Summation and saturation properties in the rewarding effect of brain stimulation.

We used a two-lever self-stimulation chamber and rats with central grey and hypothalamic electrodes to obtain a choice-frequency (C/F) function, which plots the relative choice of an electrical stimulus of fixed pulse frequency (the standard stimulus) as a function of the frequency of a second competing stimulus (the alternative stimulus). A family of C/F functions was obtained using increasing frequencies for the standard. A choice index, varying from -1.0 (exclusive choice of the fixed stimulus) to 1.0 (exclusive choice of the alternative stimulus) was computed by using the barpressing rates on the two levers. Reward saturation was assumed to occur when the C/F function obtained with the largest standard reached an asymptote below the value of 1.0. For the hypothalamic subjects, the pulse frequency at the point of reward saturation was twice as high as the frequency required for the maximum rate of self-stimulation in the usual single-lever chamber. Decreasing the pulse intensity always increased the saturation frequency, indicating that the saturation was not due to a frequency blocking effect in the directly activated neurons. Reward saturation occurred with a considerably higher frequency in the central grey than in the hypothalamus. Thus, the asymptotic rate of self-stimulation in the usual single-lever chamber is not conditioned by the processes that summate the central grey and hypothalamic rewarding effects. From the central grey data we obtained the relation between the slope and the position of the C/F function on the frequency axis. We used the slope/frequency profile to delimit the most probable profile of summation rate in the rewarding process. We found that the slope/frequency function has an early plateau followed by a rapidly decelerating phase. We proposed that these two distinct phases reflect an early accelerating rate of summation, followed by a decelerating rate of summation. In other words, reward summation would be predicted by a sigmoid growth model.