during 3 h per day. After acquisition of lever pressing (>20 responses/session), they were tested on a modified discrete-choice procedure which consisted of a continuous access to lever C alone for 1 hour, followed by 8 discrete choice trials under the S+/C+ condition. Though rats responded each day on lever C to self-administer cocaine during the hour preceding choice (), they nevertheless rapidly acquired a robust preference for lever S (). As shown in 3 representative individuals, there was an abrupt, within-session shift in behavior from lever C to lever S during choice ().
Choice between saccharin and cocaine as a function of drug history.