Quantifying the molecular structure of behavior: separate effects of caffeine, cocaine, and adenosine agonists on...

One difficulty in analyzing the molecular structure of behavior lies in reducing the data to a manageable size so that they can be described concisely but without a loss of important information. An approach to quantifying and comparing distributions of interresponse times (IRTs) and lever-press durations is described and then used to examine the acute effects of caffeine (an adenosine antagonist), cocaine, and three adenosine agonists in rats chronically consuming either tap water or water containing 1.0 mg/ml caffeine. The adenosine agonists used were R(-)N6-(2-phenylisopropyl)adenosine [R-PIA (preferential A1 receptor agonist)], 5'-(N-cyclopropyl)-carboxamidoadenosine [CPCA (preferential A2 agonist)], and 5'-N-ethylcarboxamidoadenosine [NECA (about equal agonist activity at A1 and A2 receptors)]. The rats' behavior was maintained under a Multiple Fixed-Interval (FI) 120 s, Duration > 5 s schedule of reinforcement. Under the FI schedule, the first lever-press after 120 s resulted in pellet delivery. Under the Duration > 5 s schedule, all lever-press durations greater than 5 s were reinforced. Molecular analyses of distributions of true IRTs (exclusive of lever-press durations) and lever-press durations were conducted by regressing percentiles of the distribution obtained from drug conditions against percentiles obtained from control conditions, a technique called empirical percentile-percentile analysis (or quantile-quantile analysis). The pattern of effects on response durations differed from that seen in IRTs. After acute administration of adenosine agonists, the distributions of lever-press durations under the FI schedule and subcriterion durations under the Duration > 5 s schedule were shifted rightward by a constant proportion, indicative of a generalized slowing of responding by these drugs. The effects of adenosine agonists on IRTs could be described by a power-function relationship whereby long IRTs were increased more than shorter ones. A rate-increasing dose of caffeine (10 mg/kg) did nothing to the molecular structure of lever-press durations or of IRTs, indicating that rate increases seen after this dose were due to an earlier onset of responding in the fixed-interval and were unaccompanied by disruptions in the physical execution of the response. The molecular structure of interresponse time distributions seen at the rate-decreasing dose of caffeine (60 mg/kg) resembled that of a rate-decreasing dose of cocaine, but not of the adenosine agonists. Both caffeine and cocaine produced a curvilinear relationship between drug and control percentiles such that very short IRTs were unaffected, but long ones were lengthened five- to 10-fold. This same dose of caffeine lengthened lever-press durations in a manner that resembled adenosine agonists but not cocaine. Percentile-percentile analyses reveal details about drug effects on the temporal structure of behavior and its physical execution that are not visible in molar analyses.